Integrative analysis of metabolite and transcript abundance during the short-term response to saline and oxidative stress in the brown alga Ectocarpus siliculosus



    1. UPMC Univ Paris 6, UMR 7139 Marine Plants and Biomolecules, Station Biologique, F-29680, Roscoff, France
    2. CNRS, UMR 7139 Marine Plants and Biomolecules, Station Biologique, F-29680, Roscoff, France
    Search for more papers by this author

    1. UMR 118 INRA, Agrocampus Ouest, Université de Rennes 1, Amélioration des Plantes et Biotechnologies Végétales, Campus de Beaulieu, bât. 14A, F-35042 Rennes Cedex, France
    Search for more papers by this author

    1. Univ Rennes 1, CNRS, UMR 6553, Equipe Paysages Changements Climat Biodiversité, F-35042 Rennes Cedex, France
    Search for more papers by this author

    1. Laboratoire de Biochimie ‘Epissage, Cancer, Lipides et Apoptose’, INSERM U613, Université de Bretagne Occidentale, Faculté de Médecine, F-29285 Brest, France
    Search for more papers by this author

    1. Leibniz Institute for Baltic Sea Research Warnemünde, Physical Oceanography and Instrumentation, Seestraße 15, D-18119 Rostock, Germany
    Search for more papers by this author

    1. UMR 118 INRA, Agrocampus Ouest, Université de Rennes 1, Amélioration des Plantes et Biotechnologies Végétales, Campus de Beaulieu, bât. 14A, F-35042 Rennes Cedex, France
    Search for more papers by this author

    1. UPMC Univ Paris 6, UMR 7139 Marine Plants and Biomolecules, Station Biologique, F-29680, Roscoff, France
    2. CNRS, UMR 7139 Marine Plants and Biomolecules, Station Biologique, F-29680, Roscoff, France
    Search for more papers by this author

    Corresponding author
    1. UPMC Univ Paris 6, UMR 7139 Marine Plants and Biomolecules, Station Biologique, F-29680, Roscoff, France
    2. CNRS, UMR 7139 Marine Plants and Biomolecules, Station Biologique, F-29680, Roscoff, France
    Search for more papers by this author

T. Tonon. e-mail:


The model brown alga Ectocarpus siliculosus undergoes extensive transcriptomic changes in response to abiotic stress, many of them related to primary metabolism and particularly to amino acid biosynthesis and degradation. In this study we seek to improve our knowledge of the mechanisms underlying the stress tolerance of this alga, in particular with regard to compatible osmolytes, by examining the effects of these changes on metabolite concentrations. We performed extensive metabolic profiling (urea, amino acids, sugars, polyols, organic acids, fatty acids) of Ectocarpus samples subjected to short-term hyposaline, hypersaline and oxidative stress, and integrated the results with previously published transcriptomic data. The most pronounced changes in metabolite concentrations occurred under hypersaline stress: both mannitol and proline were accumulated, but their low final concentrations indicate that, in this stress condition, both compounds are not likely to significantly contribute to osmoregulation at the level of the entire cell. Urea and trehalose were not detected in any of our samples. We also observed a shift in fatty acid composition from n-3 to n-6 fatty acids under high salinities, and demonstrated the salt stress-induced accumulation of small amounts of γ-aminobutyric acid (GABA). GABA could be synthesized in E. siliculosus through a salt stress-induced putrescine-degradation pathway.


Brown algae are an important component of coastal ecosystems as they are the dominant vegetation in the intertidal and subtidal zone. They can be found from tropical to polar regions, and some species form large kelp forests providing habitats for thousands of other species. Being sessile, intertidal seaweeds in general must be able to withstand many stresses such as desiccation, high photosynthetically active radiation (PAR), ultraviolet radiation, changes in salinity (rain, evaporation), mechanical stress (waves), as well as anthropogenic stresses such as heavy metal pollution. The capacity of algae to tolerate different types of abiotic stress has been related to their vertical distribution along the intertidal gradient (reviewed in Davison & Pearson 1996). Despite the broad scientific interest in physiological responses of algae to environmental stresses, the biochemical pathways involved and the underlying genetics still remain largely unexplored. The sequencing of algal genomes including the heterokont Ectocarpus siliculosus genome (Cock et al. 2010) has enabled a new suite of approaches to study the mechanisms involved in their stress tolerance.

Increasing amounts of transcriptomic data on the acclimation of seaweeds to changes in their environments are being gathered (Roeder et al. 2005; Collén et al. 2007; Dittami et al. 2009; Pearson et al. 2010). Within the heterokont lineage, Krell et al. (2007) monitored gene expression levels and the accumulation of proline in the salinity-stressed Antarctic diatom Fragilariopsis cylindris, and Ritter et al. (2008) evaluated the expression of genes and the production of oxylipins in copper-stressed individuals of the brown alga Laminaria digitata. However, very little is known about how transcriptomic changes translate into metabolite changes in algae, underlining the need for comparative analysis of data sets corresponding to different types of profiling.

In this report we examined changes in metabolite profiles in the cosmopolitan brown alga Ectocarpus siliculosus (Charrier et al. 2008) upon exposure to mild hyposaline, hypersaline and oxidative stress. The same stress conditions have previously been examined at a transcriptomic level in the same organism (Dittami et al. 2009), revealing large-scale transcriptomic reprogramming including the down-regulation of several genes related to primary metabolism in response to these abiotic stresses. We took advantage of this previous study to link metabolic with transcriptomic changes, and tested several hypotheses about the accumulation of possible compatible osmolytes such as urea and proline. Finally, this study permitted the discovery of new stress-related metabolites, such as γ-aminobutyric acid (GABA), which was detected only in traces under normal growth conditions in E. siliculosus (Gravot et al. 2010).


Culture conditions

Ectocarpus siliculosus (Dilwyn) Lyngbye (Ectocarpales, Phaeophyceae) unialgal strain 32 (accession CCAP 1310/4, origin San Juan de Marcona, Peru, 2002) was cultivated in 10 L plastic flasks in a culture room at 14 °C, using filtered and autoclaved seawater (SW) with a salinity of 33 psu; SW was enriched in Provasoli nutrients (Starr & Zeikus 1993), that is 0.8 mm nitrate, 50 µm phosphate, 0.45 mm boric acid, 6 µm iron, 18 µm manganese, 1.9 µm zinc, 0.48 µm cobalt, 1.5 nm vitamin B12, 0.3 nm thiamine, 4 nm biotin, 0.8 mm Tris, 75 µm EDTA (all concentrations are the final concentrations in the medium), and adjusted to a pH of 8. These nutrients are adequate to grow E. siliculosus even in otherwise nutrient-free artificial SW. PAR (400–700 nm) was provided by Philips daylight fluorescence tubes at a photon flux density (PFD) of 40 µmol m−2 s−1 for 14 h per day, a PFD that was sufficient for cultures to grow well, but low enough not to produce high light stress. Cultures were aerated with filtered (0.22 µm) compressed air. One week before the experiment, several thalli per condition, time point, and replicate (three replicates per condition, about 0.5 g fresh weight each) were transferred to individual Petri dishes containing 100 mL of fresh culture medium and maintained as described earlier, but without aeration.

Experimental setup

Stress experiments were started about 30 min after the beginning of the light phase by replacing the culture media with pre-prepared stress media. To be able to compare this study with a previous transcriptomic study (Dittami et al. 2009), identical stress conditions were chosen. For hyposaline stress, 33 psu SW was diluted to 12.5% of its original concentration with distilled water (i.e. final salinity: 4 psu). For hypersaline stress, 60 g NaCl were added per liter of 33 psu SW (i.e. final salinity: 93 psu). For oxidative stress, H2O2 (Sigma-Aldrich, St. Louis, MO, USA) was added immediately before beginning the stress experiment at a final concentration of 1 mm. For the control condition, 33 psu SW was used. Identical final quantities of Provasoli nutrients were added to each of these media. All three tested stress conditions were previously shown to be sub-lethal for the examined strain of E. siliculosus (Dittami et al. 2009), as it was able to survive indefinitely in stress media and even resumed growth (data not shown). However, the treatments reduced its photosynthetic quantum yield (measured by pulse amplitude modulation fluorometry) by 50–70% after 6 and 24 h stress exposure. In addition, after restoring control conditions for algae submitted to saline stresses, it took cultures between 3 and 6 d to completely recover their original photosynthetic quantum yield (Dittami et al. 2009). This was also true for the H2O2 treatments, although H2O2 concentrations in the medium probably decreased significantly throughout the experiment.

Transcriptomic changes were monitored after 6 h in our previous study (Dittami et al. 2009), as this corresponds to the time interval between high and low tide. For metabolite profiling, two time points (6 h and 24 h) were examined as transcriptomic changes might also translate into variations of metabolites later. Cultures were harvested by concentrating the culture on 40 µm nylon mesh filters (Cell Strainer, BD, NJ, USA), quickly drying the algae with a paper towel, and immediately freezing them in liquid nitrogen. This whole process did not take more than 1 min.

Metabolite analysis

For the analysis of amino acids, carbohydrates, polyols, organic acids and urea, samples were ground in liquid nitrogen and freeze-dried. Amino acids, non-structural carbohydrates and organic acids were extracted and quantified as follows: between 10.0 and 12.0 mg of freeze-dried sample, corresponding to approximately 50 mg fresh weight, were weighed and ground using a ball mill. The powder was suspended in 400 µL of a methanolic solution containing 100 µm of DL-3-aminobutyric acid (i.e. β-aminobutyric acid) and 200 µm ribitol, followed by 15 min of agitation at room temperature. Then, 200 µL of chloroform were added, followed by a 5 min agitation step. Finally, 400 µL of water were added, and samples were vortexed vigorously before centrifugation at 13 000 g for 5 min. Fifty µL and 200 µL aliquots of the upper phase, which contained polar metabolites including amino acids, polyols and carbohydrates, were transferred to clean vials and vacuum-dried for subsequent chromatographic analysis.

For amino acid profiling, the vacuum-dried polar phase aliquots were resuspended in 50 µL of ultra-pure water, and 10 µL was used for the derivatization using the AccQ-Tag Ultra derivatization kit (Waters, Milford, MA, USA). Derivatized amino acids were analysed using an Acquity UPLC system (Waters) according to Jubault et al. (2008), using DL-3-aminobutyric acid as internal standard. Peaks were identified according to their retention time by comparison with commercial standards. When changing the elution gradient, all of the identified peaks in the E. siliculosus samples showed identical shifts in retention time compared with the standard (data not shown), further confirming their identity.

For gas chromatography/mass spectrometry (GC-MS) profiling of non-structural carbohydrates and organic acids, the 200 µL vacuum-dried polar phase aliquots were resuspended in 40 µL of 20 g L−1 methoxyamine-hydrochloride (Sigma-Aldrich, St. Louis, MO, USA) in pyridine before incubation under orbital shaking at 30 °C for 90 min. After addition of one volume of N,O-bis(trimethylsisyl)trifluoroacetamide (Sigma-Aldrich), samples were incubated 37 min at 30 °C, transferred to glass vials and incubated at room temperature overnight before injection. GC-MS analysis was performed according to Roessner et al. (2001). The GC-MS system consisted of a TriPlus autosampler, a Trace GC Ultra chromatograph and a Trace DSQII quadrupole mass spectrometer (Thermo Fischer Scientific Inc, Waltham, MA, USA). Chromatograms were deconvoluted using the AMDIS software v2.65 ( associated with NIST libraries. Metabolite levels were quantified using ribitol as internal standard and by comparison with individual external standards.

In addition, mannitol concentrations were confirmed by an alternative method, based on an extraction in 1 mL of 70% aqueous ethanol (v/v) at 70 °C for 4 h according to Karsten et al. (1991). After centrifugation for 5 min at 5000 g, 700 µL of the supernatant were evaporated to dryness under vacuum, re-dissolved in the same volume of distilled water and analysed with an isocratic Agilent HPLC system equipped with a differential refractive index detector, a Bio-Rad resin-based column (Aminex Fast Carbohydrate Analysis, Bio-Rad, Hercules, CA, USA; 100 × 7.8 mm) and a Phenomenex Carbo-Pb2+ (4 × 3 mm) guard cartridge. Mannitol was eluted with water at a flow rate of 1 mL min−1 at 70 °C, identified by comparison of the retention time with that of a commercial standard prepared as 1 mm aqueous solution and quantified by peak area.

Urea was determined following the protocol outlined by Beale & Croft (1961). Thirty milligrams [dry weight (DW)] of sample were resuspended in 500 µL of zinc/sodium sulphate solution (9 mm ZnSO4, 84 mm Na2SO4). Then 7.5 µL of 1 m NaOH were added, and the mixture was thoroughly vortexed for 2 min before centrifugation for 10 min at 20 000 g. Two 120 µL-samples of the supernatant (technical replicates) were each mixed with 1 volume of DAM-PAA reagent (1 volume of 1% w/v diacetylmonoxime in 0.02% acetic acid, 1 volume of phenylanthranilic acid in 20% v/v ethanol with 120 mm NaCO3). One mL of activated acid phosphate (1.3 m NaH2PO4, 10 mm MnCl2, 0.4 mm NaNO3, 0.2 m HCl, in 31% v/v H2SO4) was added before incubation in boiling water for 12 min. The tubes were left to cool at room temperature and centrifuged for 1 min at 20 000 g before measuring absorption at 535 nm using a spectrophotometer. A standard curve was created using 0, 50, 100, 250 nmol urea dissolved in 10 µL of distilled water instead of the freeze-dried samples. Some Ectocarpus samples spiked with 50, 100 and 250 nmol urea were also included as positive controls to test for possible urease activity in the extract.

For the measurement of total fatty acids, approximately 400 mg (fresh weight) of sample were ground in liquid nitrogen and extracted with 2 mL of ethyl acetate as previously described (Küpper et al. 2006). As an internal standard, 250 ng of 12-OH-lauric acid were added. Extracts were evaporated under a stream of nitrogen, resuspended in 1 mL of BF3 10% in methanol for esterification and resuspended in 100 µL hexane before GC-MS analysis as described in Le Quéréet al. (2004). Fatty acids were measured only for the 6 h data point.

All metabolite concentrations were calculated per g DW, except fatty acids, which were determined as percentage of total fatty acids. As E. siliculosus is known to accumulate high concentrations of NaCl in response to salt stress, dry weights were corrected for the quantity of NaCl in each sample. This quantity was calculated based on the ratio dry weight to fresh weight (on average 20.1%) and the intracellular Na+ concentrations determined by flame photometry as described in our previous study (Dittami et al. 2009). We assumed that changes in Na+ also correspond to changes in Cl-, as this is the predominant anion in E. siliculosus, about five times more abundant than nitrate (Dittami et al. unpublished data). All analyses were performed both with and without this correction. Although the observed differences between the two analyses were only quantitative and not qualitative, we chose to show the corrected data, as it is biologically more relevant.

Genome annotation and gene expression data

Genes involved in the metabolism of the examined metabolites were manually annotated in the version 2 of the E. siliculosus genome (Cock et al. 2010), using characterized reference sequences from plants, animals and bacteria. Expression data for these genes were obtained from a previously published transcriptomic data set examining the same stress conditions (Dittami et al. 2009). This data set is available through the ArrayExpress database ( under accession number E-TABM-578.

Statistical analysis

A two-way analysis of variance (anova) with stress and time as categorical predictors was applied to the non-structural carbohydrate and the amino acid data. The latter data was square root transformed prior to the anova in order to comply with a normal distribution. A separate anova was performed to compare each of the three stress conditions to the control condition. A Bonferroni correction of the type 1 error was applied to set the familywise error rate < 0.05. As fatty acids were measured only after 6 h, a t-test was used instead of the two-way anova to compare each stress condition to the control. All tests were performed using Statistica 7 (Statsoft, Tulsa, OK, USA). Microarray data were analysed as previously described (Dittami et al. 2009), using t-tests and log2-transformed data.


Stress affects mannitol levels but not other polyols, organic acids and sugars

The concentration of mannitol varied between 15 and 300 µmol g−1 DW in all experimental conditions: it decreased strongly (by 95% after 24 h, Fig. 1) under hyposaline stress, and increased under hypersaline stress (approximately by 60% after 24 h, Fig. 1), although this increase was not statistically significant. These changes in mannitol concentration correlate with previously reported variations in the expression of two genes suggested to catalyse the first step of the synthesis of mannitol (Michel et al. 2010; Rousvoal et al. 2011): a sixfold down-regulation of the strongly expressed mannitol-1-phosphate dehydrogenase 1 (Esi0017_0062) under hyposaline stress and a fivefold up-regulation of the mannitol-1-phosphate dehydrogenase 2 (Esi0020_0181) under hypersaline stress (Dittami et al. 2009). In contrast to salinity stress, oxidative stress did not cause significant changes in mannitol concentration (Fig. 1).

Figure 1.

Ratio of mannitol between stress and control conditions. Bars represent the ratio of mannitol in the stress to the control condition (mean of three replicates ± SD). *** indicates significantly different values (alpha-error of the familywise error rate < 0.001) ^ indicates a trend (alpha-error of the familywise error rate < 0.1). hypo, hyposaline stress; hyper, hypersaline stress; oxi, oxidative stress.

GC-MS analysis of concentrated extracts allowed the detection of low levels of glycolate, glycerol, succinate, fumarate, malate, citrate, isocitrate and glucose. The concentration of these compounds ranged from a few nmol g−1 DW for malate and fumarate to over 22 µmol g−1 DW for citrate. None of these compounds exhibited significant changes in response to stress. Their concentrations are listed in Supporting Information Table S1.

Changes in the n-3 to n-6 ratio of polyunsaturated fatty acids (PUFAs) occur only in hypersaline conditions

None of the fatty acid concentrations examined changed significantly between control and stress conditions after correcting for multiple testing (see Supporting Information Table S1 for a complete profile of fatty acids). However, when considering the total amount of n-3 and n-6 PUFAs, we observed a significant decrease in the ratio of n-3 to n-6 PUFAs in hypersaline conditions (Fig. 2).

Figure 2.

Proportion of n-6 (light grey bars) and n-3 PUFAs (black bars) within total fatty acids as well as the ratio of n-3 to n-6 fatty acids (dark grey bars). Bars represent mean values of three replicates ± SD after 6 h of stress. * indicates significantly different values between the stress treatment and the control (alpha-error of the familywise error rate < 0.05). hypo, hyposaline stress; hyper, hypersaline stress; oxi, oxidative stress.

The Ectocarpus genome contains four genes possibly involved in these changes. Two of them are delta-12 desaturases, one is microsomal (Esi0207_0012) and one chloroplastic (Esi0393_0016), as well as two delta-15 desaturases, again one microsomal (Esi0231_0023) and one chloroplastic (Esi0073_0116). Delta-12 desaturases were generally up-regulated under stress, while delta-15 desaturases were generally down-regulated (Supporting Information Table S2). Interestingly, the genes coding for microsomal desaturases were regulated exclusively under hypersaline stress, while chloroplast copies were regulated in all stress conditions.

The concentration of almost all examined amino acids changed in the stress treatments

The summarized results for the amino acid profiling are shown in Table 1, and for most amino acids additional information is available in the context of metabolic pathways and of expression profiles of the genes involved in these pathways in Figs. 3–7. More detailed gene expression profiles can also be found in Supporting Information Table S2.

Table 1.  Concentrations of amino acids and related compounds. The absolute concentration of amino acids in the control condition is shown in the first column, followed by the relative changes after 6 and 24 h of exposure to stress as percentage of the concentration in the control condition (mean of three replicates ± SD). Both 6 h and 24 h samples were tested in the same anova with time and treatment (stress, control) as predictors. Boldface formatting indicates significant differences between the stress and the control treatment (alpha-error of the familywise error rate < 0.05)
 ControlHyposaline stressHypersaline stressOxidative stress
[µmol g−1 DW]6 h24 h6 h24 h6 h24 h
  1. anova, analysis of variance; GABA, γ-aminobutyric acid.

Total76.551% ± 3%71% ± 15%78% ± 7%83% ± 5%80% ± 7%82% ± 3%
Alanine23.336% ± 4%81% ± 36%66% ± 6%50% ± 2%67% ± 5%97% ± 4%
Glutamate14.830% ± 2%42% ± 3%61% ± 6%56% ± 3%77% ± 7%76% ± 7%
Glutamine10.632% ± 3%46% ± 15%69% ± 7%90% ± 15%57% ± 10%54% ± 5%
Aspartate10.572% ± 5%70% ± 8%111% ± 14%53% ± 2%123% ± 11%84% ± 4%
Asparagine0.863% ± 4%73% ± 12%104% ± 17%120% ± 8%109% ± 10%152% ± 14%
Threonine0.6173% ± 10%184% ± 54%129% ± 13%176% ± 10%125% ± 4%95% ± 3%
Valine1.0123% ± 8%165% ± 25%117% ± 8%139% ± 20%108% ± 6%98% ± 9%
Leucine0.3247% ± 18%179% ± 5%232% ± 19%343% ± 21%159% ± 5%102% ± 10%
Isoleucine0.2192% ± 12%152% ± 2%210% ± 18%283% ± 18%150% ± 5%97% ± 8%
Phenylalanine0.5120% ± 12%131% ± 8%146% ± 23%176% ± 11%130% ± 10%104% ± 4%
Tyrosine0.2203% ± 38%108% ± 6%167% ± 23%278% ± 40%165% ± 34%117% ± 27%
Tryptophan0.2145% ± 15%153% ± 2%139% ± 27%173% ± 15%118% ± 19%95% ± 12%
Arginine0.1188% ± 14%81% ± 8%323% ± 36%526% ± 46%163% ± 4%100% ± 11%
Ornithine0.1218% ± 208%73% ± 16%109% ± 43%241% ± 91%128% ± 9%52% ± 12%
GABA< 0.1130% ± 91%71% ± 28%365% ± 125%451% ± 49%117% ± 26%80% ± 25%
Proline1.468% ± 6%118% ± 54%310% ± 19%1047% ± 94%99% ± 6%78% ± 5%
Serine3.6114% ± 9%125% ± 44%58% ± 8%119% ± 15%71% ± 7%74% ± 4%
Glycine5.935% ± 2%64% ± 21%19% ± 3%26% ± 4%57% ± 8%59% ± 3%
Lysine0.4151% ± 16%98% ± 20%215% ± 11%272% ± 11%174% ± 15%109% ± 4%
Cysteine0.3103% ± 13%92% ± 121%66% ± 55%124% ± 14%130% ± 16%244% ± 25%
Methionine0.363% ± 26%48% ± 16%80% ± 31%152% ± 36%92% ± 32%90% ± 44%
Histidine0.1169% ± 227%15% ± 1%192% ± 137%52% ± 6%36% ± 4%44% ± 55%
Figure 3.

Synthesis of primary amino acids and asparagine. Bars represent the mean quantity of an amino acid (three replicates) ± SD. Lowercase letters represent enzymes as indicated below; in some cases two enzymes exhibit similar (a, b and j, k) or identical (h, i) activities. All accession numbers are available in Supporting Information Table S2. The three symbols next to the letters summarize the relative expression of this gene after 6 h of hyposaline (hypo), hypersaline (hyper) and oxidative stress (oxi) in this order. ‘↗’ indicates that a gene was up-regulated, ‘inline image’ that a gene was down-regulated, ‘–’ that there was no detectable regulation on a transcript level (<twofold change and P > 0.05), and ‘?’ that no conclusion can be drawn (see Supporting Information Table S2). Grey arrows indicate a tendency of this gene to be regulated, which was either not significant (change < twofold or P > 0.05), or not all probes followed the same profile. The statistical significance of metabolite changes after the stress treatment and with respect to the control is indicated for each stress condition under the name of the metabolite ( = P < 0.1, * = P < 0.05, ** = P < 0.01, *** = P < 0.001). ‘2x’ indicates that one molecule of glutamine produces two molecules of glutamate in reaction ‘f’. a, glutamate dehydrogenase (NAD) (EC; b, glutamate dehydrogenase (NADP) (EC; c, alanine transaminase (EC; d, aspartate transaminase (EC; e, glutamine synthetase (EC; f, glutamate synthase (EC;; g, asparagine synthase (EC; h, nitrate reductase (NADH) 1 (EC; i, nitrate reductase (NADH) 2 (EC; k, nitrite reductase [NAD(P)H] (EC; j, nitrite reductase (ferredoxin) (EC

Figure 4.

Biosynthesis and degradation of branched chain and related amino acids. Please refer to Fig. 3 for a detailed description of all symbols used. Enzymes are represented by Arabic numbers: 1, homoserine kinase (EC; 2, threonine synthase (EC; 3, threonine deaminase (EC; 4, pyruvate dehydrogenase (EC; 5, acetolactate synthase (EC; 6, ketol acid reductoisomerase (EC; 7, dihydroxy-acid dehydratase (EC; 8, isopropylmalate synthase (EC; 9, 3-isopropylmalate dehydratase (EC; 10, 3-isopropylmalate dehydrogenase (EC; 11, branched-chain amino acid transaminase (EC; 12, leucyl-tRNA synthetase (EC; 13, isoleucyl-tRNA synthetase (EC; 14, valine–tRNA ligase (EC; 15, 3-methyl-2-oxobutanoate dehydrogenase (EC; 16, dihydrolipoyl transacylase (EC; 17, isovaleryl-CoA dehydrogenase (EC

Figure 5.

Biosynthesis of aromatic amino acids. Please refer to Fig. 3 for a detailed description of all symbols used. Enzymes are represented by Roman numerals: I, 3-deoxy-7-phosphoheptulonate synthase (EC; II, 3-dehydroquinate synthase (EC; III, bifunctional 3-dehydroquinate dehydratase/shikimate dehydrogenase; IV, shikimate kinase (EC; V, 3-phosphoshikimate 1-carboxyvinyltransferase (EC; VI, chorismate synthase (EC; VII, anthranilate synthase b (EC; VIII, anthranilate phosphoribosyltransferase (EC; IX, phosphoribosylanthranilate isomerase (EC; X, indole-3-glycerol phosphate synthase (EC; XI, tryptophan synthase (EC; XII, trifunctional chorismate mutase (EC dehydratase (EC dehydrogenase (EC; XIII, aspartate aminotransferase (EC; XIV, phenylalanine–tRNA ligase (EC; XV, tyrosine–tRNA ligase (EC; XVI, tryptophan–tRNA ligase (EC

Figure 6.

Genes and amino acids related to photorespiration. Please refer to Fig. 3 for a detailed description of all symbols used. Enzymes are represented by acronyms. RuBP, ribulose-1,5-bisphosphate; RUBISCO, RuBP carboxylase (EC; PGP, phosphoglycolate phosphatase (EC; GOX, glycolate oxidase (EC; ALT, peroxisomal alanine aminotransferase (EC; GCS, glycine cleavage system.

Figure 7.

Concentrations of arginine, ornithine, glutamate, proline and γ-aminobutyric acid (GABA). Please refer to Fig. 3 for a detailed description of all symbols used. Enzymes are represented by capital letters: A, ornithine carbamoyltransferase (EC; B, argininosuccinate synthase (EC; C, argininosuccinate lyase (EC; D, arginine–tRNA ligase (EC; E, arginase (EC; F, urease (EC; G, ornithine decarboxylase (EC; H, diamin oxidase (EC; I, aldehyde dehydrogenases; J, ornithine–oxo-acid transaminase (EC; K, pyrroline-5-carboxylate reductase (EC; L, proline dehydrogenase (EC; M, 1-pyrroline-5-carboxylate dehydrogenase (EC; N, 1-pyrroline-5-carboxylate synthase.

Generally, we observed that most amino acids (with a few exceptions, e.g. arginine and proline) exhibit similar profiles in response to all of the tested stress conditions, although the most pronounced changes were clearly observed in the hypo- and hypersaline conditions. Under oxidative stress, especially after 24 h, only minor changes were observed.

The concentration of the predominant amino acids (alanine, glutamate, glutamine and aspartate), together accounting for 77.3% of the detected amino acids under control conditions, decreased in response to all of the different stressors (Fig. 3), but most strongly under hyposaline stress. They were the main reason for the decrease in the total free amino acid concentration in the hyposaline treatment (Table 1), where glutamate and glutamine, for example, decreased by 70% after 6 h. The changes in the concentration of these amino acids correspond well to the expression profiles of the genes involved in their synthesis or conversion: practically all of these genes were also down-regulated (Fig. 3, Supporting Information Table S2).

Accumulation of aromatic and branched chain amino acids

Aromatic (phenylalanine, tyrosine and tryptophan) and branched chain amino acids (valine, leucine, isoleucine) constitute only 1.1 and 2.4% of the total amino acids, respectively. Unlike the major amino acids, their intracellular concentration, together with that of threonine, consistently increased in response to the applied stress (Figs. 4 & 5). This increase was particularly strong under hyposaline and hypersaline stress, where their concentration ranged from 108% to over 300% of the concentration in the control condition. This is particularly interesting as only a single gene involved in aromatic and branched chain amino acid synthesis [a threonine deaminase (3), Fig. 4] was up-regulated under hypersaline stress, and most other genes were down-regulated (Figs. 4 & 5; Supporting Information Table S2). Moreover, genes involved in the catabolism of branched chain amino acids were clearly induced. Genes encoding t-RNA-ligases were generally down-regulated in response to stress for all amino acids.

Decrease of glycine concentration

Glycine, serine and the ratio of glycine to serine, are considered a common marker of photorespiration in terrestrial plants (Foyer, Parry & Noctor 2003). In our study, we observed a marked decrease in the ratio of glycine to serine under hypersaline and hyposaline stress, and a slight decrease in the oxidative stress condition (Fig. 6), which was likely to be a milder stressor than the saline stresses (Dittami et al. 2009). As for the accumulation of aromatic and branched chain amino acids, this does not correlate well with the transcriptomic data, which showed genes involved in the photorespiratory pathway to be generally down-regulated in hyposaline and oxidative stress conditions, and up-regulated under hypersaline stress (Fig. 6, Supporting Information Table S2). As glycine might serve as a substrate for glutathione synthesis, we also investigated the expression of the two γ-glutamylcysteine synthetases (EC and of the single copy of a glutathione synthetase (EC in the Ectocarpus genome. One of the glutamylcysteine synthetases (Esi0184_0033 corresponding to LQ0AAB47YA16FM1.SCF on the microarray) was 1.9 to 3.9-fold down-regulated in all stress conditions, while the second copy (Esi0250_0012 corresponding to LQ0AAB39YO20FM1.SCF) was 2.6-fold up-regulated only in hypersaline conditions. The glutathione synthetase (Esi0066_0082, corresponding to CL1154Contig1) was not significantly regulated in any of the examined stress conditions (P > 0.05, fold-change < 2).

GABA and proline accumulate specifically in hypersaline conditions

The strongest changes in amino acid concentrations were the increase of proline (10-fold) and arginine (fivefold) in the hypersaline condition. This increase was parallel to an induction of genes involved in arginine degradation (Fig. 7, Supporting Information Table S2). Cellular GABA content, which was close to the detection limit in control samples, clearly increased under hypersaline stress, but overall levels remained low. A diamine oxidase [(H), Fig. 7], potentially involved in conversion of putrescine to 4-amino-butyraldehyde (an immediate precursor for the synthesis of GABA) and an aminoaldehyde dehydrogenase [(I), Fig. 7], potentially involved in the conversion of 4-amino-butyraldehyde to GABA (Sebela et al. 2000), were significantly induced in hypersaline stress conditions.

Urea and trehalose concentrations were below the detection limit

No traces of urea were detected in any of the samples, although spiked samples showed normal absorption comparable with those of our standards. We assume that our extraction method of thorough grinding in liquid nitrogen is sufficient to break the Ectocarpus cells – an assumption that is supported by the fact that nucleic acids are efficiently extracted using a similar protocol, and that centrifugation for non-freeze-dried but ground E. siliculosus leads to a separation of solid particles and intracellular medium after 5 min at 20 000 g. As our method showed clear differences between the 100 nmol standard or Ectocarpus samples spiked with 100 nmol urea and the blanks, we can conclude that urea, if present in Ectocarpus, is only present in quantities smaller than the detection limit of the method, that is 1.66 µmol g−1 DW.

Similarly, trehalose was not detected in our samples, even after 24 h of exposure to hypersaline stress (detection limit = 1 nmol g−1 DW). In addition, neither of the two trehalose 6-phosphate synthase/trehalose 6-phosphate phosphatase genes (Michel et al. 2010) represented on the microarray (Esi0039_0126 corresponding to CL80Contig7; and Esi0154_0054 corresponding to LQ0AAB78YK12FM1.SCF) were up-regulated in this stress condition, although for Esi0039_0126 a 2.6- to 3.3-fold induction was observed under hyposaline and oxidative stress, respectively.


In this study we related metabolite changes occurring in E. siliculosus submitted to three sub-lethal abiotic stress treatments to previously published transcriptional data (Dittami et al. 2009). In this previous report, samples were shown to undergo profound transcriptomic changes, including a general down-regulation of genes involved in primary metabolism and an activation of genes in the autophagy and protein degradation pathways. In addition, the expression profiles of several genes led to the hypothesis that mannitol and urea might function as compatible osmolytes in E. siliculosus. However, these hypotheses remained to be verified, considering that several studies in terrestrial plants have highlighted significant differences between total transcript abundance and polysomal mRNA abundance (Branco-Price et al. 2008), enzyme activity (Gibon et al. 2004, 2006) and final metabolite content (Kaplan et al. 2007; Kempa et al. 2008).

Changes in the primary metabolism affect mainly amino acid concentrations

Interestingly, the global down-regulation of primary metabolism-related genes in response to the tested stress conditions (Dittami et al. 2009) had no significant effect on the concentration of glucose or the examined organic acids in our samples, except for mannitol. However, the overall concentration of the former compounds in the cell was very low, indicating that they might serve mainly as intermediates in chemical reactions. In any case, the absence of changes in the concentrations of these compounds should not be interpreted as proof of constant metabolite flow through a pathway under stress.

In contrast, we observed a general decrease in the predominant amino acids alanine, glutamate, glutamine and aspartate, which was strongest under hyposaline stress (Table 1). These changes corresponded to the down-regulation of genes involved in NH4+ assimilation and the synthesis of primary amino acids both in the GS/GOGAT pathway (enzymes e, f; Fig. 3) and in the glutamate dehydrogenase pathway (enzymes a, b; Fig. 3). Similar observations were made in some terrestrial plants, for example in spinach leaves under mild salt stress (Di Martino et al. 2003). However, many plants such as Arabidopsis and Thellungiella (Gong et al. 2005; Kempa et al. 2008) as well as grapevine (Cramer et al. 2007) show constant levels or even an increase in the concentration of at least some of these amino acids under saline stress. Moreover, compared with Arabidopsis and grapevine, E. siliculosus seems to be able to tolerate far higher changes in the concentrations of these primary metabolites.

Branched chain and aromatic amino acids were present only in small quantities in Ectocarpus cells (about 3.5% of total free amino acids, Table 1), but they are likely to be comparatively abundant in the algal proteome (about 24% of the codons in the predicted coding sequences of the Ectocarpus genome code for these amino acids). Interestingly, their concentration in response to stress increased while genes involved in their synthesis were down-regulated. A similar increase was observed in terrestrial plants in response to abiotic stresses (e.g. Kaplan et al. 2004). In Ectocarpus, the reasons for these changes are likely to be both a decrease of their consumption for protein synthesis, suggested by the down-regulation of genes involved in protein synthesis, including aromatic amino acid aminoacyl t-RNA-ligases (Figs. 4 & 5), and an increase in their liberation through processes related to protein turnover and autophagy, supported by the up-regulation of genes related to these processes (Dittami et al. 2009). In this context, the induction of the branched chain amino acid degradation pathway under the tested stress conditions (Fig. 4) could represent a mechanism of keeping the intracellular levels of branched chain amino acids non-toxic, as proposed by Malatrasi et al. (2006).

Glycine, serine and glutathione metabolism

Glycine is a key metabolite in the photorespiratory process and is synthesized in large amounts in the peroxisomes. It can be oxidized and converted to serine in the mitochondria. The ratio of glycine to serine is considered a marker of photorespiration both in terrestrial plants (Foyer et al. 2003) and in Ectocarpus (Gravot et al. 2010). In terrestrial plants, photorespiration and thus the ratio of glycine to serine, generally increases in response to stress (Gong et al. 2005; Cramer et al. 2007) although exceptions, such as a study by Kempa et al. (2008) showing a temporary decrease of this ratio in the first three days of the A. thaliana salt stress response, exist.

In our study, E. siliculosus exhibited a decrease in glycine content and in the ratio of glycine to serine mainly in response to saline stress. This may be a direct consequence of a general reduction of photosynthesis and of the down-regulation of genes that code for parts of the light harvesting complex (Dittami et al. 2009), which would consequently reduce the amount of both photorespiratory and non-photorespiratory glycine.

Another possibility is that the decrease in glycine, which accounted for most of the changes in the glycine to serine ratio, was related to the synthesis of glutathione. Noctor et al. (1999) described that photorespiratory glycine can enhance glutathione production in poplar, thus proving the existence of a metabolic link between these two pathways. Glutathione is an antioxidant, which, although implied in a number of different processes, is known to accumulate in several plants (Tausz, Sircelj & Grill 2004) during the acclimation to stress. A net increase in glutathione is, however, usually only observed after several days of acclimation. In tomato plants, glutamylcysteine synthetase activity was demonstrated to be an important regulator of glutathione content under salt stress (Mittova et al. 2003). In Ectocarpus, one glutamylcysteine synthetase was transcriptionally activated under hypersaline stress while the second one was down-regulated, therefore neither supporting nor weakening the hypothesis that glutathione synthesis may be related to the observed changes in glycine and serine. Additional experimental evidence, in particular measurements of glutathione content, will be required to further investigate this hypothesis.

Changes in the total fatty acid composition

In parallel to these general changes in the amino acid composition, Ectocarpus displayed several changes specifically during hypersaline stress, such as a shift from n-3 to n-6 PUFAs (Fig. 2), and corresponding changes in gene expression. In plants, changes in the fatty acid composition of membranes are commonly observed in response to stress, mainly after temperature changes (e.g. Sanina, Goncharova & Kostetsky 2008), but also after exposure to osmotic stress, as reported for grape leaves (Aziz & Larher 1998). These changes are thought to enhance membrane integrity and fluidity in different abiotic conditions.

Proline and mannitol, but not urea or trehalose, are potential osmoprotectants or local osmolytes

Other changes that occurred specifically in hypersaline stress were related to possible osmolytes. Our previous study (Dittami et al. 2009) demonstrated that the major solute responsible for the short-term adaptation of Ectocarpus to high salinities was NaCl. Expression profiles, however, suggested that additional osmolytes such as urea, mannitol and proline might also play important roles.

In this study, urea was not detected in any of the examined conditions, strongly suggesting that this compound does not accumulate in E. siliculosus cells, and excluding its role as osmolyte, at least within the first 24 h of acclimation to different salinities. Similarly, despite the presence of the trehalose biosynthetic pathway in the E. siliculosus genome (Michel et al. 2010), no trace of this non-reducing disaccharide was detected after stress treatments, confirming previous results obtained during the diurnal cycle and under altered growth conditions (Gravot et al. 2010). In this respect, Ectocarpus is similar to vascular plants, where, with a few exceptions, trehalose levels are usually low or below detection limit, in spite of the presence of functional genes encoding the enzymes of trehalose synthesis (reviewed in Müller et al. 1995, 2001). The role of these genes is not yet fully understood, but a recent study in Arabidopsis suggested that trehalose-6-phosphate (TP6) may act as a regulator of primary metabolic pathways (Zhang et al. 2009) at very low concentrations (1–20 µm). T6P quantification, which requires specific analytical developments (Lunn et al. 2006), may therefore help to elucidate the role of trehalose metabolism in Ectocarpus in future studies.

Proline, which is strongly accumulated in response to hypersaline stress, may be synthesized both from arginine via ornithine and from glutamate (Fig. 7). Unfortunately, the transcriptomic data do not provide clear indications as to which of these pathways is activated. Furthermore, even after 24 h of exposure to hypersaline stress, the overall proline concentration reaches only about 18 µmol g−1 DW. Considering that 1 g of dry weight corresponds roughly to 5 g of fresh weight, this is equivalent to an average intracellular concentration of only about 4.5 mm. Facing changes in the extracellular osmolarity of about 2000 mOsmol L−1, the contribution of proline to the adjustment of the overall intracellular osmolarity would be negligible. To a lesser extent this is also true for mannitol, which, under the same assumptions, can account for about 75 mOsmol L−1 of the total intracellular osmolarity after 24 h of exposure to stress.

Nevertheless, both mannitol and proline might be important osmolytes for longer-term acclimation (>1 d). This was illustrated by Thomas & Kirst (1991), who observed mannitol concentrations of up to 100 mm after 1 week of exposure of E. siliculosus to high salinity. Furthermore, mannitol is presumably not evenly distributed throughout the cell but localized in the cytoplasm as has been proposed for marine brown algae (Davison & Reed 1985; Reed et al. 1985) and also for the red macroalga Caloglossa leprieurii (Mostaert, Karsten & King 1995). In the short-term salt stress response, both mannitol and proline might therefore still function as local osmolytes in specific cell compartments. In the case of mannitol, this hypothesis is supported by the transcriptional regulation of the two examined mannitol-1-phosphate dehydrogenase genes, which code for enzymes involved in the synthesis of this compound (Michel et al. 2010; Rousvoal et al. 2011). In addition, both proline and mannitol were reported to act as oxygen radical scavengers (Smirnoff & Cumbes 1989), and to increase protein stability (Soderquist & Lee 2005; Takagi 2008), and are therefore likely to act also as osmoprotectors.

GABA: a potential signalling compound

GABA is closely related and structurally similar to proline, and is considered a neurotransmitter in the mammalian central nervous system (Li & Xu 2008). It is also known to have many different functions in plants, including a pivotal role in carbon and nitrogen metabolism (Fait et al. 2008), stress response (Kinnersley 2000) and signalling (Bouché & Fromm 2004). In A. thaliana GABA is accumulated under saline stress (e.g. Kempa et al. 2008), and a loss of function mutant in the GABA transaminase (GABA-T), the first step in GABA catabolism, was shown to be oversensitive to salt but not to other osmotic stress. This accumulation is thought to be mediated by post-translational activation of a glutamate decarboxylase (GAD), probably induced by an increase in Ca2+ (Renault et al. 2010); yet, despite ongoing efforts, the exact role of GABA in the stress response of terrestrial plants is still unknown.

In a previous study (Gravot et al. 2010), we proposed that in E. siliculosus the absence of GAD and GABA-T genes is likely to be responsible for the very low levels of GABA found in this organism. However, despite the absence of these genes, E. siliculosus exhibited a marked increase in GABA concentration under hypersaline stress, although absolute levels remained low (< 0.05 µmol g−1 DW) when compared with about 8 µmol g−1 DW classically observed in stressed leaves of Arabidopsis (Renault et al. 2010). This raises the question of the biosynthetic pathway for GABA in E. siliculosus. In terrestrial plants, GABA can also be derived from polyamines such as putrescine (Terano 1978; Kumar & Thorpe 1989; Rastogi & Davies 1990; Bouchereau et al. 1999; Petrivalskýet al. 2007). Parallel to the increase in GABA concentration under hypersaline stress, our data show the up-regulation of genes likely to be involved in its synthesis from putrescine.

This observation provides a first indication that GABA synthesis in Ectocarpus may occur from polyamines, and that, as in terrestrial plants, this compound may play a role in stress signalling in this alga. From an evolutionary point of view, it is interesting to compare the occurrence of the pathways involved in the production of GABA in different phylogenetic lineages, as GABA seems to have conserved or convergently acquired similar functions in green plants, heterokonts and, to a certain extent, also metazoans.


Our study presents extensive metabolite profiling under saline and oxidative stress, as well as manual genome annotation facilitating the integration of the metabolite data with existing transcriptomic data obtained under the same conditions. It demonstrated several metabolic changes in response to stress, most of which correspond well to observations made in the transcriptome. Nevertheless, caution needs to be taken when interpreting transcriptomic data, especially in the context of complex metabolic networks.

Our study allowed us to reject the hypothesis of urea as compatible osmolyte in brown algae, and showed that, in the short-term response to salt stress, proline and mannitol concentrations are too low to support a role as primary osmolytes in the entire cell. Moreover, it demonstrated changes in PUFA composition and that, even though the primary pathway for GABA synthesis was not found in the Ectocarpus genome, GABA synthesis is likely to take place from polyamines in response to hypersaline stress. These findings pave the way for further studies of the biosynthetic pathways of these compounds and of the role of these potential signalling molecules, including their implication in the abiotic stress response in brown algae.


We would like to thank Hugues Renault and François Larher for helpful discussions, and acknowledge Rennes Métropole and the Région Bretagne for financial support for the acquisition of the UPLC equipment. SMD received funding from the European community's Sixth Framework Programme (contract no. MESTCT 2005-020737).