|Home | About | Journals | Submit | Contact Us | Français|
The brown algae (Class Phaeophyceae) are morphologically and geographically diverse. The multicellular Phaeophyceae range from microscopic filaments to the largest alga on earth, the giant kelp, Macrocystis pyrifera, reaching tens of meters in length. Macrocystis pyrifera is a dominant space competitor on temperate rocky reefs and provides a three-dimensional structure that supports many fish and invertebrate species (Dayton, 1985). Throughout its large geographic range, M. pyrifera is exposed to depth-dependent gradients of light, temperature, and nitrate conditions that vary depending on multiple temporal and spatial scales. Single individuals spanning the water column may experience seasonal to hourly variations in these conditions (Dean, 1985; Konotchick et al., 2012). Despite reaching great heights, M. pyrifera does not possess a vascular system for metabolite transport as plants do. Instead, organic material (i.e. mannitol and amino acids) is transported via specialized conducting cells with large pores called sieve tube cells (Parker, 1966; Manley, 1983). Macrocystis pyrifera may balance internal carbon and nitrogen through mannitol transport downward and amino acid transport upwards, especially under limiting conditions such as during typical southern California summer stratified water column conditions when upper parts of the kelp are not exposed to nitrate (Colombo-Pallotta et al., 2006).
The photosynthetically available light for kelp is greatest at the surface and exponentially decreases with depth with c. 1% of surface light levels typically reaching 20 m in the kelp forest (Gerard, 1984). Canopy shading by the kelp itself and particulate matter (both organic and inorganic) negatively affect light penetration, leading to decreased carbon fixation and growth by M. pyrifera blades (Towle & Pearse, 1973; Dean, 1985; Wing et al., 1993). While low light limits photosynthesis, excess light can damage photosystem II, leading to oxidative stress. In response, algae have adaptations to optimize light absorption under variable light availability through differential expression of light-harvesting complexes, some of which are involved in both light capture and photoprotection (Savard et al., 1996; Peers et al., 2009).
Biological processes in kelp affected by nitrate include recruitment, growth, survivorship, reproductive output and stress tolerance (Mann, 1973; Jackson, 1977; Zimmerman & Kremer, 1986). Thus, the amount of nitrate exposure can have a large effect on the condition and the existence of a kelp bed. There is a strong linear relationship between nitrate concentration and temperature (°C) in the southern California nearshore at temperatures below 14.5°C, with colder waters possessing more nitrate (Kamykowski & Zentara, 1986; Dayton et al., 1999; Konotchick et al., 2012); at temperatures above 14.5°C, nitrate concentrations are not generally detectable. Seasonally, in southern California, surface nitrate concentrations are low for most of the year but increase during the winter when the water column is well mixed or upwelling occurs (Jackson, 1977).
Kelps respond physiologically on different temporal and spatial scales to their physical environment. For example, during summer months and El Niño years, temperatures rise, nutrients decrease and ultimately kelp densities decrease (Jackson, 1997; Dayton et al., 1999). Outward phenotypic signs of stress may include sloughing of tissue, increased epiphytic growth, disease and ultimately senescence. These are the final result of the sum of kelp physiological responses to stressors over an integrated amount of time (e.g. months to years). Physiological responses of M. pyrifera on much shorter time-scales (i.e. time scales of minutes to hours), such as gene expression responses to vertical nitracline shifts of 10 m or more in a few hours (Konotchick et al., 2012), before these conspicuous phenotypic changes manifest is unknown.
Genomic information is sparse for macroalgae (Jamers et al., 2009) and for the Phaeophyceae in particular (Phillips et al., 2008). Macrocystis and the Phaeophyceae belong to the Stramenopile lineage which is thought to have diverged from other major eukaryotic groups over a billion years ago (Douzery et al., 2004; Yoon et al., 2004). As a result, unique metabolic and developmental features have evolved (Cock et al., 2010a,b; Michel et al., 2010). Phaeophytes are evolutionarily distant from other photoautotrophs, thus the applicability of knowledge learned from terrestrial model species such as the plant Arabidopsis remain somewhat limited for this group (Peters et al., 2004; Keeling et al., 2005). Recognizing the need for model brown algal species, researchers sequenced the genome of the filamentous brown alga Ectocarpus siliculosus (Peters et al., 2004; Waaland et al., 2004; Charrier et al., 2008; Cock et al., 2010b). In brown algae, expressed sequence tags (ESTs) have been used to identify specific genes and expression patterns involved in various processes such as specific life stages or various experimentally manipulated conditions such as stressful light or temperature (Crepineau et al., 2000; Roeder et al., 2005), determine phylogenetic analysis/evolutionary relationships (Phillips et al., 2008; Nam et al., 2011), investigate carbon metabolism and develop sequence-based molecular toolkits (Moulin et al., 1999).
The current increasing trend in genomic information in the marine realm provides an opportunity to expand our knowledge of the Phaeophyceae. Decreased sequencing costs and an increase in sequencing capabilities make it possible to examine species beyond the typical or traditional model and laboratory organisms. A major assumption of comparative biology is that the more closely two organisms are related, the more they will share molecular, biochemical and morphological features (Keeling et al., 2005). Thus, the presence of both single-celled and multicellular heterokont genomes (i.e. the diatoms Phaeodactylum tricornutum, Thalassiosira pseudonana and Aureococcus anophagefferens and the phaeophyte Ectocarpus siliculosus) provides phylogenetic context, potential model systems for detailed functional information and a new opportunity for comparative genomics and gene discovery in this group (Armbrust et al., 2004; Bowler et al., 2008; Cock et al., 2010b; Gobler et al., 2011).
In order to investigate gene expression in the giant kelp and make tangible functional links to its closest relatives we needed species-specific data. For example, while the canonical pathways for nitrate uptake, reduction and incorporation into proteins are well known, the gene models and transcript sequences for these important enzymes in M. pyrifera are not. In this study, we use a pyrosequencing-based RNA-Seq approach to investigate the transcripts from four M. pyrifera libraries spanning the water column and seasons. Objectives of this study were to: increase the number of annotated transcriptional units (TUs) for M. pyrifera, develop sequence-based tools for ecophysiological study, and examine physiological patterns in response to light, temperature and nitrate, with depth.
Blade tissue was sampled from M. pyrifera (Linnaeus) C. Agardh individuals in La Jolla, California, USA (32°51.0 N; 117°17.5 W) on 7 January 2009 and 31 July 2009 using SCUBA. On each date, pieces of blade tissue were collected at the sea surface (0 m) and at 18 m depth, along the same stipe, for a total of four pieces of blade tissue used for transcriptomic library preparation. Surface blades were collected at least 1 m away from the apical growing region and the sampling at 18 m avoided the reproductive sporophylls near the base of an individual so that only the sporophyte generation was sampled. For consistency and to minimize within-blade variability, all samples were collected near the base of each blade. Any wounding effects were assumed to be consistent across samples. At the end of each dive, blades were immediately cleaned of any visible epiphytes by scrubbing with 100% ethanol and cheesecloth and then frozen on dry ice for transport to the laboratory.
To measure temperature changes and to infer nutrient concentration, thermistor chain data were collected at 10-min intervals using TidBit temperature data loggers with c. 0.2° resolution, and c. 5 min response time (Onset, Bourne, MA, USA). TidBits were placed on the bottom (located in 22 m water depth) and at 2, 6, 10, 14 and 18 m above the bottom (mab).
RNA extraction followed a modified protocol (Apt et al., 1995). Frozen algal tissue was ground to a powder on liquid nitrogen then added to an extraction buffer (100 mM Tris–HCl pH 8.0, 1.5 M NaCl, 20 mM ethylenediaminetetraacetic acid (EDTA), 20 mM DTT and 2% cetyltrimethylammonium bromide (CTAB)) at 1 : 1 w : v ratio and mixed at room temperature (RT) for 15 min then heated to 65°C for 20 min. This was followed by a ½-volume chloroform extraction, 5 min at RT, centrifugation at 10 000 g for 30 min at 4°C and collection of the supernatant. Addition of 1/3 volume ethanol was used to precipitate polysaccharides, which was followed by a second chloroform extraction. 3.0 M LiCl and 10% v : v β-mercaptoethanol was added to the aqueous phase and placed at −20°C overnight. RNA was precipitated by centrifugation at 14 000 g for 30 min at 4°C and followed by two 75% ethanol washes before resuspension.
Total RNA was cleaned using RNeasy mini Kit and the optional DNase digestion (Qiagen). RNA was amplified using MessageAmp II Kit (Ambion) with a second round of amplification. Single-strand cDNA was synthesized using superscript III (Invitrogen, Carlsbad, CA, USA) and oligo(dT) primers, then cleaned using RNAClean to remove salts, unincorporated primers and dNTPs (Agencourt, Beckman Coulter Genomics, Beverly, MA, USA). CloneMiner kit (Invitrogen) was used to synthesize second-strand cDNA and cleaned with AMPure (Agencourt). Size-selected cDNA (0.5–1 kb) from the four libraries was purified using QiaQuick gel extraction kit (Qiagen). A high-sensitivity DNA Assay chip was used to assess quality (Agilent, Santa Clara, CA, USA).
Each of the four libraries was sequenced using pyrosequencing technology (454 Life Sciences, Roche, Branford, CT, USA). Reads were filtered using cd-hit-454 (Teal & Schmidt, 2010) and assembled using newbler 2.3 (454 Life Sciences). Open reading frames (ORFs) were called using fraggenescan (Rho et al., 2010) and functionally annotated for pfams and tigrfams using hidden Markov model (HMM) searches (e-value cut-off 1e−5). Kegg orthologs (KO) annotations resulted from blast against Kegg (e-value cutoff 1e−5; v. 20110830). Phylogenetic annotation was accomplished using an internal JCVI database, PhyloDB, which contains all completed and draft algal genomes as well as all Phaeophyceae EST libraries available in NCBI GenBank for annotation. Differential expression of ORFs was evaluated using an empirical Bayes method (Wu et al., 2010).
For each sample, 1 μg RNA was purified using RNeasy (Qiagen) and 200 ng was reverse transcribed using QuantiTect (Qiagen) and diluted 50 fold. RNA for qPCR came from the same extraction used to create the transcriptomic libraries. The PCR products were quantified from the same batch of cDNA to minimize experimental variation caused by the cDNA synthesis process. Samples were run on a 7900HT Fast Real-Time PCR system and 7500 Fast Real-Time PCR systems (Applied Biosystems, Carlsbad, CA, USA). Using the annotations of the M. pyrifera ESTs to identify target TUs, we created primers for a variety of potential housekeeper genes and genes of interest (see the Supporting Information Table S1; Le Bail et al., 2008). After exclusion of those with poorly amplified products, 16 potential reference genes were evaluated to select the most stable reference genes (Vandesompele et al., 2002). We used eukaryotic initiation factor 2 alpha subunit (IF2A) and a protein required for 18S rRNA maturation and 40S ribosome biogenesis (18Smat) for housekeeper genes based on this analysis and their Ct range. The majority of candidate M. pyrifera sequences were aligned with the corresponding E. siliculosus sequence and primers were created in the region of overlap using primer 3 (Rozen & Skaletsky, 2000) with a target length of 100–150 bp and a max Tm difference < 8°C. We quantified expression levels using quantitative real-time PCR via the delta-delta cycle threshold method (ΔΔCt) and reported the data as fold-change comparisons between the surface and 18 m depth; significance was assessed using a two-tailed Welch's t-test, df = 2, P ≤ 0.05.
To comparatively assess the currently available Phaeophyceae genomic repertoire and the addition of Macrocystis pyrifera dataset, we clustered sequences using the OrthoMCL algorithm, as it was reviewed as exhibiting the best overall balance between sensitivity and specificity for eukaryotic orthology detection, which can be difficult because of the complex protein domain structure in eukaryotes (Li et al., 2003; Chen et al., 2007). The species included in each group in these analyses are listed in Table S2.
Sequences from four transcriptomic libraries totaling 228 Mbp yielded an average of 118 000 reads per library after filtering artificial replicates (Gomez-Alvarez et al., 2009), an average read length of 323 bp and a GC content of 48% (Table 1). The GC content falls within the range of reported values of 39.9–54% for the Phaeophyceae (Le Gall et al., 1993). Reads (71% of total) were assembled de novo into 9147 isotigs, or transcript variants, with an N50 size of 817 bp. The average isotig size was 642 bp with a maximum of 2872 bp. For comparison, E. siliculosus, a small filamentous intertidal brown alga, has c. 16 000 genes, of which 9601 had EST support from six cDNA libraries corresponding to different developmental stages and growth conditions (Cock et al., 2010b). We found 67.8% of isotigs in all four libraries (Fig. 1), indicating a large active transcriptional core across environmental gradients, providing confidence in the de novo assembly and in the ability to examine expression differences in transcripts common to all libraries. Only 2.3% of isotigs were only found in one library (Fig. 1).
Of 11 844 ORFs called, 5.6% had HMM-derived pfam annotations (Table 1); 11.2% of ORFs had taxon ID matches in PhyloDB and E. siliculosus, the only phaeophyte with a sequenced genome, had the highest percentage of matches (72.6%). The next highest percentage of best matches was to Laminaria digitata (11.3%). There is about c. 20-fold less genomic data available for L. digitata than there is for E. siliculosus (Table 1); however, L. digitata is taxonomically related to M. pyrifera at the family level, while E. siliculosus is related at a class level. The similarity of orthologous proteins between M. pyrifera and L. digitata and E. siliculosus was 86.8% and 79.5%, respectively (Fig. S1).
Open reading frame-level clustering as a preliminary analysis of phaeophyte protein diversity reveals novel clusters distinct from their closest relatives (Fig. 2). To assess the level of Phaeophyceae sequence data to date, we first compared EST sequences from the Phaeophyceae (not including M. pyrifera sequences) with the Ectocarpus genome, Stramenopiles, green algae and red algae (Fig. 2a, Table S2). Phaeophyceae ESTs grouped into 2026 clusters, 543 of which were unique to Phaeophyceae ESTs (exclusive of Ectocarpus and Macrocystis) and not shared with Ectocarpus, Stramenopiles, green algae or red algae; annotations for these clusters are listed in Table S3. Five hundred and sixty-two clusters were shared between all groups; examples of these shared annotations include ubiquitin, ribosomal proteins, ATP synthases and photosynthesis-related genes. Phaeophyceae ESTs also shared a large number of clusters with only Ectocarpus (313), and with clusters shared by Ectocarpus and Stramenopiles (176).
We then compared the addition of the M. pyrifera EST data generated in this study. In this analysis, the Phaeophyceae includes all the brown algal ESTs and the Ectocarpus genome (Fig. 2b). Macrocystis pyrifera ORFs were grouped into 1692 clusters, 925 of which were only found in M. pyrifera and not in the other groups; none of these clusters had an annotation (Table S3). The distributions with the highest numbers of clusters were shared with all groups (334), shared with only the brown algae (183) and shared with only the brown algae and the Stramenopiles (88). Comparison of the M. pyrifera clusters with only the brown algae ESTs (and not the Ectocarpus genome) in a similar Venn analysis is shown in Fig. S2 and Table S3. Novel clusters from Figs 2 and S2 with differential expression patterns are also indicated in Table S3. Comparisons of the phaeophyte sequences generated to date (including this study) indicate a high level of novel protein discovery in this group.
Many of the top ORF pfam descriptions (in terms of normalized summed read counts across all libraries) were to ribosomal proteins. A fold-change comparison of ORFs with ribosomal pfam descriptions between surface and depth for both seasons is shown in Fig. 3, indicating a high level of protein synthesis at depth. Top nonribosomal ORF pfams included a photosynthetic reaction center protein, chaperonin, ATP synthase and heat shock protein; for a more complete list see Table 2 (top 30) and Table S4 (top 31–100). Taxon matches were to mainly to Phaeophyceae members and to a few diatom species. Patterns between surface and depth emerged through differential comparison of expression with each season (Table S5). Transcripts upregulated in the surface included those encoding photosynthesis and damage repair proteins (e.g. photosynthetic reaction center protein, chlorophyll-binding protein, Hsp 70 protein, PsbP), while at depth ribosomal proteins dominated.
Light-harvesting genes were examined because their ORF counts showed differences between libraries in the pyrosequencing transcriptome and there is more knowledge about these genes than other genes in the Phaeophyceae class (Green et al., 1991; Green & Durnford, 1996; Dittami et al., 2010). blast searches of the transcriptome assembly identified eight contigs coding for putative light-harvesting complex (LHC) proteins (Table S6). For comparison, 53 LHCs were found in the genome of E. siliculosus (Dittami et al., 2010). A phylogenetic analysis of the M. pyrifera LHC and the LHC from other heterokonts grouped the M. pyrifera LHC isotigs into canonical LHC groups FCP, LHCR and LHCSR (Table S6). In both qPCR experiments across depth gradients, most MpLHC showed significantly greater qPCR expression at the surface than at depth (up to 530 fold change; Fig. 4).
Because of the large environmental differences in light and temperature/nutrient conditions with depth (Fig. 5), we evaluated various stress response, photosynthesis, carbon metabolism and nutrient uptake genes using qPCR (Table S1). Antioxidant genes showed patterns of higher expression at the surface during both seasons for a protein of the periredoxin group (ATPRX_Q), l-ascorbate peroxidase and vanadium bromoperoxidase (Fig. 6). ATPRX_Q showed the greatest fold-change in the surface compared with depth in both seasons. Two photosynthesis/carbon fixation genes showed higher surface expression in both seasons: ribose-5-phosphate isomerase (r5pi), which is involved in the pentose phosphate pathway and produces sugars and NADPH, and the glycine decarboxylase complex (GCS). In January, several other photosynthesis-related genes showed upregulation in the surface: uroporphyrinogen decarboxylase (involved in porphyrin and chlorophyll metabolism), PsbP, the photosystem II oxygen evolution complex protein required for PSII to be fully operational, and malate dehydrogenase. Mannitol-1-phosphate 5-dehydrogenase, involved in metabolism of mannitol, one of the main storage carbohydrates in M. pyrifera, was upregulated in the surface in both seasons. Spermidine/spermine synthase, a regulator in stress-signaling pathways (Kasukabe et al., 2004), showed higher expression in the surface in January. By contrast, nitrite reductase, the enzyme catalysing the conversion of nitrite into ammonia, showed higher expression at depth in both seasons. Nitrite reductase is usually regulated in coordination with nitrate reductase (catalyses the nitrate → nitrite conversion), as was the case in the January libraries.
The presence of ‘novel’ gene families in Phaeophyceae is supported by existing and new data (Figs 2, S2, Table S3). The sequences generated in this study clustered more often with the most closely related groups (Figs 2b, S2). The low level of pfam annotation (5.6%) and taxon ID match (11.3%) in our de novo assembly might suggest spurious gene discovery. However, the detection of the majority of isotigs in all libraries (Fig. 1), high identity level to other Phaeophyceae members (Fig. S1) and GC content in the range expected for Phaeophyceae suggests otherwise. Low annotation percentages based on blast approaches for brown algal ESTs have been seen before (i.e. 6.5% for Choristocarpus; Phillips et al., 2008). In addition, transcripts with assigned annotations and examined with qPCR showed patterns consistent with their environmental context (discussed below). Knowledge remains limited because the majority of genes (including those in novel clusters) cannot be assigned a function through traditional blast-based approaches. Linking ecological and physiological studies with gene expression data in an iterative manner can improve future annotations. Specifically, the categorization of unannotated clusters could serve as a basis for comparison with other populations on the California coast that experience different physical regimes. Insights from comparative genomics will continue to improve with the addition of more reference genomes and more annotated genes.
The transcripts from this study were collected from the blades of the M. pyrifera sporophyte generation. With additional cDNA sequencing of the gametophyte generation (Roeder et al., 2005), reproductive sporophylls, the apical growing region or other morphological parts, it is reasonable to expect that the number of annotated TUs can further expand the knowledge of the gene repertoire of M. pyrifera. Based on the high protein discovery seen thus far in the Phaeophyceae, is reasonable to assume that exploration of additional taxonomic members could yield insights into potentially unique evolutionary and physiological traits.
Our broad untargeted transcriptomic view allows us to begin to identify key mechanisms in the physiological and metabolic reactions to a variable environment. The identification of transcriptional units and development of primer sets for examining expression of select metabolic genes enables targeted hypothesis testing of physiological response to environmental conditions. Both transcriptional count data (Table S5) and quantitative PCR validation (Figs 4, ,6)6) provide support for the hypothesis of Colombo-Pallotta et al. (2006) where there is localization of different metabolic processes occurring in different parts of the kelp (Fig. 5).
At the surface, where irradiance levels are highest and the potential for oxidative damage is most intense, physiological processes are focused on protection from the damaging effects of the sun as well as capture of light energy for photosynthesis. Several stress-related proteins showed higher qPCR expression levels at the surface in both seasons, including antioxidant proteins (i.e. ATPRX_Q) and peroxide proteins (i.e. l-ascorbate peroxidase and vanadium bromoperoxidase). Ascorbate peroxidases are associated with photo-oxidative stress in algae (Ishikawa & Shigeoka, 2008). Vanadium bromoperoxidases are found in several species of brown algae and are also thought to be involved in biotic and abiotic stress responses (La Barre et al., 2010). The intense and variable light environment at the surface upregulates stress-response genes.
Light appears to regulate gene expression strongly in M. pyrifera. The high expression at the surface of the LHCSR MpLHCs is consistent with a role in photoprotection (e.g. nonphotochemical quenching, or NPQ) in a highly variable environment (Peers et al., 2009). The NPQ mechanisms in M. pyrifera differ from those of higher plants and may be linked with this LHC group (García Mendoza & Colombo-Pallotta, 2007; García Mendoza et al., 2011). Having multiple LHCs is thought to facilitate the ability to fine-tune control of photosynthesis electron flow for blades exposed to intermittent high and low irradiances at the surface. More than half of the light-harvesting complex genes showed significantly higher expression in the surface than at depth, with those in the LHCSR clade having the largest fold-change differences (Fig. 4). Other studies show functional differentiation between surface and basal blades in terms of photobiology, including enhanced photoprotection, higher maximum photosynthetic rate, higher photosystem II (PSII) electron transport rate, decreased pigment concentration and lower photosynthetic efficiency in surface blades compared with basal blades (Gerard, 1986; Smith & Melis, 1987; Colombo-Pallotta et al., 2006).
Deeper blades tend to be shaded by the canopy and are generally acclimatized to low light conditions. Physiological adaptations of M. pyrifera to gradients in light include higher relative amounts of chlorophyll and fucoxanthin and larger antennae sizes in blades found at 20 m compared with surface blades (Smith & Melis, 1987; Colombo-Pallotta et al., 2006). Blade relocation and canopy removal experiments have shown that differences in light use efficiency likely resulted from acclimatization to light conditions rather than age of blades (Gerard, 1986). The majority of photosynthesis occurs in the surface canopy with half of the standing crop lying between the surface and 1.5 m as fronds reaching and spreading along the surface (McFarland & Prescott, 1959; Towle & Pearse, 1973; Gerard, 1986). The daily increase in the dry weight of the growing tip (apical region) is greater than the products of its photosynthesis; in addition, blades further down the stipe have rates of photosynthesis that exceed their rate of growth, implying likely transport of materials (Sargent & Lantrip, 1952).
At the base of the kelp, where nutrient levels tend to be much higher than at the surface, especially during the stratified summer months, incorporation of nitrogen into amino acids and ultimately proteins may be a dominant process (Table S5). There was upregulation of nitrite reductase (NiR) at depth during both seasons, which catalyses the conversion of nitrite into ammonium (Fig. 6). M. pyrifera has the capacity for long-term storage of carbon and nitrogen in excess of immediate demands, which would allow for the decoupling of growth from immediate ambient conditions (Gerard, 1982). Thus, nutrient uptake may not only be dependent on current nutrient availability but also previous nutrient exposure. Although dissolved nitrate levels are typically elevated at depth, temperatures below 14.5°C occurred at both the surface and depth in January, suggesting the presence of biologically available levels of nitrate at both depths (Fig. 5). Expression of NR in January was higher at depth, perhaps indicating regulation by light (Dohler et al., 1995), elevated nutrient levels or other factors. Factors such as nutrient uptake metabolites, time of day, light and CO2 concentration can affect nutrient uptake, transcription and activity of nitrate reductase (Crawford et al., 2000). Increased dissolved inorganic nitrogen pools at depth could drive higher levels of protein synthesis, as suggested by upregulation of genes encoding ribosomal proteins and associated increases in cellular nitrogen demand and transcript levels for genes encoding proteins involved in nitrate assimilation.
It is difficult to say whether the individuals sampled in this study are representative of general patterns or what is driving the observed expression patterns without follow-up experiments of sufficient spatial and temporal coverage; these experiments are now possible using the reference transcriptome generated here. In addition, owing to our method of sampling along a stipe, surface blades are of younger age than the deeper blades, which may have an affect on gene expression; to decouple the effects of age with physiological response, blade translocation experiments could be conducted using these molecular tools. Finally, it is important to note that this study focused on transcriptomic data, and important insights from genomic or proteomic modifications are not addressed in this study.
Natural populations of M. pyrifera inhabit a variable physical environment and yet we were able to see strong gene expression patterns emerge. Differences in depth of dominant metabolic processes were apparent in these natural samples. Transcript levels were consistent with environment-regulated ecophysiology and this study provides a starting point for future studies coupling gene expression with finer-scale environmental measurements and other populations. Efforts to tease apart variables that co-vary together in nature (e.g. temperature and nitrate) can begin to be addressed using these tools. Through transcriptional profiling of the giant kelp, we have created a sequence-based tool kit for further physiological study of this species and demonstrate differences in metabolic function across environmental gradients with depth.
The work was made possible by funding support from a National Science Foundation Graduate Research Fellowship, the Mia J. Tegner Fellowship for Coastal Ecology Fieldwork, and a Sigma Xi Grant-in-Aid of Research (T.K.). Work in the laboratory of A.E.A. was supported by NSF-OCE-1136477, NSF-MCB-1024913, DOE-DE-SC0006719, and by the Beyster Foundation. The Bioinformatics Ghent Online Genome Annotation Service (BOGAS) for the E. siliculosus genome annotation greatly aided analyses of M. pyrifera. We thank Scripps Institution of Oceanography's scientific diving program for diving support in the field. We acknowledge the efforts of J. Bai, H. Zheng and M. Lehmann. Comments from and discussion with G. Peers, J. Leichter, L. Levin and three anonymous reviewers significantly improved the manuscript.
Additional supporting information may be found in the online version of this article.
The sequences in this study have been deposited in NCBI's Sequence Read Archive (SRA) under the following accession numbers: SRX220272, SRX220273, SRX220275, and SRX220277, found in SRA Study SRP018250).
Fig. S1 Percent similarity at the protein level between Macrocystis pyrifera and Ectocarpus siliculosus (blue) and Laminaria digitata (green) using BLAST for alignments with > 75% of the total read length.
Fig. S2 Distribution of clusters of all EST-derived Macrocystis pyrifera ORFs with the brown alage split up by genus: Laminaria, Fucus (2 species), Sargassum and Ectocarpus. Annotations (when available) for clusters on the outside of the Venn are listed in Table S3.
Table S1 Macrocystis pyrifera quantitative PCR targets, function, primer sets, and closest Ectocarpus siliculosus sequence match
Table S2 Genomes and ESTs used in comparative cluster analysis and their sources
Table S4 Continuation of Table 2: the top 31–100 open reading frames (ORFs) with non-ribosomal pfam annotation ranked by total reads across the four libraries, the pfam description, the best species match, and significantly differential expression patterns in January and July
Table S5 Differentially expressed open reading frames (ORFs) with pfam annotation and summed read counts across the four libraries, the best species match, and significant differential expression patterns in January and July
Table S6 Light-harvesting complexes identified from isotigs in Macrocystis pyrifera
Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.