Expression of Castor LPAT2 Enhances Ricinoleic Acid Content at the sn-2 Position of Triacylglycerols in Lesquerella Seed

Lesquerella is a potential industrial oilseed crop that makes hydroxy fatty acid (HFA). Unlike castor its seeds are not poisonous but accumulate lesquerolic acid mostly at the sn-1 and sn-3 positions of triacylglycerol (TAG), whereas castor contains ricinoleic acid (18:1OH) at all three positions. To investigate whether lesquerella can be engineered to accumulate HFAs in the sn-2 position, multiple transgenic lines were made that express castor lysophosphatidic acid acyltransferase 2 (RcLPAT2) in the seed. RcLPAT2 increased 18:1OH at the sn-2 position of TAGs from 2% to 14%–17%, which resulted in an increase of tri-HFA-TAGs from 5% to 13%–14%. Our result is the first example of using a LPAT to increase ricinoleic acid at the sn-2 position of seed TAG. This work provides insights to the mechanism of HFA-containing TAG assembly in lesquerella and directs future research to optimize this plant for HFA production.


Introduction
Lesquerella (Physaria fendleri, formerly Lesquerella fendleri (Gray) Wats.) (Brassicaceae) [1] is valued for the unusual hydroxy fatty acids (HFAs) in its seed oil. The major HFA in lesquerella is lesquerolic acid (20:1OH) [2], comprising 55%-60% of total seed fatty acids (FAs) [3][4][5][6][7][8]. The main source of HFAs is castor (Ricinus communis), 90% of its seed oil is ricinoleic acid (18:1OH). 18:1OH is used as a feedstock for the manufacture of a range of products, such as lubricants, plasticizers and surfactants [9]. However, castor cultivation is limited by the presence of the toxin ricin in its seed. Lesquerella seeds do not contain ricin and therefore its oil provides a safe source of HFAs. Efforts have been made to improve lesquerella through plant breeding [4]. Since lesquerella can be transformed by Agrobacterium [10], genetic engineering also provides an alternative means to improve this potential crop.
Lesquerella TAGs contain~60% 20:1OH and almost all of it is esterified to the sn-1 and sn-3 positions [3,7] (Figure 1). The absence of HFAs at the sn-2 position of TAG might be caused by the selectivity of LPATs in lesquerella for common FAs. This is supported by the fact that LPAT enzymes from plants are some of the most stringent acyltransferases regarding substrate discrimination [41]. Based on several studies of plant microsomal (putative ER-localized) LPATs, Frentzen (1998) proposed the presence of at least two distinct classes of LPATs, class-A and -B, sharing little sequence homology between them. The class-A LPATs had selectivity for unsaturated C 18 acyl groups, and they are ubiquitously expressed in higher plants for membrane lipid and TAG synthesis [41]. The class-B LPATs are present in Limnanthes [42] and Cocos [43]. Their seed oils are enriched with the unusual FAs (UFAs), erucic acid (22:1) and lauric acid (12:0), respectively, at the sn-2 position of TAGs. Their corresponding members of class-B LPATs possessed substrate preference for UFAs and were used to improve FA composition in industrial oilseeds by genetic engineering [41,[44][45][46]. Interestingly, a yeast LPAT sharing little homology to plant LPATs also has preference for esterifying unusual very long chain FAs at the sn-2 position of seed TAGs [47]. Most plant species do not have a representative of the class-B gene, including Arabidopsis where five LPATs have been identified [48,49]. Arabidopsis LPAT2 (AtLPAT2) is the lone gene that encodes the ubiquitous and ER-located LPAT, and is thus suggested to be the key enzyme involved in membrane and seed oil biosynthesis [49]. Recently, a highly seed-expressed LPAT was identified from Sterculia foetida whose seeds accumulate up to 60% of desaturated cyclopropane FA [50]. SfLPAT showed strong homology to AtLPAT2 and enabled the incorporation of the unusual cyclopropane FA at the sn-2 position of LPA in transgenic Arabidopsis [50]. In castor, whole genome data [51] allowed the identification and analysis of the LPAT multigene family [51][52][53][54][55]. Among them, two genes, castor lysophosphatidic acid acyltransferase 2 (RcLPAT2) and castor lysophosphatidic acid acyltransferase B (RcLPATB), representing class-A and -B respectively, were found to express highly in seed tissues and thus were characterized in more detail [55]. Spatial and temporal gene expression profiling indicated that RcLPAT2 exhibited a generalized constitutive pattern typical to that of class-A members. In vitro enzyme assays indicated that RcLPAT2 had a preference for 18:1OH-CoA when sn-1-ricinoleate-sn-glycerol-3-phosphate (18:1OH-LPA) was used as an acyl acceptor, whereas when sn-1-oleoyl-sn-glycerol-3-phosphate (18:1-LPA) was used as the acyl-acceptor, RcLPAT2 preferred 18:1-CoA as a substrate [55]. These results suggest that RcLPAT2 might participate in both TAG and membrane glycerolipid synthesis. However, unlike the LPATBs from Limnanthes and Cocos, the expression of RcLPATB was not restricted to seeds as it was also expressed at a significant level in leaf, stem and floral organs. However, in vitro assays indicated that RcLPATB had a low enzyme activity and a broad specificity for acyl-CoAs compared with that of RcLPAT2. RcLPATB catalyzed the incorporation of saturated FAs, 12:0-16:0, at rates surpassing those of monounsaturated FAs including 18:1-CoA and 18:1OH-CoA, independently of the acceptors 18:1-LPA or 18:1OH-LPA [55]. Therefore, it remains questionable if RcLPATB is involved in HFA-containing TAG synthesis. We note that before the completion of castor genome sequencing, expression of a castor gene (GenBank ID, EU391594) designated as a "putative castor LPAT" [37] did not change HFA composition in transgenic Arabidopsis expressing RcFAH12. The sequence (EU391594) shows no significant similarity to RcLPAT2 (GenBank ID, JQ796917) and is now annotated as a putative castor 1-acyl-sn-glycerol-3-phosphate acyltransferase zeta precursor [51] (GenBank ID, XP_002525812).
Castor oil has long been used by industry as a raw material for manufacturing numerous established products. If lesquerella oil can be reengineered to resemble castor oil, this would provide an alternative source of castor oil that is safe, cost-competitive, and readily adaptable by existing industrial technologies. Increasing the levels of tri-HFA-TAG and 18:1OH, are key steps towards achieving this goal. We therefore over-expressed RcLPAT2 in lesquerella seeds. Analysis of seed TAGs suggests that RcLPAT2 is able to acylate the sn-2 position of LPA with 18:1OH and as a result the amounts of 18:1OH and tri-HFA-TAGs were increased. Additional genetic modifications for engineering a castor-oil producing lesquerella crop are discussed. To our knowledge this is the first time that castor RcLPAT2 has been expressed in another oilseed to modify HFA content and also the first example of metabolic engineering in lesquerella.

Castor Lysophosphatidic Acid Acyltransferase 2 (RcLPAT2) Increased 18:1OH Accumulation in Transgenic Seeds
We have produced selfed T 1 seeds from 16 independent primary transgenic plants (T 0 ); expressing RcLPAT2 under the control of the seed-specific napin (napA) promoter. Analysis of transgene RcLPAT2 copy number in T 0 plants indicated that 12 of them have one copy and four have two copies (Table 1). Among the lines, the FA composition exhibited some variations in each non-HFA (Table S1) as well as HFA (Table 1), which can be explained by the heterozygous nature of lesquerella, as it is nearly a wild species, and similar seed-to-seed variations were reported [5]. However, the average levels of each non-HFAs, palmitic (16:0), palmitoleic (16:1), stearic (18:0), oleic (18:1), linoleic (18:2), linolenic (18:3), and eicosenoic (20:1) acids, from transgenic lines are similar to that of wild type (wt) seeds (Table S1). Interestingly, all transgenic lines had increased 18:1OH content ranging from 1.35%-4.08% compared with that of wt at 1.20% (Table 1). Among them, four T 1 seed populations (lines 1, 3, 4 and 5) showed relative high 18:1OH contents from 3.79%-4.08%. The correlation between 18:1OH content and transgene RcLPAT2 copy number is weak (r = 0.1, p = 0.713). About 63% (10/16) of the lines also showed an increase in 18:2OH (Table 1), despite the overall low contents of 18:2OH (0.15%-0.75%) in transgenic lines and wt (0.24%) ( Table 1). The correlation between 18:1OH and 18:2OH contents was strong (r = 0.91, p < 0.00001). With the increase of 18:1OH, we observed a small decrease in 20:1OH levels showing an average of 50.03% compared to 51.50% in wt (Table 1). For 20:2OH, its average level was 3.22%, which is slightly higher than the 3.07% measured for wt (Table 1). Total HFA levels varied from 52.56% (line 8) to 60.44% (line 9) among transgenic lines with an average of 56.08% which is comparable to 56.18% of wt (Table 1). To examine the inheritance of the RcLPAT2 transgene, we selected hygromycin resistant T 1 seedlings and examined RcLPAT2 copy number in those relatively high 18:1OH T 1 populations. We identified two homozygous lines, 3-1 and 4-5, and produced selfed T 2 seeds for further analysis. For FA composition, we observed profiles similar to that of T 1 seeds (Table 2). We note that the 18:2OH content in wt was 0.24% in Table 1 and 0.02% in Table 2. The discrepancy was likely caused by the heterozygous nature of lesquerella seeds and by small sample size used for the analysis (10 seeds per sample for Table 1, and 15 seeds per sample for Table 2). The increased 18:1OH accumulation phenotype in lines 3-1 and 4-5 was inherited into the T 2 generation, although the levels of 18:1OH in homozygous lines increased only to 5.13% and 4.72%, respectively (Table 2). In general, there was no obvious alteration on growth phenotype in transgenic lines associated with the expression of RcLPAT2. Each HFA composition was measured as percentage of total fatty acids. Triplicates of 10-seed samples were measured for wild type and each independent transgenic line. All data are averages of three measurements˘SE. Fatty acid legend: 18:1OH is ricinoleic; 18:2OH is densipolic; 20:1OH is lesquerolic; and 20:2OH is auricolic acid. Two-tailed student's t-test. * p < 0.05; ** p < 0.01; *** p < 0.001. There are four possible molecular species of TAG based on the number of HFA chains esterified to the glycerol backbone: 0-, 1-, 2-, and 3-HFA-TAG. To investigate if there is any change in TAG species composition in the transgenic seeds, we determined the relative amount of each TAG species as a percentage of total TAGs in wt and transgenic lines 3-1 and 4-5. In wt, the relative amount of 0-, 1-, 2-, and 3-HFA-TAG was 4.79%, 12.17%, 79.13% and 4.79% respectively (Figure 2), which are comparable to reported levels [3,56]. In both transgenic lines, there are increases in 0-, 1-, and 3-HFA-TAGs showing 6.13%, 26.06%, and 14.38% in line 3-1 and 8.01%, 22.64%, and 12.93% in line 4-5, respectively ( Figure 2). However, we observed that 2-HFA-TAG decreased 29%-33%, showing 53.43% and 56.41% in lines 3-1 and 4-5, respectively ( Figure 2). Regiochemical analysis was performed for lines 3-1 and 4-5 to determine whether overexpression of RcLPAT2 increases the amount of HFAs at the sn-2 position. As shown in Figure 3, sn-2 18:1OH content increased from 2.19% to 17.29% or 14.85% in lines 3-1 and 4-5 respectively. Only very small increases were observed for sn-2 20:1OH from 0.93% to 1.64% or 1.53% in lines 3-1 and 4-5 respectively.

Discussion
To evaluate the utility of RcLPAT2 for the engineering of a lesquerella crop producing castor oil, we have produced transgenic lesquerella expressing RcLPAT2 under the control of a seed specific promoter. Our results indicate that RcLPAT2 enables the incorporation of 18:1OH at sn-2 position of LPA which increases the accumulation of 18:1OH and also tri-HFA-TAGs in lesquerella. This is the first successful step in the engineering of lesquerella seeds in order to produce castor oil. However, the overexpression of RcLPAT2 did not lead to an overall increase in total HFAs (Tables 1 and 2, Figure 2). In fact, we observed small decreases in 20:1OH (Tables 1 and 2), indicating that in part, the increase in 18:1OH was at the expense of 20:1OH. Since RcLPAT2 and PfKCS3 utilize the same substrate 18:1OH-CoA, it is possible that RcLPAT2 is competing with PfKCS3 for 18:1OH-CoA substrate, limiting the availability of 18:1OH-CoA for the elongation reaction. Therefore, silencing PfKCS3 would block the formation of 20:1OH and allow more 18:1OH-CoA to be directly incorporated into TAGs. In addition, the endogenous PfFAH12 is a bifunctional oleate 12 hydroxylase:desaturase [32], that might not be strong enough to keep up with the demand of 18:1OH-CoA during the seed development. If this is the case, co-expression of a strict 12-hydroxylase gene from castor (RcFAH12) [31] or from Physaria lindheimeri (PlFAH12) [57] might be necessary to boost 18:1OH-CoA availability for incorporation into seed TAGs. On the other hand, lesquerella produces the polyunsaturated FA 18:2 at 8%-9% (Table S1,  Table 2), which indicates that it must have an active desaturase PfFAD2 [58] that coverts 18:1 to 18:2. Since 18:1 is also the substrate for FAH12 to synthesize 18:1OH, suppressing PfFAD2 activity might increase the availability of 18:1 for FAH12 to produce 18:1OH. Moreover, a microsomal ∆-15 desaturase (PfFAD3) could be silenced to further suppress the conversion of 18:1OH to 18:2OH. Our transgenic lines expressing RcLPAT2 can be used as genetic backgrounds for studying the effect of additional transgenes such as RcFAH12 or PlFAH12 over-expression and/or PfKCS3/PfFAD2/PfFAD3 silencing in order to develop a castor oil-producing lesquerella.
In transgenic lesquerella expressing RcLPAT2, we observed an increase in 0-, 1-, and 3-HFA-TAG levels and a reduction in 2-HFA-TAG species (Figure 2). Since the total HFA content remained unchanged, it is likely that HFAs were redistributed from sn-1 and/or sn-3 to the sn-2 position of TAG due to the limitation of the HFA supply. Regiochemical analysis showed that sn-2 18:1OH or 20:1OH increased 6-7-fold or 1.5-fold, respectively (Figure 3). This indicates that RcLPAT2 allows for a more efficient acylation of 18:1OH than 20:1OH to the sn-2 position of TAG in vivo.
Although a number of mechanisms may facilitate accumulation of 18:1OH at the sn-2 position of TAG, the Kennedy pathway can be perceived as a major route for the channeling of 18:1OH into the sn-2 position of LPA by RcLPAT2. Figure 4 summarizes how RcLPAT2 might influence TAG biosynthesis through the Kennedy pathway in transgenic lesquerella seeds. A lesquerella LPAT (PfLPAT) could acylate LPA with HFAs at a low level since we detected about 2% of 18:1OH and 1% of 20:1OH at the sn-2 position of wt TAGs (Figures 3 and 4). We have identified a PfLPAT2 with 92% sequence identity to AtLPAT2 [59]. Gene expression analysis showed that PfLPAT2 is ubiquitously expressed in leaf, stem, flower bud, and developing seeds, suggesting it has a house-keeping role in membrane and storage lipid biosynthesis throughout plant life. In lines 3-1 and 4-5, small increase of 2% of 20:1OH at the sn-2 position of TAG were detected, however, significant increases of 18:1OH up to 15%-17% were detected (Figures 3 and 4). Thus unlike PfLPAT2, RcLPAT2 has evolved to acylate 18:1OH to the sn-2 position of LPA in seeds. Our transgenic experiments demonstrate that RcLPAT2 can be used to produce tri-HFA-TAGs in transgenic plants. In order to further increase the HFA level at the sn-2 position, the expression of PfLPAT2 could be silenced, since substrate competition between endogenous and transgenic acyltransferases can limit the accumulation of UFAs [40,60].   (Figure 3). However, the total amount of HFAs in the seed oil is not increased, which leads to a decrease in acylation of 20:1-OH the sn-1 and sn-3 positions (Tables 1 and 2). This indicates that RcLPAT2 has a high specificity for acylating 18:1-OH to the sn-2 position of LPA in lesquerella, which results in an increased amount of tri-HFA-TAGs in transgenic seed oil (Figure 2). 18:n means 18:1, 18:2, or 18:3, described as in Table 2. Other abbreviations are as in Figure 1. Red words show the effect of RcLPAT2.
In lesquerella, acyl editing may provide 18:1OH for TAG assembly. Recent studies have suggested that LPCATs are responsible for incorporation of newly synthesized FA into PC (forward reaction), and transferring mFA such as PUFA or HFA produced on PC back to the acyl-CoA pool (reverse reaction) [61][62][63][64] (Figure 1). When the reverse reactions were measured, a ricinoleoyl group at the sn-2 position of PC was removed three-to six-fold faster than an oleoyl group by seven LPCATs from five species tested, including lesquerella [64]. Our previous results showed that that there were no HFAs detectable in the PC fraction of total lipids isolated from developing lesquerella seeds [6], it is likely that acyl editing facilitates 18:1OH to be immediately released from PC and then utilized by RcLPAT2 in these transgenic lesquerella lines.
We cannot exclude that the PC-derived DAG pathway may also contributed to the accumulation of 18:1OH at the sn-2 position of TAG (Figure 1). There is strong evidence that plants enriched with PUFAs in seed TAG may utilize the PC-derived pathway [62]. Lesquerella seed TAGs contain about 22% PUFAs (18:2 and 18:3) (Table 2) (Figure 3). However, the total amount of HFAs in the seed oil is not increased, which leads to a decrease in acylation of 20:1-OH the sn-1 and sn-3 positions (Tables 1 and 2). This indicates that RcLPAT2 has a high specificity for acylating 18:1-OH to the sn-2 position of LPA in lesquerella, which results in an increased amount of tri-HFA-TAGs in transgenic seed oil (Figure 2). 18:n means 18:1, 18:2, or 18:3, described as in Table 2. Other abbreviations are as in Figure 1. Red words show the effect of RcLPAT2.
In lesquerella, acyl editing may provide 18:1OH for TAG assembly. Recent studies have suggested that LPCATs are responsible for incorporation of newly synthesized FA into PC (forward reaction), and transferring mFA such as PUFA or HFA produced on PC back to the acyl-CoA pool (reverse reaction) [61][62][63][64] (Figure 1). When the reverse reactions were measured, a ricinoleoyl group at the sn-2 position of PC was removed three-to six-fold faster than an oleoyl group by seven LPCATs from five species tested, including lesquerella [64]. Our previous results showed that that there were no HFAs detectable in the PC fraction of total lipids isolated from developing lesquerella seeds [6], it is likely that acyl editing facilitates 18:1OH to be immediately released from PC and then utilized by RcLPAT2 in these transgenic lesquerella lines.
We cannot exclude that the PC-derived DAG pathway may also contributed to the accumulation of 18:1OH at the sn-2 position of TAG (Figure 1). There is strong evidence that plants enriched with PUFAs in seed TAG may utilize the PC-derived pathway [62]. Lesquerella seed TAGs contain about 22% PUFAs (18:2 and 18:3) (Table 2), Therefore, it is likely that PC-derived DAGs are utilized in TAG assembly in lesquerella. Overexpression of RcLPAT2 in developing seeds of lesquerella could enrich the de novo DAG pool with sn-2 18:1OH-DAGs. The 18:1OH on the sn-2 position of the de novo DAG could end up at the sn-2 position of PC by the action of CPT, generating sn-2 18:1OH-PC. If a significant FA flux through PC-derived DAG exists in lesquerella, some of the sn-2 18:1OH-PC could be converted by PDCT to sn-2 18:1OH-DAG for TAG assembly. The generation of 18:1OH-PC could also provide an opportunity for a ∆-15 desaturase to desaturate 18:1OH-PC to form 18:2OH-PC, resulting in observed small increases of 18:2OH in the transgenic lines.
In conclusion, this is the first demonstration that RcLPAT2 can be used to acylate 18:1OH at the sn-2 position of 20:1OH-LPA to generate tri-HFA-TAG, a valuable property for the engineering of a new castor oil-producing crop, such as lesquerella. This work also represents the first successful attempt in genetic engineering of lipid metabolic pathways in lesquerella.

Construction of pGPro4-PnapA-RcLPAT2 Binary Vector
The map and sequence of the pGPro4 vector was described [65]. The promoter sequence of napA was cloned into the pGPro4 vector, resulting in pGPro4-PnapA. The activity of the napA promoter is seed-specific in lesquerella [66]. RcLPAT2 was PCR amplified from castor cDNA with KOD polymerase (Novagen, Billerica, MA, USA) (Forward: 5 1 CACCATGGCTGTTGCAGCTGTAGC3 1 , Reverse: 5 1 CTAGTCCTGTTTGTTTTCTG3 1 ) and cloned in the pENTR-D-TOPO vector (Life Technologies Ltd., Paisley, UK). Sequencing of RcLPAT2 showed that there was one nucleotide change from A to G at position 1154 in the cloned RcLPAT2 in comparison to the RcLPAT2 sequence (GenBank ID: JQ796917) from the genome sequence in NCBI. This nucleotide change led to an amino acid change (ArgÑGly). This difference is most likely due to the different ecotypes used. To generate pGPro4-PnapA-RcLPAT2, RcLPAT2 in pENTR-D-TOPO was PCR-amplified with gene-specific primers containing Sal I and BstE II restriction enzyme sites (Forward: 5 1 ACTGAGTCGACGGATCCATGGCTGTTGCAGCTGTAGC3 1 , Reverse: 5 1 ACGTGTCTAGAGGTGACCCTAGTCCTGTTTGTTTTCTGC3 1 ) using Roche High fidelity DNA polymerase (Roche, Indianapolis, IN, USA). The PCR product was ligated in the pGPro4-PnapA vector digested with Sal I and BstE II.

Plant Transformation, and Estimation of Transgene Copy Number
The lesquerella seeds, WCL-LY2 [67], were kindly provided by Dave Dierig (USDA-ARS, Arid-Land Agricultural Research Center, Maricopa, AZ, USA). Plant transformation was performed using the Agrobacterium tumefaciens strain AGL1 [68] carrying the binary vector pGPro4-PnapA-RcLPAT2. Tissue culture and plant growth conditions were as described before [10]. We have produced a total of 17 primary transgenic plants (T 0 ), plant #6 was lost before maturity. In the greenhouse, we generated selfed T 1 seeds by hand-pollination between different flowers from the same plant. Transgenic T 1 plants were obtained by germinating T 1 seeds in basal medium supplemented with 50 mg/L hygromycin for two weeks and healthy T 1 seedlings were transplanted into soil for T 2 seed production. Transgene copy numbers was estimated for T 0 plants, and T 1 lines 3-1 and 4-5 using a qPCR method described before [69]. Optimized primer sequences for RcPLAT2 are 5 1 TAGGCTGGTCCATGTGGTTTT3 1 (forward) and 5 1 TTCATCCTTGGCCCAGCTT3 1 (reverse). For internal control KCS4/5, the primer sets are the same as previously reported [69].

Lipid Extraction, TAG Species, Regiochemical and GC Analysis
The GC analysis for Table 1 and Table S1 was performed as described [38]. For Table 2, GC analysis was performed in the following way. One milliliter of hexanes and one mL of methanol containing 2.5% sulfuric acid (v/v) were added to seeds and incubated for 20 min at 100˝C. The top layer was extracted, dried under nitrogen and dissolved in 100 µL hexanes. GC was performed on a Restek RTX-2330 column (30 mˆ0.25 mm inner diameterˆ0.2 µm) using the following GC parameters: 120˝C for 1 min, a ramp to 180˝C at 40˝C¨min´1, a ramp to 190˝C at 7˝C¨min´1 with a 1 min hold, followed by a ramp to 250˝C at 30˝C¨min´1 with a 6 min hold. Lipid extractions, analysis of TAG species and regiochemical analysis were performed as described [38]. However, the solvent system used for separating MAGs and 1-HFA-MAGs was based on Bates and Browse, 2011 [20]. In addition, regiochemical analysis was performed on total TAGs instead of separate TAG species as previously described [38]. In brief, TAG species analysis was performed in the following way. 1.5 mg of total lipids from wild type and transgenic lines were run in triplicate on TLC plates (Z292974-1PAK, Sigma-Aldrich Company, Dorset, UK) alongside castor oil and flax oil standards. The TLC plates were first developed for 12 cm in CHCl 3 :methanol:HOAc (93:3:0.5, v/v/v), dried for 15 min, followed by a full development in CHCl 3 :methanol:HOAc (99:0.5:0.5, v/v/v). Lipids were visualized under UV after staining with the fluorescent dye primuline. Bands representing 3-HFA-, 2-HFA-, 1-HFA-and 0-HFA-TAGs were scraped from the TLC plates. The identity of the TAG species in each band was confirmed by measuring the percentage HFAs in each of these TAG fractions by GC (100%, 66%, 33%, and 0% HFAs in the 3-HFA-, 2-HFA-, 1-HFA-and 0-HFA-TAG fractions respectively). The amount of FAs in the 3-HFA-, 2-HFA-, 1-HFA-and 0-HFA-TAG fractions was calculated based on a 15:0 TAG internal standard added to each sample before sample preparation for GC. The relative percentage of each TAG species was calculated in the following way: µg FAs in the individual 3-HFA-, 2-HFA-, 1-HFA-or 0-HFA-TAG fractions/µg FAs in the combined 3-HFA-, 2-HFA-, 1-HFA-and 0-HFA-TAG fractions = relative % of each TAG species as a fraction of total TAG. GC analysis was performed by transmethylating FAs in 1 N methanolic HCl (Sigma-Aldrich Company) for two hours at 85˝C. FAMEs were extracted by adding 500 µL of hexane, and 1 mL KCl, vortexing and centrifugation to separate the hexane layer from the aqueous phase. The hexane layer was collected and dried down under nitrogen and 200 µL BSTFA + TMCS (BSTFA + TMCS 99:1 (Sylon BFT), Sigma-Aldrich Company) was added and incubated at 80˝C for two hours to silylate the FAMEs. The BSTFA + TMCS was evaporated and the silylated FAMEs were dissolved in hexane. Quantification was performed by GC with flame ionization detection on a HP-1MS column (30 mˆ0.25 mm i.d.ˆ0.25 µm, Agilent Technologies LDA UK Limited, Stockport, Cheshire, UK) using the following GC parameters: 50˝C for 1 min, a ramp to 200˝C at 50˝C¨min´1, followed by a ramp to 325˝C at 3˝C¨min´1. The identity of each silylated FAME was determined by GC-MS. In contrast to the GC results obtained for Tables 1 and 2  and Table S1,