Wax Ester Synthase/Diacylglycerol Acyltransferase Isoenzymes Play a Pivotal Role in Wax Ester Biosynthesis in Euglena gracilis

Wax ester fermentation is a unique energy gaining pathway for a unicellular phytoflagellated protozoan, Euglena gracilis, to survive under anaerobiosis. Wax esters produced in E. gracilis are composed of saturated fatty acids and alcohols, which are the major constituents of myristic acid and myristyl alcohol. Thus, wax esters can be promising alternative biofuels. Here, we report the identification and characterization of wax ester synthase/diacylglycerol acyltrasferase (WSD) isoenzymes as the terminal enzymes of wax ester production in E. gracilis. Among six possible Euglena WSD orthologs predicted by BLASTX search, gene expression analysis and in vivo evaluation for enzyme activity with yeast expressing individual recombinant WSDs indicated that two of them (EgWSD2 and EgWSD5) predominantly function as wax ester synthase. Furthermore, experiments with gene silencing demonstrated a pivotal role of both EgWSD2 and EgWSD5 in wax ester synthesis, as evidenced by remarkably reduced wax ester contents in EgWSD2/5-double knockdown E. gracilis cells treated with anaerobic conditions. Interestingly, the decreased ability to produce wax ester did not affect adaptation of E. gracilis to anaerobiosis. Lipid profile analysis suggested allocation of metabolites to other compounds including triacylglycerol instead of wax esters.


Results
Putative WSD orthologs in E. gracilis. To identify potential orthologs whose products are expected to show wax ester synthesis activity, an amino acid sequence of A. calcoaceticus WSD was used as a query for BLASTX search against E. gracilis RNA-Seq database constructed previously (DRA accession number: SRP060591) 20 . Six components with relatively high similarity to A.calcoaceticus WSD ranging from 43.3% (7.1% identity) to 64.8% (20.2% identity) were selected as the potential candidates encoding Euglena WSD (Table 1). We designated them as EgWSD1 to EgWSD6. With a search against the Pfam domain database (http://pfam. xfam.org/) 21 , we identified wax ester synthase-like acyl-CoA acyltransferase domain (PF03007) and DUF1298 domain (PF06974), the latter of which is a conserved domain of hypothetical plant proteins of unknown function, in the N-and C-terminal halves, respectively, of all six EgWSDs (Fig. 1A). In the former domain, a predicted active site motif responsible for ester bond 22 , HHXXXDG, is highly conserved in all EgWSDs, except for EgWSD4 whose first His residue is replaced by glutamate (Fig. 1B). These protein structures including the active site motif are well consistent with other WSDs from A.calcoaceticus and Arabidopsis. In addition, four EgWSDs, except for EgWSD5 and EgWSD6, were predicted to have one or two transmembrane domains, using TMHMM server V.2.0 (http://www.cbs.dtu.dk/services/TMHMM/), SOSUI program (http://harrier.nagahama-i-bio.ac.jp/ sosui/), DAS-Transmembrane Prediction server (http://www.sbc.su.se/~miklos/DAS/tmdas.cgi), and TMpred (http://www.ch.embnet.org/software/TMPRED_form.html), expected to be membrane bound ( Supplementary  Fig. S1). Given the absence of obvious organellar targeting signals, the most likely location of EgWSDs, excluding EgWSD5 and EgWSD6, is likely to be the ER membrane, which is the main place of wax ester synthesis in E. gracilis 7 . On the other hand, as for EgWSD5 and EgWSD6, prediction of their hydrophobicity varied depending on programs used, indicating they will be of ambiguous protein with relation to localization. Further observations using specific antibody will be needed for understanding their accurate localization in the future work. EgWSD orthologs were compared with various WSDs from bacteria and plants by generating phylogenetic tree based on ClustalW analysis (http://clustalw.ddbj.nig.ac.jp/) (Fig. 1C). The EgWSDs were rather closely related to bacterial WSD families, but not plant ones, and formed their own clades, suggesting they evolved uniquely.

Expression of WSD orthologous genes in E. gracilis. The active biosynthesis of wax ester in E. gracilis
is evoked by encountering hypoxic conditions 5,6 . Previous comprehensive E. gracilis gene expression analysis showed that the expression levels of the six WSDs varied considerably (Table 1) 20 . We then performed qPCR analysis to anticipate which EgWSD would function predominantly in response to anaerobic conditions. As shown in Fig. 2, under ordinary aerobic conditions, the expression level of EgWSD2 was markedly higher than that of other EgWSDs and WS genes. Following EgWSD2, the expression levels of EgWSD3 and EgWSD5 were ranked high, in that order. The expression level of EgWSD1 gene was very low and almost negligible. Importantly, a comparative expression analysis of aerobic-and anaerobic-treated E. gracilis cells showed that gene expression levels in all EgWSD genes were not significantly altered by anaerobic treatment (Fig. 2). This result was well consistent with the previous RNA-Seq data that gene expression changes in predicted components involved in wax ester metabolism were not extensive or dynamic during the anaerobic treatment 20 . In vivo evaluation of wax ester synthesis activities of recombinant EgWSDs in yeast. To verify whether enzymes produced from the EgWSD2, EgWSD3, EgWSD5, and WS genes do indeed exhibit the wax ester synthesis activities, they were heterologously expressed in Saccharomyces cerevisiae strain H1246 which was disrupted at four key TAG synthesis related genes, ARE1/2, DGA1, and LRO1 23 . This strain showed no diacylglycerol SCiEnTiFiC REPORTS | 7: 13504 | DOI:10.1038/s41598-017-14077-6 acyltransferase activity, resulting in neutral lipid-deficient phenomenon and providing suitable background for evaluating recombinant WSD activities. The transformed yeast H1246 cells harboring the recombinant pYES2 with individual EgWSDs and WS genes were induced by the addition of D-galactose. Using a semi-quantitative RT-PCR, we confirmed that all transformed yeast H1246 cells were successfully expressed in individual transgenes with almost the same expression strength after the D-galactose induction (data not shown). The cells were further incubated for 48 h after supplementation of 250 μM myristic acid and myristic alcohol, and then C28 contents were determined and quantified by gas chromatography-mass spectrometry (GC-MS) analysis. Both EgWSD2and EgWSD5-introduced yeasts exhibited remarkable levels of C28 accumulation (approx. 2,500 and 500 μg/ gFW, respectively) compared with the empty vector control (25 μg/gFW) (Fig. 3A). In EgWSD3-introduced yeast, C28 accumulation was significant, but severely impaired compared with EgWSD2-and EgWSD5-introduced yeasts. In contrast, EgWS-introduced yeast showed C28 levels equivalent to the empty vector control. As shown in Fig. 3B, a thin-layer chromatography (TLC) analysis was supported the quantitative results from GC-MS, and also showed no detectable accumulation of TAG even under the medium without providing myristic alcohol, except for EgWSD3-introduced yeast cells which showed significant spots corresponding to TAG standard under the medium conditions with/without myristic alcohol. The results suggested that at least two major WSDs, EgWSD2 and EgWSD5, were enzymatically specific toward wax ester synthesis, but not utilize diacylglycerol as substrate. On the other hand, EgWSD3 seems to be bifunctional enzyme like bacterial WSD families. To evaluate substrate specificity of EgWSD2 and EgWSD5, we added various fatty acids and fatty alcohols with carbon lengths C16 and C18. As shown in Fig. 4, both enzymes readily utilized C14 fatty acid and C14 alcohol as substrates, showed low activity with C16 substrates, and almost negligible reaction toward C18 substrates. The results clearly indicated the reaction of EgWSD2 and EgWSD5 with a high specificity toward shorter chain length fatty acids and alcohols.
Effect of EgWSD gene silencing on wax ester production. To evaluate the physiological significance of EgWSDs in the synthesis of wax ester in E. gracilis cells, we temporarily suppressed their expression using RNA-mediated interference. A double-stranded RNA (dsRNA) synthesized from part of the EgWSD sequences was introduced into Euglena cells by electroporation. Initially, we tried to silence individual EgWSDs and EgWS. The effect of silencing on expression of individual targeted genes was subsequently confirmed by RT-PCR (Supplementary Fig. 2A). However, none of the individual gene-silenced cell lines showed significant difference in wax ester accumulation after 24 h-anaerobic treatment, indicating functional compensation each other ( Supplementary Fig. 2B). Notably, the C28 level in EgWSD2-silencing cells under aerobic growth conditions was significantly lower than that in other cell lines, indicating that EgWSD2 would be the dominant enzyme for wax ester synthesis at least under ordinary aerobic conditions. Subsequently, based on the results of gene expression analysis and recombinant EgWSD activities shown in Figs 2 and 3A, respectively, we chose two WSDs, EgWSD2 and EgWSD5, as dominant WSD genes expressed in E. gracilis cells, and tried to silence them simultaneously. As a result, the double-knockdown (double-KD) cell line showed considerable loss of wax ester accumulation under anaerobic conditions (Fig. 5), indicating that they are substantial wax ester synthases. It is also worth noting that the wax ester content in the WSD2/5 double-KD cell line was negligible in aerobiosis, whereas that in mock-control cells was still detectable (Fig. 5B inlet). The WSD2/5 silencing did not affect cell growth and viability even under anaerobic conditions at least for 48 h (data not shown). We then measured paramylon and fatty acid contents to confirm the effect of gene silencing on their productivity and consumption. As shown in Fig. 5C, the amounts of paramylon did not differ significantly between WSD2/5-silenced cell lines and mock control cells before and after anaerobic treatments. In addition, the amount of myristic acid in WSD2/5 silencing cells rather increased by approximately 35% under anaerobic conditions (Fig. 5D). These facts clearly indicate that the ability for paramylon and de novo fatty acid synthesis is not affected by WSD gene silencing, and the reaction process  of wax ester synthesis is not a limitation factor for metabolic regulation of these metabolites during wax ester fermentation.

Lipid profile analyses of EgWSD2/5 double-KD cells.
We now have a simple question what the fatty acids generated in EgWSD2/5 double-KD cells under anaerobic condition were converted into in substitution for wax esters. To verify the clue, we performed detailed lipid profile analyses according to Furuhashi et al., 24 . As shown in Figs 6 and 7A, the relative amount of wax esters, including C28, was significantly low in EgWSD2/5 double-KD cells, whereas that of fatty acids did not differ between the silenced cells and mock control cells, obviously supporting the result obtained in Fig. 5. Interestingly, when we analyzed the relative amounts of fatty alcohols, the EgWSD2/5 double-KD cells had decreased levels of not only C14 alcohols but also other fatty alcohol molecules (Fig. 7B). In addition, by comparing a lipid profile between mock control and the silencing cells, we observed a remarkable difference in a certain part with high molecular weight at a retention time over 50 min, as shown in Fig. 8A and B. Their mass spectrum indicated that they were some kind of TAG molecules (Fig. 8C). Hence, TAG levels in the EgWSD2/5 double-KD cells were compared by conventional TLC analysis. As expected, the amount of TAG molecules in the silencing cells increased, in contrast to the decreased wax esters content (Fig. 9A). TAG, which co-migrated at the same position as the standard TAG, was then scraped, eluted and trans-esterified to fatty acid methyl esters (FAME). The FAME was analyzed as the lipid compositions by GC/MS. The relative amount of C14 in the EgWSD2/5 double-KD cells increased by approximately three-fold in the mock control cells, suggesting that some of free C14 molecules are allocated into TAGs, resulting in TAG accumulation (Fig. 9B).

Discussion
E. gracilis is considered to rely on wax ester fermentation to survive anaerobic conditions 5,6 . Elucidating the metabolic processes is key to understanding the cellular physiology in absence of oxygen. Such studies provide valuable information regarding whether it is necessary for E. gracilis to synthesize wax esters for the express purpose of sustaining ATP production under anaerobic conditions. Furthermore, from a biofuel perspective, it is critical to understand functions of metabolic enzymes involved in wax ester production. Wax esters in E. gracilis consist mainly of C14 acids and C14 alcohols, and Teerawanichpan and Qiu 13 previously reported WS as the enzyme catalyzing their formation, although its physiological significance at cellular level has not been assessed.
In this study, based on our previous comprehensive gene expression analysis of E. gracilis 20 and homology search for the enzyme that mediates wax ester formation in A.calcoaceticus 17 , we predicted that EgWSD isoforms are the key enzymes catalyzing the last step of wax ester biosynthesis in this microalga. Several biochemical evidences discussed below revealed that this is certainly the case. Firstly, a comprehensive gene expression analysis indicated that the expression levels of EgWSD genes were much higher than that of EgWS gene (Table 1, Fig. 2) 20 . In particular, genes that showed higher expression levels were EgWSD2, EgWSD5, and EgWSD3. Secondly, among the six WSD orthologs in E. gracilis, gene silencing of EgWSD2 having the highest gene expression level exhibited marked deficiency in the accumulation of C28 under aerobic condition (Supplementary Fig. S2). Under anaerobic conditions, although silencing of individual genes including WS showed no significant difference in C28 accumulation, simultaneous silencing of both EgWSD2 and EgWSD5, genes of two high ranks having high expression levels, exhibited significant decreases in C28 accumulation (Fig. 5B). This finding clearly indicated that EgWSD2 is the most dominant enzyme at least in aerobic wax ester production, and EgWSD5 might be the major anaerobic wax ester synthesis enzyme together with EgWSD2. The residual amount of C28 (approximately 30% in mock control cells), which was still present in the EgWSD2/5 double-KD mutants, indicated that there might be a contribution of other WSD isoforms like EgWSD3, which show significant activity for C28 synthesis (Fig. 3A). Thirdly, when EgWS and EgWSD were expressed in the TAG biosynthesis defective yeast mutant H1246 22 , both EgWSD2 and EgWSD5 showed conspicuous wax ester synthesis ability when supplemented with both myristic acid and myristic alcohol as the substrates; however, EgWS did not (Fig. 3A). Therefore, based on these results, we concluded that WSDs play a pivotal role in wax ester biosynthesis in E. gracilis.
In agreement with the catalytic activity of WSD1 in Arabidopsis 19 , the results that both EgWSD2 and EgWSD5 expressed in yeast strain H1246 showed significant activities toward wax ester synthesis, but not TAG synthesis, also suggested that EgWSDs do not exhibit DGAT activity in vivo (Fig. 3B). This fact indicated that WSD enzymes in Euglena and plants have different catalytic properties from those in prokaryotes because a representative WS/ DGAT from A.calcoaceticus ADP1 restored TAG biosynthesis when expressed in the same yeast strain H1246 22 . This assumption is also supported by the fact that Arabidopsis WSD1 mutant showed marked suppression of wax ester accumulation 19 , whereas WS/DGAT-disrupted Streptomyces strain had lower TAG amount than WT strain 25 . Although biochemical evidence to explain the relationship between primary structure and enzyme activity of WSD is still limited, in contrast to WSD in A.bayliyi sp. ADP1 18 , EgWSD isoforms do not possess similarities to the acyl-CoA:glycerol-3-phosphate acyltransferase motifs in their primary amino acid sequences, supporting our conclusion that EgWSDs are not bifunctional enzymes like typical bacterial WSDs, and function as enzymes only for wax ester synthesis. As for substrate specificity of Euglena WSDs, in agreement with the composition of Euglena wax esters 7 , they preferably utilize C14 fatty acid and C14 alcohol (Fig. 4). In contrast, Arabidopsis WSD1 showed higher activity toward longer chain acyl acceptors like C24 or C28 19 . Consequently, WSDs from Euglena are likely novel enzymes specific for wax ester synthesis that recognize middle lengths of fatty acids and alcohols.
From the perspective of metabolic regulation from paramylon degradation to wax ester production in response to anaerobic conditions, there is now considerable evidence that paramylon accumulation and its degradation, and the following de novo fatty acid production were not affected by the suppression of wax ester synthesis, clearly indicating that the last reaction of the pathway is not a limiting step for the entire metabolic process. Although further experimental evidence needs to identify the crucial step of the regulation, it is at least suggested that conversion of pyruvate to acetyl-CoA by pyruvate:NADP + oxidoreductase (PNO) is one of key processes considering that PNO functions in anaerobic conditions and its suppression caused serious cell death under the same conditions 9,26 . It is worth noting that fatty alcohol levels in EgWSD2/5 double-KD cells were significantly lower than those in the mock control, as indicated by lipid profile analysis (Fig. 7). It is likely that FAR cooperates with WSD for efficient wax ester synthesis. Regarding increasing TAG levels in EgWSD2/5 double-KD cells, incorporation of helpless fatty acids into non-toxic TAG might be a reasonable strategy to protect cells from high concentrations of free fatty acids generated under anaerobic conditions because they are considered to be potentially damaging to cell membrane 27 . As for TAG biosynthesis in Euglena, although it has never been reported so far, EgWSD3 may participate in TAG production due to its TAG synthesis activity. Or at least three putative contigs encoding DGAT (comp21783_c0_seq. 1, comp24776_c1_seq. 1, comp30904_c0_seq. 1_7) exist in the Euglena RNA-Seq database 20 . It would be interesting to know why Euglena dominantly produces wax esters as a neutral lipid under anaerobic conditions, in spite of the existence of potential TAG biosynthesis ability. Further enzymlogical and physiological analyses of enzymes involved in TAG biosynthesis need to clarify this matter. With respect to lipid metabolism, it would also be crucial to understand why Euglena produces wax esters expressly during anaerobic fermentation. Simultaneous suppression of WSD2 and WSD5 genes did not affect any phenotypical and physiological parameters such as cell mobility, cell shape, cell growth, viability, and chlorophyll content, although accumulation of wax ester was reduced by approximately 70% under anaerobic conditions. On the other hand, the ratio of paramylon degradation in the double gene suppression strain under anaerobic condition showed no significant difference from that in the mock control, suggesting ATP acquisition thorough ordinary glycolysis. Thus, it is conceivable that production of wax ester is not indispensable to adapt to anaerobic conditions, and the cause and benefit of wax ester conversion under anaerobic conditions is still unclear. From the other perspective, this formidability is an important strategy for Euglena to survive under harsh environmental conditions 28 , even if the major route for wax ester synthesis is blocked by some unexpected reason. It might be plausible that, comparing with TAG, degradation of wax ester, which has only a single bound ester, might be the easier way to utilize the resultant products, fatty acids and alcohols, for energy generation whenever sufficient oxygen is supplied again to anaerobic environments. In summary, we have identified WSDs that are key enzymes for wax ester synthesis in Euglena. We showed that Euglena WSDs were unique enzymes in terms of substrate specificity for the recognition of middle lengths of fatty acids and alcohols, mainly the carbon length of 14. The recombinant WSDs were functionally expressed in yeast, and development of such cell lines, perhaps coupled with FAR, might provide a promising feedstock for biodiesel and aviation biofuel.  Total lipid was extracted from the Euglena cells and dried using a centrifugal evaporator. Sodium methoxide in methanol was added to the dried lipid pellet to transesterify the ester-linked lipids, followed by trimethylsilylation; then the methyl esters and the TMS derivatives were subjected to GC-TOF/MS analysis 24 . Asterisks indicate statistically significant differences between mock-control and WSD2/5 KD cells in each time points (means ± SD, n = 3; biological replicate, Student t-test, *p < 0.05).

Materials and Methods
Euglena strain and culture conditions. E. gracilis SM-ZK, a chloroplast-lacking bleached mutant derived from strain Z, was used throughout this study as it is a representative strain for previously reported wax ester research 5,6 . Cells were cultured in Koren-Hutner medium 29

Cloning of Euglena WSD and WS and heterologous expression in yeast. Based on a BLASTX
analysis against the Euglena RNA-Seq database, we identified six WSD and one WS homologs, and gene-specific primers were designed to amplify their coding region as listed in Supplementary Table S1. PCR amplification was carried out using KOD-plus polymerase (Toyobo, Japan) with Euglena cDNA pool as a template. All amplicons were cloned into pYES2 expression vector (Invitrogen, Carlsbad, CA) using gap-repair cloning method 30 . For yeast transformation, S.cerevisiae strain W303-1A was routinely grown at 30 °C in rich YPAD medium (1% yeast extract, 2% peptone, 2% glucose). The yeast was transformed using LiAc protocol 31 with a DNA mixture containing 0.2 μg of a linearized pYES2 and 1.0 μg of amplified target fragments. The transformants were then selected on SD-drop agar plates (0.2% amino acid mixture, 0.67% yeast nitrogen base without amino acid, 2% glucose, 1.8% agar) without supplementation of the appropriate metabolite. After 3 d of incubation at 30 °C, colonies were selected for cultivation in SD-drop medium, and the plasmids were extracted, amplified in E. coli DH5α, and verified by DNA sequence. The complete sequence information has been deposited in the DDBJ databank under accession number LC069357 to LC069364. For heterologous expression in yeast, the resultant constructs were transformed into yeast H1246 strain, and the yeast cells were cultivated in the medium supplemented with 2% (w/v) D-galactose and various fatty acids and alcohols for 48 h at 30 °C.

Quantitative real-time PCR experiments.
Euglena cells grown to stationary phase were anaerobically treated for 24 h and then collected. Total RNA was prepared using a RNAiso regent (Takara). Less than 500 ng of the total RNA was used for cDNA preparation with PrimeScript RT reagent Kit with gDNA Eraser (Takara). For quantitative real-time PCR, the reaction mixture contained 2 µL of the cDNA samples (100 ng/μL), 10 µL of the SYBR Premix EX Taq (Takara), 10 µL of forward and reverse primers, and H 2 O (up to 20 µL). The reaction was run with the Light Cycler 96 system (Roche). The relative expression level normalized to malate synthase was calculated with Light Cycler Application Software. The primers used are listed in Supplementary Table 1.

RNAi experiments.
Silencing of Euglena WSD paralogs by RNAi was performed as described previously 32,33 .
Approximately 500-bp partial Euglena WSD cDNAs were PCR-amplified with the addition of the T7 RNA polymerase promoter sequence at one end. The primers used are listed in Supplementary Table S1. Then the sense and antisense RNAs were synthesized using the PCR products as templates (MEGAscript RNAi Kit, Ambion). After purification of the transcribed RNA with DNase I digestion followed by ethanol precipitation, dsRNA was made by annealing equimolar amounts of the sense and antisense RNAs. Euglena cells of 2-d old cultures were collected and resuspended in culture medium containing 4.2 mM Ca(NO 3 ) 2 , 3.7 mM KH 2 PO 4 , and 2.1 mM MgSO 4 . One hundred fifty microliters of the cell suspension (100 µL; approximately 5 × 10 6 cells) was transferred to a 0.4-cm-gap cuvette and electroporated with 5 µl of RNA solution (15 µg of dsRNA in 50 mM Tris-HCl, pH 7.5, and 1 mM EDTA) using the NEPA21 electroporator (Nepa Gene) at 0.5 kV and 25 µF. The cell suspension was diluted with fresh KH medium and cultured at 26 °C for restoration. The FAME was analyzed as the lipid compositions by GC/MS. Values are the mean ± SD of three independent experiments. Asterisks denote statistically significant differences (*p < 0.05) compared with the mock control.