Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Identification of stable endogenous control genes for transcriptional profiling of photon, proton and carbon-ion irradiated cells

Identification of stable endogenous control genes for transcriptional profiling of photon, proton... Background: Quantitative analysis of transcriptional regulation of genes is a prerequisite for a better understanding of the molecular mechanisms of action of different radiation qualities such as photon, proton or carbon ion irradiation. Microarrays and real-time quantitative RT-PCR (qRT-PCR) are considered the two cornerstones of gene expression analysis. In interpreting these results it is critical to normalize the expression levels of the target genes by that of appropriately selected endogenous control genes (ECGs) or housekeeping genes. We sought to systematically investigate common ECG candidates for their stability after different radiation modalities in different human cell lines by qRT-PCR. We aimed to identify the most robust set of ECGs or housekeeping genes for transcriptional analysis in irradiation studies. Methods: We tested the expression stability of 32 ECGs in three human cancer cell lines. The epidermoid carcinoma cells (A431), the non small cell lung carcinoma cells (A549) and the pancreatic adenocarincoma cells (BxPC3) were irradiated with photon, proton and carbon ions. Expression Heat maps, clustering and statistic algorithms were employed using SUMO software package. The expression stability was evaluated by computing: mean, standard deviation, ANOVA, coefficient of variation and the stability measure (M) given by the geNorm algorithm. Results: Expression analysis revealed significant cell type specific regulation of 18 out of 32 ECGs (p< 0.05). A549 and A431 cells shared a similar pattern of ECG expression as the function of different radiation qualities as compared to BxPC3. Of note, the ribosomal protein 18S, one of the most frequently used ECG, was differentially regulated as the function of different radiation qualities (p ≤ 0.01). A comprehensive search for the most stable ECGs using the geNorm algorithm identified 3 ECGs for A431 and BxPC3 to be sufficient for normalization. In contrast, 6 ECGs were required to properly normalize expression data in the more variable A549 cells. Considering both variables tested, i.e. cell type and radiation qualities, 5 genes-- RPLP0, UBC, PPIA, TBP and PSMC4– were identified as the consensus set of stable ECGs. Conclusions: Caution is warranted when selecting the internal control gene for the qRT-PCR gene expression studies. Here, we provide a template of stable ECGs for investigation of radiation induced gene expression. Keywords: Endogenous control genes, Internal control genes, qRT-PCR, Photon, Proton, Carbon-ion, Tumour cells, A431, A549 and BxPC3 * Correspondence: a.amir@dkfz.de Molecular RadioOncology [E210], National Center for Tumor Disease (NCT), German Cancer Research Center (DKFZ), Im Neuenheimer Feld 460, 69120, Heidelberg, Germany Heidelberg Ion Therapy Center (HIT), Heidelberg Institute of Radiation Oncology (HIRO), Department of Radiation Oncology, University of Heidelberg Medical School, Im Neuenheimer Feld 450, 69120, Heidelberg, Germany Full list of author information is available at the end of the article © 2012 Sharungbam et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Sharungbam et al. Radiation Oncology 2012, 7:70 Page 2 of 12 http://www.ro-journal.com/content/7/1/70 Background PES1 for A549; and RPL37A, RPLPO and CASC3 for In addition to direct, e.g. DNA damaging effect, system BxPC3. A systematic analysis further revealed 5 stable level cellular responses to ionizing radiation are attributed genes among the 32 candidate ECGs tested to normalize to the initiation of intracellular signals and subsequent dif- gene expression data generated in different cells and ferential regulation of genes/pathways governing various after various radiation qualities. cellular processes [1]. Therefore, detecting differential regulation of genes is critical for a better understanding of Results radiation-induced molecular effects. Transcriptional per- Expression of the 32 ECGs turbation after cell exposure to different radiation qualities In this study, 32 ECGs (Additional file 1) were evaluated is investigated to unravel the systems biology of cellular to identify the most suitable reference genes for gene ex- response underlying, normal tissue toxicity, carcinogen- pression profiling of irradiated cell lines. This collection of esis, or anti-cancer effects of irradiation [1-3]. Therefore, genes constitutes frequently used ECGs which were these studies have ramification for a broad spectrum of selected based on their relative high abundance and con- basic and applied sciences ranging from effects of space stitutive expression determined by literature search and/or radiation to carcinogenesis to cancer therapy. whole genome microarray data. The three prototypic In contrast to conventional photon irradiation the mo- tumour cell lines-- A431, A549 and BxPC3 used in this lecular effects of proton or heavier ions (e.g. carbon study are among most commonly investigated model cell ions) are less explored yet. However, emerging data indi- lines for each tumor entity. They were irradiated with pho- cate molecular differences in transcriptional response of ton, proton and carbon ions. After total-RNA isolation cells to particles as compared to photon irradiation [4,5]. and quality control using lab-on-chip bioanalyzer, qRT- One reliable and highly sensitive tool that allows rapid PCR was performed using Taqman primer and probes. and accurate results in gene expression analysis is the qRT-PCR [6,7]. As in any gene expression analysis, selec- CT-values and ECG regulation tion of a valid normalization or endogenous control To get a better overview of the CT-values among all the to correct for differences in RNA sampling is critical to cell lines, the CT-range along with the minimum and avoid misinterpretation of results. Inter-sample variation maximum CT-values were listed in Table 1. The vari- due to sample collection, RNA preparation and quality, ation of CT-values ranged from 8.97 in A431 to 28.58 in inherent sample differences, pipetting errors, different A549. The 8.97 CT-value corresponded to 18S indicating efficacies of the radiation qualities and reverse transcrip- its high abundance in the samples, whereas, the CT- tion efficiency are common sources of variability. The value 28.58 corresponded to GADD45A indicating a ideal endogenous control should have a constant expres- moderate abundance. Moreover, RPL37A, GAPDH and sion level under different experimental conditions and RPLPO exhibited small CT-range indicating less vari- be sufficiently abundant across different samples and cell ation in expression whereas the large CT-range of ACTB, lines. Although any gene that is stably expressed under a GADD45A, IPO8 showed large variations in their defined experimental condition can be used for expression. normalization, the selection is most commonly made The CT-range and coefficient of variation for each cell from the constitutively expressed ECGs. line were listed in additional file 2. Within each individ- However, the expression levels of the commonly used ual cell line irradiated with photon, proton and carbon ECGs may not only vary in different cell lines but also ion, all the ECGs in BxPC3 except for CDKN1A, 18S, under different experimental treatments or pathological POLR2A, PES1 and HMBS with CT-range 1.18, 0.82, states [8-25]. This necessitates the selection of ECGs 0.80, 0.70 and 0.61, respectively, showed the smallest which are appropriate for each experimental system. Al- CT-range, while in A549 all the ECGs except PES1, though, there has been systematic selection of ECGs for ACTB and RPS17 showed the largest CT-range various experimental systems, such selection has not (Figure 1A). In other words, as a function of different ra- been conducted so far for studying the effects of different diation qualities in BxPC3, the expression of the ECGs radiation qualities. was considerably stable, while there was more variation Here, we investigate the expression stability of 32 in A549. Moreover, A431 and A549 shared similar ex- commonly used ECGs in three human cancer cell lines pression pattern of the ECGs. The coefficient of variation irradiated with photon, proton and carbon ions. Differ- plotted in Figure 1B also reflected the similar regulation ential regulation of ECGs was found as the function of of the 32 ECGs in the three cell lines. both variables, radiation quality and cell type, respect- ively. Reliable internal control genes for individual cell ECG regulation using heat map lines were identified such as PGK1, RPL37A and PSMC4 The expression of the 32 candidate ECGs across the for A431; RPLPO, UBC, GAPDH, MT-ATP6, CASC3 and samples were also visualized in a Heat map (Figure 2). Sharungbam et al. Radiation Oncology 2012, 7:70 Page 3 of 12 http://www.ro-journal.com/content/7/1/70 Table 1 Cycle threshold (CT) values and coefficient of while ACTB, IPO8 and GADD45A depicted high vari- variation (CV) of 32 endogenous control genes across the ation in expression. The remaining ECGs were differen- samples of the cell lines tially regulated. For example, B2M and ELF1 were up Gene Symbol CT Range CT Min. CT Max. Mean CT ± SEM CV(%) regulated in BxPC3 and A431 but down regulated in RPL37A 1.13 17.95 19.08 18.44 ± 0.29 1.6 A549. On the other hand, MRPL19 and PES1 were down regulated in BxPC3 but up regulated in A431 and A549. HMBS 1.39 23.96 25.35 24.61 ± 0.41 1.7 In Figure 3, we displayed a Heat map generated from a CASC3 1.22 22.79 24.01 23.45 ± 0.42 1.83 one way ANOVA analysis at p< 0.05 between the three RPLP0 1.36 17.22 18.58 17.83 ± 0.34 1.91 cell lines. It revealed 18 differentially regulated ECGs-- PSMC4 1.56 21.04 22.6 21.90 ± 0.46 2.12 ABL, CDKN1A, PSMC4, EIF2B1, GAPDH, PPIA, TBP, ABL 1.92 22.95 24.87 23.71 ± 0.51 2.15 RPS17, UBC, B2M, ELF1, PUM, GADD45A, ACTB, UBC 1.62 19.22 20.85 19.81 ± 0.43 2.19 MRPL19, YWHAZ, CASC3 and PES1. These could be cell type specific regulations as their radiation quality GAPDH 1.25 16.01 17.25 16.53 ± 0.37 2.27 variation was minimal. Among them, the first 13 except POLR2A 2.03 23.5 25.53 24.19 ± 0.56 2.35 for PSMC4 were up regulated in BxPC3 and the MT-ATP6 1.39 15.18 16.57 15.66 ± 0.38 2.46 remaining down regulated. In A431 and A549, 12 of PES1 1.97 21.04 23.02 22.12 ± 0.55 2.49 them were down regulated and 6 were up regulated. This TBP 2.19 23.88 26.08 24.47 ± 0.64 2.62 supports the finding that the expression levels of ECGs GUSB 1.7 22.28 23.98 23.30 ± 0.63 2.71 are different in BxPC3 as compared to A431 and A549. BxPC3 showed least variations and A549 showed max- EIF2B1 2.53 23.593 26.12 24.11 ± 0.70 2.91 imum variations in gene expression as the function of ra- RPS17 1.6 17.58 19.18 18.14 ± 0.53 2.92 diation qualities. However, expression levels of ECGs RPL30 1.89 17.24 19.14 18.05 ± 0.53 2.96 were comparable in A431 and A549 (Figure 1A). POP4 1.17 22.73 23.91 23.64 ± 0.72 3.05 ANOVA analysis between the three radiation qualities PPIA 2.15 17.43 19.58 18.08 ± 0.55 3.07 at p< 0.01 revealed one gene, the ribosomal protein 18S, PUM1 2.88 22.15 25.03 23.10 ± 0.71 3.08 to be differentially regulated after different radiation qualities (Figure 4). This gene is one of the most com- HPRT1 2.73 21.18 23.91 22.08 ± 0.68 3.1 monly used internal control genes for normalisation of CDKN1B 2.08 22.8 24.98 23.98 ± 0.81 3.41 qRT-PCR based gene expression data. Therefore, caution PGK1 2.25 18.17 20.42 19.35 ± 0.69 3.61 needs to be practised in using this gene as an internal MRPL19 3.09 22.73 25.82 24.01 ± 0.95 3.96 control gene, in particular when radiation effects are CDKN1A 2.6 21.3 23.99 22.98 ± 0.95 4.17 investigated. ACTB 5.65 16.1 21.75 17.80 ± 0.75 4.23 The results in this section corroborated the findings of many other previous studies that the ECGs might be dif- ELF1 3.59 21.81 25.39 22.76 ± 1.00 4.4 ferentially regulated depending upon the experimental IPO8 4.26 23.68 27.93 24.80 ± 1.13 4.57 set-up and the cell type [8-25]. More importantly, this GADD45A 4.2 24.37 28.57 26.29 ± 1.37 4.97 analysis seems to suggest that the ECGs are differentially YWHAZ 3.56 22.17 25.73 23.96 ± 1.26 5.27 regulated by the different cell types and radiation qual- TFRC 4.26 21.15 25.41 22.13 ± 1.18 5.37 ities. We attempted to confirm this observation employ- B2M 3.76 17.99 21.75 19.33 ± 1.21 6.29 ing a systematic analysis of the expression levels. 18 S 2.95 8.79 11.95 10.78 ± 1.02 9.5 Identification of appropriate ECGs The genes are sorted by the coefficient of variation increasing from top to bottom. Gene expression levels obtained using PCR should be Four replicates were used for each cell line. Cell lines were irradiated with appropriately normalized by one or more carefully photon, proton and carbon ions. Non-irradiated samples served as control. Standard error of the mean (SEM). selected stable internal control genes. The geNorm algo- rithm developed by Vandesompele et al. [26] can deter- Direct clustering of the ECGs expression showed that mine the expression stability of control genes on the the expression profile of A431 and A549 were more basis of non-normalized expression levels. This measure similar as compared to BxPC3. BxPC3 showed least vari- relies on the principle that the expression ratio of two in- ation whereas A549 showed maximum variation in ex- ternal control genes is constant in all samples regardless pression among the samples as well as among the ECGs of the experimental condition or cell type. This algo- indicating a cell type specific expression of ECGs. As rithm computes a gene expression stability measure (M) observed in Table 1, the Heat map also revealed low for each gene based on the average pairwise expression variation of RPL37A, RPLPO and GAPDH expression, ratio and then performs a stepwise exclusion of the least Sharungbam et al. Radiation Oncology 2012, 7:70 Page 4 of 12 http://www.ro-journal.com/content/7/1/70 Figure 1 Radiation induced variation in ECG expression. (A) CT range and (B) coefficient of variation Figure shows the variations in the expression level of each ECG in A431, A549 and BxPC3. BxPC3 showed least variation as a function of different radiation qualities as compared to A431 and A549. A549 showed highest variations. stable gene. Then the M values are computed again and showed that it has aberrantly expressed ECGs, (3) the stepwise exclusion performed until two genes are left. regular decrease in the average M value for BxPC3 might The genes with the lowest M values are considered to be mean that all the ECGs were stable. the most stable across all the samples for each cell line. Calculation of normalization factor Ranking of the 32 ECGs For each cell line, the normalization factors (NF) were The M values for all the 32 ECGs in A431, A549 and computed, first for the three most stable ECGs, by BxPC3 computed using the geNorm algorithm (inte- taking the geometric mean of their expression levels. grated into SUMO software) were sorted and ranked in This is followed by stepwise inclusion of the most stable Table 2. This table revealed that the two most stable remaining ECG. Then the pairwise variations V n(n+1) ECGs irrespective of the radiation qualities were: PGK1- were calculated for every series of NF and NF , n n+1 RPL37A for A431, RPLPO-UBC for A549, and RPL37A- reflecting the effect of adding an (n+ 1)th ECG (Figure 6) RPLPO for BxPC3. [26]. The actual stepwise exclusion of the worst-scoring Figure 6 shows that the value of V was low for 3/4 ECG was displayed in Figure 5. In this figure: (1) there A431, implying that the first 3 ECGs (PGK1, RPL37A, was a very steep decrease in the average M value for PSMC4) were sufficient to be used for normalization. A431 pointing at two unstably expressed ECGs, (2) the For A549, the low value of V indicated that the first 6 6/7 irregular decrease in the average M value for A549 ECGs (RPLPO, UBC, GAPDH, MT-ATP6, CASC3, PES1) Sharungbam et al. Radiation Oncology 2012, 7:70 Page 5 of 12 http://www.ro-journal.com/content/7/1/70 Figure 2 Direct clustering of 32 ECGs expression. This Heat map represents expression of all 32 ECGs across the three cell lines (A431, A549 and BxPC3) irradiated with photon, proton and carbon ion. Genes were hierarchically clustered by Pearson correlation coefficient using average linkage. Green denotes genes with relatively decreased expression while red denotes genes with relatively increased expression. Scale bar represent log expression level of ECGs. C = Carbon, P = Proton, X = Photon and 0 = Control. Expression profile of A431 and A549 are similar as compared to BxPC3 which showed unique expression profile with less variations among the samples. were sufficient for normalization. In BxPC3, the three Radiation-specific expression of the ECGs within the cell most stable ECGs (RPL37A, RPLPO, CASC3) were suffi- lines cient for normalization purposes. The gene stability measure M value was determined and validated, the candidate ECGs within each cell line were normalized by the appropriate stable ECGs found above Validation of the gene-stability measure M and plotted in Figure 8. The variation in regulation of According to Vandesompele et al. [26], three different 18S indicated that its expression depended upon the normalization factors were calculated based on the geo- radiation quality (Figure 8A). This observation is in metric mean of three genes with, respectively, the smallest line with other above mentioned analysis performed M value (NF ), the intermediate M value (NF ) 3(1–3) 3(11–13) (Figures 2 and 4). In addition, GADD45A was differen- and the highest M value (NF )asdetermined by 3(30–32) tially regulated by radiotherapy in all three cell lines geNorm (Table 2). Further, we determined the average (Figure 8). Together, these data confirm differential regu- gene-specific variation of the three genes with the most lation of candidate control genes as a function of differ- stable expression (i.e., the smallest coefficient of variation) ent radiation qualities. for each normalization factor within each cell line (Figure 7).It is conceivable that the gene-specific variation in all the cell lines were the least when the data are nor- Identifying the consensus set of ECGs for comparative malized to (NF ). This validated that the gene-stability 3(1–3) studies across all cell types measure effectively identified the ECGs with the most To compare the ECGs expression levels across A431, stable expression. A549 and BxPC3 cells, first a consensus set of ECGs was Sharungbam et al. Radiation Oncology 2012, 7:70 Page 6 of 12 http://www.ro-journal.com/content/7/1/70 Figure 3 Cell type specific regulation of ECGs. This Heat map represents 18 ECGs which are significantly regulated according to ANOVA between the three cell lines, p< 0.05. Relative expression of each gene were normalized to the average intensity of the gene over entire samples (virtual pool). Green denotes genes with relatively decreased expression while red denotes genes with relatively increased expression. Genes are hierarchically clustered by Pearson correlation coefficient using average linkage. Scale bar represent log expression level of ECGs. C = Carbon, P = Proton, X = Photon and 0 = Control. Expression levels of the ECGs in BxPC3 is different as compared to A431 and A549. A549 showed maximum variation among the samples. identified for normalization of expression data using the Comparative investigation of gene regulation on transcrip- algorithm suggested by Vandesompele et al. [26]. RPLPO, tional level as the function of radiation treatment constitu- UBC, PPIA, TBP and PSMC4 were selected by eliminating tes a cornerstone of these studies. Quantitative real time the ECGs with high M value to normalize and compare PCR (qRT-PCR) is considered the most sensitive method the cell type specific gene-expressions (Figure 9). Although for detection of gene expression level. One limitation of this theoverall abundanceofmostECGsamong different cell method is the need for proper endogenous control gene. To lines was relatively similar, cell-line specific gene-expression generate relative expression levels, the expression of the were identified for some candidate ECGs such as, 18S, reference gene/s needs minimally alter among different B2M, YWHAZ, PGK1, CDKN1A and GADD45A.Incon- types of cells or treatments. The goal of this study was to trast, ECGs with a relatively constant expression included identify such ECGs. GAPDH, RPLPO, RPL30A, PPIA, UBC etc. In A431, a 422- We analysed the expression levels of 32 ECGs using the fold expression difference was observed between the most clustering, statistical methods such as ANOVA and the stable gene (PGK1) and the least stable gene (18S)whereas geNorm algorithm. Global analysis lead to the finding that a 530 and 375 fold difference in expression was found in gene expression profile in pancreatic cancer cells (BxPC3) A549 and BxPC3, respectively. is different as compared to the two other epithelial cancer cells tested i.e. epidermoid and lung carcinoma cells (A431 and A549). The ECGs in BxPC3 showed least vari- Discussion ation in expression whereas A549 showed maximum vari- The emergence of a growing number of particle therapy fa- ation in expression as the function of radiation qualities. cilities worldwide will stimulate comparative studies aiming Among the three cell lines, the ECGs were more stable in to decipher the molecular mechanisms underlying differen- BxPC3. From the point of view of selecting appropriate tial biological effects of these novel radiation qualities. Sharungbam et al. Radiation Oncology 2012, 7:70 Page 7 of 12 http://www.ro-journal.com/content/7/1/70 Table 2 Control genes ranked in order of their expression stability* A431 M A549 M BxPC3 M PGK1-RPL37A 0.02 RPLPO-UBC 0.12 RPL37A-RPLPO 0.03 PSMC4 0.04 GAPDH 0.13 CASC3 0.03 GUSB 0.07 MT-ATP6 0.16 RPL30 0.08 RPLPO 0.08 CASC3 0.19 UBC 0.08 UBC 0.09 PES1 0.27 RPS17 0.08 TBP 0.12 RPS17 0.31 EIF2B1 0.10 GAPDH 0.17 ACTB 0.32 ACTB 0.12 PPIA 0.16 TBP 0.42 POP4 0.13 ABL 0.16 PPIA 0.39 PSMC4 0.13 HPRT1 0.17 HMBS 0.41 PPIA 0.13 RPL30 0.17 PSMC4 0.46 ABL 0.14 PUM1 0.19 ABL 0.51 PGK1 0.14 ELF1 0.20 POLR2A 0.54 TBP 0.14 Figure 4 Radiation induced differential regulation of ECG. POLR2A 0.22 PUM1 0.60 GUSB 0.15 Figure displays gene expression levels of 18 S in the three cell lines CDKN1A 0.23 RPL30 0.64 MRPL19 0.17 (A431, A549 and BxPC3) irradiated with photon, proton and carbon ion. Among the 32 ECGs, one gene; the ribosomal protein 18 S was POP4 0.28 EIF2B1 0.65 CDKN1A 0.17 found to be differentially regulated as the function of different PES1 0.32 HPRT1 0.66 HPRT1 0.18 radiation qualities (p< 0.01 by ANOVA). Bars indicate mean MT-ATP6 0.33 CDKN1B 0.67 PUM1 0.20 expression ± standard deviation. CASC3 0.32 MRPL19 0.65 HMBS 0.21 RPS17 0.33 RPL37A 0.68 MT-ATP6 0.21 ECGs this feature might be advantageous. On the other HMBS 0.34 ELF1 0.73 B2M 0.23 hand, it means that the ECGs in BxPC3 are less regulated EIF2B1 0.37 GUSB 0.87 GAPDH 0.25 by different radiation qualities. YWHAZ 0.39 YWHAZ 0.88 PES1 0.27 ANOVA analysis of non normalized data revealed that CDKN1B 0.40 POP4 0.92 POLR2A 0.27 18 out of the 32 ECGs showed cell type specific differen- MRPL19 0.43 PGK1 0.97 YWHAZ 0.32 tial regulation as shown by the differences in the expres- sion profile between A431, A549 and BxPC3. In addition, ACTB 0.48 GADD45A 1.04 ELF1 0.33 significant radiation quality specific regulation was shown GADD45A 0.51 B2M 1.01 TFRC 0.36 by one gene the ribosomal protein 18S. Of note, this gene IPO8 0.52 TFRC 1.22 IPO8 0.37 is one of the most commonly used internal control genes. B2M 0.52 CDKN1A 1.28 18S 0.41 Therefore, our data suggest validation of this gene prior to TFRC 0.64 18S 1.27 GADD45A 0.45 its use as internal control in radiation biology studies. Although the clustering and ANOVA analysis of the raw 18S 0.96 IPO8 1.34 CDKN1B 0.58 data provided an overall overview and information about *M values increasing from top to bottom; the two most stable control genes in each cell type, for example RPLPO and UBC in A549, cannot be ranked in order ECGs regulation, they do not allow the selection of appro- because of the required use of gene ratios for gene-stability measurements. priate stable ECGs for normalization of the qRT-PCR data. Next, the geNorm algorithm was employed to test for the stability of the 32 candidate ECGs as reference genes as well as selection of the optimal number of genes for ECGs for A549 data (RPLPO, UBC, GAPDH, MTATP6, normalization of gene expression. CASC3 and PES1) were required. Using the geNorm algorithm the two most stable The normalized gene expression for each cell line in ECGs for each cell line were determined- PGK1-RPL37A Figure 8 showed that 18S and B2M are unstable genes in A431, RPLPO-UBC in A549, and RPL37A-RPLPO in under different radiation qualities. This is in contrast to BxPC3. Further the optimal number of ECGs for the earlier studies using 18S and B2M as reference genes for normalization of gene expression in each cell line was proton therapy [4]. Besides, PPIA, ACTB and UBC for determined and validated: three ECGs each for A431 particle therapy using 0.5 Gy 4He ions in normal human (PGK1, RPL37A, and PSMC4) and BxPC3 (RPL37A, lung fibroblasts [27] and ACTB for A549 [5] were RPLPO and CASC3) were recommended. In contrast, six reported as reference genes. However, Table 2 showed Sharungbam et al. Radiation Oncology 2012, 7:70 Page 8 of 12 http://www.ro-journal.com/content/7/1/70 Figure 5 Identification of the most stable ECGs in each cell line. Most stable ECGs, i.e. not differentially regulated by different Figure 7 Validation of the gene stability measure (M) and the radiation qualities in each cell lines, are identified. Average geometric averaging of carefully selected control genes for expression stability M of all remaining control genes after stepwise normalization. The average gene-specific variation (determined as exclusion of the least stable reference genes in three cell lines are coefficient of variation in percent) for the three control genes with shown. More stably expressed genes are positioned on the right side the smallest variation within each cell line after normalization with of the diagram, less stably expressed on the left side. ECGs are three different factors calculated as the geometric mean of the three ranked in order of their expression stability and presented along control genes with the lowest (NF ), intermediate (NF ) and 3(1–3) 3(11–13) x-axis. Stability values (M) determined by geNorm algorithm are highest (NF ) gene-stability values (as determined by geNorm). 3(30–32) presented along y-axis. Low stability value (M) reflects greater The data confirmed the stability of the ECGs. The value under the stability. For the gene names with their ranking refer Table 2. star indicates the normalization factor for each cell line. Figure 6 Determination of the optimal number of control genes for normalization. Pairwise variation (V ) analysis between the n(n+1) normalization factors (NF ) and (NF ) to determine the number of control genes required for accurate normalization. Normalization factors were n n+1 computed taking 3–11 most stable genes for all the three cell lines. Pairwise variation of 0.15 was taken as a cut off value [26]. For A431 and BxPC3 three ECGs were sufficient for normalization in contrast to six ECGs for A549. Sharungbam et al. Radiation Oncology 2012, 7:70 Page 9 of 12 http://www.ro-journal.com/content/7/1/70 Figure 8 Differential regulation of the ECGs after normalization with stable genes of each cell line. A) A431 normalized by PGK1, RPL37A, and PSMC4; (B) A549 normalized by RPLPO, UBC, GAPDH, MTATP6, CASC3 and PES1; (C) BxPC3 normalized by RPL37A, RPLPO and CASC3.As compared to the raw intensity data the differential regulation of the 32 ECGs are more pronounced after normalization with the identified stable genes for each cell line with correct normalization factor (NF ). that these genes have intermediate stability within the The gene YWHAZ– involved in signal transduction by cell line examined here. binding to phosphorylated serine residues on a variety of Figure 8 demonstrates the regulation of particular signaling molecule-- is up regulated in A431- photon genes treated with different radiation qualities. For in- while minimally regulated in A549-photon. GADD45A– stance, Figure 8B showed that in A549, CDKN1A– a which binds to proliferating cell nuclear antigen, stimu- gene downstream of p53 pathway which is also impli- lates DNA excision repair in vitro and inhibits entry of cated in regulation of cell growth and cell response to cells into S phase-- was down regulated in A431 but up DNA damage-- is up regulated under all the radiation regulated in BxPC3. 18S– a component of the ribosome, qualities, with maximum up regulation under photon. the protein manufacturing machinery of all living cells-- Sharungbam et al. Radiation Oncology 2012, 7:70 Page 10 of 12 http://www.ro-journal.com/content/7/1/70 Figure 9 Logarithmic histogram of the expression levels of 32 ECGs in all the cell lines. The 32 ECGs were normalized to the geometric mean of five control genes (RPLPO, UBC, TBP, PPIA and PSMC4). In A431, a 422-fold expression difference is observed between the most stable gene (PGK1) and the least stable gene (18 S), while in A549 and BxPC3, a fold difference of 530 and 375 respectively were observed between them. In addition it shows the cell line specific differences in expression levels of particular genes (e.g. YWHAZ). is seen to be up regulated in carbon ion while its regula- specific gene expression was observed. Identification of tion varies for proton and photon in different cell lines. the best internal control gene is a prerequisite for a suc- In addition, five most stable ECGs (RPLPO, UBC, TBP, cessful quantitative measure of gene expression via RT- PPIA and PSMC4) in three cell lines were selected as in- PCR. In this paper we provide a template for the identifi- ternal control genes for the normalisation of the gene ex- cation of appropriate ECGs for the radiation induced pression independent of radiation qualities and cell type. gene expression studies. We identified reliable genes for This selection was based on the guideline of Vandesom- individual expression profiling of the cell lines, the pele et al. [26]. normalization of A431 may be done by PGK1, RPL37A The gene expression in each cell lines normalized by and PSMC4; A549 by RPLPO, UBC, GAPDH, MT-ATP6, the selected five stable ECGs was shown in Figure 9. The CASC3 and PES1; and BxPC3 by RPL37A, RPLPO and expression of ACTB showed 2.8-fold difference between CASC3. However, the 5 ECGs-- TBP, UBC, RPLPO, the highest and lowest expression levels, whereas PPIA, PSMC4– can be taken as the most suitable candi- YWHAZ, 18S, GADD45A showed 5.8, 3.14 and 4-fold date reference genes for radiation response expression difference between the highest and lowest expression profiling in the tumor models studied. Moreover, this levels. The expression Heat map (Figure 3) also illustrate robust set of the most suitable candidate ECGs for radi- the regulation of YWHAZ, GADD45A and 18S. ation experiment may be applied and validated for the Selecting the ECGs for normalization across all the cell clinicopathological analysis of cancer specimens of epithe- lines is a subtle issue. Although in [26], an algorithm to lial tumors, non-small cell lung cancer and pancreatic select the best ECGs within each specific cell line is pre- adenocarcinoma. sented, a clear method of selecting the best ECGs for all the cell lines is not given. More precisely, among the 5 Methods selected ECGs (Table 3) -- TBP, UBC, RPLPO, PPIA, Cell lines PSMC4-- one ECG could be stable in one cell line, while The three different human tumour cell lines, i.e., the lung it could be of intermediate stability in the other. How- carcinoma cells (A549), the epidermoid carcinoma cells ever, our selection is supported by the fact that, in Table 2, all these genes are of intermediate stability in Table 3 List of ECGs qualified as internal control genes each of the cell lines. across the cell and radiation qualities Symbol Name RPLPO Ribosomal protein, large, P0 Conclusions Careful selection and validation of ECGs prior to con- UBC Ubiquitin C ducting radiation biology experiment is warranted. We PSMC4 Protease 26S subunit, ATPase, 4 report that different radiation qualities induced differen- PPIA Peptidylprolyl isomerase A tial regulation of a number of ECGs among the candi- TBP TATA box binding protein date 32 “housekeeping genes”. Additional cell type Sharungbam et al. Radiation Oncology 2012, 7:70 Page 11 of 12 http://www.ro-journal.com/content/7/1/70 (A431) and pancreatic cancer cells (BxPC3) were used for amplification for 40 cycles at 95.0°C for 15 s and 60.0°C the study. A549 and A431 cell lines were obtained from for 1 min. Amplification data were collected via Se- Deutsche Sammlung von Mikroorganismen und Zellkultu- quence Detection Systems 2.3 software (Applied Biosys- ren GmbH (DSMZ) and BxPC3 from the American Type tems). The CT-values were computed with RQ Manager Culture Collection. The A549 and A431 cell lines were 2.xx (Applied Biosystems). grown in 5 ml Dulbeccos Modified Eagle's Medium (DMEM) (Biochrom), BxPC3 was grown in 5 ml RPMI Statistical analyses 1640 medium (GIBCO Invitrogen) supplemented with Statistical analysis of data was performed using SUMO 10.0% FCS in T25 flasks (Becton Dickinson). Cells were software package (http://www.oncoexpress.de/software/ cultured under standard conditions in a fully humidified sumo). ANOVA was used to detect variation in the ex- incubator with 5.0% CO at 37.0°C. pression of the ECGs across the samples according to the radiation qualities and cell lines respectively. The Irradiation average expression stability measure values (M) were Cells were irradiated in T25 flasks with 2Gy of photon, computed using the geNorm algorithm suggested by 2Gy of proton and 1Gy of carbon ion. Photon was deliv- Vandesompele et al. [26] (also incorporated in the ered by a linear accelerator at 6 Mev (Mevatron, SUMO program package). Siemens, Erlangen, Germany). Particle irradiation with proton and carbon ion was done using a pencil beam in Additional files a spread out Bragg peak with 1.5 cm width equivalent to a depth of 14.0 cm in water, at the Heidelberg Ion Ther- Additional file 1: List of 32 endogenous control genes used in the study. apy Center (HIT) [28]. After irradiation, the cells were Additional file 2: Cycle threshold range and coefficient of variation incubated for 12 h at 37.0°C. Control cells were treated (CV) of 32 ECGs in each cell line. The genes are sorted by the identically but without irradiation. Cells were scrapped coefficient of variation increasing from top to bottom. Four replicates using the cell scraper after adding 300.0 μl TRIzol were used in all the three cell lines. (Invitrogen) and collected in 1.5 ml Eppendorf tubes and subsequently stored at −20.0°C. Competing interests The authors declare that they have no competing interests. RNA isolation and cDNA synthesis Acknowledgements RNA was isolated in phase lock tubes using TRIzol (Invi- We thank Claudia Rittmüller, Christiane Rutenberg and Barbara Schwager for the excellent technical assistance. This work was supported in part by the trogen) according to the manufacturer’s protocol. To German Krebshilfe (Deutsche Krebshilfe, Max-Eder 108876), DFG National avoid genomic DNA contamination RNA was treated Priority Research Program: the Tumour-Vessel Interface “SPP1190”, NASA with Dnase I (Ambion). Purified RNA was eluted in Specialized Center of Research NNJ04HJ12G, and the German Federal Ministry of Research and Technology (Bundesministerium für Bildung und 20.0μL of nuclease-free water and stored at −20.0°C. Forschung – BMBF 03NUK004C). RNA concentration and purity was assessed using a Nanodrop ND-1000 spectrophotometer (Peqlab). Integ- Author details Molecular RadioOncology [E210], National Center for Tumor Disease (NCT), rity and concentration of RNA samples were determined German Cancer Research Center (DKFZ), Im Neuenheimer Feld 460, 69120, by using RNA 6000 Nano Lab Chip kits and a 2100 Bioa- 2 Heidelberg, Germany. Heidelberg Ion Therapy Center (HIT), Heidelberg nalyzer (Agilent). RNA (2.0 μg) was subjected to reverse Institute of Radiation Oncology (HIRO), Department of Radiation Oncology, University of Heidelberg Medical School, Im Neuenheimer Feld 450, 69120, transcription reaction using the high-capacity cDNA re- Heidelberg, Germany. Center of Cancer Systems Biology, NASA Specialized verse transcription kit (Applied Biosystems) according to Center of Research, St. Elizabeth’s Medical Center, Tufts University, 736 the manufacturer’s protocol. Cambridge Street [CBR 1], 02135, Boston, MA, USA. Authors contributions Real-time PCR GDS and AA designed the experiment, performed research, analysed data To assess the expression of Human Endogenous Control and wrote the manuscript. CS developed software and performed data analysis and statistics. SC, LH and JD analysed data and edited the gene set, real-time quantitative reverse transcription manuscript. SB and TH performed the heavy ion irradiation planning and PCR (qRT-PCR) was performed on 32 candidate genes treatment. All authors read and approved the final manuscript. using TaqMan chemistry (Applied Biosystems). Experi- Received: 1 December 2011 Accepted: 17 May 2012 ments were performed at least in triplicates for the three Published: 17 May 2012 cell lines-- A431, A549 and BxPC3. Plates were run on a 7900HT Fast Real-Time PCR Systems (Applied Biosys- References tems) using Fast 96-well blocks, Standard Fast PCR cyc- 1. Abdollahi A, Folkman J: Evading tumour evasion: current concepts and perspectives of anti-angiogenic cancer therapy. Drug Resist Updat 2010, ling protocol with 10.0μL reaction volumes. Cycling 13:16–28. conditions used were-- 1 cycle initiation at 50.0°C for 2. Abdollahi A, Li M, Ping G, Plathow C, Domhan S, Kiessling F, Lee LB, 2 min and 1 cycle at 95.0°C for 10 min, followed by McMahon G, Groene HJ, Lipson KE, Huber PE: Inhibition of platelet-derived Sharungbam et al. Radiation Oncology 2012, 7:70 Page 12 of 12 http://www.ro-journal.com/content/7/1/70 growth factor signaling attenuates pulmonary fibrosis. J Exp Med 2005, 23. Vaarala M, Porvari KS, Kyllonen AP, Mustonen MV, Lukkarinen O, Vihko PT: 201:925–935. Several genes encoding ribosomal proteins are over-expressed in 3. Abdollahi A, Schwager C, Kleef J, Esposito I, Domhan S, Peschke P, Hauser K, prostate-cancer cell lines: confirmation of L7a and L37 over-expression in Hahnfeldt P, Hlatky L, Debus J, Peters JM, Friess H, Folkmann JM and Huber prostate-cancer tissue samples. Int J Cancer 1998, 78:27–32. PE: Transcriptional network governing the angiogenetic switch in human 24. Viraj P, Melissa AH, Philippa H, Hillyer P, Navarro MB, Rabin RL: Systematic pancreatic cancer. Proc Natl Acad Sci 2007, doi: 10.1073/pnas.0705505104 method for determining an ideal housekeeping gene for real-time PCR 4. Suzuki Y, Nakano T, Ohno T, Oka K: Comparison of the radiobiological analysis. J Biomol Tech 2008, 19:342–347. effect of carbon ion beam therapy and conventional radiation therapy 25. Akerman G, Rosenzweig B, Domon O, Tsai C, Bishop M, McGarrity LJ, James on cervical cancer. J Radiat Res 2007, 49:473–479. T: Alterations in gene expression profiles and the DNA-damage response 5. Akino Y, Teshima T, Kihara A, Kodera-Suzumoto Y, Inaoka M, Higashiyama S, in ionizing radiation-exposed TK6 cells. Environ Mol Mutagen 2005, Furusawa Y, Matsuura N: Carbon-ion beam irradiation effectively 45:188–205. suppresses migration and invasion of human non small-cell lung cancer 26. Vandesompele J, De Preter K, Pattyn F, Poppe B, Roy NV, Paepe AD, cells. Int J Radiat Oncol 2009, 75:475–481. Speleman F: Accurate normalization of real-time quantitative RT-PCR data 6. Domhan S, Muschal S, Schwager C, Morath C, Wirkner U, Ansorge W, by geometric averaging of multiple internal control genes. Genome Biol Maercker C, Zeier M, Huber PE, Abdollahi A: Molecular mechanisms of the 2002, 3: research0034.1-0034.11. antiangiogenic and antitumor effects of mycophenolic acid. Mol Cancer 27. Ghandhi SA, Yaghoubian B, Amundson SA: Global gene expression Ther 2008, 7:1656–1668. analyses of bystander and alpha particle irradiated normal human lung 7. Almog N, Ma L, Raychowdhury R, Schwager C, Erber R, Short S, Hlatky L, fibroblasts: Synchronous and differential responses. BMC Medical Vajkoczy P, Huber PE, Folkman J, Abdollahi A: Transcriptional switch of Genomics 2008 1:63 doi:10.1186/1755-8794-1-63 dormant tumors to fast-growing angiogenic phenotype. Cancer Res 2009, 28. Combs S, Jakel O, Haberer T, Debus J: Particle therapy at the Heidelberg 69:836–844. Ion Therapy Center (HIT) - Integrated research-driven university-hospital- 8. Dheda K, Huggett JF, Bustin SA, Johnson MA, Rook G, Zumla A: Validation based radiation oncology service in Heidelberg, Germany. Radiother of housekeeping genes for normalizing RNA expression in real-time PCR. Oncol 2010, 95:41–44. Biotechniques 2004, 37:112–119. 9. Haberhausen G, Pinsl J, Kuhn C, Markert-Hahn C: Comparative study of doi:10.1186/1748-717X-7-70 different standardization concepts in quantitative competitive reverse Cite this article as: Sharungbam et al.: Identification of stable transcription-PCR assays. J Clin Microbiol 1998, 36:628–633. endogenous control genes for transcriptional profiling of photon, proton and carbon-ion irradiated cells. Radiation Oncology 2012 7:70. 10. Hamalainen H, Tubman J, Vikman S, Kyrola T, Ylikoski E, Warrington JA, Lahesmaa R: Identification and validation of endogenous reference genes for expression profiling of T helper cell differentiation by quantitative real-time RT-PCR. Anal Biochem 2001, 299:63–70. 11. Huggett J, Dheda K, Bustin S, Zumla A: Real-time RT-PCR normalisation; strategies and considerations. Genes Immun 2005, 6:279–284. 12. Mahoney D, Carey K, Fu M, Snow R, Cameron-Smith D, Gianni Parise G, Tarnopolsky MA: Real-time RT-PCR analysis of housekeeping genes in human skeletal muscle following acute exercise. Physiol Genomics 2004, 18:226–231. 13. Barnard G, Staniunas R, Bao S, Mafune K, Steele GD Jr, John JL, Gollan JL, Chen LB: Increased expression of human ribosomal phosphoprotein P0 messenger RNA in hepatocellular carcinoma and colon carcinoma. Cancer Res 1992, 52:3067–3072. 14. Henry J, Coggin D, King C: High-level expression of the ribosomal protein L19 in human breast tumors that overexpress erbB-2. Cancer Res 1993, 53:1403–1408. 15. Suzuki T, Higgins P, Crawford D: Control selection for RNA quantitation. Biotechniques 2000, 29:332–337. 16. Murphy R, Watt K, Cameron-Smith D, Gibbons CJ, Snow RJ: Effects of creatine supplementation on housekeeping genes in human skeletal muscle using real-time RT- PCR. Physiol Genomics 2003, 12:163–174. 17. Ross D, Scherf U, Eisen M, Perou CM, Rees C, Spellman P, Iyer V, Jeffrey S, de Rijn MV, Waltham M, Pergamenschikov A, Jeffrey CF, Lee JC, Lashkari D, Shalon D, Myers TG, Weinstein JN, Botstein D, Brown PO: Systematic variation in gene expression patterns in human cancer cell lines. Nat Genet 2000, 24:227–235. 18. Schmidt U, Fuessel S, Koch R, Baretton GB, Lohse A, Tomasetti S, Unversucht S, Froehner M, Wirth MP, Meye A: Quantitative multi-gene expression profiling of primary prostate cancer. Prostate 2006, 66:1521–1534. 19. Khimani A, Mhashilkar A, Mikulskis A, O’Malley M, Liao J, Golenko EE, Mayer Submit your next manuscript to BioMed Central P, Chada S, Killian JB, Lott ST: Housekeeping genes in cancer: normalization of array data. Biotechniques 2005, 38:739–745. and take full advantage of: 20. Thellin O, Zorzi W, Lakaye B, De Borman B, Coumans B, Hennen G, Grisar T, Igout A, Heinen E: Housekeeping genes as internal standards: use and • Convenient online submission limits. Biotechnol 1999, 75:291–295. • Thorough peer review 21. Tricarico C, Pinzani P, Bianchi S, Paglierani M, Distante V, Pazzagli M, Bustin SA, Orlando C: Quantitative real-time reverse transcription polymerase • No space constraints or color figure charges chain reaction: normalization to rRNA or single housekeeping genes is • Immediate publication on acceptance inappropriate for human tissue biopsies. Anal Biochem 2002, 309:293–300. • Inclusion in PubMed, CAS, Scopus and Google Scholar 22. Tsuji N, Kamagata C, Furuya M, Kobayashi D, Yagihashi A, Morita T, Horita S, Watanabe N: Selection of an internal control gene for quantitation of • Research which is freely available for redistribution mRNA in colonic tissues. Anticancer Res 2002, 22:4173–4178. Submit your manuscript at www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Radiation Oncology Springer Journals

Identification of stable endogenous control genes for transcriptional profiling of photon, proton and carbon-ion irradiated cells

Loading next page...
 
/lp/springer-journals/identification-of-stable-endogenous-control-genes-for-transcriptional-mfhn0AARSQ

References (56)

Publisher
Springer Journals
Copyright
Copyright © 2012 by Sharungbam et al.; licensee BioMed Central Ltd.
Subject
Medicine & Public Health; Oncology; Radiotherapy
eISSN
1748-717X
DOI
10.1186/1748-717X-7-70
pmid
22594372
Publisher site
See Article on Publisher Site

Abstract

Background: Quantitative analysis of transcriptional regulation of genes is a prerequisite for a better understanding of the molecular mechanisms of action of different radiation qualities such as photon, proton or carbon ion irradiation. Microarrays and real-time quantitative RT-PCR (qRT-PCR) are considered the two cornerstones of gene expression analysis. In interpreting these results it is critical to normalize the expression levels of the target genes by that of appropriately selected endogenous control genes (ECGs) or housekeeping genes. We sought to systematically investigate common ECG candidates for their stability after different radiation modalities in different human cell lines by qRT-PCR. We aimed to identify the most robust set of ECGs or housekeeping genes for transcriptional analysis in irradiation studies. Methods: We tested the expression stability of 32 ECGs in three human cancer cell lines. The epidermoid carcinoma cells (A431), the non small cell lung carcinoma cells (A549) and the pancreatic adenocarincoma cells (BxPC3) were irradiated with photon, proton and carbon ions. Expression Heat maps, clustering and statistic algorithms were employed using SUMO software package. The expression stability was evaluated by computing: mean, standard deviation, ANOVA, coefficient of variation and the stability measure (M) given by the geNorm algorithm. Results: Expression analysis revealed significant cell type specific regulation of 18 out of 32 ECGs (p< 0.05). A549 and A431 cells shared a similar pattern of ECG expression as the function of different radiation qualities as compared to BxPC3. Of note, the ribosomal protein 18S, one of the most frequently used ECG, was differentially regulated as the function of different radiation qualities (p ≤ 0.01). A comprehensive search for the most stable ECGs using the geNorm algorithm identified 3 ECGs for A431 and BxPC3 to be sufficient for normalization. In contrast, 6 ECGs were required to properly normalize expression data in the more variable A549 cells. Considering both variables tested, i.e. cell type and radiation qualities, 5 genes-- RPLP0, UBC, PPIA, TBP and PSMC4– were identified as the consensus set of stable ECGs. Conclusions: Caution is warranted when selecting the internal control gene for the qRT-PCR gene expression studies. Here, we provide a template of stable ECGs for investigation of radiation induced gene expression. Keywords: Endogenous control genes, Internal control genes, qRT-PCR, Photon, Proton, Carbon-ion, Tumour cells, A431, A549 and BxPC3 * Correspondence: a.amir@dkfz.de Molecular RadioOncology [E210], National Center for Tumor Disease (NCT), German Cancer Research Center (DKFZ), Im Neuenheimer Feld 460, 69120, Heidelberg, Germany Heidelberg Ion Therapy Center (HIT), Heidelberg Institute of Radiation Oncology (HIRO), Department of Radiation Oncology, University of Heidelberg Medical School, Im Neuenheimer Feld 450, 69120, Heidelberg, Germany Full list of author information is available at the end of the article © 2012 Sharungbam et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Sharungbam et al. Radiation Oncology 2012, 7:70 Page 2 of 12 http://www.ro-journal.com/content/7/1/70 Background PES1 for A549; and RPL37A, RPLPO and CASC3 for In addition to direct, e.g. DNA damaging effect, system BxPC3. A systematic analysis further revealed 5 stable level cellular responses to ionizing radiation are attributed genes among the 32 candidate ECGs tested to normalize to the initiation of intracellular signals and subsequent dif- gene expression data generated in different cells and ferential regulation of genes/pathways governing various after various radiation qualities. cellular processes [1]. Therefore, detecting differential regulation of genes is critical for a better understanding of Results radiation-induced molecular effects. Transcriptional per- Expression of the 32 ECGs turbation after cell exposure to different radiation qualities In this study, 32 ECGs (Additional file 1) were evaluated is investigated to unravel the systems biology of cellular to identify the most suitable reference genes for gene ex- response underlying, normal tissue toxicity, carcinogen- pression profiling of irradiated cell lines. This collection of esis, or anti-cancer effects of irradiation [1-3]. Therefore, genes constitutes frequently used ECGs which were these studies have ramification for a broad spectrum of selected based on their relative high abundance and con- basic and applied sciences ranging from effects of space stitutive expression determined by literature search and/or radiation to carcinogenesis to cancer therapy. whole genome microarray data. The three prototypic In contrast to conventional photon irradiation the mo- tumour cell lines-- A431, A549 and BxPC3 used in this lecular effects of proton or heavier ions (e.g. carbon study are among most commonly investigated model cell ions) are less explored yet. However, emerging data indi- lines for each tumor entity. They were irradiated with pho- cate molecular differences in transcriptional response of ton, proton and carbon ions. After total-RNA isolation cells to particles as compared to photon irradiation [4,5]. and quality control using lab-on-chip bioanalyzer, qRT- One reliable and highly sensitive tool that allows rapid PCR was performed using Taqman primer and probes. and accurate results in gene expression analysis is the qRT-PCR [6,7]. As in any gene expression analysis, selec- CT-values and ECG regulation tion of a valid normalization or endogenous control To get a better overview of the CT-values among all the to correct for differences in RNA sampling is critical to cell lines, the CT-range along with the minimum and avoid misinterpretation of results. Inter-sample variation maximum CT-values were listed in Table 1. The vari- due to sample collection, RNA preparation and quality, ation of CT-values ranged from 8.97 in A431 to 28.58 in inherent sample differences, pipetting errors, different A549. The 8.97 CT-value corresponded to 18S indicating efficacies of the radiation qualities and reverse transcrip- its high abundance in the samples, whereas, the CT- tion efficiency are common sources of variability. The value 28.58 corresponded to GADD45A indicating a ideal endogenous control should have a constant expres- moderate abundance. Moreover, RPL37A, GAPDH and sion level under different experimental conditions and RPLPO exhibited small CT-range indicating less vari- be sufficiently abundant across different samples and cell ation in expression whereas the large CT-range of ACTB, lines. Although any gene that is stably expressed under a GADD45A, IPO8 showed large variations in their defined experimental condition can be used for expression. normalization, the selection is most commonly made The CT-range and coefficient of variation for each cell from the constitutively expressed ECGs. line were listed in additional file 2. Within each individ- However, the expression levels of the commonly used ual cell line irradiated with photon, proton and carbon ECGs may not only vary in different cell lines but also ion, all the ECGs in BxPC3 except for CDKN1A, 18S, under different experimental treatments or pathological POLR2A, PES1 and HMBS with CT-range 1.18, 0.82, states [8-25]. This necessitates the selection of ECGs 0.80, 0.70 and 0.61, respectively, showed the smallest which are appropriate for each experimental system. Al- CT-range, while in A549 all the ECGs except PES1, though, there has been systematic selection of ECGs for ACTB and RPS17 showed the largest CT-range various experimental systems, such selection has not (Figure 1A). In other words, as a function of different ra- been conducted so far for studying the effects of different diation qualities in BxPC3, the expression of the ECGs radiation qualities. was considerably stable, while there was more variation Here, we investigate the expression stability of 32 in A549. Moreover, A431 and A549 shared similar ex- commonly used ECGs in three human cancer cell lines pression pattern of the ECGs. The coefficient of variation irradiated with photon, proton and carbon ions. Differ- plotted in Figure 1B also reflected the similar regulation ential regulation of ECGs was found as the function of of the 32 ECGs in the three cell lines. both variables, radiation quality and cell type, respect- ively. Reliable internal control genes for individual cell ECG regulation using heat map lines were identified such as PGK1, RPL37A and PSMC4 The expression of the 32 candidate ECGs across the for A431; RPLPO, UBC, GAPDH, MT-ATP6, CASC3 and samples were also visualized in a Heat map (Figure 2). Sharungbam et al. Radiation Oncology 2012, 7:70 Page 3 of 12 http://www.ro-journal.com/content/7/1/70 Table 1 Cycle threshold (CT) values and coefficient of while ACTB, IPO8 and GADD45A depicted high vari- variation (CV) of 32 endogenous control genes across the ation in expression. The remaining ECGs were differen- samples of the cell lines tially regulated. For example, B2M and ELF1 were up Gene Symbol CT Range CT Min. CT Max. Mean CT ± SEM CV(%) regulated in BxPC3 and A431 but down regulated in RPL37A 1.13 17.95 19.08 18.44 ± 0.29 1.6 A549. On the other hand, MRPL19 and PES1 were down regulated in BxPC3 but up regulated in A431 and A549. HMBS 1.39 23.96 25.35 24.61 ± 0.41 1.7 In Figure 3, we displayed a Heat map generated from a CASC3 1.22 22.79 24.01 23.45 ± 0.42 1.83 one way ANOVA analysis at p< 0.05 between the three RPLP0 1.36 17.22 18.58 17.83 ± 0.34 1.91 cell lines. It revealed 18 differentially regulated ECGs-- PSMC4 1.56 21.04 22.6 21.90 ± 0.46 2.12 ABL, CDKN1A, PSMC4, EIF2B1, GAPDH, PPIA, TBP, ABL 1.92 22.95 24.87 23.71 ± 0.51 2.15 RPS17, UBC, B2M, ELF1, PUM, GADD45A, ACTB, UBC 1.62 19.22 20.85 19.81 ± 0.43 2.19 MRPL19, YWHAZ, CASC3 and PES1. These could be cell type specific regulations as their radiation quality GAPDH 1.25 16.01 17.25 16.53 ± 0.37 2.27 variation was minimal. Among them, the first 13 except POLR2A 2.03 23.5 25.53 24.19 ± 0.56 2.35 for PSMC4 were up regulated in BxPC3 and the MT-ATP6 1.39 15.18 16.57 15.66 ± 0.38 2.46 remaining down regulated. In A431 and A549, 12 of PES1 1.97 21.04 23.02 22.12 ± 0.55 2.49 them were down regulated and 6 were up regulated. This TBP 2.19 23.88 26.08 24.47 ± 0.64 2.62 supports the finding that the expression levels of ECGs GUSB 1.7 22.28 23.98 23.30 ± 0.63 2.71 are different in BxPC3 as compared to A431 and A549. BxPC3 showed least variations and A549 showed max- EIF2B1 2.53 23.593 26.12 24.11 ± 0.70 2.91 imum variations in gene expression as the function of ra- RPS17 1.6 17.58 19.18 18.14 ± 0.53 2.92 diation qualities. However, expression levels of ECGs RPL30 1.89 17.24 19.14 18.05 ± 0.53 2.96 were comparable in A431 and A549 (Figure 1A). POP4 1.17 22.73 23.91 23.64 ± 0.72 3.05 ANOVA analysis between the three radiation qualities PPIA 2.15 17.43 19.58 18.08 ± 0.55 3.07 at p< 0.01 revealed one gene, the ribosomal protein 18S, PUM1 2.88 22.15 25.03 23.10 ± 0.71 3.08 to be differentially regulated after different radiation qualities (Figure 4). This gene is one of the most com- HPRT1 2.73 21.18 23.91 22.08 ± 0.68 3.1 monly used internal control genes for normalisation of CDKN1B 2.08 22.8 24.98 23.98 ± 0.81 3.41 qRT-PCR based gene expression data. Therefore, caution PGK1 2.25 18.17 20.42 19.35 ± 0.69 3.61 needs to be practised in using this gene as an internal MRPL19 3.09 22.73 25.82 24.01 ± 0.95 3.96 control gene, in particular when radiation effects are CDKN1A 2.6 21.3 23.99 22.98 ± 0.95 4.17 investigated. ACTB 5.65 16.1 21.75 17.80 ± 0.75 4.23 The results in this section corroborated the findings of many other previous studies that the ECGs might be dif- ELF1 3.59 21.81 25.39 22.76 ± 1.00 4.4 ferentially regulated depending upon the experimental IPO8 4.26 23.68 27.93 24.80 ± 1.13 4.57 set-up and the cell type [8-25]. More importantly, this GADD45A 4.2 24.37 28.57 26.29 ± 1.37 4.97 analysis seems to suggest that the ECGs are differentially YWHAZ 3.56 22.17 25.73 23.96 ± 1.26 5.27 regulated by the different cell types and radiation qual- TFRC 4.26 21.15 25.41 22.13 ± 1.18 5.37 ities. We attempted to confirm this observation employ- B2M 3.76 17.99 21.75 19.33 ± 1.21 6.29 ing a systematic analysis of the expression levels. 18 S 2.95 8.79 11.95 10.78 ± 1.02 9.5 Identification of appropriate ECGs The genes are sorted by the coefficient of variation increasing from top to bottom. Gene expression levels obtained using PCR should be Four replicates were used for each cell line. Cell lines were irradiated with appropriately normalized by one or more carefully photon, proton and carbon ions. Non-irradiated samples served as control. Standard error of the mean (SEM). selected stable internal control genes. The geNorm algo- rithm developed by Vandesompele et al. [26] can deter- Direct clustering of the ECGs expression showed that mine the expression stability of control genes on the the expression profile of A431 and A549 were more basis of non-normalized expression levels. This measure similar as compared to BxPC3. BxPC3 showed least vari- relies on the principle that the expression ratio of two in- ation whereas A549 showed maximum variation in ex- ternal control genes is constant in all samples regardless pression among the samples as well as among the ECGs of the experimental condition or cell type. This algo- indicating a cell type specific expression of ECGs. As rithm computes a gene expression stability measure (M) observed in Table 1, the Heat map also revealed low for each gene based on the average pairwise expression variation of RPL37A, RPLPO and GAPDH expression, ratio and then performs a stepwise exclusion of the least Sharungbam et al. Radiation Oncology 2012, 7:70 Page 4 of 12 http://www.ro-journal.com/content/7/1/70 Figure 1 Radiation induced variation in ECG expression. (A) CT range and (B) coefficient of variation Figure shows the variations in the expression level of each ECG in A431, A549 and BxPC3. BxPC3 showed least variation as a function of different radiation qualities as compared to A431 and A549. A549 showed highest variations. stable gene. Then the M values are computed again and showed that it has aberrantly expressed ECGs, (3) the stepwise exclusion performed until two genes are left. regular decrease in the average M value for BxPC3 might The genes with the lowest M values are considered to be mean that all the ECGs were stable. the most stable across all the samples for each cell line. Calculation of normalization factor Ranking of the 32 ECGs For each cell line, the normalization factors (NF) were The M values for all the 32 ECGs in A431, A549 and computed, first for the three most stable ECGs, by BxPC3 computed using the geNorm algorithm (inte- taking the geometric mean of their expression levels. grated into SUMO software) were sorted and ranked in This is followed by stepwise inclusion of the most stable Table 2. This table revealed that the two most stable remaining ECG. Then the pairwise variations V n(n+1) ECGs irrespective of the radiation qualities were: PGK1- were calculated for every series of NF and NF , n n+1 RPL37A for A431, RPLPO-UBC for A549, and RPL37A- reflecting the effect of adding an (n+ 1)th ECG (Figure 6) RPLPO for BxPC3. [26]. The actual stepwise exclusion of the worst-scoring Figure 6 shows that the value of V was low for 3/4 ECG was displayed in Figure 5. In this figure: (1) there A431, implying that the first 3 ECGs (PGK1, RPL37A, was a very steep decrease in the average M value for PSMC4) were sufficient to be used for normalization. A431 pointing at two unstably expressed ECGs, (2) the For A549, the low value of V indicated that the first 6 6/7 irregular decrease in the average M value for A549 ECGs (RPLPO, UBC, GAPDH, MT-ATP6, CASC3, PES1) Sharungbam et al. Radiation Oncology 2012, 7:70 Page 5 of 12 http://www.ro-journal.com/content/7/1/70 Figure 2 Direct clustering of 32 ECGs expression. This Heat map represents expression of all 32 ECGs across the three cell lines (A431, A549 and BxPC3) irradiated with photon, proton and carbon ion. Genes were hierarchically clustered by Pearson correlation coefficient using average linkage. Green denotes genes with relatively decreased expression while red denotes genes with relatively increased expression. Scale bar represent log expression level of ECGs. C = Carbon, P = Proton, X = Photon and 0 = Control. Expression profile of A431 and A549 are similar as compared to BxPC3 which showed unique expression profile with less variations among the samples. were sufficient for normalization. In BxPC3, the three Radiation-specific expression of the ECGs within the cell most stable ECGs (RPL37A, RPLPO, CASC3) were suffi- lines cient for normalization purposes. The gene stability measure M value was determined and validated, the candidate ECGs within each cell line were normalized by the appropriate stable ECGs found above Validation of the gene-stability measure M and plotted in Figure 8. The variation in regulation of According to Vandesompele et al. [26], three different 18S indicated that its expression depended upon the normalization factors were calculated based on the geo- radiation quality (Figure 8A). This observation is in metric mean of three genes with, respectively, the smallest line with other above mentioned analysis performed M value (NF ), the intermediate M value (NF ) 3(1–3) 3(11–13) (Figures 2 and 4). In addition, GADD45A was differen- and the highest M value (NF )asdetermined by 3(30–32) tially regulated by radiotherapy in all three cell lines geNorm (Table 2). Further, we determined the average (Figure 8). Together, these data confirm differential regu- gene-specific variation of the three genes with the most lation of candidate control genes as a function of differ- stable expression (i.e., the smallest coefficient of variation) ent radiation qualities. for each normalization factor within each cell line (Figure 7).It is conceivable that the gene-specific variation in all the cell lines were the least when the data are nor- Identifying the consensus set of ECGs for comparative malized to (NF ). This validated that the gene-stability 3(1–3) studies across all cell types measure effectively identified the ECGs with the most To compare the ECGs expression levels across A431, stable expression. A549 and BxPC3 cells, first a consensus set of ECGs was Sharungbam et al. Radiation Oncology 2012, 7:70 Page 6 of 12 http://www.ro-journal.com/content/7/1/70 Figure 3 Cell type specific regulation of ECGs. This Heat map represents 18 ECGs which are significantly regulated according to ANOVA between the three cell lines, p< 0.05. Relative expression of each gene were normalized to the average intensity of the gene over entire samples (virtual pool). Green denotes genes with relatively decreased expression while red denotes genes with relatively increased expression. Genes are hierarchically clustered by Pearson correlation coefficient using average linkage. Scale bar represent log expression level of ECGs. C = Carbon, P = Proton, X = Photon and 0 = Control. Expression levels of the ECGs in BxPC3 is different as compared to A431 and A549. A549 showed maximum variation among the samples. identified for normalization of expression data using the Comparative investigation of gene regulation on transcrip- algorithm suggested by Vandesompele et al. [26]. RPLPO, tional level as the function of radiation treatment constitu- UBC, PPIA, TBP and PSMC4 were selected by eliminating tes a cornerstone of these studies. Quantitative real time the ECGs with high M value to normalize and compare PCR (qRT-PCR) is considered the most sensitive method the cell type specific gene-expressions (Figure 9). Although for detection of gene expression level. One limitation of this theoverall abundanceofmostECGsamong different cell method is the need for proper endogenous control gene. To lines was relatively similar, cell-line specific gene-expression generate relative expression levels, the expression of the were identified for some candidate ECGs such as, 18S, reference gene/s needs minimally alter among different B2M, YWHAZ, PGK1, CDKN1A and GADD45A.Incon- types of cells or treatments. The goal of this study was to trast, ECGs with a relatively constant expression included identify such ECGs. GAPDH, RPLPO, RPL30A, PPIA, UBC etc. In A431, a 422- We analysed the expression levels of 32 ECGs using the fold expression difference was observed between the most clustering, statistical methods such as ANOVA and the stable gene (PGK1) and the least stable gene (18S)whereas geNorm algorithm. Global analysis lead to the finding that a 530 and 375 fold difference in expression was found in gene expression profile in pancreatic cancer cells (BxPC3) A549 and BxPC3, respectively. is different as compared to the two other epithelial cancer cells tested i.e. epidermoid and lung carcinoma cells (A431 and A549). The ECGs in BxPC3 showed least vari- Discussion ation in expression whereas A549 showed maximum vari- The emergence of a growing number of particle therapy fa- ation in expression as the function of radiation qualities. cilities worldwide will stimulate comparative studies aiming Among the three cell lines, the ECGs were more stable in to decipher the molecular mechanisms underlying differen- BxPC3. From the point of view of selecting appropriate tial biological effects of these novel radiation qualities. Sharungbam et al. Radiation Oncology 2012, 7:70 Page 7 of 12 http://www.ro-journal.com/content/7/1/70 Table 2 Control genes ranked in order of their expression stability* A431 M A549 M BxPC3 M PGK1-RPL37A 0.02 RPLPO-UBC 0.12 RPL37A-RPLPO 0.03 PSMC4 0.04 GAPDH 0.13 CASC3 0.03 GUSB 0.07 MT-ATP6 0.16 RPL30 0.08 RPLPO 0.08 CASC3 0.19 UBC 0.08 UBC 0.09 PES1 0.27 RPS17 0.08 TBP 0.12 RPS17 0.31 EIF2B1 0.10 GAPDH 0.17 ACTB 0.32 ACTB 0.12 PPIA 0.16 TBP 0.42 POP4 0.13 ABL 0.16 PPIA 0.39 PSMC4 0.13 HPRT1 0.17 HMBS 0.41 PPIA 0.13 RPL30 0.17 PSMC4 0.46 ABL 0.14 PUM1 0.19 ABL 0.51 PGK1 0.14 ELF1 0.20 POLR2A 0.54 TBP 0.14 Figure 4 Radiation induced differential regulation of ECG. POLR2A 0.22 PUM1 0.60 GUSB 0.15 Figure displays gene expression levels of 18 S in the three cell lines CDKN1A 0.23 RPL30 0.64 MRPL19 0.17 (A431, A549 and BxPC3) irradiated with photon, proton and carbon ion. Among the 32 ECGs, one gene; the ribosomal protein 18 S was POP4 0.28 EIF2B1 0.65 CDKN1A 0.17 found to be differentially regulated as the function of different PES1 0.32 HPRT1 0.66 HPRT1 0.18 radiation qualities (p< 0.01 by ANOVA). Bars indicate mean MT-ATP6 0.33 CDKN1B 0.67 PUM1 0.20 expression ± standard deviation. CASC3 0.32 MRPL19 0.65 HMBS 0.21 RPS17 0.33 RPL37A 0.68 MT-ATP6 0.21 ECGs this feature might be advantageous. On the other HMBS 0.34 ELF1 0.73 B2M 0.23 hand, it means that the ECGs in BxPC3 are less regulated EIF2B1 0.37 GUSB 0.87 GAPDH 0.25 by different radiation qualities. YWHAZ 0.39 YWHAZ 0.88 PES1 0.27 ANOVA analysis of non normalized data revealed that CDKN1B 0.40 POP4 0.92 POLR2A 0.27 18 out of the 32 ECGs showed cell type specific differen- MRPL19 0.43 PGK1 0.97 YWHAZ 0.32 tial regulation as shown by the differences in the expres- sion profile between A431, A549 and BxPC3. In addition, ACTB 0.48 GADD45A 1.04 ELF1 0.33 significant radiation quality specific regulation was shown GADD45A 0.51 B2M 1.01 TFRC 0.36 by one gene the ribosomal protein 18S. Of note, this gene IPO8 0.52 TFRC 1.22 IPO8 0.37 is one of the most commonly used internal control genes. B2M 0.52 CDKN1A 1.28 18S 0.41 Therefore, our data suggest validation of this gene prior to TFRC 0.64 18S 1.27 GADD45A 0.45 its use as internal control in radiation biology studies. Although the clustering and ANOVA analysis of the raw 18S 0.96 IPO8 1.34 CDKN1B 0.58 data provided an overall overview and information about *M values increasing from top to bottom; the two most stable control genes in each cell type, for example RPLPO and UBC in A549, cannot be ranked in order ECGs regulation, they do not allow the selection of appro- because of the required use of gene ratios for gene-stability measurements. priate stable ECGs for normalization of the qRT-PCR data. Next, the geNorm algorithm was employed to test for the stability of the 32 candidate ECGs as reference genes as well as selection of the optimal number of genes for ECGs for A549 data (RPLPO, UBC, GAPDH, MTATP6, normalization of gene expression. CASC3 and PES1) were required. Using the geNorm algorithm the two most stable The normalized gene expression for each cell line in ECGs for each cell line were determined- PGK1-RPL37A Figure 8 showed that 18S and B2M are unstable genes in A431, RPLPO-UBC in A549, and RPL37A-RPLPO in under different radiation qualities. This is in contrast to BxPC3. Further the optimal number of ECGs for the earlier studies using 18S and B2M as reference genes for normalization of gene expression in each cell line was proton therapy [4]. Besides, PPIA, ACTB and UBC for determined and validated: three ECGs each for A431 particle therapy using 0.5 Gy 4He ions in normal human (PGK1, RPL37A, and PSMC4) and BxPC3 (RPL37A, lung fibroblasts [27] and ACTB for A549 [5] were RPLPO and CASC3) were recommended. In contrast, six reported as reference genes. However, Table 2 showed Sharungbam et al. Radiation Oncology 2012, 7:70 Page 8 of 12 http://www.ro-journal.com/content/7/1/70 Figure 5 Identification of the most stable ECGs in each cell line. Most stable ECGs, i.e. not differentially regulated by different Figure 7 Validation of the gene stability measure (M) and the radiation qualities in each cell lines, are identified. Average geometric averaging of carefully selected control genes for expression stability M of all remaining control genes after stepwise normalization. The average gene-specific variation (determined as exclusion of the least stable reference genes in three cell lines are coefficient of variation in percent) for the three control genes with shown. More stably expressed genes are positioned on the right side the smallest variation within each cell line after normalization with of the diagram, less stably expressed on the left side. ECGs are three different factors calculated as the geometric mean of the three ranked in order of their expression stability and presented along control genes with the lowest (NF ), intermediate (NF ) and 3(1–3) 3(11–13) x-axis. Stability values (M) determined by geNorm algorithm are highest (NF ) gene-stability values (as determined by geNorm). 3(30–32) presented along y-axis. Low stability value (M) reflects greater The data confirmed the stability of the ECGs. The value under the stability. For the gene names with their ranking refer Table 2. star indicates the normalization factor for each cell line. Figure 6 Determination of the optimal number of control genes for normalization. Pairwise variation (V ) analysis between the n(n+1) normalization factors (NF ) and (NF ) to determine the number of control genes required for accurate normalization. Normalization factors were n n+1 computed taking 3–11 most stable genes for all the three cell lines. Pairwise variation of 0.15 was taken as a cut off value [26]. For A431 and BxPC3 three ECGs were sufficient for normalization in contrast to six ECGs for A549. Sharungbam et al. Radiation Oncology 2012, 7:70 Page 9 of 12 http://www.ro-journal.com/content/7/1/70 Figure 8 Differential regulation of the ECGs after normalization with stable genes of each cell line. A) A431 normalized by PGK1, RPL37A, and PSMC4; (B) A549 normalized by RPLPO, UBC, GAPDH, MTATP6, CASC3 and PES1; (C) BxPC3 normalized by RPL37A, RPLPO and CASC3.As compared to the raw intensity data the differential regulation of the 32 ECGs are more pronounced after normalization with the identified stable genes for each cell line with correct normalization factor (NF ). that these genes have intermediate stability within the The gene YWHAZ– involved in signal transduction by cell line examined here. binding to phosphorylated serine residues on a variety of Figure 8 demonstrates the regulation of particular signaling molecule-- is up regulated in A431- photon genes treated with different radiation qualities. For in- while minimally regulated in A549-photon. GADD45A– stance, Figure 8B showed that in A549, CDKN1A– a which binds to proliferating cell nuclear antigen, stimu- gene downstream of p53 pathway which is also impli- lates DNA excision repair in vitro and inhibits entry of cated in regulation of cell growth and cell response to cells into S phase-- was down regulated in A431 but up DNA damage-- is up regulated under all the radiation regulated in BxPC3. 18S– a component of the ribosome, qualities, with maximum up regulation under photon. the protein manufacturing machinery of all living cells-- Sharungbam et al. Radiation Oncology 2012, 7:70 Page 10 of 12 http://www.ro-journal.com/content/7/1/70 Figure 9 Logarithmic histogram of the expression levels of 32 ECGs in all the cell lines. The 32 ECGs were normalized to the geometric mean of five control genes (RPLPO, UBC, TBP, PPIA and PSMC4). In A431, a 422-fold expression difference is observed between the most stable gene (PGK1) and the least stable gene (18 S), while in A549 and BxPC3, a fold difference of 530 and 375 respectively were observed between them. In addition it shows the cell line specific differences in expression levels of particular genes (e.g. YWHAZ). is seen to be up regulated in carbon ion while its regula- specific gene expression was observed. Identification of tion varies for proton and photon in different cell lines. the best internal control gene is a prerequisite for a suc- In addition, five most stable ECGs (RPLPO, UBC, TBP, cessful quantitative measure of gene expression via RT- PPIA and PSMC4) in three cell lines were selected as in- PCR. In this paper we provide a template for the identifi- ternal control genes for the normalisation of the gene ex- cation of appropriate ECGs for the radiation induced pression independent of radiation qualities and cell type. gene expression studies. We identified reliable genes for This selection was based on the guideline of Vandesom- individual expression profiling of the cell lines, the pele et al. [26]. normalization of A431 may be done by PGK1, RPL37A The gene expression in each cell lines normalized by and PSMC4; A549 by RPLPO, UBC, GAPDH, MT-ATP6, the selected five stable ECGs was shown in Figure 9. The CASC3 and PES1; and BxPC3 by RPL37A, RPLPO and expression of ACTB showed 2.8-fold difference between CASC3. However, the 5 ECGs-- TBP, UBC, RPLPO, the highest and lowest expression levels, whereas PPIA, PSMC4– can be taken as the most suitable candi- YWHAZ, 18S, GADD45A showed 5.8, 3.14 and 4-fold date reference genes for radiation response expression difference between the highest and lowest expression profiling in the tumor models studied. Moreover, this levels. The expression Heat map (Figure 3) also illustrate robust set of the most suitable candidate ECGs for radi- the regulation of YWHAZ, GADD45A and 18S. ation experiment may be applied and validated for the Selecting the ECGs for normalization across all the cell clinicopathological analysis of cancer specimens of epithe- lines is a subtle issue. Although in [26], an algorithm to lial tumors, non-small cell lung cancer and pancreatic select the best ECGs within each specific cell line is pre- adenocarcinoma. sented, a clear method of selecting the best ECGs for all the cell lines is not given. More precisely, among the 5 Methods selected ECGs (Table 3) -- TBP, UBC, RPLPO, PPIA, Cell lines PSMC4-- one ECG could be stable in one cell line, while The three different human tumour cell lines, i.e., the lung it could be of intermediate stability in the other. How- carcinoma cells (A549), the epidermoid carcinoma cells ever, our selection is supported by the fact that, in Table 2, all these genes are of intermediate stability in Table 3 List of ECGs qualified as internal control genes each of the cell lines. across the cell and radiation qualities Symbol Name RPLPO Ribosomal protein, large, P0 Conclusions Careful selection and validation of ECGs prior to con- UBC Ubiquitin C ducting radiation biology experiment is warranted. We PSMC4 Protease 26S subunit, ATPase, 4 report that different radiation qualities induced differen- PPIA Peptidylprolyl isomerase A tial regulation of a number of ECGs among the candi- TBP TATA box binding protein date 32 “housekeeping genes”. Additional cell type Sharungbam et al. Radiation Oncology 2012, 7:70 Page 11 of 12 http://www.ro-journal.com/content/7/1/70 (A431) and pancreatic cancer cells (BxPC3) were used for amplification for 40 cycles at 95.0°C for 15 s and 60.0°C the study. A549 and A431 cell lines were obtained from for 1 min. Amplification data were collected via Se- Deutsche Sammlung von Mikroorganismen und Zellkultu- quence Detection Systems 2.3 software (Applied Biosys- ren GmbH (DSMZ) and BxPC3 from the American Type tems). The CT-values were computed with RQ Manager Culture Collection. The A549 and A431 cell lines were 2.xx (Applied Biosystems). grown in 5 ml Dulbeccos Modified Eagle's Medium (DMEM) (Biochrom), BxPC3 was grown in 5 ml RPMI Statistical analyses 1640 medium (GIBCO Invitrogen) supplemented with Statistical analysis of data was performed using SUMO 10.0% FCS in T25 flasks (Becton Dickinson). Cells were software package (http://www.oncoexpress.de/software/ cultured under standard conditions in a fully humidified sumo). ANOVA was used to detect variation in the ex- incubator with 5.0% CO at 37.0°C. pression of the ECGs across the samples according to the radiation qualities and cell lines respectively. The Irradiation average expression stability measure values (M) were Cells were irradiated in T25 flasks with 2Gy of photon, computed using the geNorm algorithm suggested by 2Gy of proton and 1Gy of carbon ion. Photon was deliv- Vandesompele et al. [26] (also incorporated in the ered by a linear accelerator at 6 Mev (Mevatron, SUMO program package). Siemens, Erlangen, Germany). Particle irradiation with proton and carbon ion was done using a pencil beam in Additional files a spread out Bragg peak with 1.5 cm width equivalent to a depth of 14.0 cm in water, at the Heidelberg Ion Ther- Additional file 1: List of 32 endogenous control genes used in the study. apy Center (HIT) [28]. After irradiation, the cells were Additional file 2: Cycle threshold range and coefficient of variation incubated for 12 h at 37.0°C. Control cells were treated (CV) of 32 ECGs in each cell line. The genes are sorted by the identically but without irradiation. Cells were scrapped coefficient of variation increasing from top to bottom. Four replicates using the cell scraper after adding 300.0 μl TRIzol were used in all the three cell lines. (Invitrogen) and collected in 1.5 ml Eppendorf tubes and subsequently stored at −20.0°C. Competing interests The authors declare that they have no competing interests. RNA isolation and cDNA synthesis Acknowledgements RNA was isolated in phase lock tubes using TRIzol (Invi- We thank Claudia Rittmüller, Christiane Rutenberg and Barbara Schwager for the excellent technical assistance. This work was supported in part by the trogen) according to the manufacturer’s protocol. To German Krebshilfe (Deutsche Krebshilfe, Max-Eder 108876), DFG National avoid genomic DNA contamination RNA was treated Priority Research Program: the Tumour-Vessel Interface “SPP1190”, NASA with Dnase I (Ambion). Purified RNA was eluted in Specialized Center of Research NNJ04HJ12G, and the German Federal Ministry of Research and Technology (Bundesministerium für Bildung und 20.0μL of nuclease-free water and stored at −20.0°C. Forschung – BMBF 03NUK004C). RNA concentration and purity was assessed using a Nanodrop ND-1000 spectrophotometer (Peqlab). Integ- Author details Molecular RadioOncology [E210], National Center for Tumor Disease (NCT), rity and concentration of RNA samples were determined German Cancer Research Center (DKFZ), Im Neuenheimer Feld 460, 69120, by using RNA 6000 Nano Lab Chip kits and a 2100 Bioa- 2 Heidelberg, Germany. Heidelberg Ion Therapy Center (HIT), Heidelberg nalyzer (Agilent). RNA (2.0 μg) was subjected to reverse Institute of Radiation Oncology (HIRO), Department of Radiation Oncology, University of Heidelberg Medical School, Im Neuenheimer Feld 450, 69120, transcription reaction using the high-capacity cDNA re- Heidelberg, Germany. Center of Cancer Systems Biology, NASA Specialized verse transcription kit (Applied Biosystems) according to Center of Research, St. Elizabeth’s Medical Center, Tufts University, 736 the manufacturer’s protocol. Cambridge Street [CBR 1], 02135, Boston, MA, USA. Authors contributions Real-time PCR GDS and AA designed the experiment, performed research, analysed data To assess the expression of Human Endogenous Control and wrote the manuscript. CS developed software and performed data analysis and statistics. SC, LH and JD analysed data and edited the gene set, real-time quantitative reverse transcription manuscript. SB and TH performed the heavy ion irradiation planning and PCR (qRT-PCR) was performed on 32 candidate genes treatment. All authors read and approved the final manuscript. using TaqMan chemistry (Applied Biosystems). Experi- Received: 1 December 2011 Accepted: 17 May 2012 ments were performed at least in triplicates for the three Published: 17 May 2012 cell lines-- A431, A549 and BxPC3. Plates were run on a 7900HT Fast Real-Time PCR Systems (Applied Biosys- References tems) using Fast 96-well blocks, Standard Fast PCR cyc- 1. Abdollahi A, Folkman J: Evading tumour evasion: current concepts and perspectives of anti-angiogenic cancer therapy. Drug Resist Updat 2010, ling protocol with 10.0μL reaction volumes. Cycling 13:16–28. conditions used were-- 1 cycle initiation at 50.0°C for 2. Abdollahi A, Li M, Ping G, Plathow C, Domhan S, Kiessling F, Lee LB, 2 min and 1 cycle at 95.0°C for 10 min, followed by McMahon G, Groene HJ, Lipson KE, Huber PE: Inhibition of platelet-derived Sharungbam et al. Radiation Oncology 2012, 7:70 Page 12 of 12 http://www.ro-journal.com/content/7/1/70 growth factor signaling attenuates pulmonary fibrosis. J Exp Med 2005, 23. Vaarala M, Porvari KS, Kyllonen AP, Mustonen MV, Lukkarinen O, Vihko PT: 201:925–935. Several genes encoding ribosomal proteins are over-expressed in 3. Abdollahi A, Schwager C, Kleef J, Esposito I, Domhan S, Peschke P, Hauser K, prostate-cancer cell lines: confirmation of L7a and L37 over-expression in Hahnfeldt P, Hlatky L, Debus J, Peters JM, Friess H, Folkmann JM and Huber prostate-cancer tissue samples. Int J Cancer 1998, 78:27–32. PE: Transcriptional network governing the angiogenetic switch in human 24. Viraj P, Melissa AH, Philippa H, Hillyer P, Navarro MB, Rabin RL: Systematic pancreatic cancer. Proc Natl Acad Sci 2007, doi: 10.1073/pnas.0705505104 method for determining an ideal housekeeping gene for real-time PCR 4. Suzuki Y, Nakano T, Ohno T, Oka K: Comparison of the radiobiological analysis. J Biomol Tech 2008, 19:342–347. effect of carbon ion beam therapy and conventional radiation therapy 25. Akerman G, Rosenzweig B, Domon O, Tsai C, Bishop M, McGarrity LJ, James on cervical cancer. J Radiat Res 2007, 49:473–479. T: Alterations in gene expression profiles and the DNA-damage response 5. Akino Y, Teshima T, Kihara A, Kodera-Suzumoto Y, Inaoka M, Higashiyama S, in ionizing radiation-exposed TK6 cells. Environ Mol Mutagen 2005, Furusawa Y, Matsuura N: Carbon-ion beam irradiation effectively 45:188–205. suppresses migration and invasion of human non small-cell lung cancer 26. Vandesompele J, De Preter K, Pattyn F, Poppe B, Roy NV, Paepe AD, cells. Int J Radiat Oncol 2009, 75:475–481. Speleman F: Accurate normalization of real-time quantitative RT-PCR data 6. Domhan S, Muschal S, Schwager C, Morath C, Wirkner U, Ansorge W, by geometric averaging of multiple internal control genes. Genome Biol Maercker C, Zeier M, Huber PE, Abdollahi A: Molecular mechanisms of the 2002, 3: research0034.1-0034.11. antiangiogenic and antitumor effects of mycophenolic acid. Mol Cancer 27. Ghandhi SA, Yaghoubian B, Amundson SA: Global gene expression Ther 2008, 7:1656–1668. analyses of bystander and alpha particle irradiated normal human lung 7. Almog N, Ma L, Raychowdhury R, Schwager C, Erber R, Short S, Hlatky L, fibroblasts: Synchronous and differential responses. BMC Medical Vajkoczy P, Huber PE, Folkman J, Abdollahi A: Transcriptional switch of Genomics 2008 1:63 doi:10.1186/1755-8794-1-63 dormant tumors to fast-growing angiogenic phenotype. Cancer Res 2009, 28. Combs S, Jakel O, Haberer T, Debus J: Particle therapy at the Heidelberg 69:836–844. Ion Therapy Center (HIT) - Integrated research-driven university-hospital- 8. Dheda K, Huggett JF, Bustin SA, Johnson MA, Rook G, Zumla A: Validation based radiation oncology service in Heidelberg, Germany. Radiother of housekeeping genes for normalizing RNA expression in real-time PCR. Oncol 2010, 95:41–44. Biotechniques 2004, 37:112–119. 9. Haberhausen G, Pinsl J, Kuhn C, Markert-Hahn C: Comparative study of doi:10.1186/1748-717X-7-70 different standardization concepts in quantitative competitive reverse Cite this article as: Sharungbam et al.: Identification of stable transcription-PCR assays. J Clin Microbiol 1998, 36:628–633. endogenous control genes for transcriptional profiling of photon, proton and carbon-ion irradiated cells. Radiation Oncology 2012 7:70. 10. Hamalainen H, Tubman J, Vikman S, Kyrola T, Ylikoski E, Warrington JA, Lahesmaa R: Identification and validation of endogenous reference genes for expression profiling of T helper cell differentiation by quantitative real-time RT-PCR. Anal Biochem 2001, 299:63–70. 11. Huggett J, Dheda K, Bustin S, Zumla A: Real-time RT-PCR normalisation; strategies and considerations. Genes Immun 2005, 6:279–284. 12. Mahoney D, Carey K, Fu M, Snow R, Cameron-Smith D, Gianni Parise G, Tarnopolsky MA: Real-time RT-PCR analysis of housekeeping genes in human skeletal muscle following acute exercise. Physiol Genomics 2004, 18:226–231. 13. Barnard G, Staniunas R, Bao S, Mafune K, Steele GD Jr, John JL, Gollan JL, Chen LB: Increased expression of human ribosomal phosphoprotein P0 messenger RNA in hepatocellular carcinoma and colon carcinoma. Cancer Res 1992, 52:3067–3072. 14. Henry J, Coggin D, King C: High-level expression of the ribosomal protein L19 in human breast tumors that overexpress erbB-2. Cancer Res 1993, 53:1403–1408. 15. Suzuki T, Higgins P, Crawford D: Control selection for RNA quantitation. Biotechniques 2000, 29:332–337. 16. Murphy R, Watt K, Cameron-Smith D, Gibbons CJ, Snow RJ: Effects of creatine supplementation on housekeeping genes in human skeletal muscle using real-time RT- PCR. Physiol Genomics 2003, 12:163–174. 17. Ross D, Scherf U, Eisen M, Perou CM, Rees C, Spellman P, Iyer V, Jeffrey S, de Rijn MV, Waltham M, Pergamenschikov A, Jeffrey CF, Lee JC, Lashkari D, Shalon D, Myers TG, Weinstein JN, Botstein D, Brown PO: Systematic variation in gene expression patterns in human cancer cell lines. Nat Genet 2000, 24:227–235. 18. Schmidt U, Fuessel S, Koch R, Baretton GB, Lohse A, Tomasetti S, Unversucht S, Froehner M, Wirth MP, Meye A: Quantitative multi-gene expression profiling of primary prostate cancer. Prostate 2006, 66:1521–1534. 19. Khimani A, Mhashilkar A, Mikulskis A, O’Malley M, Liao J, Golenko EE, Mayer Submit your next manuscript to BioMed Central P, Chada S, Killian JB, Lott ST: Housekeeping genes in cancer: normalization of array data. Biotechniques 2005, 38:739–745. and take full advantage of: 20. Thellin O, Zorzi W, Lakaye B, De Borman B, Coumans B, Hennen G, Grisar T, Igout A, Heinen E: Housekeeping genes as internal standards: use and • Convenient online submission limits. Biotechnol 1999, 75:291–295. • Thorough peer review 21. Tricarico C, Pinzani P, Bianchi S, Paglierani M, Distante V, Pazzagli M, Bustin SA, Orlando C: Quantitative real-time reverse transcription polymerase • No space constraints or color figure charges chain reaction: normalization to rRNA or single housekeeping genes is • Immediate publication on acceptance inappropriate for human tissue biopsies. Anal Biochem 2002, 309:293–300. • Inclusion in PubMed, CAS, Scopus and Google Scholar 22. Tsuji N, Kamagata C, Furuya M, Kobayashi D, Yagihashi A, Morita T, Horita S, Watanabe N: Selection of an internal control gene for quantitation of • Research which is freely available for redistribution mRNA in colonic tissues. Anticancer Res 2002, 22:4173–4178. Submit your manuscript at www.biomedcentral.com/submit

Journal

Radiation OncologySpringer Journals

Published: May 17, 2012

There are no references for this article.