Infection of the brown alga E ctocarpus siliculosus by the oomycete E urychasma dicksonii induces oxidative stress and halogen metabolism

Abstract Pathogens are increasingly being recognized as key evolutionary and ecological drivers in marine ecosystems. Defence mechanisms of seaweeds, however, have mostly been investigated by mimicking infection using elicitors. We have established an experimental pathosystem between the genome brown model seaweed E ctocarpus siliculosus and the oomycete E urychasma dicksonii as a powerful new tool to investigate algal responses to infection. Using proteomics, we identified 21 algal proteins differentially accumulated in response to E u. dicksonii infection. These include classical algal stress response proteins such as a manganese superoxide dismutase, heat shock proteins 70 and a vanadium bromoperoxidase. Transcriptional profiling by qPCR confirmed the induction of the latter during infection. The accumulation of hydrogen peroxide was observed at different infection stages via histochemical staining. Inhibitor studies confirmed that the main source of hydrogen peroxide is superoxide converted by superoxide dismutase. Our data give an unprecedented global overview of brown algal responses to pathogen infection, and highlight the importance of oxidative stress and halogen metabolism in these interactions. This suggests overlapping defence pathways with herbivores and abiotic stresses. We also identify previously unreported actors, in particular a Rad23 and a plastid–lipid‐associated protein, providing novel insights into the infection and defence processes in brown algae.


INTRODUCTION
Macroalgae represent important components of cold and temperate rocky shore communities. Comparable with other organisms, they are subject to numerous biotic stresses Gachon et al. 2010). Those stressors include viruses, bacteria, fungi, oomycetes, chytrids and algae in the form of endo-or epiphytes. The exploration of pathogens in Rhodophyceae, Chlorophyceae and Phaeophyceae macroalgae is steadily growing and numerous interactions between algae and stressors have been described over the last two decades. Algal endophytism is a prominent example, with the pathosystem between the red alga Chondrus crispus and the endophyte Acrochaete operculata being especially well described. Whereas the sporophytic phase of Chondrus is susceptible to infection by the endophyte, the gametophyte phase is resistant due to a different cell wall composition (Bouarab et al. 1999;Weinberger et al. 2005). With the recent completion of the C. crispus genome (Collén et al. 2013), new lines of research could be developed on this pathosystem.
The brown algal endophytes Laminariocolax aecidiodes and Laminarionema elsbetiae infect several kelp species via zoospores that germinate on the host surface and then penetrate its host (Heesch & Peters 1999). The oomycete Olpidiopsis porphyrae and the chytrid Pythium porphyrae challenge the economic important seaweed Pyropia (Porphyra) (Uppalapati & Fujita 2000;West et al. 2006). Recently, it was shown that recognition and attachment of the pathogenic zoospore to its host most likely involve lectincarbohydrate binding (Klochkova et al. 2012). Despite the increasing knowledge on macroalgal host-pathogen interactions, algal response and defence reactions to such pathogens on a molecular level are little studied. Moreover, both in brown and red algae the current knowledge of algal response to pathogen attack in many instances derives from shortterm (1 h time scale) experiments on simplified models with elicitors (e.g. Weinberger & Friedlander 2000;Küpper et al. 2001). In particular, algal cell wall-derived oligosaccharides, free fatty acids, methyl jasmonate and bacterial lipopolysaccharides were shown to induce an oxidative burst in the kelp Laminaria digitata (Hudson) J.V. Lamouroux (Küpper et al. 2001(Küpper et al. , 2006a). This accumulation of reactive oxygen species (ROS) is considered both a direct antimicrobial defence reaction and an activator of downstream defence mechanisms such as oxylipin accumulation and the production of halogenated secondary metabolites (reviewed by Weinberger 2007). Since their original identification, elicitors have widely been used to mimic infection in Laminaria digitata to gain a deeper understanding of the molecular processes underlying defence responses Cosse et al. 2009;Goulitquer et al. 2009). These studies suggested a role for aldehyde accumulation in algal defences and identified defence-related transcripts. Among the latter, vanadium-dependent haloperoxidases were prominent and showed a tight regulation of expression. Despite the important defence and signalling role of ROS in Laminaria digitata, it appears that ROS-independent signal transduction also occurs during the defence response ). The elicitor-based defence response evokes a priming effect in Laminaria digitata (Thomas et al. 2011). Attacked individuals send out waterborne cues that condition neighbouring plants for a faster gene induction.
The capability to induce an oxidative burst by oligoguluronate elicitation varies between different brown algal species and notably is restricted to the sporophytic life stages of the Laminariales (Küpper et al. 2002). Whether similar signal transduction pathways occur in brown algal species where no oxidative burst has been observed is currently unknown and it appears plausible that those species might have different defence pathways (Zambounis et al. 2013).
One of those species is the filamentous seaweed Ec. siliculosus (Dillwyn) which, over the last few years, has emerged as brown algal genome model, culminating in the recently completed genome project (Cock et al. 2010(Cock et al. , 2012. The development of tools such as microarrays and proteomics enables a comprehensive analysis of changes at the transcriptional and protein level. Importantly, this alga is host to the obligate-biotrophic oomycete pathogen Eurychasma dicksonii (E.P. Wright) Magnus.
In coastal ecosystems, Eu. dicksonii has been recorded in epidemic outbreaks throughout the world, and an exceptionally broad host spectrum of over 40 algal species is documented (Küpper & Müller 1999;Gachon et al. 2009;Strittmatter et al. 2009. It is an intracellular pathogen for most of its life cycle, and in contrast to eucarpic counterparts, infects single algal cells . It represents one of the very few eukaryotic algal pathogens that can be cultivated under laboratory conditions. Therefore, the Eu. dicksonii-Ec. siliculosus pathosystem is among the very few marine algal models amenable to molecular investigation Grenville-Briggs et al. 2011). Eu. dicksonii is particularly interesting due to its basal phylogenetic position in the oomycete tree (Küpper et al. 2006b;Sekimoto et al. 2008), many of which are devastating pathogens in aquaculture (e.g. Saprolegnia, Aphanomyces) and agriculture (e.g. Phytophthora, Plasmopara, Pythium).
Yet, so far, the brown algal response to pathogen infection has never been investigated on a real pathosystem using a comprehensive, global 'omics' approach. In this context, we chose to investigate the interaction between Ec. siliculosus and Eu. dicksonii via 2-DE-based comparative proteomics and histochemical studies with the aim of identifying the cellular processes involved in the disease, and the algal responses to biotic stress. Proteomic tools have already been used in a number of studies investigating host-pathogen interactions in various systems like angiosperms and vertebrates including humans (e.g. Kav et al. 2007;Bhadauria et al. 2010;Roy et al. 2010;Schmidt & Völker 2011). In contrast to transcriptomic analysis, proteomics reflect more closely the abundance or activity of effector molecules responsible for a given phenotype (Tan et al. 2009).
In the present study, we present novel insights derived from a real host-pathogen interaction and show the involvement of ROS and putative halogen metabolism as response of Ec. siliculosus to pathogen infection.
For the proteomic investigation 50 mg fresh weight (FW) of Ec. siliculosus was distributed into 140 mm diameter Petri dishes containing 110 mL of one-half strength modified Provasoli-enriched natural seawater (PES; West & McBride 1999). The M. pyrifera female gametophyte was chosen as a convenient alternate host for the mass production of  Eu. dicksonii inoculum because it is highly susceptible to the oomycete pathogen and does not produce any motile zoospores. Infected M. pyrifera was placed in two 70 μm pore-sized cell strainers (Falcon), a set-up that avoids contamination of the target inoculated alga (here Ec. siliculosus) with infected inoculum (Fig. 2). The progress of infection was regularly monitored using an inverted bright field microscope. Controls were performed to check that the cell strainers allowed the free passage of Eu. dicksonii spores into the surrounding medium containing Ec. siliculosus ). Mock-inoculated controls (uninfected Ec. siliculosus) were prepared in parallel by co-incubating uninfected M. pyrifera with 50 mg of Ec. siliculosus.Throughout the incubation time 80% of the medium was replaced on a 3 week basis in order to avoid nutrient depletion. The algal material was harvested when the highest possible density of symptoms was observed (approximately 1 algal cell in 10 infected, typically after 6-8 weeks). Ec. siliculosus was soaked dry on filter paper, weighed, frozen in liquid nitrogen and stored at −80°C. Four Petri dishes of either Eu. dicksonii-or mock-inoculated algal material were pooled yielding 180-200 mg (FW) for each of the mock and infected samples. In total three independent experiments were performed.

Protein extraction, quantification and clean-up for two-dimensional electrophoresis
The protein extraction protocol developed by Contreras et al. (2008) was optimized to increase protein yield, as detailed in Supporting Information Table S1. Proteins were quantified using the 2-D Quant™ kit (GE Healthcare, UK) according to the instruction manual (standard procedure). The protocol is based on the precipitation of polypeptides prior to quantification in order to remove interfering substances. Quantification is achieved by colorimetric detection of unbound cupric ions. The overall accuracy of the quantification was improved by including an additional precipitation step before adding the copper solution to the pellet.
Three technical replicates were run per protein extract. Prior to running the second dimension (SDS-PAGE), the IPG strips were incubated in equilibration buffer [0.05 m Tris/ HCl pH 8.8, 6 m urea, 30% (v/v) glycerol, 2% (w/v) SDS, 0.002% (v/v) bromophenol blue solution] containing 10 mg mL −1 DTT for 25 min and subsequently in equilibration buffer containing 25 mg mL −1 iodoacetamide for 15 min. The equilibrated strips were then placed on a single well 12% Bis-Tris Mini gel (NuPAGE Novex, Invitrogen, Carlsbad, CA, USA) and fixed with 0.35% (w/v) molten agarose NA (GE Healthcare, UK). Electrophoresis was performed in MOPS-SDS running buffer (NuPAGE, Invitrogen) with a two-step protocol (70 V for 30 min; 200 V for 55 min). Next, gels were fixed in 50% (v/v) MeOH, 7% (v/v) acetic acid for 10 min under shaking, followed by two washing steps in MQ-H2O and were then incubated in GelCode Blue stain (Pierce, Thermo Scientific, USA) overnight. Gels were washed in MQ-H2O to remove unbound stain and were scanned on an ImageScanner III (GE Healthcare, UK) using a resolution of 600 dots per inch (transparent mode, bits depth: 16 bits per channel, green filter). Gels were preserved by incubation in 0.5% (v/v) glycerol for 1 h with subsequent drying in cellophane until spot cutting.

Image analysis
Image alignment, background subtraction and normalization were automatically performed using the Progenesis SameSpot™ software v3.0 (Nonlinear Dynamics, UK). Spot alignment and detection were manually improved as necessary. Spot densities were expressed as mean normalized volumes and fold changes between infection and control material were calculated. Based on the program's statistical analysis (one-way analysis of variance), spots with a P-value <0.05 and a fold change of ≥1.5 were selected for subsequent mass spectrometry.
In total, nine 2-DE gels of control protein extracts and 2-DE gels of infection protein extracts were analysed originating from three independent experiments.

In-gel digestion and liquid chromatographytandem mass spectrometry (LC-MS/MS)
Spot cutting and the tryptic digestion of proteins and LC-MS/MS analysis was performed as previously described (Grenville-Briggs et al. 2005, 2008. Peptide peaks were detected and deconvoluted automatically using the DataAnalysis™ software (Bruker, USA). Mass lists in the form of Mascot Generic Files were created automatically and used as the input for Mascot MS/MS Ions searches of the NCBI nr database and the Ec. siliculosus protein database (version 'Ectsi_Prot_latest ' Cock et al. 2010) on the Matrix Science web server (www.matrixscience.com). The default search parameters used were as follows: enzyme = trypsin, max missed cleavages = 1; fixed modifications = carbamidomethyl (C); variable modifications = oxidation (M); peptide tolerance ± 1.5 Da; MS/MS tolerance ±0.5 Da; peptide charge = 2 + and 3 + ; instrument = ESI-TRAP. Probability-based MOWSE scores above the threshold value (P < 0.05) were considered for protein identification. The mass spectrometry proteomics data and accompanying experimental conditions have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository (Vizcaíno et al. 2013) with the dataset identifier PXD001458.

RNA extraction and transcriptional analysis via quantitative real-time PCR
As for the proteomics analysis, Eu. dicksonii-infected Ec. siliculosus tissue was harvested for subsequent RNA extraction once a dense infection in the late thallus stage was visible. Extraction of RNA, DNaseI treatment and cDNA synthesis was performed as previously described (Zambounis et al. 2013).
Oligonucleotides targeting selected genes were designed using the Beacon Designer (Premier Biosoft) for SYBR Green PCR assays with the following parameters Tannealing 60 ± 1°C; length: 18-24 bp; amplicon length: 75-200 bp. A BlastN search of the primer sequences against the Ec. siliculosus nucleotide database (Cock et al. 2010) and a local Eurychasma EST database (Grenville-Briggs et al. 2011) was performed to rule out possible cross-reactions between genes. Their sequences are given in Supporting Information Table S2.
Real-time PCR was run in three technical replicates (20 μL each), with 1× Mesa Green Mastermix (Eurogentec, Seraing, Belgium), 300 nM of each primer and 2 ng of cDNA described in Zambounis et al. (2013). Relative expression ratios of tested genes were normalized against the constitutive reference gene EEF1α2 (Le Bail et al. 2008;Dittami et al. 2009). The relative expression ratios were calculated based on the PCR efficiencies and the crossing point [C(t) values] differences of each sample versus a mock-treated (control) sample using the geometric mean of the technical replicates. The statistical significance of the results was tested with the pairwise fixed reallocation randomization test (2000 iterations) implemented in the software REST 2009 (Pfaffl et al. 2002). Whether a gene was significantly induced during any one treatment was judged by calculating the probability [P(H1)] that the observed difference between the sample and control groups was due only to chance [P(H1) < 0.05]. Five independent biological experiments were performed. Each encompassed two treatments (infected and mock-inoculated control).

Detection of ROS
In order to detect the accumulation of hydrogen peroxide in Eurychasma-infected Ectocarpus tissue, the cultures were incubated in 3,3-diaminobenzidine (DAB) dissolved in PES at a concentration of 1 mg mL −1 overnight at 15°C in the dark. The following day, the samples were destained two times in 75% (v/v) ethanol and once in 100% (v/v) methanol for 5 min at 95°C and observed by differential interference contrast (DIC) microscopy. Images were taken with an AxioCam™ HR3 (Zeiss, Germany). For the detection of superoxide radicals (O2 -), DAB was replaced by nitrotetrazolium blue (NBT) at a final concentration of 1 mg mL −1 . The superoxide dismutase inhibitor diethyldithiocarbamate (DDC) was used as previously described (Küpper et al. 2001). Briefly, uninfected and Eu. dicksonii-infected algal tissue was incubated in 1 mm DDC (from a 100 mm stock solution dissolved in ethanol) for 3 h at 15°C in the dark before addition of NBT or DAB as described earlier. As positive controls, uninfected Ec. siliculosus was incubated under medium light (30 μE m −2 s −1 ). Experiments were performed at least three times independently.

Improved protein extraction protocol and protein map of uninfected Ec. siliculosus
Brown seaweeds are a challenging biological material for protein extraction, typically leading to rather poor yields.
Due to the intracellular nature of the pathogen, the amount of infected tissue was limiting to perform the proteomics analysis. Therefore, a previously established protocol (Contreras et al. 2008) was modified (detailed in Supporting Information Table S1), leading to increased protein yields of 2-10 μg mg −1 FW compared with 0.5-1 μg mg −1 FW with the initial protocol. Hence, a total amount of 200 mg algal material was sufficient to run at least three 7 cm 2-DE gels (150 μg total protein per gel).
Approximately 600 Ec. siliculosus proteins were resolved by 2-DE. In order to obtain an overview of the proteins expressed in healthy Ec. siliculosus tissue, 26 protein spots from control gels were selected across acidic, basic, high and low molecular weight ranges and were subjected to LC-MS/MS analysis (Supporting Information Fig. S1a). All proteins were matched to Ec. siliculosus sequences (Supporting Information Table S3). They represent a broad spectrum of housekeeping proteins expected to be expressed in healthy, uninfected Ectocarpus cells, such as photosynthesisrelated proteins (e.g. light-harvesting complex protein, cytochrome b6f complex and ferredoxin-NADP oxidoreductase). Furthermore, several enzymes involved in primary metabolism were identified, for example, pyruvate carboxylase (regeneration of citric acid cycle metabolites), glyceraldehyde-3-phosphate dehydrogenase (Calvin cycle), transketolase (Calvin cycle and oxidative pentose pathway) and triosephosphate isomerase/glyceraldehyde-3-phosphate dehydrogenase (glycolysis). Note that several of these proteins are integral to the membrane (e.g. light-harvesting complex protein, cytochrome b6f complex) indicative of efficient protein recovery and good overall representation.

Eurychasma infection
The densely infected Ec. siliculosus biomass contained mostly intracellular pathogenic thalli and a few mature dehiscent Eu. dicksonii sporangia (Fig. 1c,d). During this stage, the host nucleus and microtubules are still intact (Tsirigoti et al. 2013(Tsirigoti et al. , 2014, and a therefore an active host response can be monitored. The comparison of protein profiles revealed 21 spots whose relative abundance varied significantly between healthy and infected Ec. siliculosus (Supporting Information Figs S1b & S2). Among these, 18 protein spots showed a significantly higher expression compared with the control and three protein spots a lower expression (Table 1). All spots were attributed to Ec. siliculosus proteins, in agreement with the observation that no newly expressed protein spots were detected from the gels of infected algal material.
The spot with the highest differential expression (spot ID 11; fold change 2.2) had highest matches to a Rad23-like protein/UV excision repair protein. Two protein spots (spot ID 23 and 24) with a 1.9-fold induction in infected material were attributed to a manganese superoxide dismutase (MnSOD) and a vanadium-dependent bromoperoxidase (vBPO), respectively. Other noticeable induced proteins were members of the heat shock protein 70 (HSP70) family [spot ID 31 and spot ID 33 (HSP70_1)], a disulfide isomerase (spot ID 66), a chaperonin (spot ID 120) and an aspartyl protease (spot ID 48). Interestingly, several plastidial proteins were induced: a plastid-lipid-associated protein (PAP) matched two spots (spot IDs 46 and 53) with slightly different isoelectric points, which suggests different posttranslational modifications (PTM) as well as another PAP/fibrillin domaincontaining protein (PAPfib, spot ID 61). Different PTMs might also be attributed to spot IDs 36 and 130, which match to a chloroplast-localized protein of Ectocarpus that is highly supported by EST data (Cock et al. 2010) but does not have any conserved domains. Spot ID 88 showed a positive fold change of 1.5; the conserved domain within this protein was assigned to the α-subunit of a nascent polypeptide-associated complex.

Transcript profiling of selected stress marker genes by qPCR
Five genes encoding proteins differentially expressed during infection were selected for transcriptional profiling: vBPO, MnSOD, HSP70_1, Papfib and Rad23 (Fig. 3).
The vBPO gene was significantly up-regulated (P < 0.05) in 4 out of 5 independent experiments and showed a fold change from 7.5 to 35. HSP70_1 was induced in two independent experiments (fold change from 1.4 to 16) but repressed in one experiment (fold change 0.5). In contrast to our proteomics data, transcript levels of MnSOD were significantly reduced in four experiments (expression ratio between 0.2 and 1.4). The remaining two genes Rad23 and Pabfib were down-regulated in two independent samples with a fold change between 0.07 and 0.9 for Rad23 and 0.18 and 0.27 for Pabfib, whereas in the remaining three independent experiments those two genes were not differentially expressed.

Infected Ec. siliculosus and Eu. dicksonii produce ROS
Because the proteomic analysis revealed two proteins, MnSOD and vBPO, potentially involved in the detoxification of ROS, we further investigated the presence of ROS in infected tissue via DAB and NBT histochemical staining. The former compound is oxidized in the presence of hydrogen peroxide (H2O2) by peroxidases, forming a brown precipitate. The latter is reduced in the presence of superoxide radicals (O2 -•), forming a blue precipitate. DDC is an inhibitor of SOD that catalyses the conversion of O2 -• into H2O2. As a positive control, we checked that inhibition of SOD with DDC leads to a build-up of O2 -• in the chloroplasts of uninfected Ec. siliculosus cells during exposure to moderate light (30 μE m −2 s 1 , Fig. 4a), whereas only a weak accumulation of O2 -• is detected in the absence of DDC (Fig. 4b). Conversely, a systemic plastidial generation of hydrogen peroxide is detected with DAB in those samples (Fig. 4d), and lost in DDC-treated samples (Fig. 4c). A weaker, yet still detectable build-up of O2 -• is likewise observed in dark-incubated, DDC-treated, uninfected algal tissue (Fig. 4e). In the absence of DDC, the observed superoxide build-up is even less, in agreement with the reported short life span of this free radical (Fig. 4f). In contrast, the build-up of H2O2 in dark-incubated, DDCuntreated and uninfected Ec. siliculosus was predominantly restricted to apical cells (Fig. 4h). In the corresponding DDCtreated controls, H2O2 could not be detected at all (Fig. 4g). Ec. siliculosus mitospores within plurilocular sporangia were also a light-independent source of ROS (Fig. 4i,j).
Infection triggered a mild increase in the lightindependent superoxide accumulation on top of the constitutive production described earlier (Fig. 5a). O2 -• was generally produced in Eu. dicksonii secondary zoospores within the pathogen sporangium (Fig. 5b). During earlier, intracellular infection stages, NBT staining of infected algal cells was more heterogeneous. Algal-derived blue labelling was sometimes observed around young walled Eu. dicksonii thalli (Fig. 5b-d), but also within the thalli (Fig. 5d), indicating that during this infection stage the alga as well as the pathogen may generate superoxide. In some instances, increased cytoplasmic superoxide production was also seen in unchallenged Ec. siliculosus cells (Fig. 5e).
Eu. dicksonii-infected host cells were DAB positive (Fig. 5f), and the staining was totally abolished in the presence of DDC, indicating that superoxide dismutation is the source of H2O2 generation in response to infection (Fig. 5g).
Despite some variability, typical patterns of H 2O2 were apparent: before penetration, encysted Eu. dicksonii spores were labelled (Fig. 5h). Thereafter, the complete host cell content around the intracellular pathogen structures was most often stained (Fig. 5i,j). Overall, during the late thallus phase, H2O2 was highly present. Even when the developing pathogen sporangium had already filled the host cell, H2O2 was still produced by Ec. siliculosus (Fig. 5k). Consistent with NBT labelling, Eu. dicksonii secondary spores accumulate H2O2 (Fig. 5k).

DISCUSSION
This study represents the first proteomic investigation of the biotic stress response in macroalgae. In contrast to previous investigations, this study is based on an actual host-pathogen interaction rather than mimicking infection via elicitor treatment. It reveals a complex, integrated picture of the response to infection that combines classical stress markers such as oxidative stress and halogen metabolism, with novel protagonists such as plastidial proteins (Fig. 6). The maximum protein induction (2.2 fold change) detected between control and infected Ec. siliculosus was rather low compared with similar proteome studies on plantpathogen interactions (e.g. Zhou et al. 2006). This is likely caused by a dilution effect in our protein samples as Eu. dicksonii infection is restricted to individual host cells and is asynchronous. Therefore, our proteomic results encompass a more localized and probably systemic response which recently has also been described in Laminaria digitata (Thomas et al. 2014). The highest fold change is close to the 1.5 cut-off that we applied to detect significant changes in protein abundance, suggesting that our study underestimates the cohort of truly differentially accumulated proteins.Therefore, the application of mass spectrometry imaging methods (Matros & Mock 2013) is a tantalizing avenue to further the study of systemic response and localization of specific proteins during the course of infection.
As a result of this dilution, it is unsurprising that no oomycete proteins were detected by comparative 2-DE.

The Ectocarpus response to Eurychasma infection
Several proteins identified in the course of this study have previously been reported as induced or differentially expressed by a range of abiotic stressors in macroalgae including Ec. siliculosus (Ritter et al. 2010), the kelp species Laminaria digitata ), M. pyrifera (Konotchick et al. 2013) and Fucus sp. (Pearson et al. 2010).
These proteins include HSP70s, which prevent aggregation and are major components of the protein folding process in a cell and are typical stress markers across phyla (Mayer & Bukau 2005). MnSOD is a classical ROS-scavenging enzyme, which converts highly reactive superoxide into hydrogen peroxide. Among all proteins identified, vanadium-dependent haloperoxidases are prominent brown algal stress markers in Laminaria and Ectocarpus (Crépineau et al. 2000;Roeder et al. 2005;Cosse et al. 2009). Significantly, our study constitutes the first direct evidence of a role for a vanadium haloperoxidase during pathogen infection. Based on the detection of the ROS H2O2 (a substrate for bromoperoxidase) in Eurychasma-infected Ectocarpus, it is tempting to speculate on a role of the bromoperoxidase in scavenging of ROS and producing antimicrobial halogenated compounds. Our hypothesis is supported by the fact that expression of the bromoperoxidase also increases in Ectocarpus upon copper stress and elicitation with H2O2 and linolenic acid (Mithöfer et al. 2004;Ritter et al. 2010;Zambounis et al. 2013). A role of iodide and, to some extent, of bromide as antioxidants similar to Laminaria  can be hypothesized, given the speciation of bromine and concentrations of both halogens in Ectocarpus established recently . Furthermore, halogenated organic compounds are produced upon both oxidative and simulated biotic stress in Laminaria (Palmer et al. 2005;Goulitquer et al. 2009;Thomas et al. 2011), and have a demonstrated antimicrobial action in red and brown algae (Butler & Carter-Franklin 2004;Paul et al. 2006;La Barre et al. 2010). VBPO, which has a predicted extracellular location (Cock et al. 2010), may also have a role in modifying the algal cell wall. It has recently been demonstrated that a bromoperoxidase is involved in the in vitro cross-linking of phenolic substances and alginate in the presence of H2O2 (Salgado et al. 2009), and Eu. dicksonii infection triggers cell wall modifications (Tsirigoti et al. 2013(Tsirigoti et al. , 2014. Even though the identification of a vBPO among the 21 differentially expressed proteins of Ec. siliculosus in response to infection with Eu. dicksonii suggests parallels to the elicitor-based transcriptome studies in Laminaria digitata ), there are clear differences between the two brown algal species. In Laminaria digitata different members of the multigene haloperoxidase families (iodo-and bromoperoxidase) show differential expression under various stress conditions (Crépineau et al. 2000;Roeder et al. 2005;Cosse et al. 2009). Isoforms of these Laminaria digitata haloperoxidases show tightly coordinated expression in response to oligoguluronate elicitor treatment ), suggestive of functional specialization. In contrast to the kelp Laminaria digitata, the genome of Ec. siliculosus contains only one haloperoxidase gene encoding a vBPO (Cock et al. 2010), suggesting that halogen metabolism in Ectocarpus is less elaborate compared with kelp species. Ultimately, reverse genetic approaches, which are currently being developed in brown algae, will help to elucidate specific functions of this important protein family in Laminaria and Ectocarpus (Cock et al. 2012;Farnham et al. 2013).
Whether the presence of ROS produced by Ectocarpus in response to Eurychasma infection has a function in defence signal transduction and intracellular communication (Orozco-Cárdenas et al. 2001) or in defence needs to be validated but data from Laminaria strongly suggest such a role Thomas et al. 2011). However, despite repeated attempts, we were unable to identify a trigger for Eu. dicksonii spore release, and to synchronize the infection of algal cells. Hence, we cannot investigate events that might occur within minutes after Eu. dicksonii adhesion or penetration in its host. Due to this limitation, we cannot rule out the possible occurrence of a transient H2O2 burst, fuelled by NADPH oxidase-dependent superoxide production, as has been reported in Laminaria digitata (Küpper et al. 2002). Our data simply point to the existence of a sustained H2O2 source in infected algal cells throughout the development of the pathogen. We found a constitutive superoxide production in all Ectocarpus cells irrespective of the infection status, which only partially correlates with hydrogen peroxide accumulation in the dark. This indicates that whatever the light conditions, the algal cell type and the infection status, superoxide is by far the main source of H2O2 in Ec. siliculosus. However, in vegetative cells kept in the dark, there seem to be hydrogen peroxide detoxification routes (e.g. glutathione-Stransferase, glutaredoxin, catalase). Transcripts of classical ROS scavenger enzymes have been detected under abiotic stress conditions in Ec. siliculosus Dittami et al. 2009;Zambounis et al. 2013). It is also likely that the observed ROS generation upon high light stress and infection originates from different cellular compartments. Eu. dicksonii structures are outlined with black and grey lines, respectively. H2O2 production is highlighted in brown; chloroplasts are shown as ribbon structures. (a) Constitutive, light-independent H2O2 generation in Ec. siliculosus tips and plurilocular sporangia, originating from the dismutation of O2 -• radicals. Light-dependent plastidial O2 production also leads to H2O2 accumulation (dark-brown undulated structures). (b) Cytoplasmic, light-independent H2O2 generation in the infected algal cell coincides with the accumulation of the MnSOD protein, and the strong transcriptional induction of vBPO, likely leading to reactive oxygen species detoxification, production of antimicrobial halogenated organic compounds and/or cell wall reinforcement. During zoosporogenesis, punctate cytosolic sources of O2 -• and H2O2 (probably mitochondria) are also observed in Eu. dicksonii. ↑, significantly (P < 0.05) increased gene transcript or protein abundance; ↓, significantly (P < 0.05) decreased gene transcript or protein abundance; -, no change in gene expression; PST, posttranslational modification.
Whereas during light stress the chloroplast is the primary source of ROS production (Asada 2006), the source during infection appears to be different. We could demonstrate that H2O2 accumulates preferentially in infected cells. This observation is in agreement with the increased protein abundance of MnSOD seen in our proteomics analysis. Additionally, the increased abundance of vBPO utilizing H2O2 during Eu. dicksonii infection would explain the increased ROS levels. As shown in Laminaria digitata, H2O2 might, however, not represent the only signal mediating defence responses ).

Novel stress-related proteins
In addition to the classic features of algal stress responses discussed earlier, our data also highlighted novel proteins that have not been described in the context of algal stress responses so far. A number of proteins (spot IDs: 53, 46, 36, 130 and 61) with increased abundance in Ec. siliculosus in response to infection have been predicted as chloroplast localized (Gschloessl et al. 2008). Among those, PAP/fibrillins appear as prominent stress proteins induced under various stress conditions such as high light, oxidative stress, drought and notably pathogen infection and wounding in higher plants (Langenkämper et al. 2001;Yang et al. 2006;Chen et al. 2007;Youssef et al. 2010). They constitute the main protein component of plastoglobules (lipoprotein structures) that are associated with the chloroplast thylakoid membrane via a half-lipid bilayer in plants, and contain lipids and a variety of metabolic enzymes (Austin et al. 2006;Ytterberg et al. 2006;Bréhélin et al. 2007). In Arabidopsis, a knockdown of fibrillin expression leads to increased stress sensitivity including enhanced susceptibility to pathogens (Singh et al. 2010) and reduced accumulation of triacylglycerols and jasmonate (Youssef et al. 2010) but the exact functions are still unclear. Assuming that the increased expression of PAPs upon infection with Eurychasma signifies an accumulation of plastoglobules in Ec. siliculosus, the latter might either protect the chloroplast against increased levels of ROS (Vidi et al. 2006) or contribute to the elevated production of lipids and oxylipin precursors involved in defensive signalling (Santino et al. 2013). Further quantitative electron microscopy studies are warranted to investigate infection-related plastoglobule accumulation. PAPs are often phosphorylated and differential phosphorylation of those proteins could explain the presence of two spots (ID 46 and ID 53) attributed to the same PAP protein in the present study. Differential phosphorylation may also explain the occurrence of two different spots (ID 36 and ID 130) as the same hypothetical, chloroplast-located protein. Additionally, we found the MnSOD with slightly different isoelectric points. Whereas one spot (ID 724) showed unchanged abundance in response to biotic stress, a second protein spot (ID 23) assigned to the same protein had a lower isoelectric point of 5.61. The observation of potential posttranslational modification is indicative of regulation at the protein level rather than the gene expression level that is in agreement with our qPCR data on Papfib and MnSOD. Regulation at the posttranscriptional level allows a rapid response as it does not involve new protein biosynthesis and is commonly observed as response to pathogen attack in plants (Park et al. 2012).
The induction of a Rad23-like protein in an infectionrelated context was presented in two studies of the host response to infection with the protozoan pathogen Toxoplasma gondii in human cells (Nelson et al. 2008) and a plant pathogenic nematode Meloidogyne sp. (Fosu-Nyarko et al. 2009). One important function of Rad23 proteins, demonstrated for example in yeast and Arabidopsis, is the association of polyubiquitinated proteins and the protein degradation machinery (26S proteasome) (Chen & Madura 2002;Dantuma et al. 2009;Farmer et al. 2010) via the two binding domains, ubiquitin-associated and ubiquitin-like of Rad23. In this context, the increased expression of a Rad23like protein in Ec. siliculosus in response to pathogen infection might signal increased protein degradation via the ubiquitin-26S proteasome pathway that is in agreement with the increased abundance of folding proteins (e.g. HSP70, protein disulfide isomerase, nascent polypeptide-associated complex protein, chaperonin) upon Eu. dicksonii infection.
In conclusion, the investigation of the Ectocarpus-Eurychasma host-pathogen interaction gives unprecedented insight into the algal host defence response, linking our results with previous studies on simulated biotic stress Zambounis et al. 2013). The final outcome of the interaction, however, is the successful invasion of the algal cell by the pathogen and therefore host defences, at least in the challenged cell, are ultimately overcome. Interestingly however, ROS are also detected during the late infection stages of Eu. dicksonii suggesting that the host response is still, at least partially, active. In this regard, the ongoing investigation of Ectocarpus strains resistant to Eu. dicksonii infection ) will shed more light on defence and immunity in macroalgae.
acknowledged. Finally, we would like to thank the two anonymous referees for constructive suggestions to improve our manuscript.

SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Figure S1. Representative 2-DE gels (pI 4-7) of uninfected Ectocarpus siliculosus (control, a) and Eurychasma dicksonii-infected Ectocarpus siliculosus (infection, b). The indicated spot IDs match those of Table 1 (infection) and  Supporting Information Table S3 (control). Figure S2. Protein spots with significant (P < 0.05; one-way anova) different abundance during Eurychasma dicksonii infection of the brown alga Ectocarpus siliculosus. The pictures show representative sections of control (C) and infection (I) of one 2-DE gel. Table S1. Modifications made to the original protocol (Contreras et al. 2008) for protein extraction of Eurychasma dicksonii-infected Ectocarpus siliculosus. Table S2. Oligonucleotide sequence information of Ectocarpus siliculosus genes analysed by qPCR. Table S3. Protein expression in uninfected Ectocarpus siliculosus CCAP 1310/4 ('controls').