Introduction

The analysis of gene expression provides a useful means of understanding the regulatory networks present in higher organisms [1]. One of the leading methods for achieving this is real time quantitative PCR (RT-qPCR), since this platform is relatively simple to use, has a reasonably broad dynamic range, and a good level of sensitivity [2]. The quantification of gene expression using RT-qPCR requires the simultaneous assay of a reference gene(s), in order to ensure that any variation in signal of the target is not due to variation in the amount of template used to set the PCR. An ideal reference gene is expressed at a level which is largely independent of the experimental conditions to which the subject has been exposed; however, a very few, if any, genes meet this criterion. The choice of an appropriate reference thus requires that they are stably expressed in the actual experimental conditions to be used [3]. Because even this can be difficult to ensure, it has been suggested that a set of reference genes be deployed [4]. The most frequently used reference genes in plant gene expression studies are actin, GAPDH (glyceraldehyde-3-phosphate dehydrogenase), tubulin, and 18s ribosomal RNA [5–7].

Although gene expression has been assessed by RT-qPCR in a wide range of both crop and non-crop plants [6, 8, 9], the technique as yet has not been applied to the important ornamental species chrysanthemum. In this article, we report the stability of eight potential reference genes in chrysanthemum plants exposed to aphid infestation or abiotic (heat or waterlogging) stress.

Materials and Methods

Plant Materials and Treatments

Plants of the chrysanthemum cultivar “Zhongshanzigui” were grown in pots in a greenhouse held at 32°C/27°C, with a light source providing 160 μmol/m2/s photosynthetically active radiation over a 12 h photoperiod. Seedlings at the 6–8 leaf stage were exposed to either a period at 45°C, or infestation with five second instar Macrosiphoniella sanbourni nymphs on each stem tip [10], or a waterlogging treatment consisting of flooding the pots to 2.5 cm above the soil surface for various time periods [11]. The apical shoot of each of three plants was harvested after a 0-, 3-, 6-, 12-, or 24-h exposure to either heat or waterlogging. The aphid-infested plants were similarly sampled in triplicate after 0, 1, 3, 5, and 7 days. After harvesting, the shoots were snap-frozen in liquid nitrogen and stored at −80°C until required.

RNA Extraction

Total RNA was extracted from leaves using the TaKaRa RNAiso reagent, and treated with RNase-free DNaseI (TaKaRa, Japan) [12]. Its concentration was assessed spectrophotometrically at 230, 260 and 280 nm, with samples only being retained where their 260/280 ratio was in the range of 1.80–2.05, and their 260/230 ratio was in the range of 2.00–2.60. The integrity of the RNA was further checked by electrophoresis through 1% agarose gels. The first strand cDNA was synthesized based on 1 μg total RNA using M-MLV reverse transcription system (Promega, USA), according to the manufacturer’s instructions.

Primer Design

The primers directed at the tubulin, UBC, GAPDH, PP2Acs, and PGK sequences (Table 1) were based on chrysanthemum EST sequences [13], which were functionally assigned on the basis of BLAST similarity with homologs from other species. The sequences used for the design of primers targeting actin were obtained from GenBank accessions AB205087. EF1α and psaA were amplified by the degenerate primers (GenBank accessions AB548817 and AB548887, respectively). All the eight primer pairs were designed using Primer Premier v5.0 software (Premier Biosoft International), and their sequences are listed in Table 2.

Table 1 Description of chrysanthemum reference genes from EST database for RT-qPCR
Full size table
Table 2 Primer sequences and amplicon characteristics for each of the eight reference genes
Full size table

RT-qPCR

RT-qPCRs were performed in a 25 μl volume, containing 10 μl SYBR Green PCR master mix (TOYOBO, Japan), 10 ng of cDNA template and 0.2 μM of each primer. The cycling conditions comprised an initial denaturation of 95°C/60 s, followed by 40 cycles of 95°C/15 s, 55°C/15 s, and 72°C/45 s. At the end of the cycling process, the temperature was raised from 55 to 95°C at a rate of 0.5°C/s to derive the necessary denaturation curve. Amplicon purity was assumed where a single melting peak was produced. Each qRT-PCR was run in triplicate. PCR efficiency was calculated for each target from the raw fluorescence data taken from Bio-Rad thermal cycler using the LinRegPCR program [14]. Results from the LinRegPCR and Bio-Rad software were imported into Microsoft Excel 2003 and transformed using the comparative Cq method and specific efficiencies for each target [15]. Expression stability was assessed using geNorm software [4].

Results

The transcription profiles of the eight reference genes are shown in Fig. 1. The C T values recorded from the plants subjected to stress or aphid infestation were similar to one another. The geNorm software package (http://medgen.ugent.be/genorm) for Microsoft Excel® was applied to identify which of the eight genes would be suitable for normalization purposes. Vandesompele defined the two parameters M (mean expression stability of each individual gene) and V (optimal number of reference genes) using geNorm [4]. Genes associated with a low M values are the most stable, and genes associated with M values >0.15 indicated the need for additional reference gene(s). For aphid infestation, the best performing genes were EF1α and tubulin, having an M value of 0.176. When Used together, the V value was 0.080, indicating no need to add a third reference gene. When the plants were subjected to waterlogging, however, EF1α was no longer stably expressed, and instead PP2Acs with PGK was the most appropriate choice (M = 0.263); curiously, the former gene performed particularly poorly in aphid infested plants (Fig. 2a, c). The most stable gene combination for heat stress was PP2Acs and GAPDH (M = 0.276). Across the entire data set, psaA, EF1α, UBC, and PGK were the most stably expressed genes (Fig. 2d). The V value of 0.185 obtained by the pair of PP2Acs and GAPDH was above the cut-off value, implying that at least one more reference gene was needed. Addition of EF1α lowered the V value to 0.118, so that a fourth reference gene was unnecessary. GeNorm analysis indicated that actin was the least reliable reference gene for heat and waterlogging stress experiments (Fig. 2b, c).

Fig. 1
figure 1

RNA transcription levels of the tested reference genes, presented as absolute C T values in the different samples

Full size image
Fig. 2
figure 2

Average expression stability values of control genes by geNorm analysis: a aphid infestation; b heat stress exposure; c waterlogging stress exposure; d all stresses combined. The least stable genes are on the left, and the most stable on the right; e determination of number of reference genes for normalization by geNorm analysis. On the left-most side is the pairwise variation when the number of reference genes is increased from two to three (V2/3). Stepwise inclusion of less stable genes generates the next data points. A decrease in the V value indicates a positive effect and means that the added gene should preferably be included for calculation of a reliable normalization factor

Full size image

Discussion

Reference genes selection have been reported in a number of plants, such as the following: in potato during late blight exposure, cold, and salt treatment [16], in tomato subjected to nitrogen, cold, and light stress [17], in longan tree during somatic embryogenesis [18], in sunflower throughout six differently controlled leaf senescence conditions [19], and in populus across various developmental stages [20]. In this study, we also find that eight candidate reference genes performed differently upon stresses to which chrysanthemum plants were subjected.

The poor performance of actin in chrysanthemum plants subjected to either heat or waterlogging stress, or aphid infestation was surprising though this gene has been much used as a reference in earlier gene expression studies [21–24]. Recently, several studies have also shown that the use of actin for normalization is not reliable in potato [16], in peach [25], and in bovine muscular tissue [26]. This may be that the actin gene product is a major component of eukaryotic cytoplasmic microfilaments, and it participates in cytoplasmic streaming, cell division and the distribution of the plasma membrane proteins [27].

Genes responsible for elongation factors appear to be stably expressed over a wide range of experimental conditions in potato, peach, and rice [7, 16, 25]. In the present experiments, EF1α was stably expressed under aphid infestation and heat stress, but under waterlogging stress, it was one of the least stable of the eight genes tested (Fig. 1c), indicating that it is not a suitable reference gene under waterlogging stress. However, it was interesting to note that EF1α was always stably expressed in cold-stressed potato and tomato [16, 17], cold and heat- treated rice [7], and similarly, EF1α was stable in chrysanthemum under heat stress. It is inferred that EF1α may be a suitable control gene under cold and heat stress in spite of species.

GAPDH, which has been quite widely applied as a reference gene [28], was indeed the most stable of the genes tested in chrysanthemum plants subjected to heat stress, but was only moderately stable under conditions of either aphid infestation (the fifth most stable gene of the eight genes) or waterlogging stress (the third most stable gene). This gene is therefore also inadequate as a sole reference gene, as has been pointed out on previous examples, and it may lead to wrong results in specific experimental conditions [7, 28]. It may be that GAPDH not only acted as a component of the glycolytic pathway, but also was involved in other processes such as cell proliferation and carcinogenisis [29]. PGK is collocated with GAPDH and also plays important roles in the glycolytic pathway. PGK and GAPDH are potentially co-regulated [30]. Andersen and coworkers mentioned that geNorm may be insensitive to coregulated reference genes [31]. However in our data, potential co-regulation between them was not significant (Fig. 2).

In abiotic (heat or waterlogging) stress, tubulin displayed variable expression patterns. Surprisingly, tubulin showed highly stable expression under aphid infestation among the eight reference genes. However, in perennial ryegrass [32], forage grass Brachiaria brizantha [33], and flax [34], tubuline showed an unacceptable variable expression, but was found to be an appropriate reference gene in cucumber [35], longan tree [18], and in sunflower [19], whereas PP2Acs performed worst under aphid infestation (Fig. 2a), but well under heat or waterlogging stress (Fig. 2b, c). PsaA was the most stably expressed control genes in all combined, but it was not suitable as the best under three different conditions. All these results suggested that reliability of reference genes is highly specific to a particular experimental condition.

The choice of reference genes is an important one, and our results did validate that it is necessary to select suitable reference gene for each experimental condition. Thellin and Vandesompele have recommended the requirement of at least three reference genes [4, 36]. In this study, we have investigated eight reference genes and recommend the use of at least two of the top-ranked reference genes for normalizing gene expression in chrysanthemum leaves under stresses tested here. To our knowledge, this is the first attempt to identify reference genes for gene expression studies in chrysanthemum leaves.