•Brown algal cell walls share some components with plants (cellulose) and animals (sulfated fucans), but they also contain some unique polysaccharides (alginates). Analysis of the Ectocarpus genome provides a unique opportunity to decipher the molecular bases of these crucial metabolisms.
•An extensive bioinformatic census of the enzymes potentially involved in the biogenesis and remodeling of cellulose, alginate and fucans was performed, and completed by phylogenetic analyses of key enzymes.
•The routes for the biosynthesis of cellulose, alginates and sulfated fucans were reconstructed. Surprisingly, known families of cellulases, expansins and alginate lyases are absent in Ectocarpus, suggesting the existence of novel mechanisms and/or proteins for cell wall expansion in brown algae.
•Altogether, our data depict a complex evolutionary history for the main components of brown algal cell walls. Cellulose synthesis was inherited from the ancestral red algal endosymbiont, whereas the terminal steps for alginate biosynthesis were acquired by horizontal gene transfer from an Actinobacterium. This horizontal gene transfer event also contributed genes for hemicellulose biosynthesis. By contrast, the biosynthetic route for sulfated fucans is an ancestral pathway, conserved with animals. These findings shine a new light on the origin and evolution of cell wall polysaccharides in other Eukaryotes.
Brown algae (Phaeophyceae) are photoautotrophic marine organisms which belong to the phylum of Stramenopiles. This eukaryotic supergroup arose c. 1 billion yr ago on a secondary endosymbiotic event, by which a unicellular red alga was captured by an ancestral protist (Reyes-Prieto et al., 2007). In phylogenetic analyses, chromalveolates form a distinct group from animals and fungi (Opisthokonts) on the one hand, and red algae, green algae and plants (Archaeplastida) on the other (Baldauf, 2008). Thus, brown algae evolved complex multicellularity independently from Opisthokonts and Archaeplastida (Grosberg & Strathmann, 2007). As such, they represent an interesting outgroup to investigate the evolution of multicellularity in Eukaryotes. One key step in this evolutionary process was the development of an adherent extracellular matrix (ECM), allowing for the transition from cellular autonomy to cellular cooperation. Multicellular Eukaryotes share other common characteristics, such as the ability to strictly control cell differentiation in space and time (development), as well as the evolution of sophisticated defense systems (innate immunity). The ECM plays an important role in both of these functions, as a first boundary for nonself recognition and as the seat of cell–cell signaling (Brownlee, 2002).
The ECM of extant Eukaryotes is typically organized as a three-dimensional network of crystalline fibers embedded in amorphous components. In fungi and photosynthetic organisms, cells are bound by a rigid cell wall, essentially made up of polysaccharides, including crystalline β-1,3-d- and β-1,4-d-glucans, respectively. By contrast, the metazoan ECM is essentially made up of fibrillar proteins (collagens) interconnected by protein-bound sulfated polysaccharides, referred to as glycosaminoglycans (GAGs; Sugahara & Kitagawa, 2002). Brown algae have evolved a cell wall (Fig. 1) which shares features with both plants and animals and which also exhibits some unique characteristics (Kloareg & Quatrano, 1988). Like plants, brown algae produce cellulose, but these crystalline fibers account for only a small proportion of the cell wall, the amount of cellulose ranging between 1% and 8% of the dry weight of the thallus (Cronshaw et al., 1958). The main cell wall components are anionic polysaccharides, namely alginates and fucoidans (Kloareg & Quatrano, 1988). Alginates consist of two uronic acids, β-1,4-d-mannuronate and α-1,4-l-guluronate, arranged in blocks along the polysaccharide chain (Fig. 1a). Alginates are initially polymerized as mannuronan and the guluronate residues are made at the polymer level by the action of mannuronate C5-epimerases (MC5Es; Haug & Larsen, 1969). Alginates form gels in the presence of divalent cations and their self-assembling properties depend on the relative proportion of guluronate homopolymeric blocks (Haug et al., 1974).
Fucoidans are sulfated polysaccharides containing α-l-fucose residues and occur in brown algae, but also in the body wall of sea cucumbers and in the egg jelly coat of sea urchins. Although echinoderm fucoidans are linear and repetitive polysaccharides (Pomin & Mourao, 2008), brown algal fucoidans encompass a continuous spectrum of highly ramified polysaccharides, ranging from high-uronic-acid, low-sulfate-containing polymers with significant proportions of d-xylose, d-galactose and d-mannose (xylo-fuco-glucuronan and xylo-fuco-glucans) to highly sulfated homofucan molecules (Mabeau et al., 1990). The sulfated fucans from Fucales contain long stretches of the disaccharide repeating unit (→4)-α-l-fucose-2,3-disulfate-(1→3)-α-l-fucose-2-sulfate-(1→) (Fig. 1b) (Chevolot et al., 2001; Colin et al., 2006). The sulfated fucans from Laminariales (Nishino et al., 1991) and Ectocarpales (Ponce et al., 2003) display a structure mainly based on 3-linked α-l-fucose residues mostly sulfated at C4 (Fig. 1c).
Alginates, fucoidans and cellulose are in an average weight ratio of 3 : 1 : 1 in mature intertidal brown algae (Kloareg & Mabeau, 1987). In addition, brown algal cell walls contain phlorotannins, which consist of halogenated and/or sulfated phenolic compounds (Vreeland et al., 1998; Schoenwaelder & Wiencke, 2000) and c. 5% of proteins (Quatrano & Stevens, 1976). Microchemical imaging analyses have recently demonstrated that the cell walls of Laminariales bind huge amounts of iodine (Verhaeghe et al., 2008), probably through the involvement of apoplastic, vanadate-dependent iodoperoxidases (Colin et al., 2005). In the presence of halides, vanadate haloperoxidases also catalyze the cross-linking of alginates with phlorotannins (Berglin et al., 2004), suggesting an important role of the halogenated and phenolic compounds in cell wall cohesion. Alginates, sulfated fucans and phlorotannins are synthesized in the Golgi, transported in vesicles to the plasma membranes and secreted into the expanding cell wall (Callow et al., 1978; Schoenwaelder & Wiencke, 2000). Cellulose microfibrils are produced and deposited in situ by cellulose synthase complexes (terminal complexes) localized in the plasma membrane (Peng & Jaffe, 1976). In contrast with plants, however, terminal complexes do not form rosettes in brown algae, but single rows, comprising between 10 and 100 subunits. This linear arrangement results in a flat ribbon-like shape of cellulose microfibrils, with a uniform thickness of c. 2.6 nm and a variable width in the range 2.6–30 nm (Tamura et al., 1996; Tsekos, 1999). Another difference from terrestrial plants is that the terminal complexes are guided by F-actin filaments, and not by microtubules (Bisgrove & Kropf, 2001). Cell morphogenesis and cell wall deposits have been essentially studied in Fucales zygotes (Kropf et al., 1988; Fowler & Quatrano, 1997). At germination, tip growth is initiated at a determined site at the surface of the zygote and a rhizoid emerges. The apical cell of the rhizoid will then elongate by tip growth, whereas cells in the thallus will elongate by diffuse growth (Bisgrove & Kropf, 2001). As in plants, cell expansion is driven by turgor pressure in brown algae, but the biochemical mechanisms underlying cell wall relaxation are currently unknown.
The molecular bases of cell wall biogenesis in brown algae have received little attention so far. Our current knowledge is essentially limited to the terminal step of alginate biosynthesis, the conversion of d-mannuronate into l-guluronate. Two full-length cDNAs, homologous to bacterial MC5E, were isolated from Laminaria digitata (Nyvall et al., 2003) and their expression pattern confirmed that these cDNAs encode functional MC5E. Based on phylogenetic analyses, it was proposed that MC5E, and probably the entire pathway for the biosynthesis of alginates, was acquired by brown algae from an ancestral alginate-producing bacterium (Nyvall et al., 2003). MC5Es form a large multigenic family of at least 45 genes in L. digitata, which can be divided into several subgroups according to their expression profile, suggesting a fine tuning of alginate structure in response to developmental and environmental conditions (Roeder et al., 2005; Tonon et al., 2008).
Ectocarpus siliculosus is a filamentous brown alga from the order Ectocarpales (Charrier et al., 2008), a sister group of Laminariales which includes the most complex brown algae (Bartsch et al., 2008). The sequencing of its complete, 214-Mbp genome yielded a wealth of data that provide inferences on the adaptation of this organism to the intertidal environment and on the evolution of muticellularity, such as the presence of a rich array of signal transduction genes, including a family of receptor kinases specific to brown algae (Cock et al., 2010). The genome sequence also provides an in-depth, comprehensive view of the carbohydrate metabolism of brown algae (Michel et al., 2010). In this article, we reconstruct the metabolic pathways for the biosynthesis and remodeling of cellulose, alginate and sulfated fucans in brown algae, and discuss the origin and evolution of ECM polysaccharides in photosynthetic Eukaryotes.
Materials and Methods
Identification and bioinformatics analyses of carbohydrate-related proteins
Ectocarpus siliculosus genes with predicted roles in carbohydrate metabolism were identified by homology with biochemically characterized proteins selected from the CAZY (http://www.cazy.org/, Cantarel et al., 2009) and UniProt databases. For each identified protein, evidence of conserved protein modules was queried against the Pfam database (Bateman et al., 2004). The presence of additional, orphan modules was detected by BlastP searches against the UniProt database. The homology threshold was chosen as c. 30% sequence identity on the entire module sequence. Signal peptide and transmembrane helices were predicted using HECTAR (Gschloessl et al., 2008) and TMHMM (Krogh et al., 2001), respectively. Many of the proteins that are predicted to be involved in carbohydrate metabolism belong to families encompassing several enzymatic activities or substrate specificities. To clarify the function of these proteins, they were further analyzed using a phylogenetic approach. For each different activity of a polyspecific family, a set of experimentally characterized proteins was selected in the UniProt database. These representative proteins were aligned with their homologs from E. siliculosus using MAFFT with the iterative refinement method and the scoring matrix Blosum62 (Katoh et al., 2002). Phylogenetic trees were derived from these refined alignments using the maximum likelihood method with the program PhyML (Guindon & Gascuel, 2003). The reliability of the trees was tested by bootstrap analysis, using 100 re-samplings of the dataset. The trees were displayed with MEGA (Kumar et al., 2004). The functional annotation of E. siliculosus proteins was based on the proximity to specific characterized proteins in the phylogenetic trees. Genomic comparisons were performed using the genomic BLAST server at the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi). Throughout the text, the term ‘plants’ will designate the class Embryophyta, as defined by recent phylogenetic and taxonomic studies (Lewis & McCourt, 2004; Adl et al., 2005).
The census of the glycoside hydrolases (GHs) and glycosyltransferases (GTs) of E. siliculosus has been detailed in the first part of this study (Michel et al., 2010; Supporting Information Table S2). To summarize, the genome of this brown alga encodes 41 GHs and 88 GTs, representing 18 GH and 32 GT families. This seaweed possesses at least six times less genes than terrestrial plants. Arabidopsis thaliana, for example, has 730 GH/GT genes (Henrissat et al., 2001). Ectocarpus features less functional redundancy, with fewer genes in each CAZY family.
Synthesis and remodeling of cellulose and associated glycans
Among the 11 GT2 enzymes identified in the Ectocarpus genome, nine consist of large membrane proteins homologous to the cellulose synthases (CESA) and cellulose synthase-like (CSL) proteins from plants (Fig. 2). Plant CSL proteins are divided into two distinct clades related to cyanobacteria and green algae, respectively. The cellulose synthases of the CESA type recently characterized in the red alga Porphyra yezoensis (Roberts & Roberts, 2009) also belong in this cluster. Brown algal CSL proteins cluster into two distinct clades. The first clade is located within the large group made up of the CESA from Porphyra yezoensis, cyanobacteria, plants and Oomycetes (Fig. 2). The other CSL proteins from Ectocarpus are similar (c. 40% sequence identity) to CSL proteins from Actinobacteria, fungi and two Phycodnaviruses. These CSL proteins form a robust group with the actinobacterial proteins at the base of this cluster (Fig. 2).
Ectocarpus possesses a variety of other GTs which are likely to be involved in the biosynthesis of cell wall polysaccharides. These include five GT14, eight GT47 and two GT64 proteins. Animal GT14s initiate the biosynthesis of GAGs by transferring a xylose residue to the core proteins (Gotting et al., 2000). In poplar, some GT14s are specifically expressed in the xylem and have been proposed to participate in secondary cell wall biogenesis (Aspeborg et al., 2005). In animals, heparan sulfate (HS) synthases are made up of a GT47 fused to a GT64, which alternately add β-1,4-d-glucuronate and α-1,4-N-acetlyl glucosamine residues to the HS chain (Sugahara & Kitagawa, 2002). By contrast, GT47 and GT64 are stand-alone proteins in plants, as observed for Ectocarpus. Plant GT47 proteins transfer various types of sugar and participate in the synthesis of the carbohydrate branches of xyloglucans and pectins (Iwai et al., 2002; Madson et al., 2003). The exact functions of plant GT64 proteins are uncertain, but mutation of the EPC1 gene, which encodes a GT64 in A. thaliana, results in a severely dwarfed phenotype and a 50% reduction in the level of β-1,4-galactan in wall pectins (Bown et al., 2007). Altogether, the occurrence of these three GT families in Ectocarpus suggests the presence of yet unidentified hemicellulose-like polysaccharides in the cell walls of brown algae.
In plants, selective modifications of cellulose and hemicelluloses occur during wall assembly and cell elongation. Wall expansion mainly involves the action of cellulases (family GH9), xyloglucan endotransglycosylases/hydrolases (XTH, family GH16) and expansins, which are believed to disrupt noncovalent linkages between wall polysaccharides (Henrissat et al., 2001; Cosgrove, 2005). The Ectocarpus genome does not encode any canonical cellulase even though these proteins encompass as many as 12 CAZY families (Cantarel et al., 2009). Expansins and XTHs are also absent from the genome of this brown alga. The absence of XTH genes confirms the results of a recent XTH activity screening in various algae, including Phaeophyceae (Van Sandt et al., 2007). By contrast, Ectocarpus possesses a GH10 family xylanase (Esi0176_0036), which is conserved with characterized xylanases from Actinobacteria (35% sequence identity; Canals et al., 2003). Brown algae are currently not known to produce xylans, as found in land plants. However, they synthesize fuco-glucurono-xylans, which have been proposed to cross-link cellulose fibers and alginate gels (Kloareg & Quatrano, 1988). Therefore, Esi0176_0036 may be involved in the remodeling of such hemicellulose-like polysaccharides.
Alginate synthesis and remodeling
The molecular bases for alginate biosynthesis are well known in the Gammaproteobacteria Pseudomonas aeruginosa and Azotobacter vinelandii (Rehm & Valla, 1997; Ramsey & Wozniak, 2005). The enzymes AlgA, AlgC and AlgD are responsible for the synthesis of the alginate precursor, guanosine diphosphate mannuronic acid (GDP-ManA). This compound is then polymerized by Alg8, a GT2 family GT. After polymerization, some of the d-mannuronate residues are finally epimerized into l-guluronate by the MC5E AlgG (Fig. 3). The biosynthesis of alginate in brown algae involves equivalent biochemical reactions (Lin & Hassid, 1966; Haug & Larsen, 1969), but only the final C5-epimerization step has been characterized at the molecular level so far (Nyvall et al., 2003). In bacteria, eight additional proteins have been shown to participate in the formation and export of alginate (Ramsey & Wozniak, 2005).
Of the 13 proteins involved in alginate biosynthesis in bacteria (Ramsey & Wozniak, 2005), only three, the phosphomannomutase AlgC (PMM, EC 188.8.131.52), the GDP-mannose 6-dehydrogenase AlgD (GMD, EC 184.108.40.206) and the MC5E AlgG (EC 5.1.3.-), could be unambiguously identified in Ectocarpus by homology searches. The crucial bifunctional enzyme AlgA (EC 220.127.116.11/18.104.22.168), which catalyzes the first and third steps of alginate pathway in bacteria, was not found in Ectocarpus. Four proteins (Esi0195_0002, Esi0000_0207, Esi0120_0009_0010 and Esi0120_0005), however, could catalyze the first step of alginate biosynthesis (Fig. 3). These proteins are homologous to animal mannose-6-phosphate isomerases (c. 40% sequence identity), which convert d-fructose-6-phosphate into d-mannose-6-phosphate (MPI, EC 22.214.171.124). The phylogenetic tree of MPI is congruent with the overall eukaryotic taxonomy, suggesting that these enzymes were present in the last eukaryotic common ancestor (Fig. 4a). The second step is likely to be catalyzed by Esi0149_0031, which has 32% sequence identity with AlgC from P. aeruginosa (Ye et al., 1994). This protein is, however, more closely related to the bifunctional phosphomannomutase/phosphoglucomutase (PMM/PGM) enzymes from Firmicutes, Cyanobacteria and plants (c. 50% sequence identity; Fig. 4b). In the absence of an AlgA homolog, mannose-1-phosphate guanylyltransferase (MPG, EC 126.96.36.199) could catalyze the third step, the production of GDP-mannose from GTP and d-mannose-1-phosphate (Fig. 3). These monofunctional enzymes are indeed well conserved in animals, fungi and plants. However, no homolog of eukaryotic MPG has been found either in the Ectocarpus genome (Cock et al., 2010) or in the 90,637 ESTs produced for Ectocarpus (Dittami et al., 2009). MPGs are also missing from the sequences obtained from cDNA libraries of Laminaria digitata and Sargassum binderi (Roeder et al., 2005; Wong et al., 2007). To establish whether this unexpected absence of MPG genes in brown algae is a common trend in Stramenopiles, we also searched for MPG homologs in the genome of the Oomycete Phytophthora infestans (Haas et al., 2009) and those of the diatoms Thalassiosira pseudonana (Armbrust et al., 2004) and Phaeodactylum tricornutum (Bowler et al., 2008). In Phytophthora infestans, two proteins, PITG_11097 and PITG_14229, similar to human MPG (43% and 30% sequence identity, respectively) were identified, whereas no ortholog was found in the diatom genomes.
Ectocarpus contains four proteins (Esi0051_0092, Esi0051_0113, Esi0101_0043 and Esi0164_0053) which exhibit c. 27% sequence identity with the GDP-mannose 6-dehydrogenase AlgD from P. aeruginosa (Ramsey & Wozniak, 2005). Phylogenetic analysis suggests that Esi0101_0043 encodes a UDP-glucose 6-dehydrogenase (UGD, EC 188.8.131.52), which is conserved with Oomycetes and diatoms (Fig. 4c). The three other AlgD-like proteins from Ectocarpus cluster in a distinct clade which comprises bacterial proteins only, including AlgD from P. aeruginosa and A. vinelandii. Therefore, these brown algal proteins are good candidates for being GDP-mannose-6-dehydrogenases (GMD, step 4 in Fig. 3). Ectocarpus also contains a large family of MC5Es (28 genes, step 6). These proteins display c. 35% sequence identity with AlgG from P. aeruginosa. By contrast, Ectocarpus does not possess a significant homolog of the mannuronan synthase Alg8 (step 5; Ramsey & Wozniak, 2005). The closest orthologs of the GMD and MC5E from Ectocarpus are from Frankia species (Actinobacteria; Fig. 4c,d). As in Ectocarpus, no significant homolog of Alg8 has been identified in Actinobacteria. Yet, in the genome of Frankia sp. Eul1c (sequenced by US Department of Energy Joint Genome Institute), the candidate genes for GMD (FraEuI1c_6214) and for MC5E (FraEuI1c_6212) are colocalized with genes encoding a family PL7 alginate lyase (FraEuI1c_6213) and a GT2 family GT (FraEuI1c_6210). This gene cluster is reminiscent of the alginate synthesis locus found in P. aeruginosa and A. vinelandii (Rehm & Valla, 1997). Even though FraEuI1c_6210 is highly divergent (23% sequence identity) from the P. aeruginosa Alg8, this genomic context suggests that this protein is a mannuronan synthase. Ectocarpus contains two homologs of FraEuI1c_6210: Esi0010_0147 and Esi0086_0005 (31% and 29% sequence identity, respectively). If the above hypothesis is correct, and in the absence of a clear Alg8 homolog in Ectocarpus, these two GT2 family GTs appear to be realistic candidates for catalyzing the polymerization of activated mannuronate units in brown algae.
Three MC5E proteins (Esi0882_0001, Esi0069_0059 and Esi0010_0210) feature additional WSC domains (for cell Wall integrity and Stress response Component), which are potential carbohydrate-binding modules (CBMs) discovered in fungi (Verna et al., 1997; Cohen-Kupiec et al., 1999). As CBMs generally display similar substrate specificity to their appended catalytic modules (Michel et al., 2009), these WSC domains are likely to bind alginates. Interestingly, the WSC domain family is one of the largest protein domain families in Ectocarpus, with 115 genes containing at least one WSC module. Most of these proteins display endoplasmic reticulum signal peptides and may thus consist of cell wall proteins that attach to alginate chains. A detailed analysis of these WSC modules was presented in the article describing the Ectocarpus genome (Cock et al., 2010).
With respect to enzymes potentially involved in the recycling of alginate, the Ectocarpus genome does not encode any known alginate lyase, despite the fact that as many as six families of polysaccharide lyases (PLs) have been shown to display alginate lyase activity. However, Ectocarpus contains six proteins homologous to the Δ-4,5-unsaturated β-glucuronyl hydrolase from Bacillus sp. GL1 (family GH88), which is known to degrade unsaturated oligosaccharides released by GAG lyases (Itoh et al., 2006). This presence of GH88 family enzymes strongly suggests that brown algae possess alginate lyases, but that they belong to novel PL family(ies). Another possibility is that some MC5E could act as alginate lyases, as such bifunctional enzymes exist in A. vinelandii (Svanem et al., 2001).
Synthesis and remodeling of sulfated fucans
The metabolism of sulfated fucans in brown algae is currently uncharacterized. It is very likely that the fucan precursor is GDP-l-fucose, which is the usual activated form of l-fucose. In addition, by analogy with GAG biosynthesis (Sugahara & Kitagawa, 2002), sulfated fucans are likely to be polymerized as neutral polysaccharides by one or several fucosyltransferases and then sulfated by specific sulfotransferases (STs). In mammals, plants and most bacteria, GDP-fucose is mainly produced from GDP-mannose by the de novo pathway (Becker & Lowe, 2003). In this pathway (Fig. 5), GDP-mannose is converted into GDP-4-keto-6-deoxymannose by the GDP-mannose 4,6-dehydratase (GM46D, EC 184.108.40.206). This keto intermediate is then converted to GDP-fucose in two steps catalyzed by a bifunctional epimerase/reductase, the GDP-l-fucose synthetase (GFS, EC 220.127.116.111). Ectocarpus harbors two GM46D (Esi0009_0070 and Esi0031_0033) and one GFS (Esi0003_0195), indicating that brown algae can also produce GDP-fucose through the de novo pathway. These proteins are highly similar to their animal and fungal counterparts (c. 65% sequence identity), but are more distant from the plant enzymes (c. 50% for GMD and c. 30% for GFS). Unexpectedly, the unicellular red alga Cyanidioschyzon merolae (Matsuzaki et al., 2004) lacks GM46D and GFS genes. Phylogenetic analyses confirm that the GM46D and GFS from Stramenopiles both cluster with their orthologs from Opisthokonta (Fig. 6). The plant enzymes constitute two separate clades, which are both rooted by cyanobacterial homologs. By contrast, the green algal GM46D and GFS cluster within the large group encompassing Stramenopiles, animals and fungi (Fig. 6).
In mammalian cells, an alternative salvage pathway can yield GDP-fucose directly from l-fucose (Becker & Lowe, 2003). Free cytosolic l-fucose is first phosphorylated by l-fucokinase (FK, EC 18.104.22.168). GDP-fucose pyrophosphorylase (GFPP, EC 22.214.171.124) then catalyzes the reversible condensation of fucose-1-phosphate with GTP to form GDP-fucose (Fig. 5). These enzymes are homologous, although the human FK differs from human GFPP by the presence of an additional C-terminal module belonging to the GHMP kinase family (Hinderlich et al., 2002). This salvage pathway also exists in plants, but the conversion of l-fucose to GDP-fucose is catalyzed by a single bifunctional enzyme with similarity to both human FK and GFPP (Kotake et al., 2008). In Ectocarpus, two proteins are likely to be responsible for the salvage pathway, namely Esi0130_0058, which is distantly related to both GFPP and the N-terminal module of human FK (25% sequence identity), and Esi0130_0054, which is highly similar to the GHMP kinase module of human FK (c. 45% sequence identity). As observed for plants (Kotake et al., 2008), we propose that Esi0130_0058 encodes a bifunctional enzyme which, in combination with the GHMP kinase Esi0130_0054, catalyzes the two steps needed for the conversion of l-fucose into GDP-fucose (Fig. 5). Several fucosyltransferases, from families GT10 (Esi0050_0098), GT23 (Esi0135_0016 and Esi0540_0004) and GT65 (Esi0021_0026), could be involved in the polymerization of GDP-fucose into the elongating fucan chain.
Ectocarpus contains as many as 15 different genes encoding STs, which fall into five main clades (Fig. 7). A first group of four proteins (Esi0028_0011, Esi0197_0021, Esi0197_0023 and Esi0442_0008) is related to animal aryl STs and to the plant STs involved in the biosynthesis of various secondary metabolites (e.g. glucosinolates, sulfohydroxyjasmonate). Therefore, these brown algal STs (Clade A) are likely to act on phenolic compounds, for instance sulfated phlorotannins (Vreeland et al., 1998) and/or flavonoids (Varin et al., 1997). Interestingly, Ectocarpus features a GT1 family GT (Esi0029_0138) distantly related to the flavonoid 3-O-glucosyltransferases from plants (26% sequence identity), strengthening the hypothesis that brown algae produce sulfated flavonoids. The Esi0289_0025 gene emerges within a second clade which comprises uncharacterized STs from bacteria only (Clade B). The three other groups of STs from Ectocarpus are homologous to a variety of carbohydrate STs (Fig. 7) which, in metazoans, are known to be involved in the biosynthesis of either glycosphingolipids (galactosylceramide STs, Clade D) or GAGs (carbohydrate STs, Clade C; HS glucosamine 3-O-STs, Clade E). The six proteins in the two clades related to GAG sulfotranserases (Clades C and E: Esi0239_0035, Esi0026_0167, Esi0080_0060, Esi0037_0054, Esi0312_0029 and Esi0210_0041) are good candidates to function as genuine fucan STs. The multiplicity of these enzymes probably reflects the need for sulfation at different positions on the fucan backbone.
The remodeling of sulfated fucans is likely to require a combination of specific glycosyl hydrolases and sulfatases. The fucanase FcnA from the marine bacterium Mariniflexile fucanivorans currently is the only known GH specific for algal sulfated fucans (S. Colin et al., 2006; Barbeyron et al., 2008). This GH107 family enzyme is, however, not present in Ectocarpus. More surprisingly, α-l-fucosidases from the GH29 family are also absent, whereas these enzymes are well conserved in bacteria, animals and plants. By contrast, Ectocarpus possesses nine sulfatases which are all related to GAG sulfatases from animals (c. 35% sequence with human N-acetylgalactosamine-4-sulfatase). These hydrolytic enzymes contain a unique catalytic residue, the Cα-formylglycine, which is post-translationally generated from a conserved cysteine located in an N-terminal CXPXR motif (Schmidt et al., 1995). All of the Ectocarpus sulfatases display this conserved motif. Moreover, this brown alga contains two proteins, Esi0167_0037_0035 and Esi0195_0042, which are highly similar to the human formylglycine-generating enzymes (FGEs) SUMF1 and SUMF2 (48% and 42% sequence identity, respectively) (Dierks et al., 2003). Therefore, all the requirements for producing active, formylglycine-dependent sulfatases (fgSs) are met in brown algae, and the nine identified sulfatases are likely to be involved in the remodeling of sulfated fucans. Genomic comparisons indicate that diatoms also display fgSs (two genes) and FGE, but there is no significant homolog of these genes in terrestrial Oomycetes. With respect to the Archaeplastida, the volvocine green alga Chlamydomonas reinhardtii (Merchant et al., 2007) contains 16 fgSs and a distant relative of human SUMF1 (23% sequence identity), but sulfatases and FGE are completely absent in land plants.
Very few studies have looked at the evolution of extracellular matrices across the whole eukaryotic tree (Niklas, 2004), and those that have been carried out were focused essentially on the biosynthesis of cellulose and hemicelluloses (Nobles et al., 2001; Nakashima et al., 2004; Yin et al., 2009) and chitin (Ruiz-Herrera et al., 2002). Brown algae are the only organisms in the Stramenopile lineage to have evolved complex multicellularity, thus providing a new angle to address the relationships between the development of multicellular organisms and the differentiation of their ECM. As brown algae are marine organisms, and as life first evolved in the sea, these seaweeds also provide a reference to better understand the specific traits that were developed by plants to colonize terrestrial environments. We have predicted the metabolic routes which lead to the synthesis of main brown algal ECM polysaccharides, cellulose, hemicelluloses, fucans and alginates, based on the gene content of the recently released genome of Ectocarpus siliculosus, the first available for a multicellular Stramenopile (Cock et al., 2010). Nonetheless, the exact function of the diverse candidate genes needs to be rigorously tested by future functional approaches. Below, we discuss the possible origins of these carbohydrates and propose various evolutionary scenarios to account for the evolution of the ECM cement and fiber polysaccharides in multicellular photosynthetic Eukaryotes.
Origin and speciation of cell wall polysaccharides in brown algae
Sulfated fucans are of ancestral origin Ectocarpus possesses the complete pathway for the de novo production of GDP-fucose from GDP-mannose. The existence of a salvage pathway is less certain, although candidate genes for FK and GFPP have been identified (Fig. 5). The GTs involved in the polymerization and branching of sulfated fucans are currently unknown. Ectocarpus contains four fucosyltransferases, from families GT10, GT23 and GT65. However, in other eukaryotic phyla, these enzymes are usually involved in protein glycosylation. Therefore, it is difficult to predict whether these fucosyltransferases are genuine fucan synthase(s) or whether brown algae have evolved novel GTs which cannot be detected by sequence comparisons with the currently known activities. By contrast, the sulfation and desulfation of fucans very probably are catalyzed by the six STs and the nine fgSs, which are related to the STs and fgS involved in the biosynthesis and remodeling of animal GAGs (Fig. 5). Note that the distant homologs of plant STs may not necessarily encode enzymes specific for sulfate transfer onto carbohydrates (Fig. 7). Yet, considering the widespread distribution of sulfated polysaccharides in archaeplastidal algae (see paragraph on land colonization), and based on the occurrence in the Ectocarpus genome of animal orthologs for both the synthesis of fucose and the addition or the removal of sulfate groups onto/from polysaccharide chains, the metabolism of sulfated fucans in brown algae is probably an ancestral pathway.
The Ectocarpus GTs from the GT14, GT47 and GT64 families are conserved in both animals and plants, indicating that these enzymes are also of ancestral origin. The animal GTs are involved in GAG biosynthesis (Gotting et al., 2000; Sugahara & Kitagawa, 2002), whereas the plant enzymes catalyze the branching of xyloglucans and pectins (Iwai et al., 2002; Madson et al., 2003) and the synthesis of β-1,4-galactan branches in pectins (Bown et al., 2007). The polyspecificity of these GT families makes it difficult to predict their exact function in brown algae. They may participate in the synthesis of the xyloglycan ramifications of sulfated fucans or in the synthesis of hemicellulose-like polymers.
Cellulose biosynthesis was acquired from the red algal endosymbiont In plants, the GT2 family comprises genuine cellulose synthases (CESA), but also CSL proteins, which are mainly involved in hemicellulose biosynthesis and have been classified into several subfamilies (from CSLA to CSLH) (Lerouxel et al., 2006). Molecular evidence suggests that plants acquired cellulose synthases and most CSL families from cyanobacteria (Nobles & Brown, 2004). The families CSLA and CSLC have a different origin and are most closely related to a CSL gene specific to green algae (Yin et al., 2009). A survey of sequenced algal genomes indicates that cellulose synthases are absent from diatoms and from the red microalga C. merolae (Yin et al., 2009). However, a CESA gene has been cloned recently from the red macroalga Porphyra yezoensis and clustered with cellulose synthases from Phytophthora, suggesting that Stramenopiles acquired CESA genes from their red algal endosymbiont (Roberts & Roberts, 2009). Our own phylogenetic analysis (Fig. 2) retrieved a topology similar to that observed previously (Roberts & Roberts, 2009; Yin et al., 2009). Ectocarpus contains four proteins, which emerge within the large group comprising the characterized cellulose synthases from Porphyra yezoensis, cyanobacteria, plants and Oomycetes. This predicts that Esi0004_0105, Esi0120_0014, Esi0231_0017 and Esi0231_0020 are genuine cellulose synthases. However, the exact position of Ectocarpus CESAs within this clade must be taken with caution. The brown algal sequences are unexpectedly closer to cyanobacterial cellulose synthases (c. 30% sequence identity) than to the CESAs from Porphyra yezoensis and Oomycetes (c. 25% sequence identity). This odd phylogenetic relationship may be explained by the lack of genomic data for a typical red alga, as C. merolae lacks a cell wall (Matsuzaki et al., 2004). A more representative, multicellular red alga probably would contain several CSL proteins, as observed for plants and brown algae. If this is the case, Oomycetes and brown algae may have conserved distinct red algal CSL proteins, as found for trehalose-phosphate synthases (Michel et al., 2010). Altogether, the most parsimonious scenario is that the common ancestor of brown algae and Oomycetes acquired the pathway for cellulose biosynthesis from their rhodobiont.
Polysaccharide pathways unique to brown algae In the alginate synthesis pathway shown in Fig. 3, steps 1–3 are not exclusive to alginate metabolism. They are common to all pathways which utilize activated mannose (e.g. protein glycosylation). The reactions specific for alginate biosynthesis are steps 4–6, from the conversion of the GDP-mannose into GDP-mannuronate to the final C5-epimerization of the polymannuronan chain into alginic acid. Brown algae are essentially the only Eukaryotes that possess alginate, although the presence of alginate has been reported in some calcareous red algae of the family Corallinaceae (Okasaki et al., 1982; Usov et al., 1995). However, the alginate pathway in Corallinaceae remains uncharacterized, at both the biochemical and gene levels. In Laminaria digitata, the last step of alginate biosynthesis is catalyzed by a large multigenic family of enzymes homologous to bacterial MC5Es (Nyvall et al., 2003). It was proposed that red algae acquired the alginate pathway from cyanobacteria via the primary plastid endosymbiosis and that, subsequently, this metabolism was passed on to brown algae via the secondary endosymbiosis (Nyvall et al., 2003). However, this evolutionary scenario suffers from several weaknesses: extant cyanobacteria do not contain MC5E genes; this scenario would require multiple losses of the alginate pathway in green algae, in plants and in the vast majority of the red algae (Archaeplastida), but also in the Stramenopiles other than brown algae, diatoms and Oomycetes.
On the basis of phylogenetic relationships (Fig. 4b), only the PMM Esi0149_0031 (step 2, Fig. 3) may have been acquired from Cyanobacteria, via the rhodobiont. By contrast, the first step of alginate biosynthesis is catalyzed by an ancient eukaryotic protein family (MPI, Fig. 4a). The MPG (step 3), present in bacteria and most Eukaryotes, is not encoded by the Ectocarpus and diatom genomes. Nevertheless, the presence of GDP-mannose and MPG activity has been demonstrated experimentally in the brown alga Fucus gardneri (Lin & Hassid, 1966). Therefore, brown algae and diatoms have lost the classical MPG genes, which are conserved in Oomycetes, but they are likely to possess a novel family of enzymes which catalyze the synthesis of GDP-mannose.
The GMD (step 4) and MC5E (step 6) of brown algae are conserved with a few bacteria only, including Gamma-proteobacteria and Actinobacteria. For both of these activities, as well as for the two candidate mannuronan synthases in Ectocarpus, Esi0010_0147 and Esi0086_0005, the closest orthologs are found in Frankia species (Actinobacteria). Therefore, we propose that the biogenesis of alginate in brown algae involves a hybrid pathway, with the initial, central carbon metabolism steps having an ancient eukaryotic origin, whereas the subsequent, alginate-specific steps were acquired by horizontal gene transfer (HGT) from an actinobacterium. In terrestrial environments, extant Frankia species occupy two distinct ecological niches, the soil and plant root nodules (Benson & Silvester, 1993). Similar associations may have existed between marine Actinobacteria and ancestral brown algae that promoted the HGT of alginate biosynthesis into the host. In the absence of genomic data for Corallinales, the origin of alginate in red algae is uncertain. We now favor a scenario more parsimonious than acquisition through the primary plastid endosymbiosis (Nyvall et al., 2003), namely an independent HGT with an alginate-producing bacterium limited to the ancestor of calcareous red seaweeds. Such more recent, independent HGTs have been observed, e.g. for amoeba and animals producing cellulose (Blanton et al., 2000; Nakashima et al., 2004).
The evidence for an HGT between Actinobacteria and brown algae is not limited to alginate metabolism. Phylogenetic analyses in the first part of this study (Michel et al., 2010) suggest that the capacity of brown algae to synthesize mannitol was also acquired from Actinobacteria. In addition, Ectocarpus contains a subfamily of GT2 CSL proteins which are conserved with Actinobacteria, fungi and two Phycodnaviruses only (Fig. 2). These proteins emerge as an independent clade robustly rooted with the actinobacterial representatives of this family-2 glycosyltransferase (Fig. 2), suggesting that brown algae and fungi independently acquired these CSLs by HGT from an ancestral actinobacterium. The Phycodnaviruses, which are double-stranded DNA viruses infecting algae (Dunigan et al., 2006), probably obtained these genes from brown algae or related Stramenopiles.
The uniqueness of brown algae is further highlighted by their cell wall remodelling enzymes. Ectocarpus does not feature any homologs of known cellulases, XTHs, alginate lyases or fucanases. The different families of plant expansins are also absent. Only the carbohydrate sulfatases potentially involved in the modification of sulfated fucans are found in the Ectocarpus genome. After the acquisition of cellulose from red algae, then of alginate and hemicellulosic material from Actinobacteria, brown algae may have evolved different molecular mechanisms for wall stress relaxation and expansion. More likely, they have evolved novel cellulases, hemicellulases and alginate lyases which are too derived to be detected by comparative genomics. If this is the case, a variety of new CAZY families and/or activities awaits to be discovered in the brown algae.
Insights into the origin and evolution of ECM polysaccharides in Eukaryotes
The loss of ester-sulfate substituents in the plant matrix polysaccharides, a consequence of land colonization All multicellular marine algae feature sulfated polysaccharides as major ECM components, such as ulvans and sulfated galactans in green algae (Lahaye & Robic, 2007; Farias et al., 2008), sulfated galactans in red algae and sulfated fucans in brown algae (Kloareg & Quatrano, 1988). Irrespective of their phylogenetic position, unicellular marine algae also produce extracellular sulfated polysaccharides in large amounts, including green microalgae (Sieburth et al., 1999), red microalgae (Simon-Bercovitch et al., 1999), diatoms (Hoagland et al., 1993) and haptophytes (Fichtinger-Schepman et al., 1981). Fossil data support that the first plants appeared on emerged land c. 500 million yr ago, diverging from a pioneer green alga related to extant Charales (Bateman et al., 1998). The ancestors of land plants faced several new constraints. Terrestrial environments provide no buoyancy. In their competition for light, land plants thus have evolved several specific characters, such as secondary cell wall reinforcement by lignins (Martone et al., 2009) and stronger cellulose fibers. The arrangement of the terminal complexes in rosettes, from the Characean algae and onwards in all of the terrestrial Archaeplastida, is notably recognized as an important innovation in the evolution of upright, nonaquatic plants (Graham et al., 2000).
Another major environmental change was the lower salinity, including a dramatic decrease in the availability of sulfate ions. The concentration of sulfate in seawater is high, 28 mM, whereas the concentration of this anion varies from 0.09 to 1.40 mM in freshwater and soil interstitial waters (Schmidt-Nielsen, 1997). In contrast with marine algae, the cell walls of freshwater and land plants do not contain any sulfated polysaccharides. Here, we have shown that the synthesis of sulfated polysaccharides is an ancestral metabolism in Eukaryotes. Notably, brown algae possess carbohydrate STs, FGEs and fgSs. In Archaeplastida, the presence of FGEs and fgSs has been established in the green microalga C. reinhardtii (Merchant et al., 2007). Volvox carterii is also known to express at least one fgS gene (Hallmann & Sumper, 1994), probably involved in the metabolism of its ECM sulfated glycoproteins (Ertl et al., 1989). By contrast, FGEs, fgSs and carbohydrate-specific STs are completely absent from the available genomes of land plants. We propose that FGE and fgS, as well as carbohydrate-specific STs, were present in the common ancestor of volvocine algae and land plants, and that these enzymes, and hence the capacity to synthesize and remodel sulfated polysaccharides, were lost on adaptation to sulfate-scarce freshwater and terrestrial environments. However, it is difficult to determine whether this process was completed before or after the transition from freshwater to the land. A similar, convergent evolution probably occurred within the Stramenopiles: the matricial polysaccharides of terrestrial Oomycetes, essentially β-(1,3)- and β-(1,6) glucans (Bartnicki-Garcia, 1968), are not sulfated and the Phytophthora genomes do not contain FGE and fgS genes.
The evolution towards complex multicellularity involved the expansion of ancestral matrix polysaccharides and the acquisition of skeletal fibers Based on the phylogenetic relationships of the various ECM polysaccharides depicted above, the ECM polysaccharide components of archetypal Eukaryotes probably consisted of sulfated polysaccharides and relatively short-chained anionic or neutral glycans. The latter probably included extracellular β-1,3-glucans, which are widespread in extant Eukaryotes, and which were retained as intracellular carbon storage compounds in the Stramenopiles (Michel et al., 2010). β-1,3-Glucans are capable of self-assembling into triple helices, but these fibers provide weaker biomechanical support than cellulose fibers (Stone & Clarke, 1992; Burgert, 2006). Hence, the ECM matrix of early Eukaryotes probably was made up of relatively flexible material, which could not support the building of organisms with complex morphology (Fig. 8).
In Archaeplastida and Stramenopiles, the acquisition of cellulose through the primary and secondary plastid endosymbioses, respectively, brought the capacity to build longer and stronger crystalline fibers. This gave rise to denser and stiffer ECM materials, which made it possible to erect more rigid and complex organisms (Fig. 8). This speciation probably also involved a marked expansion of the matrix polysaccharides from the common stock inherited from the hosts, in order to provide the various cements embedding and cross-linking of the cellulose fibers. Extant multicellular marine algae and land plants indeed feature a large variety of noncrystalline polysaccharides, which is reflected in the high diversity of CAZYmes and CAZYme families in the Arabidopsis (Henrissat et al., 2001) and Ectocarpus (Michel et al., 2010) genomes. Similar conclusions have been drawn, at the ECM matrix protein level, for the green algae of the lineage of Volvocines, an interesting example of a relatively recent (c. 50 million yr) transition towards multicellularity (Kirk, 2005).
Brown algae provide another interesting case study, in which the evolution towards complex multicellularity involved an independent HGT event with an ancestral actinobacterium. Here, we show that this HGT, which must have occurred after the divergence of the ancestor of brown algae from diatoms and Oomycetes, resulted in the acquisition of alginate, the main gel-forming polysaccharide in this lineage (Fig. 8), as well as new hemicellulose-like material. This provided a wealth of new molecular combinations to bridge the cellulose microfibrils with the other wall components, including alginate and sulfated fucans (as sketched in Fig. 1), as well as molecular bases for nonself recognition following pathogen attack (Küpper et al., 2001, 2002). The importance of alginate in the general biology of extant brown algae is reflected by the large number of MC5E genes: 28 in E. siliculosus, a filamentous alga (this study), and a minimum of 45 in the more complex, parenchymatous kelps (Roeder et al., 2005; Tonon et al., 2008). Another possible illustration of the key biological functions of alginate is the high frequency in the Ectocarpus genome of WSC protein domains, which may be involved in alginate–protein interactions. Finally, the HGT from Actinobacteria also provided a crucial storage compound, mannitol (Michel et al., 2010), the carbon translocation form in extant brown algae. Altogether, this HGT event was a turning point in the evolution of brown algae towards the acquisition of complex multicellularity.