A Palmitic Acid Elongase Affects Eicosapentaenoic Acid and Plastidial Monogalactosyldiacylglycerol Levels in Nannochloropsis

A heterokont palmitic acid desaturase initiates the production of eicosapentaenoic acid in the endoplasmic reticulum of Nannochloropsis, which is specifically used for monogalactosyldiacylglycerol assembly in the secondary plastid.

Abstract

Nannochloropsis species are oleaginous eukaryotes containing a plastid limited by four membranes, deriving from a secondary endosymbiosis. In Nannochloropsis, thylakoid lipids, including monogalactosyldiacylglycerol (MGDG), are enriched in eicosapentaenoic acid (EPA). The need for EPA in MGDG is not understood. Fatty acids are de novo synthesized in the stroma, then converted into very-long-chain polyunsaturated fatty acids (FAs) at the endoplasmic reticulum (ER). The production of MGDG relies therefore on an EPA supply from the ER to the plastid, following an unknown process. We identified seven elongases and five desaturases possibly involved in EPA production in Nannochloropsis gaditana. Among the six heterokont-specific saturated FA elongases possibly acting upstream in this pathway, we characterized the highly expressed isoform Δ0-ELO1. Heterologous expression in yeast (Saccharomyces cerevisiae) showed that NgΔ0-ELO1 could elongate palmitic acid. Nannochloropsis Δ0-elo1 mutants exhibited a reduced EPA level and a specific decrease in MGDG. In NgΔ0-elo1 lines, the impairment of photosynthesis is consistent with a role of EPA-rich MGDG in nonphotochemical quenching control, possibly providing an appropriate MGDG platform for the xanthophyll cycle. Concomitantly with MGDG decrease, the level of triacylglycerol (TAG) containing medium chain FAs increased. In Nannochloropsis, part of EPA used for MGDG production is therefore biosynthesized by a channeled process initiated at the elongation step of palmitic acid by Δ0-ELO1, thus acting as a committing enzyme for galactolipid production. Based on the MGDG/TAG balance controlled by Δ0-ELO1, this study also provides novel prospects for the engineering of oleaginous microalgae for biotechnological applications.

In heterokonts, plastids have emerged from a secondary endosymbiosis event, during which a red alga was engulfed by a eukaryotic host cell. The reduction of the red algal endosymbiont has led to the formation of a photosynthetic organelle surrounded by four membranes, called a complex or secondary plastid (Petroutsos et al., 2014). The lipid composition of membranes constituting this organelle has not been characterized yet. Four glycerolipids are conserved in the photosynthetic membranes from cyanobacteria to primary plastids, i.e. monogalactosyldiacylglycerol (MGDG), digalactosyldiacylglycerol (DGDG), sufoquinovosyldiacylglycerol (SQDG), and only one phospholipid, phosphatidylglycerol (PG; Petroutsos et al., 2014). Based on their detection in whole cell extracts of secondary endosymbionts (Botté et al., 2011b; Simionato et al., 2013; Abida et al., 2015), these four lipids have been postulated to reside in the thylakoids of secondary plastids as well. A striking feature of heterokonts analyzed to date lies in their high content in very-long-chain polyunsaturated fatty acids (VLC-PUFAs), especially eicosapentaenoic acid (EPA, 20:5 ^{Δ5,8,11,14,17}; Simionato et al., 2013; Abida et al., 2015; Meksiarun et al., 2015). In Nannochloropsis, EPA is overrepresented in MGDG, DGDG, PG, phosphatidylethanolamine (PE), and diacylglyceryltrimethylhomo-Ser (DGTS) (Simionato et al., 2013; Alboresi et al., 2016). The biological function of VLC-PUFAs in a redox poise environment like photosynthetic membranes is intriguing, especially since unsaturated FAs are more susceptible to oxidation (Bielski et al., 1983).

FAs are initially synthesized in the stroma of plastids by the dissociated fatty acid synthase of type II (FASII) releasing medium chain fatty acids (MC-FAs), up to a chain length of 16 or 18 carbons. De novo synthesized MC-FAs are either saturated (16:0-ACP, 18:0-ACP) or monounsaturated by a stromal acyl-ACP Δ9-desaturase (16:1^∆9-ACP, 18:1^∆9-ACP). MC-FAs are exported to the cytosol, where they are converted into acyl-CoA (16:0-CoA, 16:1^∆9-CoA, 18:0-CoA, 18:1^∆9-CoA; Li-Beisson et al., 2010). The generation of VLC-PUFAs then occurs at the ER by multiple hetero-tetrameric elongase complexes, catalyzing stepwise reactions adding two carbons to an acyl-CoA substrate (Leonard et al., 2004; Sayanova and Napier, 2004; Hamilton et al., 2014). The term elongase (ELO) refers to the first enzyme, the β-ketoacyl-CoA synthase (KCS), of which there are two structural different enzyme classes: Elo-like (ELO) KCSs are present in every phylum, whereas fatty acid elongase 1 (FAE1)-type KCS are found only in plants. Organisms are usually equipped with a subset of KCS proteins that display a range of substrate specificities and expression pattern (Haslam and Kunst, 2013).

Elongation of 18:0 or 18:1^Δ9 has not been reported in heterokonts (Arao et al., 1994; Arao and Yamada, 1994; Simionato et al., 2013; Abida et al., 2015; Cook and Hildebrand, 2016). The biochemical steps of the elongation and desaturation pathway from 18:1^∆9 to EPA were initially studied in the diatom Phaeodactylum using pulse chase experiments (Arao et al., 1994). This study revealed the presence of two interacting routes, the ω-6 and ω-3 pathways that share 18:2^∆9,12, as precursor (Sayanova and Napier, 2004). In the ω-6 pathway, 18:2^∆9,12 is desaturated into 18:3^∆6,9,12 (18:3ω-6) by an ER-localized Δ6-fatty acid desaturase (ER∆6FAD), while in the ω-3 pathway the substrate is desaturated twice, first by an ERω3FAD into 18:3^∆9,12,15 (18:3ω-3) and then by an ER∆6FAD into 18:4^∆6,9,12,15 (18:4ω-3). The predominant route in Phaeodactylum was a mix of both pathways in which 18:3ω-6 is desaturated into 18:4ω-3 by an ERω3FAD (Arao et al., 1994). Both 18:3ω-6 and 18:4ω-3 serve as substrates for a ∆6-ELO generating 20:3^∆8,11,14 (20:3ω-6) in the ω-6 pathway and 20:4^{∆8,11,14,17} (20:4ω-3 or eicosatetraenoic acid) in the ω-3 pathway. Both products are substrates for ER∆5FAD giving rise to 20:4^∆5,8,11,14 (20:4ω-6 or arachidonic acid, ARA) and 20:5^{∆5,8,11,14,17} (20:5ω-3, EPA). In addition to this ∆5-desaturation of eicosatetraenoic acid, EPA can be obtained by desaturation of ARA by an ERω3FAD (Arao et al., 1994). The parallel existence of the cross-interacting ω-3 and ω-6 routes is not conserved among heterokonts, since the Thalassiosira pseudonana genomic data allowed the reconstruction of only an ω-3 pathway involving 20:4ω-3 (Cook and Hildebrand, 2016). In Nannochloropsis, the presence of only 20:4ω-6 was reported (Schneider et al., 1995). Once generated in the ER, it is necessary that VLC-PUFAs are reimported into the plastid for the assembly of MGDG, DGDG, SQDG, and PG according to a yet-to-be-characterized process, called the “omega pathway” (Petroutsos et al., 2014).

In previous studies, VLC-PUFA biosynthesis is considered to start from oleic acid (18:1^Δ9). However, the proportions of 16:0, 18:0, and 18:1 that are produced by FASII and exported from secondary plastids are unknown, so these three FAs should be equally considered as possible precursors. Here, we chose Nannochloropsis gaditana CCMP526 as a model to study the effects of an impairment of VLC-PUFA biosynthesis upstream 18:2^∆9,12, the branching point for the ω-3 and ω-6 pathways. We knocked out the gene coding for the major initial ER-retained elongase and analyzed the impact on glycerolipid and sphingolipid metabolism, revealing an unexpected function in EPA synthesis that was specifically used for MGDG production. With the disturbance of the secondary plastid lipid balance, photosynthesis parameters and thylakoid membrane structure were altered. This work indicates that the omega pathway relies on channeling processes starting very early in FA elongation and that this process can be considered as a committing step in galactolipid synthesis.

RESULTS

Bioinformatic and Phylogenetic Analysis of N. gaditana Elongases

We identified eight putative elongases in N. gaditana, encoded by Naga_100083g23, Naga_100162g4, Naga_100004g102, Naga_100017g49, Naga_100399g1, Naga_100162g5, Naga_100003g8, and Naga_100020g80. Elongases are ER-located transmembrane proteins with a 3-ketoacyl-CoA synthase domain (Denic and Weissman, 2007), and we could verify these features in all protein sequences except for those encoded by Naga_100399g1 and Naga_100020g80 that had no Lys-rich C terminus and were predicted to be possibly located in mitochondria. Amino acid sequences of the putative elongases were compared to homologs having annotated function. We then positioned seven of the N. gaditana ELO candidates on a reconstructed FA elongation pathway in N. gaditana (Fig. 1).

Figure 1.

Reconstructed pathway of VLC-PUFA synthesis in N. gaditana. The de novo synthesis of FAs in the chloroplast stroma can generate 16:0, 18:0, and 18:1 precursors, which can be exported to the cytosol. Desaturation of 18:0 into 18:1 can occur via either the action of a stroma stearoyl-ACP Δ9-desaturase (SAD) or an ER fatty acid desaturase (ERΔ9FAD). Seven candidate genes coding for elongases and five genes coding for desaturases were retrieved from the N. gaditana genome. Elongases were predicted to act either on a saturated substrate (Δ0-ELO) or on a polyunsaturated substrate having a double bond at position Δ6 (Δ6-ELO).

There is no consensus for a nomenclature of elongases. We propose here a simple rule where the desaturation site of the substrate would determine the name. Thus, elongases putatively acting on a saturated FA like 16:0 were labeled as Δ0-ELO (Naga_100083g23, Naga_100162g4, Naga_100004g102, Naga_100017g49, Naga_100399g1, and Naga_100162g5), with Δ0 corresponding to the absence of any desaturation in the palmitic acid (16:0) substrate. Elongases acting on a polyunsaturated substrate were labeled as Δ6-ELO (Naga_100003g8) when the substrate harbors a double bond at position Δ6 and Δ5-ELO when the substrate harbors a double bond at position Δ5. Consistently, the Δ0-ELO sequences are similar to T. pseudonana TpElo3, predicted to elongate 16:0 into 18:0 (Cook and Hildebrand, 2015). No putative ∆9- or ∆7-elongases could be found. Figure 1 also shows desaturase genes acting in VLC-PUFA biosynthesis. They were identified using corresponding FAD template sequences from Phaeodactylum (Dolch and Maréchal, 2015).

The phylogeny of Nannochloropsis sequences was reconstructed based on alignments with elongases selected in representative clades of eukaryotes’ biodiversity. All sequences were aligned using MUSCLE (Edgar, 2004), and an unrooted phylogenetic tree was constructed using the Neighbor-Joining method (Saitou and Nei, 1987) with bootstrapped confidence intervals based on 1,000 replications (Fig. 2). The elongases known to use saturated FAs as a possible substrate are labeled with a star. The six putative Δ0-ELO sequences from N. gaditana group into a cluster containing only heterokont sequences, including TpELO3 from Thalassiosira, proposed to act in the elongation of 16:0 into 18:0 (Cook and Hildebrand, 2016). The Naga_100003g8 sequence belongs to a second heterokont cluster including Δ6-ELOs from diatoms. The Naga_10002080 sequence belongs to a third heterokont cluster comprising Δ5-ELOs from diatoms (Fig. 2), but we could not characterize its function since Δ5-elongation could not be detected in any Nannochloropsis species to that date. The heterokont cluster containing Δ0-ELOs is close to a group of elongase sequences from other chromalveolata (cryptomonad and haptophyta) and excavata (kinetoplastida) and to a cluster of apicomplexa elongases of the ELOC-type. The cluster containing human saturated/monounsaturated elongases, that is ELOVL3 and ELOVL6, the latter being responsible for the elongation of 16:0 and shorter (Kihara, 2012), is close to the heterokont Δ0-ELO cluster. The obtained tree shows a dichotomy between saturated/monounsaturated and polyunsaturated FA elongases and is consistent with the position of N. gaditana sequences in the VLC-PUFA biosynthetic pathway shown in Figure 1.

Figure 2.

Unrooted phylogenetic tree of FA elongases. Selected sequences cover the biodiversity of eukaryotes, including Opisthokonts, e.g. Fungi (Saccharomyces) and Metazoa (Drosophila, Musca, Apis, Bombus, Caenorhabditis, Homo, Mus, Gallus), Heterokonts (Phaeodactylum, Thalassiosira, Ectocarpus, Phytophthora, Albugo, Saprolegnia, Aphanomyces), Apicomplexa (Toxoplasma, Neospora, Eimeria, Cryptosporidium, Plasmodium, Gregarina), Haptophytes (Emiliania), Cryptomonads (Guillardia), and Kinetoplastida (Trypanosoma, Leishmania). A star indicates sequences with a characterized (or proposed) activity of elongation of saturated FAs in previous reports (Lee et al., 2006; Jung et al., 2007; Tehlivets et al., 2007; Kihara, 2012; Ramakrishnan et al., 2012; Cook and Hildebrand, 2016). The elongase characterized in this work is shown in red.

The conservation of Δ0-ELO homologs within a heterokont-cluster and the presence of several isoenzymes within N. gaditana indicate an important function of this family of elongases. No heterokont ∆0-ELO has been functionally studied to date to our knowledge. We aimed to verify the proposed scheme of EPA synthesis in N. gaditana experimentally and addressed the question of ∆0-ELO functional redundancy or specificity. To that purpose, we first compared the gene expression profiles of the Δ0-ELO genes.

Sequence Analysis of the Protein Encoded by the Naga_100083g23 Gene (Δ0-ELO1)

We retrieved gene expression data from N. gaditana cultivated under both nitrogen repletion and deficiency (Corteggiani Carpinelli et al., 2014). Naga_100083g23 (Δ0-ELO1) had the highest expression level, and we therefore investigated its gene product in more detail. The Δ0-ELO1 amino acid sequence possesses an elongase domain (Supplemental Fig. S1) comprising seven predicted transmembrane domains (TM1-TM7). The C terminus part is Lys (K) rich, indicating an ER retention. The “HXXH” motif essential in yeast (Saccharomyces cerevisiae) elongases for 3-ketoacyl-CoA synthase activity (Denic and Weissman, 2007) is conserved in the Δ0-ELO1 sequence (highlighted in green in Supplemental Fig. S1). The same motif is present in yeast saturated/monounsaturated FA elongation protein family and is not an absolute signature for the substrate specificity (Hashimoto et al., 2008). A LYF motif (highlighted in purple) is also detected, conserved in yeast Fen1p proteins that elongate FAs of a chain length between C16 and C24 (Denic and Weissman, 2007). We then addressed the function of NgΔ0-ELO1 by heterologous expression in yeast.

Functional Characterization of Naga_100083g23 Gene Product by Heterologous Expression in Yeast

We cloned the codon optimized sequence of NgΔ0-ELO1 in the pYES expression vector and tested the activity of the pYES2-NgΔ0-ELO1 construct on PUFAs. Wild-type yeast clones containing the algal elongase or the empty vector as a control were grown in liquid cultures in the presence or absence of exogenously added long chain (C18) and very long chain (C20 and C22) PUFA substrates. No elongation product was detected for any of the PUFA substrate tested, but we did notice a systematic decrease of C16:0 in the presence of NgΔ0-ELO1 (data not shown). We then investigated a possible elongation activity of NgΔ0-ELO1 on C16:0. To eliminate potentially competing endogenous FA synthase activity, we transformed the pYES2-NgΔ0-ELO1 vector in a fas2 knockout (KO) yeast strain. In yeast, the cytosolic FAs synthase complex produces 16 and 18 carbons FAs, while the yeast scELO1 gene is involved in microsomal elongation of C14:0 to C16:0 (Toke and Martin, 1996). Expressing NgΔ0-ELO1 in the fas2-KO strain resulted in a decrease in 16:0, as previously observed in wild type, and in a small but statistically significant increase in C18:0, suggesting that NgΔ0-ELO1 does elongate 16:0 in yeast (Supplemental Fig. S2A). Since the increase in 18:0 did not match the decrease in 16:0, we suspected that 18:0 might be further elongated, either by NgΔ0-ELO or by the yeast endogenous scELO2/scELO3 elongase complex. To verify this hypothesis, FA methyl esters (FAMEs) extracted after expression of NgΔ0-ELO1 in wild type and fas2-KO strains were this time analyzed on a less polar column more suitable for long-chain hydrocarbons, allowing detection of C26:0 and 2-hydroxy-C26:0, the two major VLCFAs present in yeast. Surprisingly, in wild-type and fas2-KO cells, all VLCFAs >20 carbons and up to C30:0 accumulated at a higher level in the presence of the algal NgΔ0-ELO1 protein (Supplemental Fig. S2B), suggesting an enhancement of endogenous elongation activity and explaining the limited increase in C18:0 observed in the FAS free background (Supplemental Fig. S2A). This enhanced FA elongase activity was not detected in yeast-elo1-KO strains, suggesting that the endogenous scELO1 is required for NgΔ0-ELO1 activity in yeast (data not shown). In the complemented yeast, the presence of C28 and C30 FAs is surprising, as these compounds are normally not synthesized in yeast, with only trace amounts of 28:0 detectable in the fas2-KO/pYES2 control. To verify whether these VLCFAs were the products of NgΔ0-ELO1 itself or of the yeast endogenous elongase, we expressed it in a yeast-elo3-KO strain, which is unable to produce C26:0 (Oh et al., 1997). This confirmed that the enhanced production of VLCFAs observed in the presence of NgΔ0-ELO1 in yeast requires a functional endogenous elongase and that the C28 and C30 FAs are likely produced by the yeast scELO2/scELO3 complex (data not shown). Altogether, these experiments are consistent with the fact that the N. gaditana Δ0-ELO1 protein uses 16:0 as a substrate and that it may be involved in determining the fate of the VLCFAs produced downstream.

Cloning and Transformation of N. gaditana with a Naga_100083g23 KO Cassette

For the targeted gene KO via homologous recombination, we cloned a Naga_100083g23 disruption cassette and transformed N. gaditana wild type (NgWT; Kilian et al., 2011; Perin et al., 2015). The presence of the KO cassette was tested in 31 resistant colonies, of which 29 had both flanking sites inserted. Homologous recombination was verified by PCR in 3 of 15 colonies. In the three independent clones, the presence of a single gene insertion event was confirmed by quantitative PCR (qPCR) performed on genomic DNA (gDNA), based on the similar quantification cycle threshold of the inserted zeocin DNA with a single copy gene reference (PAP, Naga_100038g41). Consistently, Δ0-ELO1 transcript could not be amplified in qPCR performed on cDNA from the three mutant lines. We thus obtained three independent Naga_100083g23 KO lines (KO5, KO13, and KO15) that showed similar total FA profiles (Supplemental Fig. S3) and phenotypes in all assays performed. Data from the KO lines could therefore be pooled and are designated below as NgΔ0-elo1-KO.

Analysis of FAs and Glycerolipids from NgΔ0-elo1-KO Lines Highlights a Reduction of EPA and a Specific Alteration of the MGDG/TAG Balance

We investigated whether FA and lipid profiles could be altered in NgΔ0-elo1-KO mutants compared to the NgWT. Thin-layer chromatography (TLC)-separated lipid classes were introduced by direct infusion into a trap type mass spectrometer to assess all lipid structures and particularly the regioselective localization of acyl groups at the sn-1 and sn-2 positions of membrane glycerolipids and sn-1, sn-2 and sn-3 positions of triacylglycerol (TAG; Supplemental Table S1). FA, polar, and neutral lipids were then quantified by gas chromatography coupled to ion flame detection (GC-FID). NgΔ0-elo1-KO and NgWT had same total FA contents per cell, with an average of 1.72 (± 0.19 SEM) nmol FA/million cells, when harvested after 4 d of cultivation (Fig. 3A). The FA profile of NgΔ0-elo1-KO was different from that of the wild type (Fig. 3B): the proportion of EPA (20:5, Fig. 3B) was significantly lower (P = 0.0072), with a 7.97% (± 0.002 SEM) decrease in NgΔ0-elo1-KO compared to NgWT. The only significant increase was at the level of 18:1 (P = 0.0031; Fig. 3B).

Figure 3.

Comparison of FAs and glycerolipids from NgWT and NgΔ0-elo1-KO lines. A, Total amount of FAs. FA content is expressed in nmol per 1.10⁶ cells. B, FA profiles. The proportion of each FA is indicated based on the carbon chain-length and number of desaturations. Statistically significant alterations were an 8% reduction of EPA (P = 0.0072) and a 0.2% increase of 18:1 (P = 0.0031; two-tailed t test). Error bars correspond to sem (n = 3–9). C, Glycerolipid classes. Glycerolipids from NgWT and NgΔ0-elo1-KO lines were extracted and separated by two-dimensional TLC for membrane polar lipids and 1D-TLC for neutral lipids. FAs from each glycerolipid class were analyzed and quantified. Plastid membrane glycerolipids correspond to PG, MGDG, DGDG, and SQDG. ER-synthesized membrane glycerolipids correspond to DGTS, lyso-DGTS (LDGTS), PC, phosphatidylinositol (PI), PE, TAG, and DAG. Significant alterations in NgΔ0-elo1-KO compared to NgWT were observed at the levels of MGDG (P = 0.0132) and TAG (P = 0.0216) with a 43.8% reduction and a 71% increase, respectively. Other changes were not significant in two-tailed t test. NgWT FA analyses are shown in green; NgΔ0-elo1-KO FA analyses are shown in blue. Error bars correspond to sem (n = 3–5).

In Nannochloropsis, EPA is enriched in three chloroplast lipids (MGDG, DGDG, and PG) and two ER-synthesized lipids (PE and DGTS; Supplemental Table S1). To test if specific glycerolipid classes were reduced in abundance or altered in their FA composition, membrane polar lipids were separated by two-dimensional TLC, neutral glycerolipids (diacylglycerol [DAG], TAG, and free FA) were separated by 1D-TLC, and all TLC-resolved lipids were quantified by GC-FID (Fig. 3C). MGDG was the predominant lipid class in NgWT, representing 27.8 mol% (± 2.3 SEM) of total glycerolipids. Plastid DGDG and SQDG and the endomembrane lipid DGTS and phosphatidylcholine (PC) ranged from 9 to 19 mol%. PG, which can occur in both plastid and nonplastid membranes, represented 7.25 mol% (± 1.1 SEM) of total glycerolipids. The endomembrane lipid classes PE, carboxymethyl-PE, and phosphatidylinositol were <5 mol%. Neutral lipids, that is TAG and DAG, made 12.9 mol% (± 2.6 SEM) and 1.4 mol% (± 0.2 SEM) of total glycerolipids, respectively.

Compared to NgWT, the KO of the Δ0-ELO1 gene led to a 43.8% reduction of MGDG (P = 0.0132) and a 71% increase of TAG (P = 0.0216), whereas differences between KO and wild-type proportions of other glycerolipid classes were not significant in two-tailed t test (Fig. 3C). When analyzing the FA profiles within each lipid class, we could not detect any statistically relevant alterations (Supplemental Fig. S4).

Taken together, the analyses of glycerolipid profiles show a relationship between the KO of the Δ0-ELO1 gene, a decrease of the synthesis of EPA, a specific reduction of the EPA-rich lipid class MGDG, and a concomitant increase of EPA-poor TAG. Other lipid classes also rely on FA synthesis, most importantly sphingolipids. We examined whether the KO of the Δ0-ELO1 gene could also alter sphingolipids, the other major group of acyl-lipids synthesized in eukaryotes.

Targeted Analysis of Sphingolipid Precursors Did Not Show Any Alteration in NgΔ0-elo1-KO Lines

No exhaustive sphingolipid profile of a unicellular alga has been reported, yet. Sphingolipids consist of a long chain base (LCB), an FA linked by an amide bond, and a polar head. In plants, LCBs comprise dihydroxylated and trihydroxylated C18-species (d18 and t18) with or without unsaturation. These C18-LCB could theoretically arise from the action of a Δ0-ELO operating on a 16-carbon substrate. We therefore focused our analysis on this category of LCB. When LCBs are N-acylated by a LC-FA (labeled c16 or c18) or a VLC-FA (<c20), a ceramide (Cer) structure is formed. Hydroxylation of the FA (labeled h16, h18, and >h20) generates hydroxyl-ceramides (h-Cers; Markham, 2013). A variety of possible head groups gives rise to a diversity of sphingolipid classes deriving from Cers and h-Cers, so we focused our analysis on these common procursors (Sperling and Heinz, 2003). Using a targeted mass spectrometry approach, the d18:0, d18:1, t18:0, and t18:1 forms were identified in N. gaditana. In this targeted analysis, acyl-profiles in Cers and h-Cers in NgΔ0-elo1-KO and NgWT were similar (Fig. 4), providing no evidence for another effect of Naga_100083g23 KO (NgΔ0-elo1-KO) besides the lowering of EPA production and the alteration of the MGDG versus TAG balance.

Figure 4.

FA profile of sphingolipid precursors from NgWT and NgΔ0-elo1-KO lines. A targeted analysis of sphingolipids was performed, focusing on LCBs with dihydroxylated (d) and trihydroxylated (t) 18-carbon species, linked to Cers or h-Cers, and with different chain lengths (16–26 carbonyl residues) of saturated or monounsaturated FAs. NgWT sphingolipid analyses are shown in green and NgΔ0-elo1-KO in blue. No significant differences were detected.

Loss of Δ0-ELO1 Could Be Partially Compensated by Overexpression of Δ0-ELO2 in NgΔ0-elo1-KO Lines

We analyzed the expression of genes that could theoretically compensate a loss of Δ0-ELO1, that is (1) the five other Δ0-ELO genes, (2) genes involved in EPA-biosynthesis, and (3) genes involved in the synthesis of MGDG and TAG, and normalized the expression in NgΔ0-elo1-KO to the NgWT levels of the respective gene (Fig. 5). Due to biological variations in algae cultures, we set an empirical threshold of a minimal 0.5-/1.5-fold change to be considered as relevant (Fig. 5, gray).

Figure 5.

Relative mRNA levels of elongases and desaturases in NgΔ0-elo1 mutants compared to the NgWT. RNA was extracted from NgWT and NgΔ0-elo1-KO cells and reverse transcribed. qRT-PCR was performed and curves were evaluated using theΔΔCq method provided with the BioRad CFX software. NgΔ0-elo1-KO relative gene expression levels were normalized to the mRNA level in the NgWT, both corrected to the Cq of the housekeeping genes PsbA and TUB. Data are shown as boxplots with the line at the median, representing the fold change of gene expression in three biological replicates of both NgΔ0-elo1-KO and NgWT. A confidence threshold was set at a minimal fold change of 0.5 (gray shade). Significantly up-regulated in NgΔ0-elo1-KO compared to NgWT were the FA synthase protein FABZ (1.95-fold ± 0.2338, P = 0.0306 in t test) and Δ0-ELO2 (2.1-fold ± 0.1923, P = 0.031 in t test). Genes analyzed correspond to those presented in Figure 1.

As expected, Δ0-ELO1 was not detected in qPCR in NgΔ0-elo1-KO lines. The FASII subunit FABZ was 1.95-fold ± 0.2338 up-regulated, but as mentioned above, this did not affect the total FA content per cell. Among the other putative Δ0-ELO isoforms, only Δ0-ELO2 was up-regulated above the 1.5-fold change ratio, with an induction of 2.1-fold (± 0.1923, P = 0.031), and we can speculate that this isoform might have a partly overlapping role with Δ0-ELO1. According to available RNAseq data on NgWT, Δ0-ELO2 expression is two to four times lower than Δ0-ELO1 expression in 3- or 6-d-old cultures, respectively (Corteggiani Carpinelli et al., 2014).

In N. gaditana, the galactolipid synthases are encoded by single copy genes, that is Naga_100092g5 (MGD) and Naga_100010g107 (DGD). Expression levels of MGD, DGD, and chloroplastic phosphatidic acid phosphatase (PAP), which could provide the DAG backbone for MGDG and DGDG assembly, were similar in NgΔ0-elo1-KO and NgWT, with tendencies of up-regulation for MGD and PAP (Supplemental Fig. S5A). The observed decrease of MGDG in the KO line was therefore not attributable to a decrease of MGD gene expression. In the TAG biosynthesis pathways, the mRNA levels of DGAT2 and LPAAT in NgΔ0-elo1-KO normalized to NgWT levels were at 1, whereas PDAT was up-regulated by 1.4-fold (Supplemental Fig. S5A). None of the genes tested in qPCR had a fold change >1.5 in NgΔ0-elo1-KO compared to NgWT, so we were unable to provide evidence for a possible increase of TAG level by a gene expression control.

Taking lipid and qPCR analyses together, the KO of the Δ0-ELO1 gene was not fully compensated by the overexpression of Δ0-ELO2, and the MGDG decrease accompanied by TAG increase could not be explained by changes in the expression of galactolipid or TAG synthesis genes. Given the very important role of galactolipids on plastid ultrastructure and function (Boudière et al., 2014; Demé et al., 2014; Petroutsos et al., 2014), we sought whether this specific alteration of MGDG level could have an impact on the biogenesis and function of the secondary plastid in N. gaditana Δ0-elo1 KO lines.

Altered Ultrastructure of Thylakoid Membranes in NgΔ0-elo1-KO Mutants

We conducted transmission electron microscopy imaging on midlog phase grown NgΔ0-elo1-KO and NgWT cells (Fig. 6). N. gaditana wild-type cells were spherical or subspherical cells of 3 to 5 µm on average, as reported earlier (Hibberd and Leedale, 1970; Hibberd, 1981; Fig. 6A). Most of the cell volume was occupied by a single plastid containing in general (95%) six thylakoid lamellae, each consisting of three to four membrane stacks (Fig. 6, B and C). The outermost secondary plastid membrane formed a continuum with the outer nuclear envelope membrane (Fig. 6D). In NgWT, thylakoid lamellae are quite uniform and parallel, and extend between two, sometimes three, poles where they are associated with the plastid-limiting membrane system (Fig. 6, A–D). In NgΔ0-elo1-KO lines, the cell surface was less regular (Fig. 6, E and F). Chloroplast cross-sections exhibited similar surfaces and number of thylakoid lamellae, but the shape of the stacks displayed a high membrane curvature (Fig. 6, E–H) and some lamellae did not span the whole chloroplast but were shorter (Fig. 6H). Most strikingly, NgΔ0-elo1-KO lamellae displayed frequent expansions with enlarged thylakoid lumen (Fig. 6, E–H). These structures correspond to the expansion of one membrane of a lamellae stack, which is no more cohesive, as seen by the reduced lamella width (Fig. 6H).

Figure 6.

Transmission electron micrographs of NgWT (A–D) and NgΔ0-elo1-KO (E–H) reveal bent thylakoid structures and lumen expansions in the mutant. Different magnifications of NgWT (A–D) and NgΔ0-elo1-KO (E–H) sections are shown. A, NgWT cells are heterogeneous in size, shape, and organelle proportions, partially caused by different cutting planes. B, Wild-type thylakoid membranes are devoid of grana stacks. Lamellae derive from two to three chloroplast poles. C, Wild-type chloroplasts consist of five to seven thylakoid lamellae, each consisting of a stack of three membrane bilayers. D, Wild-typeT plastid outer-limiting membrane is in continuum with nucleus. E, NgΔ0-elo1-KO cells display higher thylakoid membrane curvature and two or more thylakoid membrane poles. F to H, Interlamellar space (lumen) appears more expanded (black arrows) and less parallelism between membranes is observed; NgΔ0-elo1-KO thylakoids develop a more pronounced curvature. H, NgΔ0-elo1-KO polar growth of thylakoid lamellae appears partially aborted in some cells. Mit, mitochondria; N, nucleus; Pl, plastid.

We sought whether such alterations of chloroplast ultrastructure might have an impact on the function of the photosynthetic machinery.

Fast-Fluorescence Kinetics Indicated Impaired Photosynthesis in NgΔ0-elo1-KO Mutants

Physiological studies were conducted on the NgWT and NgΔ0-elo1-KO mutants in different culture volumes and light conditions. In our cultivation conditions, we could not detect any growth or biomass production phenotype in NgΔ0-elo1-KO in any of the cultivation regimes (Supplemental Fig. S6). Even considering that our cultures are CO₂-limited, this suggests that the alteration of thylakoid morphology has no major impact on photosynthetic growth. This was confirmed by the more detailed investigation of photosynthesis parameters like chlorophyll content, Fv/Fm, electron transfer rate (ETR), and quantum yield of PSII [Y(II)] and nonphotochemical quenching (NPQ). Relative chlorophyll fluorescence per million cells appeared similar, based on a t test, in NgWT (mean 5904 RU ± 206 SEM) and NgΔ0-elo1-KO mutants (mean 5758 RU ± 190 SEM). Although quantification of chlorophyll based on fluorescence could be altered by self-absorption effect, this indicates that chlorophyll levels are likely little affected by the mutation. The maximum photosynthetic efficiency measured by Fv/Fm was 5% reduced in NgΔ0-elo1-KO compared to the NgWT (Supplemental Fig. S7).

The transfer of electrons generated by PSII to PSI is dependent on the light energy and was tested with different photosynthetically active radiation (PAR) levels (Fig. 7A). Maximum ETR was observed at 310 µE/m²s in both NgWT and NgΔ0-elo1-KO, but transfer was lower in the mutant with a 4.1% reduction at 41 µE.m⁻²s⁻², 7.9% at 90 µE.m⁻²s⁻², 14.8% at 110 µE.m⁻²s⁻², 18.6% at 310 µE.m⁻²s⁻², 26% at 440 µE.m⁻²s⁻², and 19.3% at 600 µE.m⁻²s⁻². These data indicated that in our cultivation system and range of illumination, with increasing PAR, NgΔ0-elo1-KO has a reduction in ETR attributable to a higher light-sensitive or altered NPQ with respect to NgWT.

Figure 7.

Impact of Δ0-ELO1 on photosynthesis. A, Lower photosynthetic ETR, calculated via the YII, in NgΔ0-elo1-KO compared to NgWT. From three 3-d-old 2-mL cultures per cell line, a 160-µL aliquot was used for ETR estimates. YII was measured via room temperature chlorophyll fluorescence kinetics with different photosynthetic active radiations between 41 µE.m⁻²s⁻² and 600 µE.m⁻²s⁻², and the ETR was calculated accordingly. ETR was impaired in mutant lines, and the differences between the strains as well as between the different light intensities and their interactions were significant based on a two-way ANOVA (P < 0.0001). B, Increased NPQ in NgΔ0-elo1-KO compared to NgWT. Room temperature chlorophyll fluorescence kinetics were measured to calculate NPQ with constant photosynthetic active radiation of 600 µE.m⁻²s⁻². The yellow shade indicates the irradiation time frame. Representative data from a 160-µL aliquot of three 7-d-old 50-mL cultures per strain are shown. Differences between mutant and NgWT were significant based on a 2-way ANOVA (P < 0.0001). C, Increased protein level of LHCX1 in NgΔ0-elo1-KO compared to NgWT. The same cultures were harvested for protein extraction, and 30 µg total extract was separated via SDS-PAGE. Gels were stained with Coomassie Brilliant Blue as loading control or transferred to a membrane for LHCX1 antibody detection after western blotting (arrow). Antibody-labeled bands were visualized and intensities determined and normalized to the intensities of the respective loading control lanes. Corrected LHCX1 signal was 18% (± 1.5 SEM) stronger in NgΔ0-elo1-KO than in NgWT (P = 0.0072 in two-tailed t test).

We measured NPQ under constant irradiation with 600 µE.m⁻²s⁻¹ actinic light on dark-adapted cells that were grown for 7 d at 30 µE.m⁻²s⁻¹ white fluorescence light (Fig. 7B). NPQ measurements showed higher NPQ levels in the mutant than in the wild type. After 5 min of prolonged irradiation, a steady level of NPQ was reached, which was about 20% higher in NgΔ0-elo1-KO lines compared to the NgWT. NPQ is composed of different components that can be distinguished following its relaxation after light is switched off. The thermal energy dissipation (qE) component, dependent on PSII antennae-aggregation and the accumulation of zeaxanthin (Zx) and antheraxanthin (Gentile and Blanch, 2001), is rapidly relaxed, while qI, dependent from photoinhibition of PSII, is much slower (Eberhard et al., 2008). During the relaxation phase, fluorescence rapidly decreased in NgΔ0-elo1-KO and NgWT and curves superimposed, indicating an increased qE but not a qI phenotype in NgΔ0-elo1-KO. When cells were grown under continuous irradiation at the same light intensity (30 µE.m⁻²s⁻¹ white light), cell growth and maximal NPQ levels were 2-fold higher compared to 12-h-light/-dark cycle-grown samples after 7 d of cultivation in 50 mL. In all these experiments performed at different light regimes, the 20% NPQ difference between the NgWT and NgΔ0-elo1-KO level was maintained.

In Nannochloropsis, the LHCSR-like protein LHCX1 acts in the photoprotective mechanism for excess excitation energy (Bailleul et al., 2010), involving the activation of the violaxanthin (Vx) de-epoxidase (VDE), whose activity determines qE. Zx epoxidase (ZEP) reverts VDE reaction. Vx-chlorophyll a binding protein (VCP) is the predominant light harvesting protein binding chlorophyll and carotenoids (most likely Vx; Carbonera et al., 2014). We addressed the mRNA level of VDE, ZEP, VCP, and LHCX1 by quantitative real time PCR (qRT-PCR), using tubulin alpha (TUB) and PSBA as reference genes, and plotted the gene expression in NgΔ0-elo1-KO normalized with that measured in NgWT (Supplemental Fig. S5B). None of the tested genes displayed any significant transcriptional change. Protein levels were also assessed, since LHCSR proteins have been previously reported to be translationally regulated (Li et al., 2009). A western-blot analysis revealed an 18% increase of the LHCX1 protein level, correlating with the 20% increase of NPQ values (Fig. 7C) in NgΔ0-elo1-KO compared to NgWT. This correlation is consistent with the qE phenotype we observed.

DISCUSSION

Δ0-ELO1 Controls EPA and MGDG Levels in Nannochloropsis

We identified sequences in the N. gaditana genome involved in VLC-PUFA biosynthesis pathway starting from a 16:0 substrate. We confirmed the existence of a unique ω-6 pathway based on the presence of 20:4ω-6 and absence of 20-4ω-3 (Schneider et al., 1995). While most of the lipid-related enzymes in N. gaditana appear to be encoded by single-copy genes (e.g. desaturases, elongases, galactolipid synthases), we found six putative Δ0-elongases that grouped in a heterokont cluster, and none of them seemed to be a pseudo-gene since we could confirm their expression. We considered the most plausible substrate of these elongases to be 16:0, since elongation of 16:1 would generate 18:1Δ7 and elongation of 18:0 would generate 20:0, two FAs that were not detected in glycerolipids.

The specificity of the most expressed isoform, Δ0-ELO1, was assayed in yeast heterologous system and showed an activity on 16:0 but not myristic acid (14:0). The presence of multiple Δ0-ELO sequences appeared initially surprising, since elongation of 16:0 into 18:0 in the ER would be theoretically redundant with the FAs produced by FAS II in the plastid (Li-Beisson et al., 2010). This raised our interest in these genes to investigate aspects of FAs’ fate determination. If so many putative Δ0-ELO isoforms are involved in upstream reactions, could the branch point defining how FAs are processed and in which glycerolipids they end up already lie here?

This work and the recent study on ERΔ12FAD in Nannochloropsis oceanica (Kaye et al., 2015) pioneer in this question and point out that the EPA fate is determined during its synthesis by elongases and desaturases. In the logic of the EPA synthesis scheme, Δ0-ELOs and ERΔ12FAD operate in the same pathway and in case 18:0 or 18:1^Δ9 could be provided by the plastid, one would expect little impact of a genetic modification of a Δ0-ELO gene. However, the phenotypes of NgΔ0-elo1-KO compared with ERΔ12FAD overexpression in N. oceanica (Kaye et al., 2015) are not consistent with this hypothesis. On one hand, the overexpression of ERΔ12FAD led to an overall reduction of 18:1 and an increase of 18:2, consistent with an increased Δ12-desaturation. A specific modification of PC profile was also observed along with a moderate redirection of EPA and ARA from chloroplast lipids into TAG, whose abundance was marginally elevated (Kaye et al., 2015). On the other hand, following Δ0-ELO1 KO, the overall FA profile showed a slight increase of 16:0, consistent with the KO of a 16:0 elongase, and a significant decrease in 20:5, the end product of EPA biosynthesis. The different lipid classes were little affected except for a specific disturbance of EPA and a down-regulation of MGDG. This study therefore shows that an N. gaditana Δ0-ELO protein plays a role in defining the fate of 16:0 for the production of EPA required for a specific lipid class (here MGDG) at the very early step of elongation/desaturation. Other Δ0-ELO isoforms could have overlapping function, as suggested in the NgΔ0-elo1-KO by the 2-fold up-regulation of Δ0-ELO2. Available expression data (Corteggiani Carpinelli et al., 2014) indicated that in NgWT, the expression level of Δ0-ELO2 was 2-fold lower than that of Δ0-ELO1 in young cultures. Therefore, it is likely that we could not reveal the full Δ0-ELO1 function due to partial compensation by Δ0-ELO2.

Could Δ0-ELO1 Be a Component of the Omega Pathway?

The import of EPA and other ω3-VLC-PUFAs from the ER to secondary plastids is unknown and has been called the omega pathway (Petroutsos et al., 2014). Based on radiolabeling experiments, it is considered that in the primary plastid of plants, ER-located PC serves as a precursor molecule for MGDG (Roughan et al., 1980), in the so-called eukaryotic pathway of galactolipid synthesis (Heinz, 1977). The nature of the precursor imported in primary plastids is not clearly identified, although it is known to derive from a diacyl-glycerolipid, should it be PC itself (Miquel et al., 1988) and /or lyso-PC (Mongrand et al., 2000; Botella et al., 2016) and/or PC-derived PA (Benning, 2009) and/or PC-derived DAG (Nakamura et al., 2009; Maréchal and Bastien, 2014). We were unable to identify orthologs of any of the known components of the plant chloroplast lipid transporters TGD, that is TGD1-5 (Fan et al., 2015). N. gaditana PC is poor in EPA and enriched in C18 FA species, which does not comply with a conservation of FA-precursor import in the secondary endosymbiont following the same machinery as that known in primary plastids. One could assume that a high turnover from PC to MGDG would prevent the accumulation of similar FA profiles, but our data on NgΔ0-elo1-KO revealed that MGDG reduction did not affect PC content or quality. Similarly, in N. oceanica ERΔ12FAD overexpression lines, the altered FA composition in PC did not induce comparable changes in the MGDG profile. It is thus unlikely that PC has a similar precursor role for galactolipids in Nannochloropsis.

As an alternative, EPA could be imported as a nonesterified FA. The homologous sequences of LACS9, a long chain acyl-CoA synthetase acting in FA trafficking between plastid and cytosol in Arabidopsis (Arabidopsis thaliana; Jessen et al., 2015), shared only 30% amino acid similarity, suggesting that the primary plastid components were not conserved in heterokonts. Unlike Arabidopsis and other primary endosymbionts, where chloroplasts membranes are disconnected from the ER and rely on specific chloroplast-ER binding sites for material exchanges (Block and Jouhet, 2015), the outermost membrane of Nannochloropsis plastid is in continuum with the outer membrane of the nuclear envelope and the ER, which could possibly allow a transport of FA (Watson, 1955; Falciatore and Bowler, 2002). The observed decrease of MGDG in the NgΔ0-elo1-KO lines was not attributable to a decrease of MGD gene expression. This work therefore shows that if EPA is imported via the omega pathway as a nonesterified FA, at least part of EPA directed to MGDG involves elongation/desaturation components stemming from 16:0 elongation. NgΔ0-ELO1 would therefore be part of a channeling process and a committing entry point into the omega pathway dedicated to the synthesis of MGDG.

MGDG Decrease Versus TAG Increase

In the NgΔ0-elo1-KO lines, the MGDG decrease is concomitant with a TAG increase. TAG can be produced in two general ways: by de novo synthesis or lipid remodeling. On the one hand, the ER located lysophosphatidate acyltransferases, Naga_100002g46 (LPAAT), and of the acyl-CoA:DAG acyltransferases, Naga_100343g1 (annotated DGAT2) are indicators of a de novo TAG synthesis via the Kennedy pathway. On the other hand, a putative phospholipid:DAG acyltransferase, Naga_100065g17 (PDAT), could also operate on galactolipids and display lipase activities that would indicate remodeling (Li et al., 2014). A lipase called PGD1 has been reported to be involved in MGDG breakdown to feed TAG pools in Chlamydomonas, and a homologous protein, PSD1, was reported in N. oceanica (Li et al., 2012). We did not detect any PGD1/PSD1 homolog in N. gaditana. LPAAT, DGAT2, and PDAT mRNA levels were not significantly changed in the KO line. Taken together, these results support that the increase in TAG is most likely due to a metabolic diversion of glycerolipid intermediates otherwise used to produce MGDG.

Role of EPA-Rich MGDG in Thylakoid Morphology and Photosynthesis Control

A lower level of MGDG in NgΔ0-elo1-KO strongly altered the thylakoid membrane morphology. In Arabidopsis, abnormal thylakoids are also observed when MGDG or DGDG synthesis is genetically or chemically reduced (Dörmann et al., 1995; Kobayashi et al., 2007; Botté et al., 2011a). Membrane curvature is stabilized by unequal distribution of lipids in both leaflets (Garab et al., 2000). A defined MGDG/DGDG ratio is therefore likely required to stabilize proper membrane morphology and stacking, most likely due to the physical properties of MGDG as a hexagonal II phase building (HII) lipid versus as a lamellar-forming (L_α) lipid (Garab et al., 2000; Jouhet, 2013; Demé et al., 2014). Thus, the lowered MGDG/DGDG ratio in NgΔ0-elo1-KO could lead to the distorted thylakoid architecture we observed.

The membrane structural phenotype in NgΔ0-elo1-KO correlated with an impairment of photosynthesis. During oxidative photosynthesis, light energy is harvested by antenna complexes containing light harvesting proteins, called VCPs in Nannochloropsis (Basso et al., 2014; Carbonera et al., 2014). Briefly, in VCPs, light is converted into excitation energy, transferred to a chlorophyll a dimer in PSII reaction center. A valent electron is released, accepted by plastoquinone, and donated to the electron transport chain and finally to PSI reaction center, in which it receives new light energy captured by PSI antenna. Electrons are transported through the PSI complex from the luminal to the stromal side, where they are accepted by ferredoxin (Fd). Fd-NADP⁺-oxidoreductase mediates the reduction of NADP⁺ with electrons donated by Fd. The electron transport promotes translocation of protons, building up the proton motif force driving the ATP synthetase. NADPH and ATP are then injected in the Calvin cycle (Nelson and Ben-Shem, 2004). While chlorophyll contents remained stable in NgΔ0-elo1-KO and energy conversion at PSII was only marginally affected, cells were affected in electron transport and more sensitive to high PAR intensity, reflected by lower energy proportions used for photochemistry and stronger induction of NPQ. When absorbed, light energy is too high to be photochemically quenched, and NPQ is induced within seconds. In this photo-acclimation mechanism, photo-energy is emitted as qE, and photo-damaged PSII reaction center protein D1 is substituted by a new translated protein in a relatively slow mechanism, during which PSII is disassembled and thus inactive (qI, photoinhibition; Kanervo et al., 2005). The qE fluorescence is the major part of NPQ and involves the luminal LHCX1 in heterokonts for the activation of VDE (Lepetit et al., 2013). In Nannochloropsis, VDE activity converts light-harvesting Vx by 76% into heat-emitting antheraxanthin and 24% further into Zx (Gentile and Blanch, 2001).

Our study shows that in NgΔ0-elo1-KO, both LHCX1 protein level and qE-dependent fluorescence were increased, whereas qI was not affected. Together with the lower ETR in the NgΔ0-elo1-KO mutant, this further underlined disturbed thylakoid structures and dynamics. Notably, in contrast to higher plants, there is no antiproportional correlation between NPQ and ETR in heterokonts (Nymark et al., 2009). A similar phenotype of the ETR/NPQ couple in NgΔ0-elo1-KO was observed in Phaeodactylum during iron starvation, a condition where PSI levels are more impaired than PSII and in which chlorophyll was degraded (Allen et al., 2008). Here, chlorophyll a content seemed stable based on fluorescence measurements, so future studies include refined chlorophyll analyses, an assessment of the stoichiometry of PSI versus PSII, and the determination of xanthophyll cycle enzyme levels and the Vx of accessory pigments.

MGDG was found in vitro and in vivo to be beneficial for VDE action due to its HII property, solubilizing Vx, and allowing VDE integration into the lipid phase (Goss et al., 2005; Jahns et al., 2009; Goss and Jakob, 2010; Schaller et al., 2010). Upon qE induction, LHCs form aggregates and Vx is released, a process thought to trigger the formation of MGDG HII-phase within the bilayer, acting as a platform for VDE activity (Jahns et al., 2009). Vx disassembly from LHCII complexes and diffusion into the HII phase is believed to be the rate-limiting step in Vx de-epoxidation (Latowski et al., 2000). Maximum conversion was observed in LHCII extractions at a MGDG/Vx ratio of 3.5:5; a higher ratio led to a decreased activity (Schaller et al., 2010, 2011). Despite this evidence of a beneficial role of MGDG in the onset of NPQ, lower abundance of this galactolipid is observed under photoprotection-inducing conditions, such as in the Ng-Δ0-elo-KO line. Interestingly, the exposure of N. gaditana to high light, known to induce NPQ, is also marked by a decrease of the MGDG content (Alboresi et al., 2016). Thus, the reduction of MGDG could be the trigger for enforced NPQ in both NgΔ0-elo1-KO and high light. The molecular mechanism could lie in the physicochemical properties of MGDG: the decreased MGDG level during high light or in NgΔ0-elo1-KO might increase the lipid/protein ratio, thereby limiting Zx solubilization (Garab et al., 2000). Additionally, MGDG could stabilize the antenna-PSII connection. In Phaeodactylum, the quenching of chlorophyll a by Zx in LHCs requires Zx-dependent antenna protein aggregation (Lepetit et al., 2012; Valle et al., 2014). In N. gaditana, PSII-VCP bonds are very weak compared to other species (Basso et al., 2014). Thus, if MGDG was having a stabilizing function, a reduction could facilitate PSII-VCP disconnection, promoting the building up of the NPQ platform in aggregated antenna proteins and thereby lowering the ETR. Similar chlorophyll levels in NgΔ0-elo1-KO and NgWT are consistent with this hypothesis, since in heterokonts, unlike plants, PSII antenna sizes are not responsive to light intensities (Simionato et al., 2011).

Eventually, although MGDG levels are reduced at high light, gene expression of MGD is induced (Alboresi et al., 2016). This observation suggests that the turnover of EPA-rich MGDG is higher under NPQ-inducing conditions. Evidence supports a role of VLC-PUFAs as antioxidants rather than prooxidants (Richard et al., 2008), and a recent study suggests that PUFA-MGDGs might act as supramolecular antioxidants capturing oxygen species, limiting damage to proteins (Schmid-Siegert et al., 2016). It is therefore possible that MGDG molecules in the vicinity of PSII scavenge electrons generated when the capacity of the ETR is exceeded. This process could be correlated with an increased turnover of MGDG. Thus, released free VLC-PUFAs or VLC-PUFA-MGDG might be involved in a redox stress-sensing system as part of a photoprotection mechanism during high light.

Perspectives

This study leads to some interesting biotechnological developments. EPA is known to be beneficial for health (Wen and Chen, 2003; Ji et al., 2015). High biomass-producing Nannochloropsis is a candidate for EPA and TAG production, but productivity has not yet reached industrial feasibility in spite of efforts put into strain selection and culture optimization (Van Vooren et al., 2012; Camacho-Rodríguez et al., 2013, 2014, 2015; Chen et al., 2013; Ma et al., 2014; Meng et al., 2015) and first metabolic engineering trials (on N. oculata or salina; Radakovits et al., 2012; Vieler et al., 2012; Iwai et al., 2015; Kang et al., 2015). The branching point at the level of 16:0 elongation, directing the fate of the EPA end product for specific lipid classes like MGDG, should be taken into account. The function of EPA-rich MGDG in photosynthesis control should also be considered in future strain optimization strategies. This study should now be expanded to other Δ0-ELO isoforms.

METHODS

Sequence Analyses

The nucleotide and amino acid sequences of putative elongases from Nannochloropsis gaditana were retrieved from the Nannochloropsis Genome Portal (http://www.nannochloropsis.org), combining ontology searches and comparisons with known elongase sequences from other organisms. Homologous sequences of representative groups of eukaryotes’ biodiversity were collected from the NCBI nr databases. Selected sequences cover the biodiversity of eukaryotes including opisthokonts, for example fungi (Saccharomyces ELO1, NP_012339; ELO2, NP_009963; ELO3, NP_013476) and metazoa (Drosophila Baldspot, XP_005187861; Musca Baldspot-Like, XP_005187861; Apis Baldspot-Like, XP_003251850; Bombus Baldspot-Like, XP_003401825; Caenorhabditis ELO3, NP_001255291; Homo ELOVL1, XP_002040; ELOVL2, NP_060240; ELOVL3, NP_689523; ELOVL4, NP_073563; ELOVL5, NP_068586; ELOVL6, NP_569717; ELOVL7, NP_079206; Mus ELOVL6, NP_569717; Gallus ELO6; NP_001026710), heterokonts (Phaeodactylum putative Δ0-ELO/ELO3, XP_002184740; Δ6-ELO/ELO6, XP_002180428; Δ5-ELO/ELO5, XP_002176686; Δ5b-ELO/ELO5b, XP_002179048; Thalassiosira Δ0-ELO/ELO3; XP_002293395; Δ6-ELO/ELO1, XP_002288481; Δ5-ELO/ELO2, XP_002291938; Ectocarpus ELO, CBN78890; Phytophthora ELO, ETM45813; Albugo ELO, CCI42963; Saprolegnia ELO, XP_008620377; Aphanomyces ELO, XP_008865117), apicomplexa (Toxoplasma ELOA, TGME49_053880; ELOB TGME49_242380; ELOC TGME49_205350; Neospora ELOC-Like, XP_003882280; Eimeria ELOC-Like, CDJ36988; Cryptosporidium ELOC-Like, AAO34582; Plasmodium ELOC-Like, XP_001351023; Gregarina, ELOC-Like XP_011131654), Haptophytes (Emiliania ELO2-Like, XP_005769239), cryptomonads (Guillardia ELO, XP_005838572), and kinetoplastida (Trypanosoma ELO1, Tb927.7.4180; ELO2, Tb927.7.4170; ELO3, Tb927.7.4160; ELO4, Tb927.5.4530; Leishmania ELO2, LbrM.14.0670). Since heterokonts are a superphylum comprising a plastid-containing phylum, Ochrophyta, and nonplastid-containing phyla, Pseudofungi/Oomycota and Bigyra (Cavalier-Smith and Chao, 2006; Cavalier-Smith et al., 2015), we selected sequences in both plastid- and nonplastid-containing species. We included elongases previously characterized experimentally in yeast (Saccharomyces cerevisiae [Tehlivets et al., 2007]), Homo sapiens (Kihara, 2012), Drosophila melanogaster (Jung et al., 2007), Trypanosoma brucei (Lee et al., 2006), Toxoplasma gondii (Ramakrishnan et al., 2012), Toxoplasma pseudonana (Cook and Hildebrand, 2016), and Phaeodactylum tricornutum (Jiang et al., 2014). The amino acid sequences were aligned using the MUSCLE program (Edgar, 2004). An unrooted phylogenetic tree was constructed based on the alignment results using the Neighbor-Joining method (Saitou and Nei, 1987) implemented in the phylogeny.fr platform (Dereeper et al., 2008), with gamma correction and bootstrapped confidence intervals based on 1,000 replications.

N. gaditana Strains and Cell Culture Conditions

NgWT and mutant lines were maintained in F/2 medium (Guillard and Ryther, 1962) containing modified sea salts (NaCl, 21.194 g.L⁻¹; Na₂SO₄, 3.55 g.L⁻¹; KCl, 0.599 g.L⁻¹; NaHCO₃, 0.174 g.L⁻¹; KBr, 0.0863 g.L⁻¹; H₃BO₃, 0.023 g.L⁻¹; NaF, 0.0028 g.L⁻¹; MgCl₂0.6H₂O, 9.592 g.L⁻¹; CaCl₂0.2H₂O, 1.344 g.L⁻¹; and SrCl₂0.6H₂O, 0.0218 g.L⁻¹; NaNO_3, 46.67 mg.L⁻¹ and NaH₂PO_4, 3.094 mg.L⁻¹), with a temperature set at 20°C and a light regime of either 12-h-light/-dark night cycle or continuous irradiation (photon flux of 30 µmol m⁻²s⁻¹ white light). Cultures of 50 to 100 mL were grown in 250-mL Erlenmeyer flasks, and 2 mL-cultures on 24-well culture plates (Thermo Fisher), inoculated with a cell density of 20⁶ cells.mL⁻¹. For long-term storage at −80°C, 10⁷ cells were frozen stepwise (30 min at 4°C, 60 min at −20°C) in 15% dimethyl sulfoxide (Sigma-Aldrich). Cell densities were measured by the A₇₃₀ of a 300-µL culture aliquot in a transparent 96-well plate (Thermo Fisher) using a TECAN infinite M1000Pro plate reader and, for most experiments, cells were collected at midlog phase. All experiments were performed with at least three biological replicates, each representing an individual culture of a given strain.

Cloning of a Naga_100083g23 (Δ0-ELO1) into a Yeast Expression Vector and Functional Characterization in Yeast

The codon optimized open reading frame encoding Naga_100083g23 (Δ0-ELO1) was synthesized by Life-Technologies Services France and cloned into the pYES2 vector to generate the pYES2-NgΔ0ELO1 construct. With the help of the added restriction sites EcoRI and XbaI, the yeast (Saccharomyces cerevisiae) optimized sequence was cloned into cut pYES2 vector to generate the pYES2-NgΔ0ELO1 construct. The pYES2-NgΔ0ELO1 vector was transformed in yeast wild type (W303-1A: MATα, ade2-1, his3-11,15, leu2-3,112, trp1-1, ura3-1, and can1-100; Thomas and Rothstein, 1989) and fas2Δ KO strain (DTY-10a2: MATα, fas2Δ::LEU2, can1-100, ura3-1, ade2-1, and his3-11,his3-15; Toke and Martin, 1996) using a lithium acetate-based method (Elble, 1992) and transformed clones selected on SD-URA (MP Biochemicals). Expression of the transgene was induced in liquid cultured by addition of 2% (w/v) Gal in the presence or absence of exogenously supplied FA substrates as described previously (Beaudoin et al., 2000). Total FAs extracted from yeast cultures were analyzed by gas chromatography of methyl ester derivatives. Lipids were extracted and transmethylated with methanolic HCl and the resulting FAMEs were separated using either a DB-23 (15 m, 0.25 mm, 0.25 µm; Agilent J&W) or a HP-1MS (HP-1MS Ultra Inert 30 m, 0.32 mm, 0.25 µm; Agilent) column coupled to either a flame ionization detector or a mass spectrometer.

Cloning of a Δ0-ELO1-KO Cassette and Transformation of N. gaditana CCMP526

The transformation vector UEP-p35S-loxP BSD FL1-FL2 526 comprises a p35S-loxP cassette, a zeocin resistance gene (ZEO, CDS 3078..3448) under the control of the ubiquitin promotor, and the P. tricornutum FcpA terminator. The two flanking regions up- and downstream of the zeocin resistance gene were substituted by homologous sequences of N. gaditana gDNA, surrounding the coding sequence of Naga_100083g23 (Δ0-ELO1) to induce homologous recombination after nuclear transformation (9). Respective gDNA fragments were amplified by PCR using the oligonucleotide pair 5′-gttgggaataatgcgggacc-3′ and 5′-ccgctttggtttcacagtca-3′ for the terminal flank, and 5′-acgatgggtatgttgcttgc-3′ and 5′-tgtacagggcggatttcact-3′ for the upstream flank. Flanks were inserted into the PCR BLUNT vector (Invitrogen) and subcloned into the UEP-P35S-loxP vector using the restriction enzyme activity of BamHI/SacI for the downstream flank and KpnI/EcoRI (NEB) for the upstream flank. Nuclear transformation was performed as described earlier (Perin et al., 2015) with the following modifications: 10⁸ NgWT cells were harvested during exponential growth at a concentration of 3.10⁶ cells.mL⁻¹, washed two times with 375 mm d-sorbitol, and resuspended in 100 µL final volume. The recombination cassette was digested from the vector (SacI/KpnI), and 1 µg of the digestion product was applied to the cell suspension and mixed gently. After 15 min of incubation on ice, cells were single-pulse electroporated at 500 Ω and 50 μF capacitance (NEPA21 Type II, Sonidel Limited) with 2400 V, (MicroPulser, BioRad). The transformation mix was transferred to 5 mL fresh F/2 medium in a 50-mL Falcon tube and incubated for 16 h under continuous light irradiation. Cells were then plated on 1.5% F/2 agar plates containing 7 µg.mL⁻¹ zeocin, and colonies were obtained after 3 to 4 weeks of incubation with continuous irradiation.

Genotyping of Naga_100083g23 KO Mutants

PCR-genotyping was carried out on a part of transformed N. gaditana colonies resuspended in 10 µL distilled water transferred to a tube with 90 µL water. Cells were thermo-lysed (95°C, 5 min) and 1 µL was used in PCR analyses. KO lines were assessed based on the presence of the flanking sequences (primers detailed above) and the zeocin resistance gene (oligonucleotide pair binding in promotor: 5′-gaggaatgtgtgtggttggg-3′, and terminator: 5′-gccgtattgttggagtggac-3′), and on the absence of the Naga_100083g23 (Δ0-ELO1) sequence (5′-gacacttctctgcctttgcc-3′ and 5′-atggtggtaccagtggagga-3′). The number of cassettes inserted into the N. gaditana genome in three independent clones was verified by qPCR on gDNA extracted from thermo-mechanically broken pellets of 10⁸ cells using the chloroform-phenol method (Pacific Biosciences of CA, Inc, 2012). gDNA (25 ng per reaction) was mixed in SYBR Green qPCR (Agilent) reagent with 1.2 nm oligonucleotides annealing with TUB (10) as loading control, PAP (Naga_100038g41) as a single copy gene control, and ZEO and Δ0-ELO1 as genes of interest. After 10 min of initial denaturation, 40 thermo-cycles were performed at 95°C, then 55°C and 72°C, each temperature being held for 30 s. We sought strains in which Δ0-ELO1 was not amplified and in which the detection value of ZEO equaled that of the single PAP gene, corrected to the primer efficiencies, thus revealing a single gene insertion. Primer efficiencies were calculated from the linear regression of a dilution series of the same DNA sample. The qPCR primers were the following, given in the 5′-3′ orientation (F, forward; R, reverse): Naga_100083g23F, gtgggcaccaaggttatgga; Naga_100083g23R, gaaggaggtgtggtacggtg; PAPF, aagtggtacctttgctccgt; PAPR, aaggtagccgagtagccaaa; TUBF, ttgagcataccgacgtgact; TUBR, gcgatgagcctgttcagatt; ZEOF, tgtgccaaaatcatacagcagg; ZEOR, cgaagtcgtcctccacgaag.

RNA Extraction and qPCR Expression Quantification

A 10⁸-cell pellet was frozen in liquid nitrogen and stored at −80°C until RNA extraction was performed. A volume of 1 mL TriReagent (Sigma-Aldrich) was added to the frozen pellet, transferred to a 1.5-mL tube, and mixed vigorously for 30 s (vortex). Then, 200 µL chloroform (Sigma-Aldrich) was added and the tube inverted several times during an incubation of 15 min at room temperature. Samples were centrifuged (15 min at full speed at 4°C) and the upper phase transferred to a new tube. RNA was precipitated using 1 mL 100% isopropanol (Sigma-Aldrich) after 10-min incubation at room temperature and collected by a 30 min-centrifugation (full speed, 4°C). The pellet was washed with 75% ice-cold ethanol and dried in a heat block at 60°C. RNA was suspended in 30 µL diethylpyrocarbonate-treated water (Roth), precipitated via the addition of 1 volume of NH₄⁺-acetate, 5 m and 3 volumes of 100% ethanol, and collected by centrifugation (15 min at full speed at 4°C). Washing, drying, and resuspension were repeated following the same procedure and RNA quantity was estimated by the A₂₈₀, 260 nm and 230 nm using a NanoDrop reader. RNA integrity was verified on a 2% agarose ethidium bromide gel. cDNA was generated from 1 µg RNA using the QuantiTect Reverse transcription kit (QIAGEN) and 15 to 20 ng/reaction was used for qPCR to test differential gene expression using the following oligonucleotides, given in the 5′-3′ orientation (F, forward; R, reverse): DGAT2F, tggtggtgatcctctccctt; DGAT2R, attgcaaaacgcgtcccatc; ∆0-ELO2F (Naga_100162g5), ggcccaataggaggcatgtt; ∆0-ELO2R (Naga_100162g5), cacaccacacctctccactc; ∆0-ELO3F (Naga_100162g4), gcacagccccctactacatc; ∆0-ELO3R (Naga_100162g4), ggcctacgtcccttcaaaca; ∆0-ELO4F (Naga_100004g102), tgtcgtcccccaccttatct; ∆0-ELO4R (Naga_100004g102), gcgagcttggagaggatgaa; ∆0-ELO5F (Naga_100017g49), gcaagtattcgcgtcggttc; ∆0-ELO5R (Naga_100017g49), tggaatcaacggtacgcctc; ∆0-ELO6F (Naga_100399g1), cctcttcacgcacaaggact; ∆0-ELO6R (Naga_100399g1), caggaccaggattaccgtgt; ∆6-ELOF (Naga_100003g8), tttttgacgatgaacgcgca; ∆6-ELOR (Naga_100003g8), agaggacgagaagcgagaga; ∆5-ELOF (Naga_100020g80), agagagcttgcatatcgccc; ∆5-ELOR (Naga_100020g80), ccgcacgtaagaacgaggta; ∆4FADF (Naga_100042g12), aggtcccaccgtacttctca; ∆4FADR (Naga_100042g12), gccaaaatgtcgggcgatac; ∆5FADF (Naga_100273g7), cttggcctctttcgtggtct; ∆5FADR (Naga_100273g7), tgcatgacgtgcaacaaagg; ∆6FADF (Naga_100061g21), catctttgcagccttccacc; ∆6FADR (Naga_100061g21), caagtctcgatacgctcgct; ∆9FADF (Naga_100027g27), gtactcggagacagatgcgg; ∆9FADR (Naga_100027g27), aaatccaactgtttgccgcc; ∆12/ω3FADF (Naga_100092g4), gccccatatggcgacattct; ∆12/ω3FADR (Naga_100092g4), aggtggaagaaggaggtggt; LPAATF, ccttcggagcatggcttctt; LPAATR, catccagtctagccgtgtcc; MGDF, ccggacaggaagaagggaac; MGDR, ctcatcttctcccgcacctc; PSBAF, acccaatcggacaaggtagc; PSBAR, ccaaacacaccagcaacacc; SADF, cttgaacagagacccggagg; SADR, aagtgctcgaacaggtctgg; VCPF, catgcttgccatgctccac; VCPR, cggaggtgatggcgttgat; VDEF, agggcaagtggtacatcagc; VDER, gatgcgccagttcagctttc; ZEPF, gccagatgcattcggagagt; and ZEPR, agaccttgtaggccacctct. qRT-PCR was performed using 20 ng cDNA in a BioRad CFX96 qPCR machine, and relative expression levels were calculated using the ΔΔCq (quantitative cycle) method provided with the BioRad CFX software. PSBA and TUB served as internal loading controls, and mutant Cq values were normalized to the NgWT mRNA levels.

Glycerolipid Extraction, TLC, Structural Analysis by Ion Trap Spectrometry, and Quantification by GC-FID

Glycerolipids were extracted from 5.10⁹ cells harvested from 100 mL-cultures after 4 d or 50-mL cultures after 7 d of growth, frozen in liquid nitrogen, and lyophilized. Extraction, separation, and analyses were adapted from a previously described method (Abbadi et al., 2004; Simionato et al., 2013). Cell pellets were suspended in 4 mL of boiling ethanol for 5 min, during which they were mechanically broken. Then, 2 mL methanol and 8 mL chloroform were added at room temperature, saturated with argon, and the mix was incubated for 1 h. Cell debris was removed by filtrating the mixture through glass wool, and the filter was rinsed with 3 mL chloroform/methanol 2:1 (v/v). A volume of 5 mL NaCl 1% was added to the filtrate to induce separation of the organic and the solvent phase. The later was recovered and dried under argon for storage purposes and suspended in chloroform to take aliquots. A 1/100 aliquot of the lipid extract was taken for FA quantification by GC-FID. To this end, a known quantity of 15:0 FAs was added to each sample, and lipids were methanolyzed into FAMEs for 1 h in 3 mL H₂SO₄ 2.5% in pure methanol at 100°C. The reaction was stopped by addition of 3 mL water, and phase separation was induced with 3 mL hexane. The hexane phase was collected and concentrated upon drying prior injection into the GC-FID system (Perkin Elmer). FAMEs were separated on a BPX70 (SGE) GC column and their peaks identified by comparison of their retention times with those of standards (Sigma). Quantifications were based on the elution peak of the 15:0 reference by normalization of the peak areas and correction to respective molecular masses. Separation of lipid classes was performed using 300 µg of total lipid extracts. For polar and neutral glycerolipids, two distinct resolving systems of TLC on glass-backed silica gel plates (Merck) were used, adapted from (Simionato et al., 2013). For the resolution of neutral lipids and free FAs, one-dimensional TLC is sufficient using hexane:diethylether:acetic acid (70:30:1, v/v) as solvent. Adequate separation of polar lipids was achieved by two-dimensional TLC with chloroform:methanol:water (65:25:4, v/v) as first solvent and chloroform:acetone:methanol:acetic acid:water at 50:20:10:20:5 (v/v) as second solvent. Lipid spots were sprayed with 2% 8-anilino-1-naphthalenesulfonic acid in methanol for UV visualization, and silica containing the different lipids were scraped off the plate for FAME production or GC-FID analyses of the different lipid classes or for structural analysis by mass spectrometry (MS). For MS analyses, purified lipid classes were dissolved in 10 mm ammonium acetate in pure methanol. They were introduced by direct infusion (electro spray ionization [ESI]-MS) into a trap type mass spectrometer (LTQ-XL, Thermo Scientific) and identified by comparison with standards as described previously (Abida et al., 2015). In these conditions, the produced ions were mainly present as H⁻, H⁺, NH4⁺, or Na⁺ adducts. Lipids were identified by tandem MS (MS/MS) analysis with their precursor ion or by neutral loss analyses as described previously (Abida et al., 2015). All experiments were made in triplicate.

Sphingolipid Extraction and Liquid Chromatography Coupled to MS/MS Analyses

N. gaditana wild-type and mutant cultures were harvested upon centrifugation after 7 d of cultivation (12-h-light/-dark cycle) and freeze-dried. Sphingolipids were extracted from 5 mg dry weight cells (three biological replicates per condition) according to a previously described method (Markham and Jaworski, 2007). A volume of 500 µL isopropanol-hexane-water (55:20:25) and 10 µL (0.01 nmol) of internal standard [C12-Cer: N-(dodecanoyl)-sphing-4-enine (d18:1-12:0)] was added to freeze-dried material and ground using a Polytron homogenizer. The plunger was rinsed with 500 µL of extraction solvent and the sample incubated at 60°C for 15 min. After centrifugation at 4,000 rpm for 5 min at room temperature, the supernatant was recovered and the pellet extracted once more with 1 mL of solvent. Supernatants were combined and dried using a speed Vac (Eppendorf) system. Then, the samples were incubated at 50°C for 1 h with 500 µL of 33% methylamine solution in ethanol-water (7:3). Samples were dried under nitrogen and resuspended by sonication in 100 µL of tetrahydrofuran-methanol-water (2:1:2) containing 0.1% formic acid and filtrated prior to analysis. UPLC-ESI-MS/MS analyses were carried out on a Waters Acquity UPLC system coupled to a Waters Xevo tandem quadrupole mass spectrometer equipped with an ESI source. Chromatographic conditions, mass spectrometric parameters, and multiple reaction monitoring (MRM) methods were defined as described previously (Tellier et al., 2014).

Transmission Electron Microscopy

For transmission electron microscopy, cells were pelleted (4,000 rpm, 15 min, 4°C), frozen in liquid nitrogen, and stored at −80°C until use. Sample fixation in reduced osmium tetroxide was adapted from a previously described method (Deerinck et al., 2010). To this end, cell pellets were resuspended in 0.1 m phosphate buffer (pH 7.4) and 2.5% glutaraldehyde and incubated overnight at 4°C in inverted tubes. Cells were then pelleted and washed five times in 0.1 m phosphate buffer. Cells were fixed by a 1-h incubation on ice in 500 µL 0.1 m phosphate buffer containing 1% osmium and 1.5% ferricyanide potassium red before they were pelleted and washed five times with 0.1 m phosphate buffer. Pellets were resuspended in 0.1 m phosphate buffer containing 0.1% tannic acid and incubated for 30 min in the dark at room temperature prior to centrifugation at 10,000g for 10 min at 30°C. Again, cells were pelleted and washed five times with 0.1 m phosphate buffer. The samples were dehydrated in ascending sequences of ethanol and infiltrated with ethanol/Epon resin mixture. Finally, the algal strains were embedded in Epon. Ultrathin sections (50–70 nm) were prepared with a diamond knife on a UC6 Leica ultramicrotome (Leica) and collected on 200-μm nickel grids. Ultrathin sections were examined on a Philips CM120 transmission electron microscope operating at 80 kV.

Chlorophyll Fluorescence Kinetics Measurements

To determine photosynthesis parameters in cell cultures, room temperature fast chlorophyll fluorescence kinetics were measured using a Speedzen MX fluorescence imaging system (JBeamBio) with settings previously described (Allorent et al., 2013). To this end, a 140-µL or 300-µL volume of N. gaditana culture was transferred to a transparent 96-well plate and dark-incubated for 20 to 30 min before measurements. Excitation was performed in the blue range (λ = 450 nm, F0), and actinic light pulses were given with a photosynthetic active wavelength of 520 nm. F0 is the steady-state fluorescence in dark-adapted cultures, F in light-adapted cultures, Fm is the maximal fluorescence after a saturating light pulse of dark-adapted cultures, Fm’ the same in light adapted cultures, and Fv is the difference between F0 and Fm. With these parameters, the maximum efficiency of energy conversion at PSII can be calculated as Fv/Fm, photochemical quenching capacity indicated by the quantum yield of PSII as Y(II) = (Fm’ − F)/Fm’, and NPQ as Fm − Fm’/Fm’ (Misra et al., 2012). Based on Y(II), the electron transport rate (ETR) was calculated as ETR = 0.5 × Y(II) × (photon flux density, µE) × 0.84, the latter term being a constant optimized for Arabidopsis (Arabidopsis thaliana) that might differ for Nannochloropsis (Schreiber, 2004).

Protein Biochemistry

A 50-mL N. gaditana culture was harvested in the midlog phase (3,500 rpm, 10 min, 4°C), frozen in liquid nitrogen, and thawed on ice to facilitate cell rupture. Proteins were extracted from the cell pellet by adjusting the medium to a final concentration of 6 mm CHAPS, 1 mm DTT, 50 mm MOPS, pH 7.8, KOH, and incubation for 20 min on ice. Protein concentrations were determined (Lowry et al., 1951) and protein samples (30 µg) were analyzed by SDS-PAGE (Laemmli, 1970). Protein bands were visualized by staining with Coomassie Brilliant Blue (Roth). Alternatively, proteins were transferred to a nitrocellulose membrane (Protran BA83, GE Healthcare) via western blotting. The nitrocellulose membrane was blocked with 5% skimmed dry milk in a Tris buffer (10 mmol.L⁻¹ Tris, 100 mmol.L⁻¹ NaCl, pH 7.5) containing 0.05% Tween 20 overnight at 4°C under gentle agitation. The antibody recognizing the LHCX1 protein was produced as previously described (Perin et al., 2015). A 1:80,000 dilution of the anti-LHCX1 antibody was added to buffered milk, and the nitrocellulose membrane was incubated overnight as described above. Subsequently, the membrane was washed three times with a fresh buffered milk medium prior to the addition of an antirabbit antibody (1:10,000; Peroxidase-conjugated AffiniPure Goat Anti-Rabbit IgH, Jackson Immunoresearch). After a 2-h incubation at room temperature, antibody-labeled protein bands were visualized using a BioRad Chemidoc MP system. Intensities were quantified relatively to the Coomassie-stained loading control, in the nonsaturated range, using the Adobe Photoshop software.

Accession Numbers

Sequence from Nannochloropsis genome and data used for phylogenetic tree reconstructions can be found in the N. gaditana and the GenBank data library under the following accession numbers: AAO34582; CBN78890; CCI42963; CDJ36988; ETM45813; LbrM.14.0670; Naga_100003g8; Naga_100004g102; Naga_100017g49; Naga_100020g80; Naga_100027g27; Naga_100042g12; Naga_100061g21; Naga_100092g4; Naga_100162g4; Naga_100162g5; Naga_100273g7; Naga_100273g7; Naga_100399g1; NP_001026710; NP_001255291; NP_009963; NP_012339; NP_013476; NP_060240; NP_068586; NP_073563; NP_079206; NP_569717; NP_569717; NP_689523; Tb927.5.4530; Tb927.7.4160; Tb927.7.4170; Tb927.7.4180; TGME49_053880; TGME49_205350; TGME49_242380; XP_001351023; XP_002040; XP_002176686; XP_002179048; XP_002180428; XP_002184740; XP_002288481; XP_002291938; XP_002293395; XP_003251850; XP_003401825; XP_003882280; XP_005187861; XP_005187861; XP_005769239; XP_005838572; XP_008620377; XP_008865117; and XP_011131654.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Amino acid sequence motifs in the elongase NgΔ0-El01.

• Supplemental Figure S2. Functional analysis of Δ0-NgELO1 in yeast.

• Supplemental Figure S3. Comparison of the total FA profiles of three independent mutant lines of N. gaditana obtained by knocking out the Naga_100083g23/NgΔ0-elo1 gene.

• Supplemental Figure S4. FA profile of membrane glycerolipid classes from NgWT and NgΔ0-elo1 KO lines.

• Supplemental Figure S5. Gene expression analysis.

• Supplemental Figure S6. Similar growth (A) and biomass (B) production of NgWT and NgΔ0-elo-KO.

• Supplemental Figure S7. Moderate reduction of PSII efficiency in NgΔ0-elo1-KO compared to NgWT, based on chlorophyll fluorescence (A) and Fv/Fm measures (B).

• Supplemental Table 1. Positional distribution of FAs, and molecular species found in each glycerolipid classes.

Glossary

ARA

arachidonic acid

Cer

ceramide

DAG

diacylglycerol

DGDG

digalactosyldiacylglycerol

DGTS

diacylglyceryltrimethylhomo-Ser

EPA

eicosapentaenoic acid

ER

endoplasmic reticulum

ESI

electro spray ionization

ETR

electron transfer rate

FA

fatty acid

FAME

fatty acid methyl ester

FASII

fatty acid synthase of type II

Fd

ferredoxin

GC-FID

gas chromatography coupled to ion flame detection

gDNA

genomic DNA

h-Cer

hydroxyl-ceramide

HII

hexagonal II phase building

KO

knockout

LCB

long chain base

MC-FA

medium chain fatty acid

MGDG

monogalactosyldiacylglycerol

MS

mass spectrometry

NPQ

nonphotochemical quenching

PAR

photosynthetically active radiation

PE

phosphatidylethanolamine

PG

phosphatidylglycerol

qE

photo-energy emitted as heat and fluorescence

SQDG

sufoquinovosyldiacylglycerol

TAG

triacylglycerol

TLC

thin-layer chromatography

VLC-PUFA

very-long-chain polyunsaturated fatty acid

Vx

violaxanthin

Y(II)

quantum yield of PSII

Zx

zeaxanthin