Several different stresses were tested in this study. Chemical agents tested included H₂O₂, 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 2627. 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 pathways 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.

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.

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 CO₂ 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 chemicals for 3 and/or 6 h (see Table 1) during the light phase. The treatments were 10 mM H₂O₂ (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 CuSO₄ 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 H₂O₂ 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, H₂O₂ 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.

Three biological replicates were obtained for each treatment and these were used for total 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.

Nucleic acid concentrations were measured by the absorbance at OD₂₆₀ using a NanoDrop ND-1000 spectrophotometer. The purity of the RNA samples was assessed by measuring the ratio OD₂₆₀/OD₂₈₀ and OD₂₃₀/OD₂₆₀ (see Additional file 3). RNA integrity was then verified on 1.5% agarose gel stained with ethidium bromide (see Additional file 4). From each RNA sample, 1.4 μg was reverse transcribed to cDNA using oligo(dT)_12–18 and the Superscript™ First Strand synthesis for RT-PCR (Invitrogen) according to the manufacturer's instructions, and subsequently diluted with nuclease free water to 1 ng.μl^-1 cDNA.

Frozen tissue was ground in liquid nitrogen and then in a wheaton potter with 15 ml of extraction buffer (100 mM Tris-HCl pH7.5; 1.5 M NaCl; 2% CTAB; 50 mM EDTA; 50 mM DTT) per g of wet tissue. The suspension was then mixed vigorously at room temperature for at least 30 min. Proteins were degraded with 25 units of proteinase K for 2 h at 55°C, and then extracted with chlorophorm/isoamyl alcohol for several minutes. Polysaccharides were precipitated with progressive addition of 0.2 – 0.3 V of ethanol, and then extracted with 1 V of chloroform after spinning at 10,000 g and 20°C for 20 min. Nucleic acids were recovered from the upper phase by addition of 0.25 V of 12 M LiCl and 1% of β-mercaptoethanol, incubation at -20°C overnight and spinning at 10,000 g and 4°C for 1 h. The supernatant was precipitated with 0.6 V isopropanol, 0.3 M NaAc pH 5.2, left at 4°C for 30 min, and then spun down for 30 min at 13,000 g. The pellet was then dissolved in 400 μl H₂O, and precipitated again with ethanol and NaAc. DNA was dissolved in 500 μl of TE (10 mM Tris-HCl pH 8.0; 1 mM EDTA), 5.4 M CsCl (density 1.66) and 250 μg.mL^-1 of ethidium bromide. Spinning at 90,000 g for 24 h at room temperature allowed the recovery of a band containing ultrapure genomic DNA under UV. Ethidium bromide was extracted 4–5 times with TE-saturated butanol and CsCl eliminated by successive ethanol precipitations.

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 quantification 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.