Production of ricinoleic acid-containing monoestolide triacylglycerides in an oleaginous diatom, Chaetoceros gracilis

Ricinoleic acid (RA), a hydroxyl fatty acid, is suitable for medical and industrial uses and is produced in high-oil-accumulating organisms such as castor bean and the ergot fungus Claviceps. We report here the efficient production of RA in a transgenic diatom Chaetoceros gracilis expressing the fatty acid hydroxylase gene (CpFAH) from Claviceps purpurea. In transgenic C. gracilis, RA content increased at low temperatures, reaching 2.2 pg/cell when cultured for 7 d at 15 °C, without affecting cell growth, and was enhanced (3.3 pg/cell) by the co-expression of a palmitic acid-specific elongase gene. Most of the accumulated RA was linked with monoestolide triacylglycerol (ME TAG), in which one RA molecule was esterified to the α position of the glycerol backbone and was further esterified at its hydroxy group with a fatty acid or second RA moiety, or 1-OH TAG, in which RA was esterified to the glycerol backbone. Overall, 80% of RA was accumulated as ME TAGs. Furthermore, exogenous RA-methyl ester suppressed the growth of wild-type diatoms in a dose-dependent manner and was rapidly converted to ME TAG. These results suggest that C. gracilis masks the hydroxyl group and accumulates RA as the less-toxic ME TAG.

Low-temperature-dependent ricinoleic acid production in Cp4. Optimal temperature conditions for RA accumulation were determined using Cp4 cells cultured at seven different temperatures: 10.0 °C, 12.5 °C, 15.0 °C, 17.5 °C, 20.0 °C, 22.5 °C, and 25 °C. At 10.0 °C and 12.5 °C, the cells grew poorly, and the cell density (absorbance at 730 nm) failed to reach 1.0, even after 10 d ( Supplementary Fig. S6). Meanwhile, at 15 °C, Cp4 cells grew slowly but their cell density reached the same levels as observed at optimal temperatures (20-25 °C) 21 after 7 d ( Supplementary Fig. S6), In addition, we observed that the cellular content of RA and the proportion of RA in total lipids (TLs) increased in a low-temperature-dependent manner (Fig. 2a,b), with the highest cellular RA content (2.2 pg/cell; 8.8% of TLs) observed after 7 d at 15 °C (Fig. 2a, Supplementary Fig. S7).
Furthermore, the growth rates of Cp4 and wild-type (WT) cells were not significantly different at any of the temperatures examined ( Supplementary Fig. S6), and the level of CpFAH expression in Cp4 cells increased transiently during the first 3 d at 15 °C to levels higher than those observed in cells grown at 25 °C (Fig. 2c). Conversely, the mRNA levels of endogenous Lhcr5 expression decreased gradually at both 15 °C and 25 °C (Fig. 2d), and the levels of fatty acids derived from TAG and fatty acids derived from TLs in Cp4 cells were greater than those in WT cells after 7 d at 15 °C (Fig. 2e,f).

Characterisation of estolide triacylglycerol structures containing ricinoleic acid.
To identify the lipid compounds containing RA, TLs were extracted from WT and Cp4 cells cultured for 7 d at 15 °C and separated by thin-layer chromatography (TLC). Each of the resulting lipid spots, a spot of origin, and the other portions without spots was then extracted and analysed by GC-MS ( Supplementary Fig. S8). Consequently, RA was detected only in the extracts of three spots (Nos 1, 2, and 3 in Fig. 3a), which contained 61% (1.3 pg/cell), 9% (0.2 pg/cell), and 17% (0.4 pg/cell) of the total cellular RA (2.2 pg/cell), respectively (Fig. 3b). In contrast, RA was not detected in the extracts from the origin, the other lipid spots, or any other parts of the TLC. In addition, the Rf values of spots Nos 1 and 2 were identical to those of normal TAG, in which three endogenous fatty acids were linked with a glycerol backbone, and free fatty acids (FFAs), respectively, and the signal intensities of the TAG and FFA spots from Cp4 cells were notably higher than those from WT cells, whereas spot No. 3 was only detected in Cp4 cells (Fig. 3a).
Scientific RepoRts | 6:36809 | DOI: 10.1038/srep36809 The structures of the RA-containing lipid molecules from each spot were determined using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) ( Table 1, Fig. 4, Supplementary Figs S9 and S10). Consequently, the RA-containing lipids from spots designated as Nos 1 and 2 in Fig. 3a were identified as ME TAG, which was the reported previously from ergot oil 22 . In these molecules, one RA molecule was linked to the glyceride backbone of TAG, and an additional fatty acids were esterified with the hydroxy group of the intracellular RA moiety (Fig. 3c).
In the LC-MS/MS analysis, total six species of ME TAG were identified from the spot No. 1 extract (Table 1). Among these, an endogenous 16:0 or 16:1 fatty acid was esterified with the hydroxy group of the RA moiety (Table 1), whereas in ME TAG detected from spot No. 2, the second RA was esterified with the hydroxyl group of the RA moiety (Fig. 3c, Supplementary Fig. S9). Therefore, we denoted ME TAG from spot No. 2 as 1-OH ME TAG. In addition, we also observed that ME TAG and 1-OH ME TAG co-migrated with TAG and FFAs, respectively. The RA-containing lipid molecules from spot No. 3 were identified as 1-OH TAG, in which an RA moiety was linked with the glyceride backbone of TAG (Fig. 3c, Supplementary Fig. S10).
To determine the RA-linked position on the ME TAG glycerol backbone, a lipid extract from spot No. 1 was hydrolysed using a site-specific lipase from Rhizopus arrhizus ( Supplementary Fig. S11), which hydrolyses acyl chains at the α (sn-1 and sn-3) positions of lipids. Following lipase treatment, we observed that the signal intensity of the TAG spot decreased; however, the intensities of the monoacylglycerol (MAG) and FFAs increased (lane 3 in Supplementary Fig. S11). In addition, RA and 12OH-16:1Δ 9 were detected from the FFAs spot but not from the MAG spot in the lipase-treated sample ( Supplementary Fig. S11).
Effect of exogenous hydroxyl fatty acids on cell growth of C. gracilis. Previously, RA production was shown to severely suppress the cell growth of CpFAH-expressing fission yeast 12 . However, in the present study, the growth of Cp4 and WT C. gracilis cells were similar ( Supplementary Fig. S6). In order to evaluate the toxicity of RA in C. gracilis cells, various concentrations of exogenous RA-methyl ester (RAME) was added to WT cultures at 15 °C, and oleic acid-methyl ester (OAME), which has the same structure as RAME, except for the lack of the hydroxy group, was used as a control. In addition, we used the methyl ester of each fatty acid, because the carboxy groups in FFAs are generally toxic to cells 23,24 , and acyl-CoA esters are unstable in the neutral aqueous solution, that is, the diatom's medium conditions 25 . Consequently, RAME inhibited cell growth in a dose-dependent manner (Fig. 5a), and addition of 4.0 μ g/ml RAME arrested cell growth completely, whereas OAME had no effect on cell growth (Fig. 5a).
To find out the destination of exogenously fed RAME in different lipid classes inside the WT cell, cellular lipid extracts from WT cells that had been cultured with 1.2 μ g/ml RAME were separated using TLC, and the content of RA in each spot was quantified (Fig. 5b). RA was not detected from cells harvested immediately after the addition of RAME to the medium. At d 0.5, RA was detected in four lipid spots, which corresponded to the expected migration of ME TAG, 1-OH ME TAG, 1-OH TAG co-migrating with RAME, and polar lipids. In particular, 38% and 20% of the exogenous RA was incorporated into 1-OH TAG (0.46 μ g/ml culture) and 1-OH ME TAG (0.24 μ g/ml culture), respectively. Their content decreased time-dependently and became undetectable at d 3 and d 7 (Fig. 5b). Furthermore, only the level of ME TAG increased during the first 3 d, whereas the other RA-containing lipids decreased in a time-dependent manner (Fig. 5b). Consequently, 83% of the exogenously fed RAME had been incorporated into the ME TAG (1.01 μ g/ml culture) at d 3, and this RA level in the ME TAG was still maintained at d 7 (0.89 μ g/ml culture). At d 7, the cell density of the RAME-treated culture reached 0.9, which was the same as that observed in the control and OAME-treated cultures.
Co-expression of CpFAH and MALCE1 encoding C16-fatty acid-specific elongase. Oleic acid (18:1Δ 9 ) only made a minor contribution to the fatty acid composition of transgenic C. gracilis cells (2.6% of total fatty acids in Cp4; Supplementary Fig. S7) but, nonetheless, it might limit the cellular content of RA. Similarly, stearic acid (18:0) was also identified as a minor fatty acid (4.2%) in Cp4 cells; whereas 16:0 was the most  Supplementary Fig. S4). Because oleic acid is synthesised from 16:0 by sequential fatty acid elongation and Δ 9 -desaturation, to increase the level of intra-cellularly accumulated RA, we attempted to increase the levels of C18 fatty acids, by providing additional 18:0 and 18:1Δ 9 as fatty acid substrates for CpFAH-catalysed hydroxylation.
To achieve this goal, a fungal gene encoding C16-FA-specific elongase, MALCE1 24 , was introduced into the transgenic Cp4 cells using an additional expression plasmid construct with a Zeocin-resistance gene, Streptoalloteichus hindustanus bleomycin (Sh ble), as a second selection marker, and the promoter region of an endogenous nitrate reductase gene 20 (CgNR promoter; Fig. 6a). In addition, a partial DNA fragment of CgpsbO (132 bp), which encodes a chloroplast transit peptide, was fused to MALCE1 (CgpsbO-tp-MALCE1;  The cellular growth rate of the transgenic line Cp4-ML47 was similar to those of WT and parental Cp4 cells, when cultured in the presence of nitrate (i.e. inductive conditions) or ammonium ions (i.e. non-inductive conditions; Supplementary Fig. S12). In addition, the expression of MALCE1 was upregulated during the first 3 d in inductive conditions, similar to the expression pattern observed for endogenous CgNR (Fig. 6b), and even in non-inductive conditions the level of MALCE1 expression was slightly higher at 7 d than at 1 d (Fig. 6b). In these conditions, the RA content of the Cp4-ML47 line (3.3 pg/cell) after 7 d at 15 °C and in inductive conditions was 1.4-and 1.5-fold greater than that of the Cp4-ML47 (2.4 pg/cell) and parental Cp4 (2.2 pg/cell) cells in non-inductive conditions, respectively (Fig. 6c, Supplemental Fig. S13). Consistently, the proportion of RA (11.5%) in TLs of the Cp4-ML47 line in inductive conditions was 1.8-fold greater than that of the Cp4-ML47 (6.3% of TLs) in non-inductive conditions (Supplemental Fig. S14). Conversely, the 16:0 content (7.5 pg/cell) and its proportion in the TLs (42.7%) was 0.6-and 0.8-fold of that observed in non-inductive conditions (2.4 pg/cell; 52.6%) (Supplementary Figs S13 and S14). Although the proportion of 18:0 (7.8%) and 18:1Δ 9 (2.2%) in TLs of the Cp4-ML47 line in inductive conditions was 1.3-and 3.8-fold greater than those of the Cp4-ML47, respectively (Supplemental Fig. S14). the contents of 18:0 (1.5 pg/cell) and 18:1Δ 9 (0.4 pg/cell) in the Cp4-ML47 line in inductive conditions were not significantly different from those in non-inductive conditions ( Supplementary Fig. S13).
To demonstrate the role of increased 18:1Δ 9 on RA synthesis in the FAH-expressing line, Cp4 and WT cells were cultured at 15 °C for 7 d with 4 μ g/ml OAME. TAG content and RA of the TAG in the Cp4 cells cultured with OAME were 1.9-fold greater than those in the Cp4 line cultured without OAME (Supplementary Fig. S15). However, the proportion of RA and other fatty acids in TAG in Cp4 line was not changed depending on the supply of OAME ( Supplementary Fig. S15). Similarly, by supplying OAME, the TAG content in the WT cells increased 2.2-fold, but the fatty acid composition was not affected, (Supplementary Fig. S15).

Discussion
RA is suitable for medical and industrial uses. In this study, carbon-neutral production of RA (2.2 pg/cell, 8.8% of TLs) was achieved by expression of the CpFAH gene in a transgenic oleaginous diatom C. gracilis in photoautotrophic conditions, without any apparent negative effects on cell growth. Furthermore, co-expression of 16:0-specific fatty acid elongase MALCE1 with CpFAH increased RA levels to 3.3 pg/cell (11.5% of TLs). Sixty-one percent of the synthesised RA was accumulated as ME TAGs, in which the RA hydroxyl group was masked by other fatty acids, which might explain its reduced cellular toxicity.
The cellular level of RA in CpFAH-expressing C. gracilis increased in a low-temperature-dependent manner (Fig. 2). Accordingly, maximum RA accumulation (1.7 pg/cell) was observed after 7 d at 15 °C, which was 8-fold greater than that observed when cultured at 25 °C (0.2 pg/cell), which is the optimal growth temperature for C. gracilis cells. In addition, expression analysis indicated that the abundance of CpFAH mRNA increased during the first 3 d of culture at 15 °C (optimal temperature for RA accumulation), after which it decreased. However, such temporal increases in CpFAH expression were not observed in cells cultured at 25 °C, and the expression pattern of the endogenous Lhcr5 gene 20 , the promoter of which was used to drive CpFAH expression, was not observed to increase in cells cultured at either 15 °C or 25 °C. These results suggested that the stability of the CpFAH transcript increased temporally during the first 3 d of culture at 15 °C and that it might promote RA  accumulation. An increase in the proportion of RA in the TLs occurred concomitantly with this increase in CpFAH mRNA stability during first 3 d (Fig. 2). Although the mRNA stability was not maintained after 3 d, the cellular content of RA was increased even after 3 d, and the proportion of RA in TLs (8.8-11%) was also maintained in this period, suggesting the protein stability of CpFAH and its enzymatic activity might be maintained at a high level at 15 °C even after the mRNA stability of CpFAH had declined. When WT C. gracilis was cultivated with RAME, cell growth was inhibited in a dose-dependent manner (Fig. 5a). In contrast, C. gracilis cells cultivated with OAME exhibited the same growth patterns as cells cultivated with ethanol, which indicates that the RA hydroxyl group was toxic to C. gracilis. Similar to the present results, RAME has been shown to inhibit cell proliferation and viability in yeast via intracellular hydroxyl groups 26 . In addition, when a lower concentration of RAME was added to cultures (1.2 μ g/ml), cell growth was restored at the late growth stage (3 d to 7 d). At 3 d, 83% of exogenously fed RA was incorporated into ME TAG, and the same levels of ME TAG were maintained still at 7 d without metabolic breakdown. On the other hand, at the early stage (0.5 d), 38% and 20% of exogenous RA was incorporated into 1-OH TAG and 1-OH ME TAG, respectively. Their content decreased time-dependently and became undetectable at d 3 and d 7 (Fig. 5b). These results suggest that half of the exogenous RAME was taken up at the early stage into TAG with free hydroxyl groups, and at least a large part of these lipids might be converted to ME TAG by an endogenous metabolic process. The accumulation of ME TAG might contribute to decrease the cellular toxicity of RA and restore cell growth. Feeding experiments Scientific RepoRts | 6:36809 | DOI: 10.1038/srep36809 using 13 C-labelled RAME as a tracer will be necessary to identify the metabolic process of exogenous RAME. Estolides in the biological samples can be extracted efficiently without change of its property 27 using a Bligh and Dyer liquid-liquid extraction method 28 , which was adopted in this study. However, in the industrial-scale extraction and purification of the diatom oil containing ME TAG, the effects of estolides structure on the yield at each process should be validated.
Estolide TAG has been detected in the oils of C. purpurea, from which CpFAH was isolated 22 , and also in the seed oils of Physaria spp 29 , Cardamine impatiens L. 30 , and Trewia nudiflora L. 31 . However, the pathway of estolide TAG synthesis remains unknown, and the related biosynthetic enzyme(s) have yet to be characterised. In CpFAH-expressing transgenic diatom C. gracilis, 61% of total RA was recovered from the TAG and ME TAG-co-eluting spots, whereas the remaining 9% and 17% were recovered from the 1-OH ME TAG and 1-OH TAG spots, respectively (Figs 3 and 4, Supplementary Figs S8-S10). This suggested that C. gracilis possesses an acyltransferase that catalyses the esterification of RA hydroxyl groups with the carboxy end of common fatty acids or other RA molecules. Furthermore, because RA moieties were only observed to occur at the α -position of the TAGs in CpFAH-expressing C. gracilis, ME TAG may be produced in one of two ways. First, esterification of RA hydroxyl groups with the carboxy end of common fatty acids may occur on phospholipid-linked RA moieties or on RA-CoA in the ER, after which the resulting estolide is integrated at the α -position of diacylglycerol (DAG), or alternatively, the RA is initially integrated into DAG to produce 1-OH TAG by diacylglycerol acyltransferase or phospholipid acyltransferase, after which common fatty acids are esterified with the hydroxyl groups of 1-OH TAG RA moieties. Because no hydroxyl fatty acids (HFAs), such as RA or estolide TAG, were detected in WT C. gracilis cells in the first place, it is unclear why the diatom possesses the biosynthetic activity of estolide TAG. However, we suggest that either (1) HFAs themselves, or diatom-infectious bacteria that produce HFA, are present in the habitat of C. gracilis, and so C. gracilis has a specific defence system for detoxifying exogenous HFAs; or (2) the production of estolide TAG could be a side-effect of producing acyltransferases involved in endogenous lipid metabolism. Nevertheless, the isolation and functional characterisation of the genes and enzymes involved in estolide synthesis are required to provide definitive answers to these hypotheses.
The amounts of RA and 16:0 accumulated by cells of the MALCE1-CpFAH-coexpression line Cp4-ML47 in inductive medium were 1.4-fold greater and 0.6-fold lower than that of the same line cultured in non-inductive medium, respectively. Consistently, the proportions of RA and 16:0 in TLs of the Cp4-ML47 line in inductive medium were 1.8-fold greater and 0.8-fold lower, respectively, than that of the same line cultured in non-inductive medium. Furthermore, the proportions of 18:0 and 18:1Δ 9 in TLs of the Cp4-ML47 line cultured in inductive medium were significantly higher than those in non-inductive medium. These results suggested that the expression of MALCE1 enhanced the conversion of 16:0 to 18:0 and promoted RA production. The condensing enzyme MALCE1 is normally located on ER and uses acyl-CoA as the substrate for acyl chain elongation by working with three other ER enzymes: ketoacyl-CoA reducatse, acyl-CoA dehydratase, and enoyl reductase in the fatty acid elongase complex 32 . In this study, MALCE1 was fused to a plastid-transit peptide at the N-terminus, suggesting that this enzyme can also use acyl-ACP as a substrate in cooperation with three endogenous plastidic enzymes: ketoacyl-ACP reductase, acyl-ACP dehydratase, and enoyl reductase. The accumulated levels of 18:0 and 18:1Δ 9 in the Cp4-ML47 line did not show any difference between cells cultured in inductive and non-inductive conditions, suggesting that the Δ 9 -desaturation from 18:0 to 18:1Δ 9 is not a rate-limiting step in the production of RA. Furthermore, the RA content of the Cp4-ML47 line in non-inductive conditions was slightly greater than that of the Cp4 line expressing the CpFAH gene. This might be caused by the leaky expression of MALCE1 regulated by the CgNR promoter, triggered by the depletion of ammonium ions in the medium during the late growth phase.
Supply of exogenous OAME in the Cp4 culture increased the TAG and RA contents in cells (Supplemental Fig. S15), suggesting that C. gracilis cells could take up exogenous OAME from the medium and utilize it for endogenous lipid metabolic processes. Unexpectedly, the fatty acid composition of TAG fatty acids was not changed between the Cp4 cells cultured with OAME and without OAME. This indicated that all fatty acid moieties in TAG increased with keeping each ratio. At first, we predicted that the exogenously fed OAME enhanced only the downstream lipid metabolism, that is, production of RA by CpFAH and production of the linoleic acid and other C18-to C22-polyunsaturated fatty acids, such as EPA and DHA, by endogenous fatty acid desaturation and elongation processes. We suggest that either (1) OAME was incorporated into cellular lipids or converted to OA-CoA to synthesize downstream lipid, and endogenous supply of OA from 16:0 fatty acid was suppressed to maintain the balance of the overall fatty acid ratio; or (2) when exogenous OAME was taken up into C. gracilis cells, it was digested through catabolic pathways and used for de novo fatty acid synthesis. Actually, in WT C. gracilis cells cultured with the exogenous OAME, TAG contents also increased to maintain the fatty acid ratio (Supplemental Fig. S15). This suggested that C. gracilis maintained metabolic homeostasis of fatty acids by adapting to the supply of exogenous fatty acids.
To enhance the conversion of acyl chains, including RA, from phospholipids to TAG in A. thaliana, a castor phospholipid:diacylglycerol acyltransferase (PDAT) was co-expressed with the RcFAH gene, which resulted in an RA content of 19% 33 in the seed oil. However, in CpFAH-expressing C. gracilis, TAG accounted for 87% of the TLs, and RA was only found in TAG molecules. Therefore, the co-expression of PDAT and CpFAH might not be necessary in C. gracilis considering that the TAG content of WT C. gracilis reaches 84% of TLs. Alternatively, use of more powerful promoters than Lhcr5 and NR in this study for the expression of CpFAH and MALCE1 genes, and knockout or knockdown of any endogenous Δ 12 -desaturase gene in RA-producing lines may be effective methods for further enhancing the production of RA in C. gracilis cells. Furthermore, TAG and total fatty acid contents in CpFAH-expressing C. gracilis cultured at 15 °C were 1.8-fold that in WT cells (Fig. 2), which has not been reported in other heterologous RA-producing organisms. Their production may have been promoted to compensate for the heterologous production of RA, as well as for the production of ME TAG molecules. These findings suggest that CpFAH-expressing C. gracilis can be used as a resource of biofuel production. Notably, CpFAH-expressing C. gracilis accumulates modest level of RA in estolide TAG without growth inhibition, potentially because estolide reduces the cellular toxicity of RA. Both estolide TAG 34,35 and estolides 36 have several other valuable chemical properties as well, which could be utilised for industrial and medical uses in the future. Therefore, we conclude that C. gracilis is an attractive producer of RA and estolide TAG.

Methods
Strains and culture conditions of C. gracilis. A wild-type Chaetoceros gracilis strain (UTEX LB2658) was used for transformation. It was cultured in Daigo's IMK medium (Nihon Pharmaceutical, Osaka, Japan) supplemented with sea salts (Sigma-Aldrich, St. Louis, MO, USA) and 0.2 mM Na 2 SiO 3 . The cells were grown at 20 °C in continuous light conditions at 50 μ mol photons/m 2 /s 1 . At the start of culture, the cell density at 730 nm was adjusted to 0.07. For feeding experiments, the stated concentration of RAME (Sigma-Aldrich) and OAME (Sigma-Aldrich) was added at the start of culture.
cDNA isolation from C. purpurea. To

Vectors construction and transformation of C. gracilis cells.
To obtain transgenic C. gracilis lines expressing the CpFAH gene, the CpFAH ORF derived from C. purpurea NBRC 6263 was amplified by PCR using primers CpFAH-BglII-fw and CpFAH-NsiI-rev and cloned into the BamHI-PstI site of a pCgLhcr5p plasmid (accession number; AB981621) downstream of the CgLhcr5 promoter for a light-harvesting fucoxanthin chlorophyll protein (fcp) gene 20 . The resultant plasmid was linearised by digestion using HindIII for transformation. Transformation was performed by a multi-pulse electroporation method using NEPA21 apparatus (Nepagene, Chiba, Japan) 20,39 . The transformed cells were selected on Daigo's IMK agar plates containing 1% agar and 400 μ g/ml nourseothricin (clonNat, Werner Bioagents, Thuringia, Germany) and further screened by genomic PCR (Supplementary Fig. S1).
To introduce MALCE1 into chloroplasts of the CpFAH-expressing line (Cp4), the pCgNRp plasmid (AB981622) 20 was firstly modified to have a cDNA encoding the transit peptide (tp) region of CgpsbO (accession number; AB373993) downstream of the CgNR promoter from a nitrate reductase gene (NR): A 132 bp-fragment of CgpsbO-tp amplified using the following primers: InFusion_NRp/CgpsbOtp_fw and PsbOsignal_AflII_rv, was inserted into the BamHI-XbaI site of the pCgNRp plasmid generating a new pCgNRp/CgpsbO-tp vector. Then, a fragment of the MALCE1 gene (accession number; AB468587) amplified using primers MALCEI_AflII_fw and MALCEI_XbaI_rv was inserted into the AflII-XbaI site of this vector. Furthermore, a clonNat-registrant gene (nat1) gene was replaced with a Zeocin-resistance gene (Sh ble) for the secondary transformation of the clonNat-resistant Cp4 line: a PCR-amplified fragment of the Sh ble gene using primers Sh_ble_BglII_fw and Sh_ ble_NsiI_rv was subcloned into the BamHI-PstI site of pUC118-pCgACAT-nat downstream of the acetyl-CoA acetyltransferase (CgACAT) promoter (AB981624) 20 . The resulting vector was digested with SacI and HindIII, and the fragment containing part of the CgACAT promoter and the Sh ble gene was transferred into the SacI-HindIII site of pCgNRp/CgpsbO-tp-MALCE1. The resultant plasmid was linearised by digestion using HindIII for transformation. Transformed Cp4 cells resistant to both 400 μ g/ml clonNat and 100 μ g/ml Zeocin (Sigma-Aldrich) were selected and further screened by genomic PCR. The PCR primers used for vectors construction and genomic PCR selection are listed in Supplementary Table S1.
Quantitative real-time PCR. Quantitative real-time PCR was performed using SYBR Premix Ex Taq GC (Takara Bio) and a LightCycler 480 Instrument (Roche, Basel, Switzerland), as described previously 40 . The Cg α -tubulin gene was used as an internal control. The primers used for qRT-PCR are described in Supplementary  Table S1. Lipid analysis. TLs were extracted from the yeast and diatom cells in a chloroform-methanol-water system 28 .
For transesterification of the TLs, the extracted TLs were incubated in 2.5% hydrogen chloride/methanol at 85 °C for 2.5 h, and fatty acid methyl esters (FAMEs) were extracted by 4 ml of petroleum ether with 0.4 ml of 5 M NaCl, and dried under N 2 gas. 20 μ l of 1 mM heptadecanoic acid (17:0) was added to the each lipid sample before the transesterification, and used as an internal standard. The resultant FAMEs dissolved in 30 μ l of acetonitrile, were trimethylsilylated by adding 30 μ l of N,O-bis(TMS)-acetamide:pyridine (1:1), and heating at 90 °C for 30 min. Then, 1 μ l of total FAMEs and TMS derivatives were analyzed using a Shimadzu gas chromatography mass spectrometry, GCMS-QP2010 system (Shimadzu, Kyoto, Japan) equipped with a DB-23 column (30 m × 0.25 mm) with 0.25-μ m film thickness (Agilent Technologies, CA, Santa Clara). The column temperature was maintained at 160 °C for 1 min, and then increased to 210 °C at a rate of 4 °C/min. For MS, the mass selective detector under electron impact conditions (70 eV) was scanning an effective m/z range of 40-500 at 1,666 amu/sec. Neutral lipids were separated into sterol esters, MAG, diacylglycerol, and triacylglycerol (TAG) by TLC using silica plates (TLC silica gel 60, 20 × 20 cm; Merck Millipore, Darmstadt, Germany) developed with n-hexane/ diethyl ether/acetic acid (70:30:1 v/v). After drying, the plate was sprayed with 80% aqueous acetone containing 0.01% primuline. TAG spots were scraped for transesterification as above. Then, 20 μ l of 1 mM 17:0 added to the sample before transesterification was used as the internal standard for quantification. The amount of TAG was quantified by gas-liquid chromatography (GC) using a Shimadzu (Kyoto, Japan) GC-1700 gas chromatograph equipped with a flame ionization detector and a split-less injection system, fitted with a capillary column (TC-70, 60 m × 0.25 mm I.D., GL Sciences, Tokyo, Japan) after derivatisation to FAMEs. The initial column temperature was maintained at 60 °C for 2.5 min, increased to 180 °C at range of 40 °C/min, maintained at 180 °C for 10 min, increased to 260 °C at range of 5 °C/min, and then maintained for 8.5 min. The injector and detector were operated at 250 °C. The fatty acid peaks were identified by comparing the retention times to known standards.
Positional analysis of TAG. Positional analysis of TAG was performed using R. arrhizus lipase (Sigma-Aldrich) as reported previously 41 . Lipids were extracted from the C. gracilis cells and resolved by TLC as described above. TAG was extracted from silica gel using the chloroform-methanol-water system as above 28 . Approximately 10 mg was dried under a N 2 gas stream and resuspended in 350 μ l of 0.1 M PBS buffer (pH 7.4) containing 4.28 mM Triton X-100. The sample was dispersed by sonication (Sonicator UR-21P, Tomy) for 6 × 10 s (output control: 8) on ice. Then, 20 μ g of R. arrhizuz lipase dissolved in the 100 μ l of PBS buffer was added to the emulsified TAG preparation, and incubated at 22 °C for 3 h. The lipase-treated lipids were extracted from the reaction mixtures, and resolved by TLC then developed with n-hexane/diethyl ether/acetic acid (70:140:3 v/v) for the separation of lipids containing HFAs, as described previously 9 . FFAs and MAG spots were scraped for transesterification as above.
Liquid chromatography coupled with tandem mass spectrometry analysis. LC-MS/MS analysis was performed using a LCMS-8040 tandem quadrupole mass spectrometer (Shimadzu, Kyoto, Japan). LabSolution LCMS software Ver. 5.65 (Shimadzu) was used to control the instrument and to process the data. The system used in the analysis consisted of a system controller CBM-20A, two pumps LC-30AD, and autosampler SIL-30AC, a column heater CTO-30AC, and a degasser DGU-20A 5 .
Lipids were extracted from the silica gel derived from the each TLC spot by the Bligh and Dyer method 28 and dried by evaporation. Then, the extracts were resolved with 1,000 μ l of 2-propanol and subjected to LC-MS/MS analysis. The liquid chromatography conditions were optimised as follows: solvent A was water, and solvent B was 2-propanol. The gradient profile was as follows: 100% B (1-25.0 min); 80% B (25.1-30 min). The flow rate was set to 0.2 ml/min, and the column temperature was 40 °C. Chromatographic separation was carried out on a Shim-pack XR-ODSII (75 × 2.0 mm, 2.2 μ m, Shimadzu GLC, Tokyo, Japan). The injection volume was 5 μ l. An LCMS-8040 tandem quadrupole mass spectrometer was operated in both positive and negative mode with APCI source in the range of m/z 200-1200. The operating parameters were optimised as follows: nebulizer gas flow, 4.0 l/min; drying gas flow, 5.0 l/min; desolvation line (DL) temperature, 200 °C; heat block temperature, 200 °C.
Accession number of the CpFAH gene used in this study. The CpFAH sequence in C. purpurea NBRC 6263 used in this study was deposited in the NCBI/EMBL/DDBJ database as accession number LC149858.