The Gram-negative bacterium Xanthomonas campestris pv. vesicatoria (Xcv) causes bacterial spot disease of pepper and tomato by direct translocation of type III effector proteins into the plant cell cytosol. Once in the plant cell the effectors interfere with host cell processes and manipulate the plant transcriptome. Quantitative RT-PCR (qRT-PCR) is usually the method of choice to analyze transcriptional changes of selected plant genes. Reliable results depend, however, on measuring stably expressed reference genes that serve as internal normalization controls. We identified the most stably expressed tomato genes based on microarray analyses of Xcv-infected tomato leaves and evaluated the reliability of 11 genes for qRT-PCR studies in comparison to four traditionally employed reference genes. Three different statistical algorithms, geNorm, NormFinder and BestKeeper, concordantly determined the superiority of the newly identified reference genes. The most suitable reference genes encode proteins with homology to PHD finger family proteins and the U6 snRNA-associated protein LSm7. In addition, we identified pepper orthologs and validated several genes as reliable normalization controls for qRT-PCR analysis of Xcv-infected pepper plants. The newly identified reference genes will be beneficial for future qRT-PCR studies of the Xcv-tomato and Xcv-pepper pathosystems, as well as for the identification of suitable normalization controls for qRT-PCR studies of other plant-pathogen interactions, especially, if related plant species are used in combination with bacterial pathogens.
Citation: Müller OA, Grau J, Thieme S, Prochaska H, Adlung N, Sorgatz A, et al. (2015) Genome-Wide Identification and Validation of Reference Genes in Infected Tomato Leaves for Quantitative RT-PCR Analyses. PLoS ONE 10(8): e0136499. https://doi.org/10.1371/journal.pone.0136499
Editor: Ya-Wen He, Shanghai Jiao Tong University, CHINA
Received: July 14, 2015; Accepted: August 4, 2015; Published: August 27, 2015
Copyright: © 2015 Müller 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
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This work was supported by a grant from the Deutsche Forschungsgemeinschaft (SFB 648) to UB. 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.
The analysis of gene transcription profiles is a powerful tool to uncover the roles of specific genes in cellular processes and to place them into regulatory networks. Quantitative reverse transcription PCR (qRT-PCR), also termed real-time RT-PCR, is the method of choice to analyze changes in gene transcription because of its high sensitivity, large dynamic range and accuracy . The reliability of results strongly depends on suitable reference genes for normalization which should be stably expressed under the experimental conditions used. Housekeeping genes encoding, e.g., actin, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and ribosomal RNAs, are generally assumed to represent suitable normalization controls . However, a number of studies reported that transcription of housekeeping genes can fluctuate considerably under certain experimental conditions, even if expression is constant in other cases ( and references therein). This illustrates the necessity to systematically validate reference genes for specific experimental conditions to avoid misinterpretation of qRT-PCR results [3, 4].
The interaction of plants with pathogens induces dramatic changes in plant transcription patterns. In most cases, the plant withstands pathogen attacks by inducing innate immune responses, associated with transcriptional reprogramming, e.g., the induction of pathogenesis-related (PR) genes [5–7]. Specialized pathogens, however, can suppress plant immunity and successfully colonize the host. Infection is accompanied by transcriptional changes of numerous plant genes including those involved in basal cell processes [7–12]. For example, in maize seeds infected by fungi genes involved in metabolism, energy and protein synthesis are prevalently down-regulated, including classical housekeeping genes like GAPDH . The bacterial pathogen Pseudomonas syringae pv. tomato represses cell wall and photosynthetic genes in Arabidopsis plants . Similar results were obtained in sweet orange and peach infected with Xanthomonas citri supsp. citri and X. arboricola pv. pruni, respectively [8, 11].
Recently, there were a number of reports validating reference genes in different plant species after infection with fungi, oomycetes, viruses or bacteria [13–31], or suffering from plant and animal parasites [32–36]. Among the genes most often found to be suitable normalization controls under biotic stress conditions were genes encoding actin [13, 23, 24, 30, 34, 35], glyceraldehyde 3-phosphate dehydrogenase (GAPDH) [15, 16, 27, 28, 30], β-tubulin [17, 25, 28, 32] and elongation factor 1α (EF-1α) [21, 34–36]. However, a major drawback of most studies is the selection of reference gene candidates based on “the usual suspects”, i.e., genes with known or suspected housekeeping roles. Such a biased approach might miss the optimal internal control. This idea is supported by whole-transcriptome analyses in different plant species and different experimental setups that, together with qRT-PCR studies, identified genes differing from the traditional housekeeping genes as most stably transcribed [37–43].
Our lab studies the interaction of the phytopathogenic γ-proteobacterium X. campestris pv. vesicatoria (Xcv) with its solanaceous hosts, tomato (Solanum lycopersicum) and pepper (Capsicum annuum). Xcv causes bacterial spot disease which results in defoliation and severely spotted fruits, both of which lead to massive yield losses, especially in regions with a warm and humid climate . An essential pathogenicity factor of Xcv is the type III secretion (T3S) system that translocates bacterial effector proteins into the plant cell cytosol. Although the molecular function of many Xcv type III effectors is unknown, several suppress host defenses elicited upon recognition of pathogen-associated molecular patterns (PAMPs), i.e., PAMP-triggered immunity (PTI) . A well-characterized effector family from Xanthomonas are TAL (transcription activator-like) effectors . The type member AvrBs3 from Xcv binds to plant gene promoters and activates the transcription of UPA (upregulated by AvrBs3) genes in pepper and other solanaceous plants resulting in hypertrophy, i.e., enlargement, of mesophyll cells [47, 48]. In resistant pepper plants, UPA genes include the Bs3 resistance gene leading to the specific elicitation of the hypersensitive response (HR), a rapid, localized programmed cell death at the infection site, that is a hallmark of effector-triggered immunity (ETI) .
Since we are interested in transcriptome changes during pathogen attack, we first analyzed the results of two genome-wide microarray screens of tomato cv. MoneyMaker (MM) to identify reference gene candidates suitable for qRT-PCR analysis of Xcv-infected (pathogenic and non-pathogenic strains) compared to unchallenged plants. Validation by qRT-PCR revealed 11 novel tomato reference genes. In addition, we identified the pepper orthologs of these genes and found several to be suitable normalization controls for qRT-PCR analyses in pepper during biotic stress.
Material and Methods
Plant material and inoculations
Tomato (Solanum lycopersicum) plants of cultivar (cv.) MoneyMaker and pepper (Capsicum annuum) cv. ECW-30R plants were grown in the greenhouse under standard conditions (day and night temperatures of 23°C and 19°C, respectively, for tomato, and 25°C and 19°C for pepper, with 16 h light and 40 to 60% humidity). For qRT-PCR studies, tomato and pepper plants were transferred to a Percival growth chamber (Percival Scientific, Perry, USA) three days before inoculation. Mature leaves of seven-week-old tomato and pepper plants were inoculated with mock (10 mM MgCl2) or Xcv (5×108 cfu/ml in 10 mM MgCl2) using a needleless syringe.
Bacterial strains and growth conditions
Xcv strains 85–10  and 85–10ΔhrcN  were grown at 30°C on NYG (nutrient yeast glycerol) agar plates  supplemented with appropriate antibiotics. Plasmids pLAT211 (avrBs4 in pLAFR6 ) and pGGX1:avrBs3  were introduced into Xcv by conjugation, using pRK2013 as helper plasmid in triparental matings .
For microarray studies, 12 tomato plants were inoculated per experiment. To minimize differences in gene expression due to leaf-to-leaf variability, Xcv strains and 10 mM MgCl2, respectively, were infiltrated into the same leaves. Four leaf discs (0.5 cm diameter) per inoculum and leaf were harvested, immediately frozen in liquid nitrogen and stored at -80°C. In the first study, Xcv 85–10 and 85–10ΔhrcN were inoculated; leaf material was harvested 45 min and 6, 10 and 24 hours post infiltration (hpi). Leaf material of four plants was pooled for each time-point (16 leaf discs per sample, three technical replicates). In the second study, 85–10ΔhrcN and 10 mM MgCl2 were infiltrated and leaf material was harvested at 0, 4, 8 and 16 hpi and pooled as above. In addition, four leaf discs per plant were harvested as control before treatment. This was performed three times independently with four plants each (biological replicates). The experimental setup is summarized in S1 Fig.
Total RNA was extracted using the QIAGEN RNeasy Plant Mini Kit (QIAGEN, Hilden, Germany) and treated with DNase I (Roche, Mannheim, Germany) for 30 min. Approximately 1.5 μg total RNA was sent to Source BioScience (Berlin, Germany) for cDNA synthesis and microarray hybridizations.
For the tomato whole-genome chip (Source BioScience), oligonucleotides for 34,383 annotated tomato genes [according to the international tomato annotation group (ITAG, version 2.3)] were spotted on Agilent custom arrays. Five 50-bp oligonucleotides per gene were tested on an Agilent custom array 4x180K, and a set of suitable oligos was chosen for the final chip. Due to space limitations (8x60K), 25,985 randomly chosen genes were represented twice with different oligonucleotides, whereas 8398 genes were represented by one oligonucleotide each. Finally, seven identical 8x60K chips were used for sample analysis. Different chips were hybridized with biological and technical replicates, respectively. cDNA synthesis, labelling, hybridization, washing, scanning and data collection was performed by Source BioScience according to Agilent standard protocols.
Data processing and statistical analyses
Microarray raw data (column "gProcessedSignal") were analyzed by the statistical software R . All experiments of one study (treatments, time points and replicates) were normalized by quantile normalization on the probe level using the "preprocessCore" R package (version 1.26.1, http://www.bioconductor.org/packages/release/bioc/html/preprocessCore.html). For each gene, values for transcript accumulation were obtained as the arithmetic mean of the intensities of all probes representing the gene. The coefficient of variation (CV) was computed for each gene as the standard deviation of its transcript levels across all experiments divided by its mean transcript level. To evaluate the similarity of expression patterns in biological and technical replicates, normalized log-expression values of the individual experiments were clustered hierarchically using the R function hclust . The distance between the expression vectors of experiments was determined as one minus the Pearson correlation of log-expression values using the R function cor.dist from the bioDist package of the Bioconductor suite . Clustering was performed using complete linkage, which yields compact clusters with high intra-cluster correlations. Dendrograms were plotted using the specific plot function of the R class hclust .
Quantitative reverse transcription polymerase chain reaction (qRT-PCR)
Templates for qRT-PCR were produced as follows: three to four leaf discs (1.3 cm diameter) from different plants infiltrated with Xcv and MgCl2, respectively, were pooled for RNA isolation using the QIAGEN RNeasy Plant Mini Kit. Oligo-dT- and random hexamer-primed cDNA was synthesized with the Maxima First Strand cDNA Synthesis Kit (Thermo Scientific, Schwerte, Germany). qRT-PCR was performed on a CFX96 thermal cycler (Bio-Rad, Munich, Germany) using a SYBR Green-based PCR reaction mixture (Absolute Blue qPCR SYBR Green Fluorescein Mix; Thermo Scientific) and 8 ng template cDNA. Oligonucleotide sequences are listed in S1 Table. To compare Ct (cycle threshold) values measured on different plates using different reaction mixtures, automatically calculated thresholds of all plates were set manually to the highest threshold obtained. The efficiency of PCR reactions was determined for each primer pair using a dilution series of template plotted into a standard curve. To ensure amplification specificity, amplicons were subjected to melting curve analysis and analyzed on 1% agarose gels. Transcript levels were determined as technical duplicates of biological triplicates.
Selection of candidate reference genes for gene expression studies in tomato
To identify reference genes suitable for the analysis of Xcv-induced changes in the mRNA levels of tomato genes we evaluated the results of two whole-genome microarray screens. For the first screen, S. lycopersicum cv. MM plants were inoculated with the Xcv wild-type (WT) strain 85–10 and the T3S-deficient derivative 85–10ΔhrcN, respectively. Leaf material was harvested 45 min and 6, 10 and 24 hours post infiltration (hpi). In the second screen, S. lycopersicum cv. MM plants challenged by 85–10ΔhrcN inoculation were compared to mock-infiltrated tomato plants, and leaf material was harvested at 0, 4, 8 and 16 hpi. Transcriptional changes of 34,383 annotated tomato genes were analyzed using “Agilent custom arrays“. Hierarchical cluster analysis illustrates similar expression patterns in biological and technical replicates confirming that the experimental treatments worked (S1 Fig). In the first screen, two samples (“85–10; 45 mpi; #2” and “85–10; 6 hpi; #2”) showed aberrant gene expression patterns compared to the corresponding replicates resulting in separate clustering (S1A Fig). Both samples were excluded from further data evaluation.
The microarray analyses revealed a high variability in the expression patterns of housekeeping genes conventionally used as references in transcript studies [30, 61–64] (Fig 1). To identify the most stably transcribed genes, the coefficient of variation (CV) was determined for each gene, which is defined as the standard deviation of its expression levels across all experiments (treatments and time-points) divided by its mean expression level. Genes with a log2 mean expression level below 7 or above 13 were excluded to account for the bigger influence of random noise on low expression values, and for saturation effects of microarrays at high mRNA levels, respectively. Genes with CV values ≤ 0.12 in both microarray studies were ranked by increasing CV in the second screen which delivers more reliable data compared to the first study (biological instead of technical replicates). The best 50 candidate reference genes are listed in Table 1. The tomato sequences were classified based on the functional categories of their A. thaliana orthologs which were identified by BLASTx  against “The Arabidopsis Information Resource” database (TAIR Blast 2.2.8; S2 Fig). Only predicted proteins that displayed minimum 40% amino-acid identity over at least 70% of the tomato sequence were taken into account. This allowed a functional classification of approximately three quarters of the sequences (74%), most of them possessing putative functions in protein expression (transcription and splicing) and turnover (ubiquitination/proteolysis; S2 Fig).
Leaves of S. lycopersicum cv. MM plants were untreated or infiltrated with 10 mM MgCl2 (mock) and 85–10ΔhrcN, respectively. To analyze mRNA accumulation of selected genes leaf material was harvested at 0, 4, 8 and 16 hours post infiltration (hpi). Relative RNA levels of housekeeping genes traditionally employed as references [30, 61–64] were determined by microarray hybridization analysis. Expression values of analyzed plant samples are plotted separately, i.e., three biological replicates per time-point per infiltration. For each gene, expression values were obtained as the mean of the intensities of all probes representing the respective gene. Blue curves represent housekeeping genes that were also analyzed in this study together with novel reference genes. In addition, the most stably expressed gene in the microarray experiment is shown (red curve). ACT (actin), TC194780a; EF-1α (elongation factor 1α), SGN-U212845; GAPDH (glyceraldehyde 3-phosphate dehydrogenase), TC198136a; UBI (ubiquitin), TC193502a; TBP (TATA binding protein), SGN-U329249; RPL8 (ribosomal protein L8), X64562; TUA (α-tubulin), AC122540; CYP (cyclophilin), AK326854; TAF6 (TFIID subunit 6), Solyc10g006100.2.1.
Evaluation of the expression stability of novel and traditional tomato reference genes
qRT-PCR analyses of the 11 most stably expressed genes (Table 1) were performed to validate their expression stability in S. lycopersicum cv. MM infected with Xcv. The genes encode a TFIID subunit (TAF6), importin β (IMP-β), a PHD finger family protein (PHD), a cytochrome c oxidase subunit (COX), polyribonucleotide 5´-hydroxyl-kinase Clp1 (CLP1), a ubiquitin carboxyl-terminal hydrolase family protein (UCH), a polypyrimidine tract-binding protein-like protein (PTBL), U6 snRNA-associated Sm-like protein LSm7 (LSM7) and an acyl carrier protein (ACP), as well as two unknown proteins (UP1 and UP2; Table 1, S2 Table). For comparison, four housekeeping genes were analyzed that are widely used as references, namely actin (ACT), EF-1α, GAPDH and ubiquitin (UBI). First, suitability of oligonucleotides (S1 Table) and target sequences was confirmed. Melting curves and gel electrophoresis revealed unique amplicons for all oligonucleotide combinations used validating their specificity (S3 Fig). PCR efficiencies ranged between 80.48 and 99.71% (S1 Table). For expression analysis of the reference gene candidates, total RNA was isolated from tomato leaves 0, 6, 10 and 24 h after treatment with 10 mM MgCl2, Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs4), respectively. The latter strain induces the ETI, i.e., the HR in S. lycopersicum cv. MM due to the Bs4-dependent recognition of the avirulence protein AvrBs4, a member of the TAL effector family . Technical duplicates of three biological replicates were subjected to qRT-PCR analysis. Average Ct (cycle threshold) values of the new reference gene candidates ranged from 27.1 (CLP1) to 31.1 (UP1; Fig 2). To select the optimal reference genes, we used three different algorithms to evaluate our qRT-PCR results: geNorm , NormFinder  and BestKeeper .
Box plot graphs of Ct values for each reference gene tested in all samples (n = 48). Ct values are inversely proportional to the amount of template. Boxes indicate the 25/75 percentiles, median values are represented by black lines. Whisker caps indicate the value range, dots represent outliers. New reference gene candidates are indicated in bold.
The geNorm software provides a ranking of the tested genes based on a stability value M which is calculated by average pairwise variation of each candidate gene combination . The lower the M value, the higher the expression stability of the gene. Eventually, the algorithm selects an optimal pair of reference genes out of the candidate set analyzed. Considering a cutoff of M ≤ 0.5, the traditional references GAPDH, ACT and UBI proved unreliable for the normalization of qRT-PCR data under the experimental conditions chosen (Fig 3). By contrast, all newly identified tomato candidate genes and EF-1α represent suitable references, with IMP-β and PHD being optimal (Fig 3).
Tomato reference genes were ranked based on expression stability calculated by geNorm. New reference gene candidates are indicated in bold. M values represent the average expression stability of each gene (n = 48). The cut-off value for reliable reference genes is indicated by a dashed line.
Next, we analyzed the qRT-PCR data using NormFinder . The stability value M calculated by this “model-based variance estimation approach” considers not only the “overall expression variation” measured in different samples, but additionally takes into account variations among and inside sample subgroups . Thus, the algorithm avoids co-regulated reference genes which display systematic intergroup variation and would lead to erroneous conclusions. Since we are interested in changes of plant gene expression levels induced by different Xanthomonas strains but also in expression level changes over a certain time period, two separate NormFinder analyses were performed with sample subgroups defined based on treatment [MgCl2, Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs4)] and time-point of sampling (0, 6, 10 and 24 hpi), respectively. As shown in Fig 4, all tested genes fulfill the minimal requirement for suitable reference genes, i.e., possess an M value below 1.5. However, the traditionally employed reference genes ACT and GAPDH were considerably less stable than the other genes, whereas EF-1α and UBI seemed more suitable under the chosen experimental conditions. The top-ranked references, however, were among the newly identified candidate genes, namely COX > PHD > CLP1 > LSM7 with respect to the grouping by treatment (Fig 4A).
Tomato reference genes were ranked based on expression stability calculated by NormFinder (n = 48). New reference gene candidates are indicated in bold. The cut-off value for reliable reference genes is indicated by a dashed line. Sample groups were defined based on (a) treatment [MgCl2, Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs4)] or (b) time-point of harvesting (0, 6, 10 and 24 hpi).
We compared the six most stable new reference genes according to NormFinder with the four classical reference genes using BestKeeper . This tool evaluates the suitability of up to 10 reference genes based on the calculation of Ct value variations, performing pair-wise correlations of all candidate gene combinations. Extreme samples (x-fold over-/under-expression) are also considered. As shown in Table 2, expression of all genes except for ACT and GAPDH fluctuated in a range compatible with standard deviations (SD) [± Ct] < 1 and SD [± x-fold] < 2, which represents an acceptable overall variation . Notably, BestKeeper evaluated all six new reference gene candidates as better suited than the four traditional housekeeping genes, with PHD > CLP1 > LSM7 > COX being the top four. Taken together, regardless of the ranking order, geNorm, NormFinder and BestKeeper evidenced the superior expression stability of the new tomato reference genes under the experimental conditions chosen.
Quantification of immunity marker genes in infected tomato leaves
We applied our findings to the analysis of two target genes previously reported to be induced during PTI and ETI, respectively, LRR22  and an UDP-glucosyltransferase gene (UGT, Solyc09g092500 ). For this, total RNA was analyzed from tomato leaves six hours after treatment with 10 mM MgCl2, Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs4), respectively. To increase the accuracy of normalization we took into account two reference genes. We compared the two best reference genes identified by geNorm (IMP-β and PHD), NormFinder (PHD and COX) and BestKeeper (CLP1 and PHD) with the two least-stable genes, GAPDH and ACT, for their ability to provide reliable relative quantification of SlLRR22 and SlUGT by qRT-PCR. As shown in Fig 5, accumulation of SlLRR22 transcript was approximately two-fold higher in the leaves treated with 85–10ΔhrcN than in the mock control if compared to any of the new reference gene combinations. By contrast, comparison to the suboptimal references revealed an apparent five-fold induction of gene expression. In addition, referring to ACT and GAPDH suggested a more than two-fold upregulation of SlLRR22 by the Xcv WT strain 85–10 and by 85-10(pavrBs4), the latter induction being significant, which was not detectable with any of the superior reference genes. Notably, standard deviations between the different biological datasets were substantially lower if one of the new reference gene combinations was employed. The analysis of the ETI marker gene, SlUGT, did not show pronounced differences in the expression pattern depending on the reference genes chosen. In all cases, transcript abundance was significantly higher in the leaves treated with the avirulent strain 85-10(pavrBs4) than in the mock-infiltrated leaves. However, a slight induction of SlUGT expression by both Xcv 85–10 and 85–10ΔhrcN was only detected when the traditional references were employed. A possible explanation for these results is downregulation of ACT and/or GAPDH by Xcv infection. To test this possibility, the expression of both genes was analyzed using the newly identified reference genes as normalization controls. As shown in S4 Fig, GAPDH transcript levels were indeed significantly lower in the leaf material inoculated with bacteria compared to the mock control, whereas ACT appeared not to be changed under these conditions.
Expression patterns of SlLRR22 (a) and SlUGT (b) in S. lycopersicum cv. MM leaves treated with 10 mM MgCl2 (mock) or 5×108 cfu/ml of Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs4), respectively, 6 hpi. qRT-PCR data were normalized with different reference gene pairs. Values are mean-fold changes in mRNA levels in Xcv-infected relative to mock-inoculated leaves for three biological replicates. Error bars indicate standard deviation (SD). Letters denote statistically significant differences (Student´s t-test, P < 0.05).
Selection and validation of pepper reference genes based on tomato orthologs
Based on the tomato microarray data, pepper orthologs of the eleven most stably expressed genes (Table 1) were identified by BLASTx against the European Nucleotide Archive (http://www.ebi.ac.uk/ena). Oligonucleotides for qRT-PCR were derived (S1 Table), and melting curve analysis and gel electrophoresis confirmed specific products for nine candidate genes (S5 Fig). PCR efficiencies ranged between 72.09 and 99.32% (S1 Table). For expression analysis, pepper ECW-30R (Bs3) leaves were infiltrated with 10 mM MgCl2, Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs3), respectively, and leaf material was harvested at 0, 6, 10 and 24 hpi. 85-10(pavrBs3) translocates the effector AvrBs3 which induces the HR in Bs3 pepper plants. Technical duplicates of three biological replicates were subjected to qRT-PCR analysis. Average Ct values of the new reference gene candidates ranged from 27.4 (UCH) to 38.8 (TAF6; Fig 6). For comparison, the four classical reference genes EF-1α, GAPDH, ACT and β-tubulin (TUB) were also analyzed.
Box plot graphs of Ct values for each reference gene tested in all samples (n = 48). Ct values are inversely proportional to the amount of template. The boxes indicate the 25/75 percentiles, median values are represented by black lines. Whisker caps indicate the value range, dots represent outliers. New reference gene candidates are indicated in bold.
The data were evaluated similarly to the analysis of the tomato reference genes described above. GeNorm analysis revealed that only three genes, UCH, LSM7 and PHD, match the cutoff-value for a reliable reference gene (M ≤ 0.5). In general, the pepper orthologs of the newly identified tomato reference genes were more stably expressed than the traditional pepper references (Fig 7).
Ranking of C. annuum reference genes based on expression stability calculated by geNorm. New reference gene candidates are indicated in bold. M values represent the average expression stability of each gene (n = 48). The cut-off value for reliable reference genes is indicated by a dashed line.
Using NormFinder, the classical reference gene EF-1α matched the requirements of a suitable reference gene (M < 1.5) when the sample subgroups were defined by treatment (Fig 8A), but turned out to be completely unreliable when the classification was based on the time-point of sampling (Fig 8B). GAPDH and TUB matched the minimal requirements of a reliable reference gene but were considerably less stable than the other genes tested, while ACT appeared more suitable. Notably, all newly identified reference genes were evaluated as reliable normalization controls with UCH > PHD > UP2 > LSM7 as the top-four when the grouping was based on treatment (Fig 8A).
Ranking of C. annuum reference genes based on expression stability calculated by NormFinder (n = 48). New reference gene candidates are indicated in bold. The cut-off value for reliable reference genes is indicated by a dashed line. Sample groups were defined based on (a) treatment [MgCl2, Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs3)] and (b) time-points of harvesting (0, 6, 10 and 24 hpi).
BestKeeper analysis of the six best new pepper reference genes according to NormFinder and the four classical references surprisingly revealed that only one gene, UCH, fulfilled both requirements for a suitable normalization control in qRT-PCR studies, i.e., SD [± Ct] < 1 and SD [± x-fold] < 2 (Table 3). Most of the other genes matched at least the threshold for SD [± x-fold], whereas EF-1α appeared to be completely unreliable as reference gene (Table 3).
Quantification of PTI and ETI marker genes in infected pepper leaves
We compared the two best reference genes from pepper identified by geNorm, i.e., UCH and LSM7 with the best reference genes according to BestKeeper, UCH and ACT, and the traditional reference genes EF-1α and GAPDH for their ability to provide reliable relative quantification of the target genes LRR22 and TFT4, which are induced during PTI and ETI, respectively [67, 69]. As shown in Fig 9, employment of different reference genes did not result in substantial differences in the expression patterns of CaLRR22 and CaTFT4. The T3S-deficient Xcv strain 85–10ΔhrcN led to significantly higher expression of CaLRR22 compared to the WT strain 85–10 and Xcv 85-10(pavrBs3). CaTFT4 was induced significantly during the incompatible interaction with Xcv 85-10(pavrBs3), similarly to the reported induction after recognition of the type III effector AvrBs2 . However, the observed differences in target gene expression levels were only judged as significant when the newly identified reference genes were used, but not with the traditional combination EF-1α/GAPDH. Utilization of the newly identified normalization controls resulted in significantly lower standard deviations underlining the higher reproducibility of the results in different experiments (Fig 9).
Expression patterns of CaLRR22 (a) and CaTFT4 (b) in C. annuum ECW-30R leaves treated with 10 mM MgCl2 (mock) or 5×108 cfu/ml of Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs3), respectively, six hpi. qRT-PCR data were normalized with different reference gene pairs. Values are mean fold changes in mRNA levels in Xcv-infected relative to mock-inoculated leaves for three biological replicates. Error bars indicate SD. Letters denote statistically significant differences (Student´s t-test, P < 0.05).
Correct normalization of gene transcripts depends on the choice of suitable reference genes. This is essential for reliable analyses of gene expression by qRT-PCR and has to be established for specific experimental conditions . Based on microarray expression analyses of >34,000 genes, we identified and validated 11 novel tomato reference genes with superior expression stability under biotic stress conditions, i.e., challenge by the bacterial pathogen Xcv. Although the new reference genes do not comprise “classical” housekeeping genes, homologies on the protein level indicate putative roles in basic cell functions, e.g., oxidation/reduction processes (COX), mRNA processing (LSM7, CLP1, PTBL), regulation of transcription/chromatin dynamics (PHD), nuclear import (IMP-β) and fatty acid biosynthesis (ACP; S2 Table). The three statistical programs we used for the evaluation of gene expression stability, geNorm, NormFinder and BestKeeper, slightly differed in the ranking of the reference gene candidates, which was also observed in previous studies and is probably due to different algorithms underlying the programs [23, 30, 31, 70]. Importantly, the newly identified genes were usually evaluated as more stable than the traditional housekeeping genes we analyzed for comparison and, notably, always included the optimal normalization control identified by the respective program. Based on our results, we recommend the use of PHD and LSM7 as reference genes for normalization in future plant gene expression studies in the Xcv-tomato pathosystem.
To the best of our knowledge, previous studies of pepper and tomato comparing reference gene stabilities selected candidates solely based on homology. It was shown that different genes, often housekeeping genes, are preferable under different conditions [23, 30, 38, 62, 71–73]. Notably, our microarray data revealed that the expression of classical tomato housekeeping genes varied considerably, confirmed by qRT-PCR studies of selected genes. In particular, GAPDH and ACT were attested a variability too high for a reliable reference gene by geNorm and BestKeeper, respectively. Therefore, we do not recommend the further employment of these genes as normalization controls in qRT-PCR analysis of tomato genes after pathogen infection, especially because we clearly showed an Xcv-dependent downregulation of GAPDH expression. Taken together, our results demonstrate the advantage of an unbiased, whole transcriptome-based approach to identify suitable reference genes. Concordantly, several whole-transcriptome analyses of different plant species and experimental setups identified other than traditional housekeeping genes as the most stably expressed genes [37–43].
It is, however, not feasible to perform microarray analyses for reference gene identification every time the experimental setup is changed. Therefore, one has to resort also to the homology-based selection of candidate genes. The identification of suitable candidates can be strongly improved by using orthologs of genes that were experimentally verified as appropriate references in related organisms under similar experimental or developmental conditions [38, 74–76]. We used such an approach to identify the pepper orthologs of our new superior tomato reference genes and determined UCH and PHD as the most suitable references for normalization of plant gene expression in the Xcv-pepper pathosystem. Interestingly, one of the traditional reference genes, ACT, also turned out to be stably expressed in our experimental setup. This contradicts the results of Wan et al. who described ACT as relatively unstable under different abiotic stresses and hormonal treatments . On the other hand, EF-1α turned out to be the most unstable pepper gene in our analyses although it was published as one of the least-variably expressed genes under abiotic stress conditions and hormone treatment . This underpins the observation that a chosen gene can be stable under certain conditions but highly variable under others . It should be noted that differences between the pepper lines used in the different studies might also play a role.
Although our selection of pepper orthologs of the new tomato reference genes surely represents an improvement compared to the selection of genes based on their known or suspected housekeeping roles, the ranking of our tomato reference genes and their pepper equivalents illustrates that the expression of gene orthologs can distinctly differ even between related plant species. In general, the M values calculated by NormFinder were lower for the tomato genes compared with their pepper orthologs. This difference appeared even more pronounced using geNorm which judged only three of the pepper genes tested as reliable reference genes. Similarly, using Bestkeeper, only one pepper gene, UCH, matched both requirements for a suitable reference gene. Therefore, we would like to emphasize that, even if our new pepper reference genes proved to be superior to most of the classical normalization controls we analyzed, a whole-transcriptome analysis of Xcv-challenged pepper plants might uncover even more suitable reference genes.
Taken together, the newly discovered tomato reference genes proved to be superior normalization controls for qRT-PCR studies of Xcv-infected tomato plants. In addition, they led to successful identification of the pepper orthologs as reliable reference genes in qRT-PCR analyses of the Xcv-pepper pathosystem. Similarly, these genes might be useful for the identification of suitable qRT-PCR normalization controls in other plant species for the analysis of plant gene expression during pathogen infection.
S1 Fig. Experimental setup and data cluster analysis of the tomato microarray screens.
(a) First microarray experiment. 12 plants were inoculated with Xcv strains 85–10 and 85–10ΔhrcN, four leaves per plant. Leaf material was harvested 45 min post infiltration (mpi) and 6, 10 and 24 hpi and pooled (four plants each). RNA was isolated, and the cDNAs used for microarray hybridizations. (b) Second microarray experiment. Three separate infiltrations of four plants each were performed with 10 mM MgCl2 (mock) and Xcv 85–10ΔhrcN. Leaf material was harvested 0, 4, 8 and 16 hpi and analyzed as described in (a). Dendrograms on the right show hierarchical cluster analysis of the respective microarray dataset (normalized log-expression values).
S2 Fig. Functional classification of the 50 most stable reference genes in Xcv-infected versus uninfected tomato plants.
Functional categories of the 50 most stably expressed tomato genes according to microarray hybridization data, based on Gene Ontology (GO) terms of the respective A. thaliana orthologs.
S3 Fig. Validation of oligonucleotide pairs of new tomato reference gene candidates for qRT-PCR analysis.
Presence of unique amplicons as a measure of PCR amplification specificity was determined (a) by electrophoresis on 1% agarose gel and (b) by melting curve analysis.
S4 Fig. Relative expression of ACT and GAPDH in Xcv-infected and mock-treated tomato plants.
Expression patterns of (a) SlACT and (b) SlGAPDH in S. lycopersicum cv. MM leaves 6 hpi of 10 mM MgCl2 (mock) or 5×108 cfu/ml of Xcv 85–10, 85–10ΔhrcN and 85-10(pavrBs4), respectively. qRT-PCR data were normalized with different reference gene pairs. Values are mean-fold changes in mRNA levels in Xcv-infected relative to mock-inoculated leaves for three biological replicates. Error bars indicate SD. Letters denote statistically significant differences (Student´s t-test, P < 0.05).
S5 Fig. Validation of oligonucleotide pairs of new pepper reference gene candidates for qRT-PCR analysis.
Presence of unique amplicons as a measure of PCR amplification specificity was determined (a) by electrophoresis on a 1% agarose gel and (b) by melting curve analysis.
S1 Table. Oligonucleotide sequences used for qRT-PCR analyses.
Conceived and designed the experiments: OAM ST HP UB. Performed the experiments: OAM HP NA AS. Analyzed the data: OAM JG ST UB. Contributed reagents/materials/analysis tools: UB. Wrote the paper: ST OAM JG UB.
- 1. Huggett J, Dheda K, Bustin S, Zumla A. Real-time RT-PCR normalisation; strategies and considerations. Genes Immun. 2005;6(4):279–84. pmid:15815687
- 2. Stürzenbaum S, Kille P. Control genes in quantitative molecular biological techniques: the variability of invariance. Comp Biochem Physiol B Biochem Mol Biol. 2001;130(3):281–9. pmid:11567890
- 3. Gutierrez L, Mauriat M, Pelloux J, Bellini C, Van Wuytswinkel O. Towards a systematic validation of references in real-time RT-PCR. Plant Cell. 2008;20(7):1734–5. pmid:18664615
- 4. Guenin S, Mauriat M, Pelloux J, Wuytswinkel OV, Bellini C, Gutierrez L. Normalization of qRT-PCR data: the necessity of adopting a systematic, experimental conditions-specific, validation of references. J Exp Bot. 2009;60(2):487–93. pmid:19264760
- 5. Eulgem T. Regulation of the Arabidopsis defense transcriptome. Trends Plant Sci. 2005;10(2):71–8. pmid:15708344
- 6. van Loon L, Rep M, Pieterse C. Significance of inducible defense-related proteins in infected plants. Annu Rev Phytopathol. 2006;44:135–62. pmid:16602946
- 7. Soria-Guerra R, Rosales-Mendoza S, Chang S, Haudenshield J, Padmanaban A, Rodriguez-Zas S, et al. Transcriptome analysis of resistant and susceptible genotypes of Glycine tomentella during Phakopsora pachyrhizi infection reveals novel rust resistance genes. Theor Appl Genet. 2010;120(7):1315–33. pmid:20058146
- 8. Fu X-Z, Gong X-Q, Zhang Y-X, Wang Y, Liu J-H. Different transcriptional response to Xanthomonas citri subsp. citri between kumquat and sweet orange with contrasting canker tolerance. PLoS ONE. 2012;7(7):e41790. pmid:22848606
- 9. Lanubile A, Pasini L, Marocco A. Differential gene expression in kernels and silks of maize lines with contrasting levels of ear rot resistance after Fusarium verticillioides infection. Plant Physiol. 2010;167(16):1398–406.
- 10. Grewal R, Gupta S, Das S. Xanthomonas oryzae pv. oryzae triggers immediate transcriptomic modulations in rice. BMC Genomics. 2012;13:49. pmid:22289642
- 11. Socquet-Juglard D, Kamber T, Pothier J, Christen D, Gessler C, Duffy B, et al. Comparative RNA-seq analysis of early-infected peach leaves by the invasive phytopathogen Xanthomonas arboricola pv. pruni. PLoS ONE. 2013;8(1):e54196. pmid:23342103
- 12. Thilmony R, Underwood W, He SY. Genome-wide transcriptional analysis of the Arabidopsis thaliana interaction with the plant pathogen Pseudomonas syringae pv. tomato DC3000 and the human pathogen Escherichia coli O157:H7. Plant J. 2006;46(1):34–53. pmid:16553894
- 13. Nicot N, Hausman J-F, Hoffmann L, Evers D. Housekeeping gene selection for real-time RT-PCR normalization in potato during biotic and abiotic stress. J Exp Bot. 2005;56(421):2907–14. pmid:16188960
- 14. Qi J, Yu S, Zhang F, Shen X, Zhao X, Yu Y, et al. Reference gene selection for real-time quantitative polymerase chain reaction of mRNA transcript levels in Chinese cabbage (Brassica rapa L. ssp. pekinensis). Plant Mol Biol Rep. 2010;28:597–604.
- 15. Chen L, Zhong H, Kuang J, Li J, Lu W, Chen J. Validation of reference genes for RT-qPCR studies of gene expression in banana fruit under different experimental conditions. Planta. 2011;234(2):377–90. pmid:21505864
- 16. Selim M, Legay S, Berkelmann-Löhnertz B, Langen G, Kogel KH, Evers D. Identification of suitable reference genes for real-time RT-PCR normalization in the grapevine-downy mildew pathosystem. Plant Cell Rep. 2012;31(1):205–16. pmid:22006104
- 17. Wei L, Miao H, Zhao R, Han X, Zhang T, Zhang H. Identification and testing of reference genes for Sesame gene expression analysis by quantitative real-time PCR. Planta. 2013;237(3):873–89. pmid:23229061
- 18. Monteiro F, Sebastiana M, Pais M, Figueiredo A. Reference gene selection and validation for the early responses to downy mildew infection in susceptible and resistant Vitis vinifera cultivars. PLoS ONE. 2013;8(9):e72998. pmid:24023800
- 19. Zhu X, Li X, Chen W, Chen J, Lu W, Chen L, et al. Evaluation of new reference genes in papaya for accurate transcript normalization under different experimental conditions. PLoS ONE. 2012;7(8):e44405. pmid:22952972
- 20. Klie M, Debener T. Identification of superior reference genes for data normalisation of expression studies via quantitative PCR in hybrid roses (Rosa hybrida). BMC Res Notes. 2011;4:518. pmid:22123042
- 21. Pinheiro T, Litholdo C, Sereno M, Leal G, Albuquerque P, Figueira A. Establishing references for gene expression analyses by RT-qPCR in Theobroma cacao tissues. Genet Mol Res. 2011;10(4):3291–305. pmid:22095481
- 22. Štajner N, Cregeen S, Javornik B. Evaluation of reference genes for RT-qPCR expression studies in hop (Humulus lupulus L.) during infection with vascular pathogen Verticillium albo-atrum. PLoS ONE. 2013;8(7):e68228. pmid:23874551
- 23. Wieczorek P, Wrzesińska B, Obrępalska-Stęplowska A. Assessment of reference gene stability influenced by extremely divergent disease symptoms in Solanum lycopersicum L. J Virol Methods. 2013;194(1–2):161–8. pmid:23994079
- 24. Saha G, Vandemark G. Stability of expression of reference genes among different lentil (Lens culinaris) genotypes subjected to cold stress, white mold disease, and Aphanomyces root rot. Plant Mol Biol Rep. 2013;31(5):1109–15.
- 25. Saha G, Vandemark G. Evaluation of expression stability of candidate references genes among green and yellow pea cultivars (Pisum sativum L.) subjected to abiotic and biotic stress. American J Plant Sci. 2012;3(2):235–42.
- 26. Kong Q, Yuan J, Niu P, Xie J, Jiang W, Huang Y, et al. Screening suitable reference genes for normalization in reverse transcription quantitative real-time PCR analysis in melon. PLoS ONE. 2014;9(1):e87197. pmid:24475250
- 27. Mafra V, Kubo K, Alves-Ferreira M, Ribeiro-Alves M, Stuart R, Boava L, et al. Reference genes for accurate transcript normalization in citrus genotypes under different experimental conditions. PLoS ONE. 2012;7(2):e31263. pmid:22347455
- 28. Jarosová J, Kundu J. Validation of reference genes as internal control for studying viral infections in cereals by quantitative real-time RT-PCR. BMC Plant Biol. 2010;10:146. pmid:20630112
- 29. Lilly S, Drummond R, Pearson M, MacDiarmid R. Identification and validation of reference genes for normalization of transcripts from virus-infected Arabidopsis thaliana. Mol Plant-Microbe Interact. 2011;24(3):294–304. pmid:21091160
- 30. Mascia T, Santovito E, Gallitelli D, Cillo F. Evaluation of reference genes for quantitative reverse-transcription polymerase chain reaction normalization in infected tomato plants. Mol Plant Pathol. 2010;11(6):805–16. pmid:21029324
- 31. Liu D, Shi L, Han C, Yu J, Li D, Zhang Y. Validation of reference genes for gene expression studies in virus-infected Nicotiana benthamiana using quantitative real-time PCR. PLoS ONE. 2012;7(9):e46451. pmid:23029521
- 32. Die J, Román B, Nadal S, González-Verdejo C. Evaluation of candidate reference genes for expression studies in Pisum sativum under different experimental conditions. Planta. 2010;232(1):145–53. pmid:20379832
- 33. Castro-Quezada P, Aarrouf J, Claverie M, Favery B, Mugniéry D, Lefebvre V, et al. Identification of reference genes for normalizing RNA expression in potato roots infected with cyst nematodes. Plant Mol Biol Rep. 2013;31(4):936–45.
- 34. Miranda V, Coelho R, Viana A, de Oliveira Neto O, Carneiro R, Rocha T, et al. Validation of reference genes aiming accurate normalization of qPCR data in soybean upon nematode parasitism and insect attack. BMC Res Notes. 2013;6:196. pmid:23668315
- 35. Kundu A, Patel A, Pal A. Defining reference genes for qPCR normalization to study biotic and abiotic stress responses in Vigna mungo. Plant Cell Rep. 2013;32(10):1647–58. pmid:23868569
- 36. Gu C, Chen S, Liu Z, Shan H, Luo H, Guan Z, et al. Reference gene selection for quantitative real-time PCR in Chrysanthemum subjected to biotic and abiotic stress. Mol Biotechnol. 2011;49(2):192–7. pmid:21416201
- 37. Czechowski T, Stitt M, Altmann T, Udvardi M, Scheible W-R. Genome-wide identification and testing of superior reference genes for transcript normalization in Arabidopsis. Plant Physiol. 2005;139(1):5–17. pmid:16166256
- 38. Dekkers B, Willems L, Bassel G, van Bolderen-Veldkamp R, Ligterink W, Hilhorst H, et al. Identification of reference genes for RT-qPCR expression analysis in Arabidopsis and tomato seeds. Plant Cell Physiol. 2012;53(1):28–37. pmid:21852359
- 39. de Oliveira L, Breton M, Bastolla F, Camargo S, Margis R, Frazzon J, et al. Reference genes for the normalization of gene expression in Eucalyptus species. Plant Cell Physiol. 2012;53(2):405–22. pmid:22197885
- 40. Demidenko NV, Logacheva MD, Penin AA. Selection and validation of reference genes for quantitative real-time PCR in buckwheat (Fagopyrum esculentum) based on transcriptome sequence data. PLoS ONE. 2011;6(5):e19434. pmid:21589908
- 41. Narsai R, Ivanova A, Ng S, Whelan J. Defining reference genes in Oryza sativa using organ, development, biotic and abiotic transcriptome datasets. BMC Plant Biol. 2010;10:56. pmid:20353606
- 42. Gamm M, Héloir M-C, Kelloniemi J, Poinssot B, Wendehenne D, Adrian M. Identification of reference genes suitable for qRT-PCR in grapevine and application for the study of the expression of genes involved in pterostilbene synthesis. mol genet genomics. 2011;285(4):273–85. pmid:21340517
- 43. Lin F, Jiang L, Liu Y, Lv Y, Dai H, Zhao H. Genome-wide identification of housekeeping genes in maize. Plant Mol Biol. 2014;86(4–5):543–54. pmid:25209110
- 44. Stall RE. Xanthomonas campestris pv. vesicatoria. In: Singh RPS U.S., and Kohmoto K., editor. Pathogenesis and Host-Parasite Specificity in Plant Diseases. I, Prokaryotes. Tarrytown, NY: Pergamon, Elsevier Science Inc.; 1995. p. 167–84.
- 45. Büttner D, Bonas U. Regulation and secretion of Xanthomonas virulence factors. FEMS Microbiol Rev. 2010;34(2):107–33. pmid:19925633
- 46. Boch J, Bonas U. Xanthomonas AvrBs3 family-type III effectors: discovery and function. Annu Rev Phytopathol. 2010;48:419–36. pmid:19400638
- 47. Kay S, Hahn S, Marois E, Hause G, Bonas U. A bacterial effector acts as a plant transcription factor and induces a cell size regulator. Science. 2007;318:648–51. pmid:17962565
- 48. Marois E, Van den Ackerveken G, Bonas U. The Xanthomonas type III effector protein AvrBs3 modulates plant gene expression and induces cell hypertrophy in the susceptible host. Mol Plant-Microbe Interact. 2002;15(7):637–46. pmid:12118879
- 49. Römer P, Hahn S, Jordan T, Strauß T, Bonas U, Lahaye T. Plant-pathogen recognition mediated by promoter activation of the pepper Bs3 resistance gene. Science. 2007;318:645–8. pmid:17962564
- 50. Thieme F, Koebnik R, Bekel T, Berger C, Boch J, Büttner D, et al. Insights into genome plasticity and pathogenicity of the plant pathogenic bacterium Xanthomonas campestris pv. vesicatoria revealed by the complete genome sequence. J Bacteriol. 2005;187(21):7254–66. pmid:16237009
- 51. Lorenz C, Büttner D. Functional characterization of the type III secretion ATPase HrcN from the plant pathogen Xanthomonas campestris pv. vesicatoria. J Bacteriol. 2009;191(5):1414–28. pmid:19114489
- 52. Daniels MJ, Barber CE, Turner PC, Sawczyc MK, Byrde RJW, Fielding AH. Cloning of genes involved in pathogenicity of Xanthomonas campestris pv. campestris using the broad host range cosmid pLAFR1. EMBO J. 1984;3:3323–8. pmid:16453595
- 53. Bonas U, Conrads-Strauch J, Balbo I. Resistance in tomato to Xanthomonas campestris pv. vesicatoria is determined by alleles of the pepper-specific avirulence gene avrBs3. Mol Gen Genet. 1993;238:261–9. pmid:8479432
- 54. Schreiber T, Sorgatz A, List F, Blüher D, Thieme S, Wilmanns M, et al. Refined requirements for protein regions important for activity of the TALE AvrBs3. PLoS ONE. 2015;10(3). Epub e0120214.
- 55. Figurski D, Helinski DR. Replication of an origin-containing derivative of plasmid RK2 is dependent on a plasmid function provided in trans. Proc Natl Acad Sci U S A. 1979;76:1648–52. pmid:377280
- 56. R-Core-Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria 2013. Available: http://www.R-project.org/.
- 57. Huber W, Carey VJ, Gentleman R, Anders S, Carlson M, Carvalho BS, et al. Orchestrating high-throughput genomic analysis with Bioconductor. Nat Meth. 2015;12(2):115–21.
- 58. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 2002;3(7):RESEARCH0034. pmid:12184808
- 59. Andersen CL, Jensen JL, Ørntoft TF. 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. 2004;64(15):5245–50. pmid:15289330
- 60. Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP. Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper–Excel-based tool using pair-wise correlations. Biotechnol Lett. 2004;26(6):509–15. pmid:15127793
- 61. Taylor KW, Kim JG, Su XB, Aakre CD, Roden JA. Tomato TFT1 is required for PAMP-triggered immunity and mutations that prevent T3S effector XopN from binding to TFT1 attenuate Xanthomonas virulence. PLoS Pathog. 2012;8(6):e1002768. pmid:22719257
- 62. Løvdal T, Lillo C. Reference gene selection for quantitative real-time PCR normalization in tomato subjected to nitrogen, cold, and light stress. Anal Biochem. 2009;387(2):238–42. pmid:19454243
- 63. Cohn JR, Martin GB. Pseudomonas syringae pv. tomato type III effectors AvrPto and AvrPtoB promote ethylene-dependent cell death in tomato. Plant J. 2005;44(1):139–54. pmid:16167902
- 64. Expósito-Rodríguez M, Borges A, Borges-Pérez A, Pérez J. Selection of internal control genes for quantitative real-time RT-PCR studies during tomato development process. BMC Plant Biol. 2008;8:131. pmid:19102748
- 65. Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W, et al. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nuc Acids Res. 1997;25(17):3389–402.
- 66. Ballvora A, Pierre M, Van den Ackerveken G, Schornack S, Rossier O, Ganal M, et al. Genetic mapping and functional analysis of the tomato Bs4 locus governing recognition of the Xanthomonas campestris pv. vesicatoria AvrBs4 protein. Mol Plant-Microbe Interact. 2001;14(5):629–38. pmid:11332727
- 67. Kim J-GG, Li X, Roden JA, Taylor KW, Aakre CD, Su B, et al. Xanthomonas T3S effector XopN suppresses PAMP-triggered immunity and interacts with a tomato atypical receptor-like kinase and TFT1. Plant Cell. 2009;21(4):1305–23. pmid:19366901
- 68. Pombo MA, Zheng Y, Fernandez-Pozo N. Transcriptomic analysis reveals tomato genes whose expression is induced specifically during effector-triggered immunity and identifies the Epk1 protein kinase which is required for the host response to three bacterial effector proteins. Genome Biol. 2014;15(10):492. pmid:25323444
- 69. Teper D, Salomon D, Sunitha S, Kim J-G, Mudgett MB, Sessa G. Xanthomonas euvesicatoria type III effector XopQ interacts with tomato and pepper 14–3–3 isoforms to suppress effector-triggered immunity. Plant J. 2014;77(2):297–309. pmid:24279912
- 70. Yang H, Liu J, Huang S, Guo T, Deng L, Hua W. Selection and evaluation of novel reference genes for quantitative reverse transcription PCR (qRT-PCR) based on genome and transcriptome data in Brassica napus L. Gene. 2014;538(1):113–22. pmid:24406618
- 71. Bin WS, Wei LK, Ping DW, Li Z, Wei G, Bing LJ, et al. Evaluation of appropriate reference genes for gene expression studies in pepper by quantitative real-time PCR. Mol Breeding. 2012;30(3):1393–400.
- 72. Wan H, Yuan W, Ruan M, Ye Q, Wang R, Li Z, et al. Identification of reference genes for reverse transcription quantitative real-time PCR normalization in pepper (Capsicum annuum L.). Biochem Biophys Res Commun. 2011;416(1–2):24–30. pmid:22086175
- 73. Gao S, Xu T, Qi M, Liu Y, Li H, Lv S, et al. Evaluation of the expression of internal control transcripts by real-time RT-PCR analysis during tomato flower abscission. African J Biotechnol. 2012;11(66):12983–9.
- 74. Graeber K, Linkies A, Wood ATA, Leubner-Metzger G. A guideline to family-wide comparative state-of-the-art quantitative RT-PCR analysis exemplified with a Brassicaceae cross-species seed germination case study. Plant Cell. 2011;23(6):2045–63. pmid:21666000
- 75. Hruz T, Wyss M, Docquier M, Pfaffl MW, Masanetz S, Borghi L, et al. RefGenes: identification of reliable and condition specific reference genes for RT-qPCR data normalization. BMC Genomics. 2011;12(1):156.
- 76. Die JV, Rowland LJ. Superior cross-species reference genes: a blueberry case study. PLoS ONE. 2013;8(9):e73354. pmid:24058469