Normalisation genes for expression analyses in the brown alga model Ectocarpus siliculosus

Background Brown algae are plant multi-cellular organisms occupying most of the world coasts and are essential actors in the constitution of ecological niches at the shoreline. Ectocarpus siliculosus is an emerging model for brown algal research. Its genome has been sequenced, and several tools are being developed to perform analyses at different levels of cell organization, including transcriptomic expression analyses. Several topics, including physiological responses to osmotic stress and to exposure to contaminants and solvents are being studied in order to better understand the adaptive capacity of brown algae to pollution and environmental changes. A series of genes that can be used to normalise expression analyses is required for these studies. Results We monitored the expression of 13 genes under 21 different culture conditions. These included genes encoding proteins and factors involved in protein translation (ribosomal protein 26S, EF1alpha, IF2A, IF4E) and protein degradation (ubiquitin, ubiquitin conjugating enzyme) or folding (cyclophilin), and proteins involved in both the structure of the cytoskeleton (tubulin alpha, actin, actin-related proteins) and its trafficking function (dynein), as well as a protein implicated in carbon metabolism (glucose 6-phosphate dehydrogenase). The stability of their expression level was assessed using the Ct range, and by applying both the geNorm and the Normfinder principles of calculation. Conclusion Comparisons of the data obtained with the three methods of calculation indicated that EF1alpha (EF1a) was the best reference gene for normalisation. The normalisation factor should be calculated with at least two genes, alpha tubulin, ubiquitin-conjugating enzyme or actin-related proteins being good partners of EF1a. Our results exclude actin as a good normalisation gene, and, in this, are in agreement with previous studies in other organisms.


Background
Brown algae (Phaeophyceae) are multi-cellular marine organisms that grow along temperate, tropical and polar coasts. Many of them are subject to frequent changes in their local environment, because they are uncovered at low tide, and are hence exposed to desiccation, and to variations in osmotic pressure due to rain or evaporation. In addition, pollution of the coasts, due to human activities, constitutes an additional source of abiotic stress, to which they must develop adaptive mechanisms. Expression analyses of genes involved in the perception of the stress, and in the establishment of the appropriate responses, provide a means to decipher the molecular mechanisms potentially involved in such adaptations. Despite the availability of medium scale cDNAs libraries for several different species of the brown algae (Laminaria, Sargassum, and Fucus), this task has been hindered by the lack of genome-scale resources. In 2004, Peters et al. [1] have compared a range of features in several species of Phaeophyceae and concluded that Ectocarpus siliculosus was the best candidate to consider for such developments. Recently, the genome of this alga has been sequenced, offering a unique opportunity to survey the expression of gene families in brown algae (Genoscope, J.M. Cock, unpublished data). E. siliculosus is a small filamentous alga, extensively studied over the last two centuries for its complex life cycle and its physiological features (reviewed in [2]). The genome is currently being annotated, allowing the initiation of both large scale and targeted surveys of the Ectocarpus genes, such as microarrays or real-time RT-PCR respectively.
Compared to high-throughput microarray techniques, real-time quantitative RT-PCR only allows assays of gene expression to be carried out at relatively low throughput (10-20 genes in 10-50 samples). Nonetheless, this technique has been adopted by a large community as a standard method for gene expression studies, because of its high reliability, and its rapidity of execution [3,4]. This technique is now widely used for a large number of animal and plant organisms, as well as for bacteria and viruses.
A few years after the emergence of this technique, a need for a reliable normalisation method became insistent. Different methods of identifying normalisation genes, such as geNorm [5], NormFinder [6] and BestKeeper [7], were then developed. This was followed by a wave of reports on the identification of the best normalisation genes from a broad range of species and specific tissues. These biological materials included human tissues [8][9][10] and viruses [11], as well as tissues from a variety of animals such as cow [12], pig [13], horse [14], dolphin [15], fishes [16,17], worm [18] and others. In parallel, a similar extension of the use of this technique was observed in plant research, for rice [19], poplar [20], potato [21], grapevine [22], and for plant pathogens [23].
In the brown alga E. siliculosus, the expression level of 20 genes specific to the two generations in the life-cycle of this alga was recently reported in a mutant impaired in development [24]. Additional developmental and physiological studies are underway, including studies aimed at assessing the resistance of this alga to environmental changes. Hence, the availability of a set of housekeeping genes for normalising the expression levels of genes of interest is a pre-requisite to any valuable conclusion, especially since this organism lives in a frequently changing environment.
In this paper, we propose optimal normalisation genes for expression analyses in E. siliculosus. Thirteen housekeeping genes that have been reported to be good potential candidates in the previously cited literature, were chosen for this task. Their expression was examined by Q-RT-PCR in a diversity of algal samples corresponding to growth kinetic series, osmotic stress experiments, and chemical treatments. Namely, the candidate genes are involved in the synthesis and the dynamics of the cytoskeleton, in the synthesis, folding and degradation of proteins, and in the metabolism of carbon, all of these processes being known to be only moderately affected by the fluctuation of growth conditions.

Treatments applied to E. siliculosus and choice of housekeeping genes
Several different stresses were tested in this study. Chemical agents tested included H 2 O 2 , a reactive oxygen species produced by many organisms, including seaweeds, under conditions of abiotic and biotic stresses [25]. We also tested heavy metals such as copper, which are among the most significant pollution actors in marine environments worldwide. Diuron and atrazine are herbicides that inhibit photosynthesis by blocking the d1 protein of photosystem II [26,27]. Diuron is also an additive of antifouling paints, which prevent growth of organisms on ships' hulls. In addition pathogen or grazer attacks were mimicked by wounding E. siliculosus tissues with a razor blade. We also tested the effect of oxylipins. These are oxygenated derivatives of polyunsaturated fatty acids which play a major role in inflammatory processes, allergies, and, in a wide sense, defensive stress responses to infection, drugs, and xenobiotics [28]. In land plants, C18 derived jasmonates play a pivotal role in defense induced mechanisms [29]. In mammals, oxidation of the C20 arachidonic acid produces derivatives such as leukotriens and prostaglandins. As brown algae are able to produce oxylipins typical of both land plants and animals [30], their putative action on the induction of stress signalling path-ways is of particular interest. Finally, as the metabolism and the physiology of brown algae are regulated to a large extent by diurnal rhythms, a time series of samples taken over a single day was also analysed.
RNA was extracted from biological triplicates of algae that had received the above treatments resulting in a total of 63 samples for 21 different treatments (see Table 1, and Methods for details on concentration and timing). Single strand cDNA was synthesised simultaneously from each of the 63 extracts in order to minimize any variation during this step of the process. The abundance of the transcripts of 13 potential housekeeping genes was then assayed on these cDNAs. The genes tested included commonly used genes such as a ribosomal protein and translation initiation or elongation factors (eIF2A, eIF4E, EF1a, 26S ribosomal protein), cytoskeleton proteins (tubulin alpha, actin and actin-related proteins), and proteins involved in the protein degradation process (ubiquitin and ubiquitin-conjugating enzyme). In addition, cyclophylin, two actin-related proteins, a tubulin molecular motor (dynein) and an enzyme involved in the pentose phosphate pathway (glucose 6-phosphate dehydrogenase) were included in this study. The genes tested are listed in Table 2.

Quantification and data analysis
In order to assess whether the transcripts of these 13 genes remained at comparable levels in the different samples tested, we calculated the variation in the Ct value for each gene. Figure 1A shows that the transcripts of these genes exhibited different levels of abundance, with CYC being expressed at the lowest level, and UBQ being expressed the most strongly. Variation in transcript accumulation across the 21 culture conditions was not the same for all the genes tested. EF1a showed the weakest variation, while G6PD expression seemed to be strongly influenced by the treatments, its range of expression level exceeding 10 Ct ( Figure 1B).
In order to test the robustness of these data, we performed the geNorm pairwise analysis, which was first described by Vandesompele et al. [5], and has since been widely used to evaluate the stability of expression of genes from many organisms. The results of this analysis ( Figure 2) were slightly different from those obtained with the Ct value calculation. The two calculation methods identified the same least and most stably expressed genes (G6PD and CYC were the least stable, and TUA, EF1a and Dyn were the most stable), but the intermediate genes were ordered differently. This was particularly striking for ARP2. 1 and UBCE, which the geNorm analysis indicated were the most stably expressed genes, and which the Ct value analysis indicated to be among the least stably expressed. In order to test whether averaging the biological triplicates had an effect on the final result, we performed the geNorm analysis using the transcript abundances measured in the 63 individual cDNA samples, as previously described [31]. This analysis identified the same genes as having either highly variable (CYC, G6PD and ACT) or very stable (ARP2.1, UBCE, TUA, EF1a and Dyn) transcript abundances, but the order of the six remaining genes was again different (data not shown). Thus, averaging the biological replicates modified the results of the analysis for the intermediate genes, but not for the most and the least stable genes. A one-way ANOVA test showed that for most of the genes considered individually, the variance between the different culture conditions is significantly higher than the variance between the biological replicates for a given condition (see Additional file 1). Therefore, averaging on the three biological replicates should not introduce any significant distortion.
NormFinder is another approach that has been used to assess the stability of expression of housekeeping genes [6]. When NormFinder was applied to the data obtained in this study, it indicated that the genes with the most stable levels of transcript abundance were TUA, ARP2.1, EF1a and Dyn (Table 3). These were almost the same genes as the ones identified by geNorm, with the exception that Dyn performed better than UBCE. Therefore, there was a very good correlation between the results obtained from geNorm and NormFinder, despite the fact that the methods of calculation are fundamentally different.
In order to test whether the normalisation genes identified above were also the best choices for specific conditions, we performed expression stability measurements on distinct series of treatments i.e. osmotic stresses, chemical treatments and diurnal rhythm. Figure 3 shows that with both the geNorm and NormFinder calculation methods (A and B) the optimal choice of normalisation gene depended on the type of experiments. Thus, although EF1a remained among the most stable genes, variations in the identity of the other very stable genes were observed depending on the treatment. This was particularly striking for the ACT gene, the abundance of its transcript showing a high level of variability following all the treatments except osmotic stresses. The transcript of the TUA gene varied significantly in abundance in the diurnal sample series (see Additional file 2), but remained at a stable level in the abiotic stress treatments. Note that CYC and G6PD, which showed the greatest variability in transcript abundance in the global analyses, were also highly variable in each of the different classes of experiment.
To calculate the number of normalisation genes necessary to obtain the normalisation factor we determined, using geNorm, the pairwise variation between sets of normalisation factors obtained when using two, three or more genes for normalisation. Figure 4 shows that the normalisation factors are only modified slightly when a third (or more) gene is added (pairwise variation of 0.13 for 3 genes).
Vandesompele et al. [5] recommended that additional normalisation genes be included if the pairwise variation between the normalisation factors is higher than 0.15. According to geNorm, measuring the expression levels of ARP2.1 and UBCE is sufficient to normalise the expression of genes of interest in these samples.

Discussion
In this study, the three methods used to identify the best normalisation genes were concordant, as previously reported in other studies [10]. Comparison of the three methods indicates that EF1a is the most reliable gene to normalise gene expression in experiments aiming at quantitatively measuring the transcriptional response to abiotic stresses and chemical treatments. The V pairwise analysis shows that two genes are sufficient for a proper expression normalisation. The choice of the second gene, however, will depend on the type of experiment that is being carried out. For osmotic stresses and chemical treat- ments, TUA can be used reliably, while for the diurnal cycle ARP2.2 is more relevant.
As a brown alga, E. siliculosus is member of the kingdom of the Heterokonta, which is phylogenetically very distant from animals and land plants [32]. Interestingly, and despite this evolutionary distance, a consensus seems to emerge from similar analyses performed in organisms belonging to distant lineages. The gene coding for the elongation factor of protein translation EF1a was shown to be the best reference genes in salmon [17] and in several plants such as in rice [19], grapevine [22] and potato [21]. The alpha tubulin encoding gene was also reported to be one of the best reference genes for horse tissues [13] and in poplar [19]. On the other hand, actin genes have been very often reported as exhibiting highly variable levels of expression in both human and animals tissues [10,12,[16][17][18], and in plants [19]. In this study, we have shown that this is also true for E. siliculosus.
Despite the fact that ubiquitin and related enzymes (UBCE) are not commonly used as normalisation genes, in this study, they were found to be quite suitable. Interestingly, Czechowski et al. [31] showed, using microarray analysis, that genes of the ubiquitin complex, comprising an ubiquitin conjugating enzyme such as UBCE, and several E3-ubiquitin protein ligases, were very stably expressed. They also pointed out that genes with fairly low levels of expression such as UBCE may be of particular interest for normalising expression levels of genes that are expressed at moderate to low levels, such as transcription factors. This latter example illustrates how microarray analyses may be useful to find additional normalisation genes, which can be then tested by Q-PCR for their suitability.

Conclusion
E. siliculosus is recognised as the genomic and genetic model of brown macroalgae [1,2]. As the genome sequence is currently in the phase of expert annotation, the community interested in E. siliculosus is likely to grow in the near future. The results presented in this paper pave The stability of expression of the 13 genes was calculated using the Normfinder method designed by Andersen et al. [6] Expression level of 13 housekeeping genes Global ranking of the 13 housekeeping genes using geNorm analysis Figure 2 Global ranking of the 13 housekeeping genes using geNorm analysis. The M value was calculated with the geNorm software [5]. Low values of M indicate that a gene is expressed very stably. the way for further studies on different aspects of E. siliculosus biology including development/morphogenesis and abiotic/biotic stress responses. In addition, they will be helpful for comparison with results from microarray hybridizations, which are currently in progress.

Culture conditions and treatments
E. siliculosus (Ectocarpales, Phaeophyceae) unialgal strain 32 (CCAP accession 1310/4, origin san Juna de Marcona, Peru) was cultivated in 10 L plastic flasks in a culture room at 14°C using filtered and autoclaved natural seawater enriched in Provasoli nutrients [33]. Light was provided by Philips daylight fluorescence tubes with a photon flux density of 40 μmol. m-2 .s -1 for 14 hours per day. Cultures were bubbled with filtered (0.22 μm) compressed air to avoid CO 2 depletion. To conduct the chemical treatment experiments, algal tissues were transferred into Petri dishes containing artificial seawater enriched with Provasoli (ASW) for at least 18 h before treatments in order to acclimatize the cultures to the change of growth conditions. They were then treated with different chemi-cals for 3 and/or 6 h (see Table 1) during the light phase. The treatments were 10 mM H 2 O 2 (final concentration), 42 μg.L -1 Diuron, 55 μg.L -1 atrazine, 5 μM 13-HOTrE and 15(S)-HEPE. Algae were also incubated in 10 μM CuSO 4 for 3 and 6 h. An equal volume of solvent (ASW or DMSO 1%) was used in each corresponding control treatment. Wounding was carried out by damaging the tissues with a razor blade. To perform saline stress and an additional H 2 O 2 treatment, tissues were transferred to ASW for one week before applying the stress. The NaCl concentration in control ASW was 450 mM, while it was 60 mM and 1.5 M in ASW used to submit the alga to hyposaline and hypersaline conditions respectively. In addition, H 2 O 2 was added at 1 mM final concentration to the control ASW for generating oxidative stress. Treatments were applied for 6 h before harvesting the tissues for RNA extraction. To collect samples through the diurnal cycle, algae were incubated in ASW and tissue harvested every 6 h during one day. The first sample was taken 30 min after the beginning of the light period. The summary of the culture conditions is presented in Table 1.
Ranking of the 13 housekeeping genes over the three different series of culture conditions Three biological replicates were obtained for each treatment and these were used for total RNA extraction.

RNA extraction
The protocol used for RNA extraction was based on the method developed by Apt et al. [34] with some modifications. Frozen tissue was ground in liquid nitrogen and immediately incubated in the presence of extraction buffer (100 mM Tris-HCl pH 7.5, 2% CTAB, 1.5 M NaCl, 50 mM EDTA, 50 mM DTT). The sample was shaken at room temperature for 30 min, then one volume of chlorophorm:isoamylic alcohol (24:1, V/V) was added and the sample shaken again for 25 min. After centrifugation, the upper phase was transferred to a new tube and mixed with 0.3 V of absolute ethanol to precipitate the polysaccharides, and extracted with 1 V of chlorophorm. After centrifugation the upper phase was transferred to a fresh tube and RNA was precipitated by addition of 0.25 V of 12 M LiCl and β-mercapto-ethanol to 1% final concentration, overnight at -20°C. After centrifugation, the pellet was resuspended for DNAse treatment by an RNAse-free DNAse I (Turbo DNAse, Ambion) according to the manufacturer's instructions, in order to eliminate any residual genomic DNA from the preparation. An extraction was then carried out by adding Phenol-Chlorophorm (1:1, V/ V). After centrifugation the upper phase was transferred to a fresh tube, and extracted with 1 V of chlorophorm:isoamylic alcohol (24:1, V/V) and centrifuged again. The upper phase was precipitated with 0.3 M NaAc pH 5.5 and 75% ice cold ethanol by incubating overnight at 20°C. After centrifugation, the supernatant was removed, and the pellet washed with 80% ethanol. After centrifugation and drying on ice, the pellet was resuspended in an appropriate volume of RNAse-free water.

Quantification of RNAs and cDNA synthesis
Nucleic acid concentrations were measured by the absorbance at OD 260 using a NanoDrop ND-1000 spectrophotometer.

Real-time PCR
All the genes were quantified on the same lot of cDNAs, to minimize experimental variation that could be due to cDNA synthesis. For each gene, a pair of oligonucleotide sequences was designed in the 3' UTR of the genes when the sequence was known, or in the 3'coding sequence using Primer Express TM1.0 (PE Applied Biosystems, Foster City, CA, USA) ( Table 2). The Q-PCR reactions were performed in a 96-well thermocycler (Biorad, Opticon) with SYBRgreen reaction mix from ABgene (AB-1162/B), for 15 min at 95°C, followed by 41 runs of 15 sec at 95°C and 1 min at 60°C. Each sample was technically duplicated. E. siliculosus genomic DNA was used as a quantifica-Determination of the optimal number of control genes for normalisation Figure 4 Determination of the optimal number of control genes for normalisation. The pairwise variation V of the normalisation factors was calculated for the 21 different culture conditions for the 13 housekeeping genes with the geNorm software [5]. 0.14 V 2 / 3 V 3 / 4 V 4 / 5 V 5 / 6 V 6 / 7 V 7 / 8 V 8 / 9 V 9 / 1 0 V 1 0 / 1 1 V 1 1 / 1 2 Pairwise variation V tion reference. A dilution series ranging from 37 to 48671 copies of the E. siliculosus genome was prepared and tested for each gene. The amplification efficiency was tested using this dilution series ( Table 2). The specificity of amplification was checked with a dissociation curve obtained by heating the samples from 65°C to 95°C (Table 2). In addition to the DNAse-I treatments of RNAs, the absence of a genomic DNA contaminants was checked, by amplifying an intron sequence on the cDNAs. The number of copies of contaminant gDNA was subtracted from all other values, prior to any further analyses.