Quantitative reverse transcription PCR (qRT-PCR) is one of the most precise and widely used methods of gene expression analysis. A necessary prerequisite of exact and reliable data is the accurate choice of reference genes. We studied the expression stability of potential reference genes in common buckwheat (Fagopyrum esculentum) in order to find the optimal reference for gene expression analysis in this economically important crop. Recently sequenced buckwheat floral transcriptome was used as source of sequence information. Expression stability of eight candidate reference genes was assessed in different plant structures (leaves and inflorescences at two stages of development and fruits). These genes are the orthologs of Arabidopsis genes identified as stable in a genome-wide survey gene of expression stability and a traditionally used housekeeping gene GAPDH. Three software applications – geNorm, NormFinder and BestKeeper - were used to estimate expression stability and provided congruent results. The orthologs of AT4G33380 (expressed protein of unknown function, Expressed1), AT2G28390 (SAND family protein, SAND) and AT5G46630 (clathrin adapter complex subunit family protein, CACS) are revealed as the most stable. We recommend using the combination of Expressed1, SAND and CACS for the normalization of gene expression data in studies on buckwheat using qRT-PCR. These genes are listed among five the most stably expressed in Arabidopsis that emphasizes utility of the studies on model plants as a framework for other species.
Citation: Demidenko NV, Logacheva MD, Penin AA (2011) Selection and Validation of Reference Genes for Quantitative Real-Time PCR in Buckwheat (Fagopyrum esculentum) Based on Transcriptome Sequence Data. PLoS ONE 6(5): e19434. https://doi.org/10.1371/journal.pone.0019434
Editor: Leonardo Mariño-Ramírez, National Institutes of Health, United States of America
Received: February 15, 2011; Accepted: March 29, 2011; Published: May 12, 2011
Copyright: © 2011 Demidenko et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The study is supported by the Russian Foundation for Basic Research (project №09-04-01363), the Ministry of Education and Science of Russian Federation (GK-16.512.11.2153 for AAP and P913 for MDL) and President's program for support of scientific schools (project №3293.2010.4) and Innovative Project Grants for Young Scientists (UMNIK-13151-9) for NVD. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Analysis of gene expression using qRT-PCR is widely used in many fields of plant research including development , , response to abiotic stress , , and to pathogen infection , . It remains the most sensitive and most accessible for wide range of researchers compared with other methods of gene expression analysis (microarray, Northern blot, whole transcriptome shotgun sequencing). Recently a set of guidelines aimed at the improvement of the reliability and reproducibility of the studies using qRT-PCR was published . Despite its numerous advantages, this method has however several issues, with one of the most important being the normalization. There are several strategies of normalization  but the most popular is the use of reference gene. This is a gene whose expression is presumably stable in control and experimental conditions. Though there are several genes traditionally used for this purpose (18S rRNA, actin, tubulin), many studies demonstrate that the expression of these genes varies greatly and the selection of alternative genes is favourable for accurate normalization , .
In this article we report the search and validation of reference genes for quantitative real-time PCR in buckwheat (Fagopyrum esculentum). Buckwheat is an important grain and honey crop widely cultivated in several countries (Canada, China, Japan, Russia and Ukraine). Its seeds are used as whole grain, groats and flour. Because of similar usage, buckwheat is usually classified as cereal. However, phylogenetically it is far from true cereals - members of the monocot family Poaceae. Buckwheat belongs to the family Polygonaceae from the order Caryophyllales - the group of flowering plants which is distant from any model plant species. By now, only few studies investigating buckwheat gene expression by real-time PCR are reported [e.g.11], [ 12].
We have evaluated the expression stability of eight candidate reference genes in five developmental stages in Fagopyrum esculentum. These genes are putative orthologs of the Arabidopsis genes identified as the most stable in ; seven of them are not widely used as reference and one is a traditionally used housekeeping gene, that of glyceraldehyde 3-phosphate dehydrogenase. To obtain the sequences of these genes we used the assembly of buckwheat floral transcriptome sequenced by 454 technology .
Large-scale sequencing of buckwheat genes reported in  allows efficient search of candidate genes for agriculturally important traits. Validation of reference genes reported in this study will enable the analysis of their expression in a variety of developmental and environmental conditions.
Selection of candidate reference genes and amplification specificity and efficiency
First, we performed a search for the orthologs of the genes listed as stably expressed in the genome-wide investigation of Arabidopsis thaliana  in buckwheat transcriptome data. This search revealed eight genes for which only one sequence with high similarity to a certain gene was found. These genes are the orthologs of AT4G33380 (unknown function, Expressed1), AT4G26410 (unknown function, Expressed2), AT5G15710 (F-box family, F-box), AT4G34270 (Tip41-like, Tip41), AT5G46630 (clathrin adaptor complex subunit, CACS), AT2G28390 (SAND family, SAND), AT3G53090 (ubiquitin protein ligase 7, UPL7) and AT1G13440 (glyceraldehyde 3-phosphate dehydrogenase, GAPDH). For other genes, either no hits or multiple sequences that have high similarity to the Arabidopsis gene but differ significantly one from another were found. Thus these eight genes were chosen as candidate reference genes and used for primer design. Agarose gel electrophoresis with SYBR Green staining and melting curve analysis revealed single products of expected length (Fig. 1a, b). The sequencing of amplicons confirmed that single products corresponding to the contigs that were used for primer design are generated (GenBank accession numbers: JF343809 - Expressed1, JF343807 - Expressed2, JF343804 - F-box, JF343806 - Tip41, JF343811 - CACS, JF343810 - SAND, JF343805 - UPL7 and JF343808 - GAPDH). No primer dimers or non-specific amplicons were revealed with no template amplification. No signals were also detected from no RT-controls, what reveals no genomic DNA contamination. Average amplification efficiency varied from 89,1% and 91,3% (Table 1).
(a) amplified fragments were separated by 3% SYBR Green staining agarose gel. Highest marker band corresponds to 500 bp; each following band is 50 bp less than preceding (b) melting curves generated for all amplicons.
Expression stability of candidate reference genes
Quantification cycle (Cq) data were obtained from each reaction with eight primer pairs. To reveal differences in transcript levels between various candidate genes, the average Cq was calculated across all samples (Fig. 2a). For seven genes Cq values ranged between 23 and 25 cycles, and only one gene – GAPDH – have the most abundant transcript level (mean Cq = 18,6). To determine genes with the least Cq values dispersal interquantile ranges were calculated. There are four genes with relatively low Cq dispersal: Expressed 1, SAND, CACS and Tip41 (Fig. 2b).
(A) Cq values for each reference gene in all buckwheat samples. A line across the box depicts the median. The box indicates the 25% and 75% percentiles. Whiskers represent the maximum and minimum values, circles represent outliers and asterisks indicate extremes. (B) interquantile ranges indicates variability of Cq values among 25% and 75%.
Three programs were applied to calculate the expression stability of candidate reference genes: geNorm , NormFinder  and BestKeeper . Expression stability values were determined across all samples. Cq values were used directly for stability calculations for BestKeeper or were transformed to relative quantities using delta-Cq method (geNorm, NormFinder).
a) GeNorm analysis.
GeNorm is a Visual Basic application tool for Microsoft Excel that operates on the assumption that expression ratio of two ideal reference genes is constant throughout the different groups of templates. Gene expression stability value (M) was calculated for all genes involved into investigation. Vandesompele et al.  recommends using M value below the threshold of 1.5. In our analysis all genes had M less than 1,5 that allows to consider genes as rather stable. Three genes: SAND, Expressed1 and CACS had the highest expression stability values (the lowest M value), F-box revealed least stability value and other four genes occupy the intermediate positions between these both groups (Fig. 3a). To determine optimal number of reference genes geNorm calculates the pairwise variation Vn/Vn+1 between two sequential normalization factors NFn and NFn+1 that contain an increasing number of reference genes. The large value of variation means that adding another gene is necessary for calculation of a reliable normalization factor. The most accurate way is to include the next gene until the variation Vn/Vn+1 drops below recommended threshold of 0.15 and current investigation revealed that pairwise variation V2/3 was 0.096, that indicates, that addition of the second gene is necessary for accurate analysis and adding the third gene is optional (Fig. 3b).
(A) expression stability of all 8 candidate reference genes from all samples. A lower average expression stability M value indicates more stable expression. (B) pairwise variantion calculated by geNorm to determine the minimum number of reference genes for accurate normalisation. Asterisk indicates the optimal number of reference genes.
b) NormFinder analysis.
NormFinder is another Excel application, which uses a model-based approach for identifying the most stable reference genes, based on combining samples into groups. In current study these five groups correspond to different types of samples (fruits, leaves and inflorescences on various developmental stages). NormFinder calculates intra- and inter-group variations and genes with the least ones are considered to be stable. As well as geNorm, NormFinder revealed SAND and Expressed1 as stable, but third gene was Tip41, while geNorm ranks it only on the sixth place (Fig. 4). However NormFinder results suggest using as references the same genes as geNorm results. Similar to geNorm, NormFinder revealed that the least stable gene is F-box.
c) BestKeeper analysis.
Third applet uses raw Cq data and determines the most stably expressed genes based on coefficient of correlation (r) to the BestKeeper Index (BI). BI is the geometric mean of Cq values of best reference genes. The BestKeeper revealed, that the best correlations were obtained for CACS (r = 0.858), Expressed1 (r = 0.818) and SAND (r = 0.812) with p value of 0.001. F-box ranked as the least stable gene with r = −0.331 (Fig. 5).
The effects of choice of reference gene
To illustrate effects of the choice of non-optimal reference gene we modeled a situation of gene expression analysis, taking as reference in the first case the gene identified as the most stable (Expressed1) and then the least stable (F-box). Two other genes, identified as stable - SAND and CACS - were considered as target genes. When the expression level of SAND and CACS was calculated relative to Expressed1 it was found to be stable (1.4–1.6 for SAND and 1.2–1.8 for CACS) in all plant structures (Fig. 6a). In contrast, when F-box was used as reference, the relative expression of these genes varied greatly: 1.9–3 for SAND and 1.9–3.2 for CACS (Fig. 6b). This contradicts to the results of evaluation of gene expression stability (see above) that demonstrate the stability of expression of SAND and CACS in different buckwheat tissues. Thus great difference between gene expression levels of SAND and CACS is an artifact caused by the wrong choice of reference gene.
(a) two most stable genes – SAND and CACS normalised by Expressed1, (b) SAND and CACS normalised by F-box. Data was obtained from all sample groups (fruits, leaves and inflorescences on various developmental stages). Error bars represent standard error of the mean.
Transcriptome sequence data as a source for robust normalization genes
Normalization is one of the key factors affecting the accuracy and reliability of the quantitative gene expression analysis. In view of this, the systematic validation of reference genes for new experimental systems was advocated  including in plant science , and several algorithms for the selection of the most stably expressed genes were developed –. These algorithms became now very widely used - for example, geNorm – the Excel plug-in, implementing one of them was downloaded about 15 thousand times . The number of articles reporting the validation of reference genes in plants also increased in past few years –. Most of them however explore the stability of traditionally used housekeeping genes such as those encoding 18S ribosomal RNA, ribosomal proteins, actins, tubulins, elongation factor 1alpha, GAPDH and so on (Table S1). In a genome-wide survey of gene expression stability in Arabidopsis thaliana it was shown that many genes are more stably expressed than these reference genes . These genes are now widely adopted for normalization in the studies on Arabidopsis , . The orthologs of these genes are obvious candidates for reference genes in other species; they were, however, tested only in few species ,  and in all these species they were found to be among the most stably expressed (Table S1).
One of the main reasons of the limited use of these genes for normalization in other species is the limited availability of their sequences. While highly conserved genes such as actin, ef1-alpha, 18S rRNA can be easily sequenced using consensus primers (or even consensus primers can be used for qPCR ) the sequences of these novel reference genes are variable between species that complicates the design of primers for sequencing or qPCR. Also most of them are expressed at much lower levels than traditional reference genes thus they are less likely to be represented in EST sequencing data (especially low-coverage). With the advent of next-generation sequencing, large-scale transcriptomic data became available for many plant species – and these data presumably have a great potential as a source of candidate reference genes. However to our knowledge they have never been used for this purpose. Our results demonstrate that 454 transcriptome sequencing data are indeed a useful source of potential reference genes.
The choice of reference gene for real-time PCR data normalization in buckwheat
Our evaluation of the stability of candidate reference genes using three different applications yielded similar results. In particular, Expressed1 and SAND were listed among as best reference genes. Since these applications use principally different algorithms for their calculations it is common to get different results , . The few discrepancies that are however observed is that CACS was considered as stable by only two programs and TIP41 is highly ranked by NormFinder, in contrast to two other programs. F-box was revealed as the worse gene by all three approaches (Table 2). This is consistent with the genome-wide investigation  where these genes are among five the most stable and supports the suggestion that the orthologs of the novel reference genes that they have identified could serve the same purposes in other species. Similar results were obtained in the study of gene expression stability in tomato  – one of the few studies adopting new reference genes proposed in . Optimal number of reference genes calculated by geNorm is two – Expressed1 and CACS. However taking into account that NormFinder does not support high stability of CACS but lists SAND as the most stable (according to two other applications SAND is on the third place) and that CACS expression is known to be affected by several treatments  we suggest using three reference genes - Expressed1, CACS and SAND. We expect that use of these genes for normalization in the future studies evaluating gene expression in buckwheat using qRT-PCR will improve the sensitivity and reproducibility of the results. In contrast, the choice of non-optimal reference gene can lead to inaccurate results.
First, this study demonstrates that the data derived from 454 sequencing of normalized cDNA libraries are useful as a source of candidate reference genes. We have evaluated the expression stability of 8 candidate reference genes in different organs of common buckwheat using three software applications – geNorm, NormFinder and BestKeeper. Despite small incongruences between different algorithms they all indicate on the stability of three buckwheat genes - the orthologs of AT4G33380, SAND and CACS. According to geNorm, the combination of Expressed1 and CACS is sufficient to provide accurate normalization however we suggest include also SAND in the set of reference genes for buckwheat. These results can be useful for the studies on gene expression in buckwheat using qRT-PCR. Interestingly, that these genes are listed among five the most stably expressed in Arabidopsis. This supports the applicability of the data inferred from the study of model plants to the non-models ones and encourages their use as a framework when performing the reference gene normalization experiments in other species.
Materials and Methods
Plant material and biological samples
Samples were collected from the buckwheat plants grown in a greenhouse at 20–25°C and relative humidity 80% under long day (16 hours light/8 hours dark) conditions. Five samples were taken: 1) fruits; 2) leaves with the developing leaf blade; 3) mature leaves; 4) developing inflorescences (5–8 mm in length) 5) inflorescences on the stage of the anthesis of first flower. All samples were collected in two replicates.
RNA extraction and cDNA synthesis
Total RNA was isolated from 805 mg of plant material using RNeasy Plant Kit (Qiagen, USA) according to the modified protocol , based on CTAB DNA extraction protocol  with further use of the RNeasy Plant Mini Kit (Qiagen, Stanford, California, USA). To eliminate genomic DNA from RNA samples they were treated twice with RNase-Free DNase (Qiagen, USA). First digestion was performed according manufacturer's instructions then columns were washed with 350 µl of RW1 and digestion was repeated. To evaluate RNA integrity RNA was visualized on 1% SYBR Green stained agarose gel. Clear bands corresponding to 18S and 28S rRNA and absence of smear were observed indicating minimal degradation of RNA. Concentration and purity of total RNA were determined with NanoDrop1000 spectrophotometer (Thermo Scientific, USA). Samples with concentration more than 100 ng/µl and optical density absorption ratio A260/A280 more than 1.8 were taken for cDNA synthesis. Individual RNA samples were stored at −70°C with the addition of RNAse inhibitor RNasin (Sileks, Russia) and then were adjusted to the concentration of 1005 ng/µl for reverse transcription. First strand cDNA was performed using “First strand cDNA synthesis kit” (Sileks, Russia) with 24T primer (0,4 nmol per reaction) in 25 µl reaction mix according manufacturer's protocol. Before each qPCR stage cDNA samples were heated (65°C–90 seconds 40°C–30 seconds) and then cDNA products were diluted 10-fold prior to use in real-time PCR.
Primer design and PCR conditions
Based on the transcriptome data sequences of eight potential reference genes were obtained. Buckwheat transcriptome database was searched with TBLASTX (e-value 10−6) using the sequences of Arabidopsis genes from genome-wide investigation of A.thaliana  as query. For all genes primer pairs were designed using PrimerQuest  with the following parameters: optimal length 25 nucleotides, melting temperature 60–65°C, product size range 79–261 base pairs, maximum self complementarity at 3′end – 3 nucleotides and then checked for the absence of stable hairpins and dimers using OligoAnalyser . To check primer specificity real-time PCR was performed on cDNA and products were analyzed on 3% SYBR Green staining agarose gel and by melting curves. Amplification efficiency was accessed using Miner 2.2 software , .
Real-time PCR reactions were run on ANK-32 thermocycler (Syntol, Russia) using 2.5× RT-PCR reaction mix (Syntol, Russia). To detect dsDNA synthesis EvaGreen dye was used since it is known to have lower inhibitory effect on PCR than commonly used SYBR Green . Each reaction was performed in 25 µl mix containing 400 nmol of each primer and 1 µl of 1∶10 diluted cDNA. The following amplification program was used: initial denaturation 95°C for 5 minutes, then 35 cycles of 95°C at 15 seconds and 62°C at 60 seconds. Each sample was analyzed in three technical replicates; mean Ct values were calculated. Mean Ct dispersal for technical replicates did not exceed 0,3 cycle. To ensure the absence of contamination or primer dimer formation for each primer pair non-template control reaction was run. Reverse transcription negative control was also included to ensure absence of genomic DNA in the template. These no-RT control reactions were run with primers to AT4G33380 ortholog as these primers anneal within one exon. To confirm the specificity of primer annealing the amplicons corresponding to each primer pair were sequenced. DNA sequencing was performed in the interinstitutional sequencing center at Engelhardt Institute of Molecular Biology (Moscow, Russia) using ABI PRISM BigDye Terminator kit v. 3.1 with following analysis on ABI PRISM 3730 genetic analyzer (Applied Biosystems, USA). Sequences of the amplicons were deposited in the GenBank (accession numbers are listed in the Results section).
Analysis of gene stability
Gene expression stability was estimated with three various approaches: geNorm (ver. 3.5) , NormFonder  and BestKeeper . Real-time PCR Ct values were transformed into Cq by standard formula: , where E is the efficiency of the amplification of each primer pair. Then for both programs Cq values were transformed to relative quantities using the delta-Cq formula , where ΔCq is the difference between the sample with lowest Cq and the Cq value of the sample of question.
The geNorm statistical algorithm is based on pairwise variation of a single reference candidate gene relative to all other genes were taken into investigation. The main assumption of this approach is that the expression ratio of two ideal reference genes is identical in all samples regardless of tissue type. In this case variation of the ratios of two candidate reference genes reflects instability in expression levels of one gene. Then reference expression stability measure (M value) is calculated with the average of pairwise variations. Vandesompele et al. recommend use only those genes, which M value does not exceed 1,5. Genes with lowest M value are the most stable and geNorm estimates the normalization factor (NFn) using geometric mean of expression levels of n best reference genes. Finally geNorm defines an optimal number of reference genes, because there are no reasons to use whole genes with low variability if NF doesn't significally change.
Another model-based approach – NormFinder – determines expression stability of candidate references according to their group origin (tissue, organ etc.) and compares expression variation between groups. With help of given group identifier the program divide data to panels, which may be differ due to experimental conditions: inflorescence, leaf, root etc. The main goal of the approach is to determine inter- and intra-group variation and combine both results in a stability value for each investigated gene. According to this algorithm genes with lowest stability will be top ranked.
Third program – BestKeeper – calculates standard deviation (SD) and the coefficient of variation (CV) based of Cq values of all reference candidate genes. Genes with SD less than 1 are considered to be stable. Then program calculates a pairwise correlation coefficient between each gene and the BestKeeper index – geometric mean between Cq values of stable genes grouped together. Genes with the highest coefficient of correlation with the BestKeeper Index indicates the highest stability.
Summary of the results of the studies reporting reference gene validation in plants. This table lists the species for which reference gene validation was reported, the candidate reference genes, the programs used, the gene(s) inferred as the most stable and the corresponding citation. The studies reporting validation of reference gene in some particular conditions (for example, under pathogen infection) are not included.
Conceived and designed the experiments: AAP MDL. Performed the experiments: NVD. Analyzed the data: NVD AAP. Contributed reagents/materials/analysis tools: AAP MDL. Wrote the paper: NVD MDL AAP. Performed all the experimental procedures, data analysis and drafted the manuscript: NVD. Participated in design of the study and revised the manuscript: MDL. Designed and supervised the study: AAP. Read and approved the final manuscript: NVD MDL AAP.
- 1. Vaucheret H, Vazquez F, Crété P, Bartel DP (2004) The action of ARGONAUTE1 in the miRNA pathway and its regulation by the miRNA pathway are crucial for plant development. Genes Dev 18(10): 1187–1197.
- 2. Koo SC, Bracko O, Park MS, Schwab R, Chun HJ, et al. (2010) Control of lateral organ development and flowering time by the Arabidopsis thaliana MADS-box gene AGAMOUS-LIKE6. The Plant Journal 62: 807–816.
- 3. Huang J, Gu M, Lai Z, Fan B, Shi K, et al. (2010) Functional analysis of the Arabidopsis PAL gene family in plant growth, development and response to environmental stress. Plant Physiology 153: 1526–1538.
- 4. Ren X, Chen Z, Liu Y, Zhang H, Zhang M, et al. (2010) ABO3, a WRKY transcription factor, mediates plant responses to abscisic acid and drought tolerance in Arabidopsis. The Plant Journal 63(3): 417–429.
- 5. Qiu JL, Fiil BK, Petersen K, Nielsen HB, Botanga CJ, et al. (2008) Arabidopsis MAP kinase 4 regulates gene expression through transcription factor release in the nucleus. EMBO J 27(16): 2214–2221.
- 6. Portieles R, Ayra C, Gonzalez E, Gallo A, Rodriguez R, et al. (2010) NmDef02, a novel antimicrobial gene isolated from Nicotiana megalosiphon confers high-level pathogen resistance under greenhouse and field conditions. Plant Biotechnol J 8(6): 678–690.
- 7. Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, et al. (2009) The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin Chem 55(4): 611–22.
- 8. Huggett J, Dheda K, Bustin S, Zumla A (2005) Real-time RT-PCR normalisation; strategies and considerations. Genes Immun 6(4): 279–284.
- 9. Czechowski T, Stitt M, Altmann T, Udvardi MK, Scheible WR (2005) Genome-wide identification and testing of superior reference genes for transcript normalization in Arabidopsis. Plant Physiol 139(1): 5–17.
- 10. Gutierrez L, Mauriat M, Guénin S, Pelloux J, Lefebvre JF, et al. (2008) The lack of a systematic validation of reference genes: a serious pitfall undervalued in reverse transcription-polymerase chain reaction (RT-PCR) analysis in plants. Plant Biotechnol J (6): 609–618.
- 11. Nikolić BD, Samardzić TJ, Bratić MA, Radin PI, Gavrilović PS, et al. (2010) Buckwheat (Fagopyrum esculentum Moench) FeMT3 gene in heavy metal stress: protective role of the protein and inducibility of the promoter region under Cu(2+) and Cd(2+) treatments. J Agric Food Chem 58(6): 3488–94.
- 12. Timotijevic GS, Milisavljevic MD, Radovic SR, Konstantinovic MM, Maksimovic VR (2010) Ubiquitous aspartic proteinase as an actor in the stress response in buckwheat. J Plant Physiol 167(1): 61–68.
- 13. Logacheva MD, Kasianov AS, Vinogradov DV, Samigullin TH, Gelfand MS, et al. (2011) De novo sequencing and characterization of floral transcriptome in two species of buckwheat (Fagopyrum). BMC Genomics 12(1): 30.
- 14. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, et al. (2002) Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol 3(7): research0034–research0034.11.
- 15. Andersen CL, Jensen JL, Ørntoft TF (2004) Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets. Cancer Res 64(15): 5245–5250.
- 16. Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP (2004) Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper-Excel-based tool using pair-wise correlations. Biotechnol Lett 26(6): 509–515.
- 17. Nolan T, Hands RE, Bustin SA (2006) Quantification of mRNA using real-time RT-PCR. Nat Protoc 1(3): 1559–1582.
- 18. geNorm Software [http://medgen.ugent.be/~jvdesomp/genorm/].
- 19. Hong SY, Seo PJ, Yang MS, Xiang F, Park CM (2008) Exploring valid reference genes for gene expression studies in Brachypodium distachyon by real-time PCR. BMC Plant Biol 8: 112.
- 20. Jarosová J, Kundu JK (2010) Validation of reference genes as internal control for studying viral infections in cereals by quantitative real-time RT-PCR. BMC Plant Biol 10: 146.
- 21. Mallona I, Lischewski S, Weiss J, Hause B, Egea-Cortines M (2010) Validation of reference genes for quantitative real-time PCR during leaf and flower development in Petunia hybrida. BMC Plant Biol 10: 4.
- 22. Wen R, Torres-Acosta JA, Pastushok L, Lai X, Pelzer L, et al. (2008) Arabidopsis UEV1D promotes Lysine-63-linked polyubiquitination and is involved in DNA damage response. Plant Cell 20(1): 213–27.
- 23. Walley JW, Kelley DR, Nestorova G, Hirschberg DL, Dehesh K (2010) Arabidopsis deadenylases AtCAF1a and AtCAF1b play overlapping and distinct roles in mediating environmental stress responses. Plant Physiol 152(2): 866–75.
- 24. Exposito-Rodriguez M, Borges A, Borges-Perez A, Perez J (2008) Selection of internal control genes for quantitative real-time RT-PCR studies during tomato development process. BMC Plant Biol 8(1): 131.
- 25. Chen X, Truksa M, Shah S, Weselake RJ (2010) A survey of quantitative real-time polymerase chain reaction internal reference genes for expression studies in Brassica napus. Anal Biochem 405(1): 138–40.
- 26. Lin YL, Lai ZX (2010) Reference gene selection for qPCR analysis during somatic embryogenesis in longan tree. Plant Science 178: 359–365.
- 27. Novaes E, Drost DR, Farmerie WG, Pappas GJ Jr, Grattapaglia D, et al. (2008) High-throughput gene and SNP discovery in Eucalyptus grandis, an uncharacterized genome. BMC Genomics 9: 312.
- 28. Desgagné-Penix I, Khan MF, Schriemer DC, Cram D, Nowak J, et al. (2010) Integration of deep transcriptome and proteome analyses reveals the components of alkaloid metabolism in opium poppy cell cultures. BMC Plant Biol 10: 252.
- 29. Riggins CW, Peng Y, Stewart CN Jr, Tranel PJ (2010) Characterization of de novo transcriptome for waterhemp (Amaranthus tuberculatus) using GS-FLX 454 pyrosequencing and its application for studies of herbicide target-site genes. Pest Manag Sci 66(10): 1042–1052.
- 30. de Almeida MR, Ruedell CM, Ricachenevsky FK, Sperotto RA, Pasquali G, et al. (2010) Reference gene selection for quantitative reverse transcription-polymerase chain reaction normalization during in vitro adventitious rooting in Eucalyptus globulus Labill. BMC Mol Biol 11: 73.
- 31. Sarosh BR, Danielsson J, Meijer J (2009) Transcript profiling of oilseed rape (Brassica napus) primed for biocontrol differentiate genes involved in microbial interactions with beneficial Bacillus amyloliquefaciens from pathogenic Botrytis cinerea. Plant Mol Biol 70(1–2): 31–45.
- 32. Kim S, Yoo MJ, Albert VA, Farris JS, Soltis PS, et al. (2004) Phylogeny and diversification of B-function MADS-box genes in angiosperms: evolutionary and functional implications of a 260-million-year-old duplication. Amer J Bot 91: 2102–2118.
- 33. Doyle JJ, Doyle JL (1987) A rapid DNA isolation from small amount of fresh leaf tissue. Phytochemical Bulletin 19: 11–15.
- 34. PrimerQuest [http://eu.idtdna.com/Scitools/Applications/Primerquest/default.aspx].
- 35. OligoAnalyzer [http://eu.idtdna.com/analyzer/Applications/OligoAnalyzer].
- 36. Real-time PCR Miner [http://www.miner.ewindup.info/version2].
- 37. Zhao S, Fernald RD (2005) Comprehensive algorithm for quantitative real-time polymerase chain reaction. J Comput Biol 12(8): 1047–1064.
- 38. Mao F, Leung WY, Xin X (2007) Characterization of EvaGreen and the implication of its physicochemical properties for qPCR applications. BMC Biotechnol 7: 76.