Variability of the reverse transcription step: practical implications.
Variability can be up to 100-fold (7), raising serious concerns about the validity of the huge number of publications that record relatively modest differences in the expression levels of mRNA and miRNAs and claim biological and clinical significance (2). When RT is used in combination with qPCR, there are additional well-publicized, but usually ignored, problems associated with normalization of RNA fold changes against reference genes (14). Not only are most reference genes never validated (15), making it difficult to assess the reliability of any conclusions (16), but their RT efficiencies also can suffer from poor reproducibility, resulting in inadequate RT robustness and lack of covariance with RT efficiencies of the genes of interest (17). The effects of this variability are exacerbated when results from different laboratories are compared, making it essential to consider appropriate quality assurance measures (18). Disregard of technical issues is not confined to RT-qPCR: unreliable quantification and false-positive results are also observed with microarrays (19, 20), and estimating relative transcript abundance by next generation sequencing is inconsistent not just at low levels of coverage but even when coverage levels are high (21).
To ascertain the reliability of currently available RTases, we set out to quantify the variability inherent in the RT step by assessing the performance of different RTases in RT-qPCR assays carried out with separate RT and qPCR steps. Our results, although based on a limited number of RTases, target genes, and sample types, lead us to conclude that reporting of results on the basis of RT should include information on empirically determined RT variability, and future experiments involving cDNA synthesis require detailed statistical evaluation of analysis methods to improve the reliability of RNA-associated data.
Materials and Methods
In-depth details of materials and methods are provided in Supplemental Methods, which accompanies the online version of this article at http://www.clinchem.org/ content/vol61/issue1. The investigational work flow was as follows. (a) We assessed uniformity of qPCR instrument blocks and reproducibility of pipetting by 48 replicate qPCR assays carried out on the Bio-Rad CFX and Illumina Eco (now PCRmax Eco48) instruments. (b) We evaluated RT variability by carrying out RT-qPCR assays targeting GAPDH (glyceraldehyde-3-phosphate dehydrogenase)  #1 on 10 cDNA replicates from of each of 6 RTases. Enzymes were scored by calculating a change in quantification cycle ([DELTA]Cq) range, which measures the difference between the highest and lowest Cq recorded for each set of replicates; the best-performing (smallest [DELTA]Cq) and worst-performing (highest [DELTA]Cq) RTases were compared further with RT-qPCR assays targeting additional markers. (c) We assessed concentration-dependent variability by RT-qPCR assays comparing the [DELTA]Cq ranges from 10 replicate RT reactions transcribed at 2 different concentrations. (d) We analyzed assay-dependent variability by RT-qPCR assays comparing the [DELTA]Cq ranges from 10 replicate RT reactions targeting different regions of 2 mRNAs. (e) We investigated sample-dependent variability by RT-qPCR assays comparing the [DELTA]Cq ranges from 10 replicate RT reactions from 4 RNA samples. (f) We mea sured the effect of RNA quality on RT variability by comparing the [DELTA]Cq ranges from 10 replicate RT reactions of 2 poor-quality RNA samples.
MINIMUM INFORMATION FOR PUBLICATION OF qPCR EXPERIMENTS
Relevant information demonstrating compliance with the Minimum Information for the Publication of Real-Time Quantitative PCR Experiments (MIQE) guidelines (22) is provided within the appropriate Excel tabs of the online Supplemental Methods; a MIQE checklist was submitted during electronic submission of the article.
Total RNA was prepared from five 30-mg breast cancer biopsy samples with the RNeasy lipid tissue mini-kit (Qiagen). Sample details as well as RNA quantity and quality assessments are shown in online Supplemental Tab 1.
PRIMERS AND PROBES
Oligonucleotide sequences, mRNA targets, and PCR efficiency details are shown in online Supplemental Tab 2.
INSTRUMENT BLOCK UNIFORMITY/PIPETTING ACCURACY
We tested block uniformity of the Bio-Rad CFX Connect (Bio-Rad) and Illumina Eco (now PCRmax Eco48) qPCR instruments by running identical 10-xL GAPDH reactions containing 300 nmol/L primers and approximately 50 ng cDNA in each alternate well of a 96-well plate (CFX) or all wells of a 48-well plate (Eco 48).
RT reactions were carried out by use of random priming within the recommended range of RNA concentrations of either 50 or 12.5 ng/[micro]L as specified by the manufacturers' RT protocols in 10 individual 5-[micro]L reactions.
One set of experiments was carried out by use of replicate 1-xL RT reactions containing 50 ng/[micro]L RNA. These were subsequently subjected to dualplex RT-qPCR analysis targeting CDH1 [cadherin 1, type 1, E-cadherin (epithelial)]/ CTNNB [catenin (cadherin-associated protein), [beta]1, 88 kDa] and CDH1/MAX (MYC associated factor X).
SYBR Green I qPCR assays were carried out in 10-[micro]L reactions containing 1 X KAPA SYBR Fast reaction mix (Anachem), primers at 300 nmol/L final concentration, and 1 [micro]L cDNA on the CFX qPCR instrument (Bio-Rad) programmed as follows: 95[degrees]C for 5 s, followed by 40 cycles of 95[degrees]C for 2 s, 60[degrees]C for 1 s, and 72[degrees]C for 1 s, with fluorescence collection at 72[degrees]C.
Dualplex assays were carried out in 10-[micro]L reactions containing 1 X NuPCR reaction mix (Illumina), NuPCR primer/probe mix (0.5 [micro]L/assay), and 1 [micro]L cDNA on the Eco thermal cycler programmed as follows: 95[degrees]C for 2 min, followed by 40 cycles of 95[degrees]C for 15 s, 50[degrees]C for 30 s, and 72[degrees]C for 30 s, with fluorescence collection at 50[degrees]C.
Standard curves were prepared with 10-, 5-, or 2-fold (depending on abundance of target) serial dilutions of C10 cDNA, diluted into 100 ng/[micro]L yeast transfer RNA (tRNA) (Life Technologies). These were run for each assay, and the slopes and y-intercepts derived from the regression equations were used to calculate sample target copy numbers. Amplification efficiencies for each assay are shown in online Supplemental Tab 2.
The statistical analyses for most data sets were analyzed and graphed with Prism 6 for Macintosh, version 6.0e (Graphpad Software), and are shown in the appropriate online supplemental tabs. The statistical analysis of the data variance shown in online Supplemental Tab 9 is provided in the online Supplemental Methods.
INSTRUMENT BLOCK UNIFORMITY/PIPETTING ACCURACY
The uniformity of the thermal cycler blocks in both qPCR instruments was high, as was the accuracy of pipetting (see online Supplemental Tab 3), with the 48 RT-qPCR assays targeting GAPDH#1 resulting in median Cqs of 20.17 (range 19.96-20.39) or 21.11 (range 20.89-21.34), respectively, for the Eco and CFX instruments.
RT-qPCR targeting GAPDH#1 resulted in [DELTA]Cqs for the 10 replicates reverse-transcribed by iScript, Vilo, Grand-script, Readyscript, Primescript, and Tetro ranging from 0.4 to 1.74 (Fig. 1; online Supplemental Tab 4A). In contrast, the [DELTA]Cqs attributable to qPCR/pipetting variability were lower and ranged from 0.07 to 0.16 (see online Supplemental Tab 4B).
A more detailed analysis of the least (Readyscript) and most (Vilo) variable RTases confirmed that Readyscript showed consistently less variation, with RT-dependent [DELTA]Cqs ranging from 0.34 to 1.74 and 0.86 to 3.05, respectively (Fig. 2A; online Supplemental Tab 5A). A similar result was obtained when a different sample (C10) was used, with Readyscript and Vilo recording median RT-dependent [DELTA]Cqs of 0.52-0.99 and 0.55-1.3, respectively (Fig. 2B; online Supplemental Tab 5B). On the other hand, Vilo recorded consistently lower Cqs for most of the mRNA targets (see online Supplemental Tab 5, C and D).
A third RNA sample (50 ng/[micro]L) was reverse-transcribed by Readyscript and subjected to dualplex CDK2 (cyclin-dependent kinase 2)/RBL1 (retinoblastoma-like 1) or MAX MYC (v-myc avian myelocytomatosis viral oncogene homolog) and singleplex UBC (ubiquitin C) #1 qPCR analysis. This resulted in RT-dependent [DELTA]Cqs ranging from 0.65 to 1.45 (Fig. 3A; online Supplemental Tab 6A), with qPCR/pipetting-associated [DELTA]Cqs from 0.25 to 0.34 (see online Supplemental Tab 6B). There was significant correlation between CDK2 and RBL1 as well as MAX and MYC, although it was higher for the former ([r.sup.2] = 0.84 vs 0.58, respectively) (see online Supplemental Tab 6, C and D).
When the experiment was repeated with the a 1:5 (5-fold) dilution of the same RNA, the RT-dependent [DELTA]Cq range increased to 2.48-4.14 (Fig. 3B; online Supplemental Tab 7), with high correlation between CDK2I RBL and MAXIMYC ([r.sup.2] = 0.94 vs 0.99, respectively) (see online Supplemental Tab 7, B--D).
RT-qPCR analysis of assays designed against different regions of UBC and GAPDH mRNA from 2 RNA samples reverse-transcribed by Readyscript and Vilo gave similar patterns. Cq values for UBC#1 and GAPDH#1 were lower than those for UBC #2 and GAPDH #2, despite similar amplification efficiencies (Fig. 4; online Supplemental Data Tab 8, A and B). As before, the RT variability across all target mRNAs was higher for Vilo than Readyscript. A repeat with Primescript of the replicate RT reactions for the GAPDH assay also gave a similar result (see online Supplemental Tab 8C).
Expression patterns of RT-qPCR assays targeting CDK2/ RB1L, MAX/MYC, or UBC #1 were similar across 4 different RNA samples reverse-transcribed by Readyscript (see online Supplemental Tab 9A). However, there was significant heterogeneity of [DELTA]Cq variance among sample types (P = 0.005) (see online Supplemental Statistical Analysis), suggesting that the observed intramarker variability is sample dependent (see online Supplemental Tab 9B). The 2 samples with RNA integrity (RIN) values of 10 showed less variability, with median RT-dependent [DELTA]Cqs of 2.2 and 2.4, as opposed to 2.8 and 3.4 for the samples with RIN values 8 and 9.5.
There was significant correlation for CDK2/RBL1 in all 4 samples, although the variability of individual RT reactions was apparent and indicated by the 95% CIs shown, which range from 0.87 to 0.99 (see online Supplemental Tab 10). Correlations for MAX/MYC were similarly significant for 3 of the RNA samples, with greater variability recorded by 1 (C10). The relative expression levels [4.1 (0.7)] and differences in fold change [1.5 (0.1)] of CDK2/RBL1 were similar in the 4 RNA samples, whereas they differed widely for MAX/MYC at [0.8 (0.5)] and [3.5 (3.0)] of MAX/MYC, respectively (Fig. 5).
We further analyzed the correlation between mRNAs expressed in sample C10 by subjecting 35 RT replicates to amplification with dualplex RT-qPCR targeting either CDH1/CTNNB or CDH1/MAX. The results are consistent with those observed previously: although there was significant correlation within both sets of markers, 1 set (CDH1IMAX) showed more variability as revealed by the lower correlation coefficient (see online Supplemental Tab 11).
RNA QUALITY DEPENDENCE OF VARIABILITY
The RNA sample with the higher integrity (C71A) recorded lower Cqs for KRT19 (keratin 19), UBC#1, and HMBS (hydroxymethylbilane synthase), but not for TP53I3 (tumor protein p53 inducible protein 3) and GAPDH (see online Supplemental Tab 12). In each series of 10 RT replicates, there was 1 (C71A sample 1, C71B sample 3) that generated significant outlier results for KRT19, UBC#1, TP53IP, and HMBS (Grubbs test, P < 0.05, 2-sided) but not for GAPDH (see online Supplemental Tab 12,AandB). The RT-dependent [DELTA]Cq range reverse-transcribed from the higher-quality RNA was narrower for all 4 targets compared to the lower-quality one (median 2.42 vs 3.70, respectively).
The probe-based assays targeting CDK2 and RBL1 recorded similar patterns, with the Cqs recorded for both targets lower in C71A (see online Supplemental Tab 13). There was 1 outlier (C71A, CDK2 sample 10) as calculated by Grubbs test, with significant correlation between the targets assessed in the dualplex reaction in sample C71A (Pearson r = 0.715, 95% CI 0.155-0.927), but none in C71B (Pearson r = 0.208, 95% CI -0.485 to 0.741). The RT-dependent [DELTA]Cq range was lower in the higher-quality RNA sample for CDK2 but not for RBL1.
We have studied the real-world variability of commercial RTases by use of RT-qPCR to quantify several mRNA targets with different RNA samples of varying concentration, integrity, and purity. Although the variability between the different enzymes was never as high as that reported previously (7), our data show that there is experimental variation and that it is sufficiently large to have important implications with regard to reproducibility, robustness, and accuracy of any molecular technique using cDNA. Variability was also sample and concentration dependent and was always higher than that of qPCR/pipetting-associated variability of technical replicates. The sample-dependent differences may be linked to differential RNA integrity and may also be affected by organic contaminants. Although the amounts of RNA used in the RT step were well within the manufacturers' specifications, we observed a significant concentration-dependent variability. A similar effect has also been identified in next-generation sequencing, where the use of limiting amounts of mRNA results in significant technical variation, with inefficient amplification of the majority of low to moderately expressed transcripts masking subtle biological differences (23). This is rather disconcerting, given the increasing interest in carrying out RNA expression analysis on single cells.
Assay-dependent variability was apparent from the results obtained with multiple assays targeting the same mRNA: RT variability differed between UBC#1 and #2 as well as between GAPDH #1 and #2. This is not easily explained by differences in amplicon secondary structures, since although there are secondary structures at the forward or reverse primer binding sites of UBC #1 and GAPDH #2, there are none in those generated by UBC #2 and GAPDH #1. Assay- and sample-dependent variability of the RT step was also apparent when comparing the expression of 5 mRNA markers across 4 RNA samples. There was statistically significant heterogeneity of [DELTA]Cq variance among sample types (Brown and Forsythe test for heterogeneity of variance, P = 0.005), which may be partly explained by differences in RNA integrity, with high RIN value samples recording lower RT-dependent [DELTA]Cqs than those with lower RIN values. Integrity-associated variability was also evident from the results obtained from a comparison of 2 RNA preparations extracted from the same sample at different times. One looked acceptable on an electropherogram although it did not yield a RIN value; the other was degraded, with a RIN value of 4.9. The latter sample recorded higher Cqs and showed more variability across all but 1 of the targets quantified by qPCR. This observation of assay-dependent contribution to data variability reemphasizes the need for empirical assay design and validation before carrying out RT-qPCR analysis (24) and indicates that choosing optimal assays can reduce data variability. It also substantiates the original MIQE requirement of submitting primer sequences with each publication, since even a small difference in primer sequence can change the performance of an RT-qPCR assay (25).
Dualplex qPCR assays targeting CDK2/RBL1 and MAX/MYC in high-quality RNA preparations indicate that the RT reactions are highly correlated, although again there is random variability apparent, as demonstrated by variable correlation coefficients and 95% CIs. For example, the [r.sup.2] value between CDK2 and RBL1 in 4 RNA samples is around 0.94, but is similarly high for MAX/MYC in only 3 of the samples, dropping to 0.76 in 1 (see online Supplemental Tab 10). An analysis of the correlation of CDK2/RBL1 Cqs in 2 poor-quality samples shows significantly lower correlation, in line with the absence of an RIN value for 1 sample and the low RIN value for the other (see online Supplemental Tab 14). This is probably associated with differential stability of different transcripts and emphasizes again the need for care when comparing RNA samples with different integrity values. The data confirm the view that a universal RIN value provides no definitive information about the integrity of individual mRNA species, and hence must be treated with caution. They also suggest that variability for RNA samples is not constant and that minimal variability and high correlation between markers in 1 sample cannot be extrapolated to other samples. This irregularity is also apparent when the correlation is analyzed between CDH1 mRNA and 2 mRNAs specifying proteins CTNNB1 and MAX, previously reported as being associated with CDH1 (26, 27). The correlation between CDH1 and CTNNB1 mRNA is much better than between CDH1 and MAX, with respective correlation coefficients of 0.96 and 0.83 and 95% CIs of 0.93-0.98 and 0.69-0.91 (see online Supplemental Tab 12).
Expression levels of genes of interest are usually reported as fold changes relative to those of 1 or more reference genes. The results reported in online Supplemental Tab 9 were used to calculate the expression levels of CDK2, RBL1, MAX, and MYC relative to the reference gene UBC #1 (see online Supplemental Tab 15). It is apparent that the relative fold changes are sample and assay dependent, as 5 of 16 assays (31%) record fold changes that are <2-fold, another 7 (44%) have changes that are between 2- and 3-fold, and 4 (25%) record fold changes >3-fold, with a maximum of nearly 7-fold. This sample- and target-dependent variability is further emphasized by the results shown in Fig. 5, which imply that small fold changes for CDK2/RBL1 can be reliably quantified, whereas the much greater differences in fold change for MAX/MYCsuggest that small fold changes for these markers cannot be reliably quantified.
The previously documented variability of the RT step (6,7) is acknowledged in the MIQE guidelines, which require a detailed description of the protocol and reagents used to convert RNA into cDNA (22). Our results extend the earlier data by demonstrating that the RT step is defined by its dependence on enzyme/buffer formulation, sample, assay, and RNA concentration, that this dependence results in significant variability in RT efficiency, and that this variability is inconsistent. For these reasons, it is essential that the reporting of results involving cDNA syntheses is as transparent and complete as possible and includes the amount of RNA reverse-transcribed, priming strategy, enzyme type, volume, temperature, and duration of the reverse transcription step. Our data also provide further support for the suggestion that the reverse-transcription step be carried out in duplicate or triplicate (22, 28). Unfortunately, the research community continues to disregard the existence of RT variability, and this recommendation remains unheeded.
In addition, there is a lack of transparency in the reporting of experimental detail (15), which calls into question the reliability of conclusions on the basis of data obtained from RT-qPCR (16,29). The implications with regard to reproducibility, robustness, and accuracy of RTase-based assays are clear. Most published RNA biomarkers report small fold changes between healthy and diseased samples, with the vast majority between 2-and 8-fold (for example, (30-37)). However, virtually none turn out to be of clinical significance (38), and around 80% of the results cannot be reproduced independently (39), likely because of serious technical issues (2), consistent with an estimate that around 85% of research funding is wasted (40). For the first time, we demonstrate that the variability of the RT step is 1 reason for this, as the variability inherent in the RT step is well within the range reported for most biomarkers (see online Supplemental Tab 16) and so calls into question many results published with regard to RNA-based biomarkers.
These conclusions support earlier observations; however, neither the complexity and unpredictability of RT variability nor the impact of these factors on data interpretation and consequences have been demonstrated in this detail before. Although the conclusions arising from the previous publications are being universally ignored, it is essential that this information be reintroduced into the consciousness of the research community. A simple work flow designed to identify the most appropriate RTase for any sample/target combination is presented in Fig. 6. This is of particular importance when considered alongside the increasing speed of technological development that makes it more and more difficult to conduct detailed troubleshooting or quality assessment of the enormous amounts of published data.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contribution to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: T. Nolan, Sigma Aldrich.
Consultant or Advisory Role: G.L. Shipley, Roche Applied Science.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: S.A. Bustin, two grants from the University Faculty of Health, Social Care and Education Support Fund.
Expert Testimony: None declared.
Patents: None declared.
Other Remuneration: S.A. Bustin, Bio-Rad, Takara, TATAA Biocentre, Bioline, Sigma, and Life Technologies provided RT enzymes free of charge.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
(1.) Sewall CH, Bell DA, Clark GC, Trltscher AM, Tully DB, Vanden Heuvel J, et al. Induced gene transcription: implications for biomarkers. Clin Chem 1995; 41: 1829-34.
(2.) Bustin SA, Murphy J. RNA biomarkers in colorectal cancer. Methods 2013; 59:116-25.
(3.) Freeman WM, Walker SJ, Vrana KE. Quantitative RTPCR: pitfalls and potential. Biotechniques 1999; 26: 112-5.
(4.) Bustin SA. Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): trends and problems. J Mol Endocrinol 2002; 29:23-39.
(5.) Chandler DP, Wagnon CA, Bolton HJ. Reverse transcriptase (RT) inhibition of PCR at low concentrations of template and its implications for quantitative RT-PCR. Appl Environ Microbiol 1998; 64:669-77.
(6.) Stahlberg A, Hakansson J, Xian X, Semb H, Kubista M. Properties of the reverse transcription reaction in mRNA quantification. Clin Chem 2004; 50:509-15.
(7.) Stahlberg A, Kubista M, Pfaffl M. Comparison of reverse transcriptases in gene expression analysis. Clin Chem 2004; 50:1678-80.
(8.) Sanders R, Mason DJ, Foy CA, Huggett JF. Evaluation of digital PCR for absolute RNA quantification. PLoS One 2013; 8:e75296.
(9.) Bustin SA, Nolan T. Pitfalls of quantitative real-time reverse-transcription polymerase chain reaction. J Biomol Tech 2004; 15:155-66.
(10.) Nam DK, Lee S, Zhou G, Cao X, Wang C, Clark T, et al. Oligo(dT) primer generates a high frequency of truncated cDNAs through internal poly(A) priming during reverse transcription. Proc Natl Acad SciUSA2002; 99: 6152-6.
(11.) Perez-Novo CA, Claeys C, Speleman F, Van Cauwenberge P, Bachert C, Vandesompele J. Impact of RNA quality on reference gene expression stability. Biotechniques. 2005; 39:52, 54, 56.
(12.) Vermeulen J, De Preter K, Lefever S, Nuytens J, De Vloed F, Derveaux S, et al. Measurable impact of RNA quality on gene expression results from quantitative PCR. Nucleic Acids Res 2011; 39:e63.
(13.) Nolan T, Hands RE, Bustin SA. Quantification of mRNA using real-time RT-PCR. Nat Protoc 2006; 1:1559-82.
(14.) Bustin SA. Real-time, fluorescence-based quantitative PCR:a snapshot of current procedures and preferences. Expert Rev Mol Diagn 2005; 5:493-8.
(15.) Bustin SA, BenesV, Garson J, Hellemans J, Huggett J, Kubista M, et al. The need for transparency and good practices in the qPCR literature. Nat Methods 2013; 10: 1063-7.
(16.) Dijkstra JR, van Kempen LC, Nagtegaal ID, Bustin SA. Critical appraisal of quantitative PCR results in colorectal cancer research: can we rely on published qPCR results? Mol Oncol 2014; 8:813-8.
(17.) Linden J, Ranta J, Pohjanvirta R. Bayesian modeling of reproducibility and robustness of RNA reverse transcription and quantitative real-time polymerase chain reaction. Anal Biochem 2012; 428:81-91.
(18.) Keilholz U, Willhauck M, Rimoldi D, Brasseur F, Dummer W, Rass K, et al. Reliability of reverse transcription polymerase chain reaction (RT-PCR)-based assays for the detection of circulating tumour cells: a quality assurance initiative of the EORTC Melanoma Cooperative Group. Eur J Cancer 1998; 34:750-3.
(19.) Miklos GL, Maleszka R. Microarray reality checks in the context of a complex disease. Nat Biotechnol 2004; 22: 615-21.
(20.) Chagovetz A, Blair S. Real-time DNA micro arrays: reality check. Biochem Soc Transact 2009; 37:471-5.
(21.) McIntyre LM, Lopiano KK, Morse AM, Amin V, Oberg AL, Young LJ, et al. RNA-seq: technical variability and sampling. BMC Genomics 2011; 12:293.
(22.) Bustin SA, BenesV, Garson JA, Hellemans J, Huggett J, Kubista M, et al. The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin Chem 2009; 55:611-22.
(23.) Bhargava V, Head SR, Ordoukhanian P, Mercola M, Subramaniam S. Technical variations in low-input RNA-seq methodologies. Sci Rep 2014; 4:3678.
(24.) Bustin SA. Why the need for qPCR publication guidelines?-The case for MIQE. Methods 2010; 50:217-26.
(25.) Bustin SA. A-Z of quantitative PCR. La Jolla, CA: IUL Press; 2004.
(26.) Hulsken J, Birchmeier W, Behrens J. E-cadherin and APC compete for the interaction with beta-catenin and the cytoskeleton. J Cell Biol 1994; 127:2061-9.
(27.) Batsche E, Muchardt C, Behrens J, Hurst HC, Cremisi C. RB and c-Myc activate expression of the E-cadherin gene in epithelial cells through interaction with transcription factor AP-2. Mol Cell Biol 1998; 18: 3647-58.
(28.) Tichopad A, Kitchen R, Riedmaier I, Becker C, Stahlberg A, Kubista M. Design and optimization of reverse-transcription quantitative PCR experiments. Clin Chem 2009; 55:1816-23.
(29.) Bustin S. Transparency of reporting in molecular diagnostics. Int J Mol Sci 2013; 14:15878-84.
(30.) Wu X, Nguyen BC, Dziunycz P, Chang S, Brooks Y, Lefort K, et al. Opposing roles for calcineurin and ATF3 in squamous skincancer. Nature 2010; 465:368-72.
(31.) Wang YK, Zhu YL, Qiu FM, Zhang T, Chen ZG, Zheng S, et al. Activation of Akt and MAPK pathways enhances the tumorigenicity of CD133+ primary colon cancer cells. Carcinogenesis 2010; 31:1376-80.
(32.) Plum L, Lin HV, Dutia R, Tanaka J, Aizawa KS, Matsumoto M, et al. The obesity susceptibility gene Cpe links Fox01 signaling in hypothalamic pro-opiomelanocortin neurons with regulation of food intake. Nat Med 2009; 15:1195-201.
(33.) Koppikar P, Bhagwat N, Kilpivaara O, Manshouri T, Adli M, HricikT, et al. Heterodimeric JAK-STAT activation as a mechanism of persistence to JAK2 inhibitor therapy. Nature 2012; 489:155-9.
(34.) Lardon J, Corbeil D, Huttner WB, Ling Z, Bouwens L. Stem cell marker prominin-1/AC133 is expressed in duct cells of the adult human pancreas. Pancreas 2008; 36:e1-6.
(35.) Liu Z, Shao Y, Tan L, Shi H, Chen S, Guo J. Clinical significance of the low expression of FER1L4 in gastric cancer patients. Tumour Biol 2014; 35:9613-7.
(36.) Ou-Yang QH, Duan ZX, Jin Z, Lei JX. OLC1 is over-expressed in breast cancer and its expression correlates with poor patient survival. Tumour Biol 2014; 35: 8823-7.
(37.) Liu T, Xia B, Lu Y, Xu Y, Lou G. TNFAIP8 over-expression is associated with platinum resistance in epithelial ovarian cancers with optimal cytoreduction. Hum Pathol 2014; 45:1251-7.
(38.) Macleod MR, Michie S, Roberts I, Dirnagl U, Chalmers I, Ioannidis JP, et al. Biomedical research: increasing value, reducing waste. Lancet 2014; 383: 101-4.
(39.) Prinz F, Schlange T, Asadullah K. Believe it or not: how much can we rely on published data on potential drug targets? Nat Rev Drug Discov 2011; 10:712.
(40.) Chalmers I, Glasziou P. Avoidable waste in the production and reporting of research evidence. Lancet 2009; 374:86-9.
Stephen Bustin,  * Harvinder S. Dhillon,  Sara Kirvell,  Christina Greenwood,  Michael Parker,  Gregory L. Shipley,  and Tania Nolan 
 Postgraduate Medical Institute, Faculty of Medical Science, Anglia Ruskin University, Chelmsford, UK;  Shipley Consulting, LLC, Austin, TX;  Institute of Population Health, Faculty of Medical and Human Sciences, University of Manchester, Manchester, UK.
* Address correspondence to this author at: Postgraduate Medical Institute, Bishop Hall Lane, Chelmsford, UK CM11SQ. E-mail email@example.com.
Received July 25, 2014; accepted September 23, 2014.
Previously published online at DOI: 10.1373/clinchem.2014.230615
 Nonstandard abbreviations: RT, reverse transcription; qPCR, real-time quantitative PCR; RT-qPCR, real-time quantitative RT-PCR; RTase, reverse transcriptase; Cq, quantification cycle; MIQE, Minimum Information for the Publication of Real-Time Quantitative PCR Experiments; tRNA, transfer RNA; RIN, RNA integrity.
 Human genes: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; CDH1, cadherin 1, type 1, E-cadherin (epithelial); CTNNB, catenin (cadherin-associated protein), [beta]1, 88 kDa; MAX, MYC associated factor X; CDK2, cyclin-dependent kinase 2; RBL1, retinoblastoma-like 1; MYC, v-myc avian myelocytomatosis viral oncogene homolog; UBC, ubiquitin C; KRT19, keratin 19; HMBS, hydroxymethylbilane synthase; TP53I3, tumor protein p53 inducible protein 3.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Molecular Diagnostics and Genetics|
|Author:||Bustin, Stephen; Dhillon, Harvinder S.; Kirvell, Sara; Greenwood, Christina; Parker, Michael; Shiple|
|Date:||Jan 1, 2015|
|Previous Article:||Circulating long noncoding RNA TapSAKI is a predictor of mortality in critically ill patients with acute kidney injury.|
|Next Article:||Solexa sequencing: decoding genomes on a population scale.|