The impact of reducing fatty acid desaturation on the composition and thermal stability of rapeseed oil

Summary Oilseed rape (Brassica napus) is the third largest source of vegetable oil globally. In addition to food uses, there are industrial applications that exploit the ability of the species to accumulate the very‐long‐chain fatty acid (VLCFA) erucic acid in its seed oil, controlled by orthologues of FATTY ACID ELONGASE 1 (Bna.FAE1.A8 and Bna.FAE1.C3). The proportion of polyunsaturated fatty acids (PUFAs) in rapeseed oil is predicted to affect its thermal stability and is controlled by orthologues of FATTY ACID DESATURASE 2, particularly Bna.FAD2.C5. Our aim was to develop rapeseed lines combining high erucic and low PUFA characters and to assess the impact on thermal stability of the oil they produce. The new type of rapeseed oil (high erucic low polyunsaturate; HELP) contained a substantially greater proportion of erucic acid (54%) compared with high erucic rapeseed oil (46%). Although the total VLCFA content was greater in oil from HELP lines (64%) than from high erucic rapeseed (57%), analysis of triacylglycerol composition showed negligible incorporation of VLCFAs into the sn‐2 position. Rancimat analysis showed that the thermal stability of rapeseed oil was improved greatly as a consequence of reduction of PUFA content, from 3.8 and 4.2 h in conventional low erucic and high erucic rapeseed oils, respectively, to 11.3 and 16.4 h in high oleic low PUFA (HOLP) and HELP oils, respectively. Our results demonstrate that engineering of the lipid biosynthetic pathway of rapeseed, using traditional approaches, enables the production of renewable industrial oils with novel composition and properties.


Introduction
Oilseed rape (OSR) is the world's third largest source of vegetable oil, after palm and soybean (FAO, 2018). It is a crop type of Brassica napus, which is a polyploid species (AACC, 2n = 38) formed by the spontaneous hybridization of B. rapa (AA, 2n = 20) and B. oleracea (CC, 2n = 18) (Nagaharu, 1935;Palmer et al., 1983;Parkin et al., 1995). Brassica is the crop genus most closely related to the widely used model plant species Arabidopsis thaliana (Yang et al., 1999), in which the lipid biosynthesis pathways leading to the complex suite of triacylglycerol (TAG) species found in storage oil have been extensively characterized (Li-Beisson et al., 2013). An overview of key components of the pathway that are responsible for fatty acid synthesis and modification is provided in Figure 1. Rapeseed oil is an edible oil, but due to the presence of very-long-chain fatty acids (VLCFAs, carbon chain length ≥20), particularly erucic acid, it also has a wide range of industrial applications (R€ obbelen, 1991;Zanetti et al., 2012). Feeding studies involving rats raised false concerns about adverse health effects of erucic acid (Beare et al., 1959;Charlton et al., 1975;Thomasson and Boldingh, 1955;de Wildt and Speijers, 1984), resulting in the breeding of low erucic acid rapeseed (LEAR) varieties for human consumption (Stefansson et al., 1961). High erucic acid rapeseed (HEAR) cultivars contain >45% erucic acid in their oil and are used as a 'green feedstock' for the oleochemical industry (Knutsen et al., 2016;Meakin, 2007). Erucic acid containing oils have high degree of lubricity and substantivity (ability to cling to the surfaces) than other oils (Piazza and Foglia, 2001). Erucic acid is derived into products such as erucamide, behenyl alcohol, brassylic acid, pelargonic acid, behenic acid, brassinolide and erucyl erucate (Caballero, 2006;Carlson et al., 1977;Molnar, 1974;Neischlag et al., 1967), which are used in the production of a wide range of industrial products, such as lubricants, slip additives, biofuels, pharmaceuticals, cosmetics, jet fuels, plasticizers, soaps, detergents, surfactants, textiles, recording material, rubber, nylon production and many more (Chang et al., 2013;Getachew et al., 2016;Iakovlieva et al., 2017;Johnson and Fritz, 1989;Neischlag et al., 1967;Nieschlag and Wolff, 1971;Pennick et al., 2012;Piazza and Foglia, 2001;Van Dyne and Blas e, 1991;Zanetti et al., 2012). Although other species have been evaluated for the production of erucic acid, the main source remains OSR (Hebard, 2016;Lalas et al., 2012;Sanyal et al., 2015).
Studies in A. thaliana have shown the b-ketoacyl-CoA synthase encoded at the FATTY ACID ELONGASE1 (FAE1) locus to be rate limiting for the conversion of oleic acid (C18:1) to VLCFAs, and its orthologues have the same function in B. napus (Cassagne et al., 1987(Cassagne et al., , 1994James and Dooner, 1990;James et al., 1995;Kunst et al., 1992;Lemieux et al., 1990;Millar and Kunst, 1997;Roscoe et al., 2001;Stumpf and Pollard, 1983;Suneja et al., 1991). The same enzyme is required for two elongation steps: oleic to eicosenoic (C20:1) and eicosenoic to erucic acid (C22:1) (Kondra and Stefansson, 1965). B. napus has two orthologues of FAE1: Bna.FAE1.A8 and Bna.FAE1.C3. They act in an additive manner, with both being functional in HEAR varieties and nonfunctional in LEAR varieties (Ecke and Breeding, 1995;Harvey and Downey, 1964;Howell et al., 1996;Jourdren et al., 1996;Qiu et al., 2006;Thormann et al., 1996). The proportion of VLCFAs that can be accumulated in storage oil in B. napus is limited by the poor affinity of its native lysophosphatidic acid acyltransferase (LPAAT) for acyl groups with more than 18 carbon atoms, meaning that they are not incorporated into the sn-2 position of the glycerol backbone during TAG biosynthesis (Frentzen, 1993;Katavic et al., 2001;Lassner et al., 1996). This results in a hypothetical limit of 66.7% VLCFAs. Up to 72% erucic acid has been achieved in B. napus by using a LPAAT transgene from Limnanthes douglasii to enable incorporation into the sn-2 position (Nath, 2008;Nath et al., 2007).
Studies in A. thaliana have shown the oleate desaturase encoded at the FATTY ACID DESATURASE 2 (FAD2) locus to control the conversion of oleic acid (C18:1) to linoleic acid (C18:2) PUFAs (Miquel and Browse, 1992;Okuley et al., 1994) (Hu et al., 2006;Scheffler et al., 1997;Schierholt et al., 2000;Smooker et al., 2011;Yang et al., 2012). A recent study (Wells et al., 2014) showed that, in rapeseed variety Cabriolet, only Bna.FAD2.C5 is functional. Thus, Bna.FAD2.C5 was targeted for chemical mutagenesis, resulting in population JBnaCAB_E, from which an allelic series of mutations of this gene was characterized. Several mutant lines produced seeds containing greatly reduced content of PUFAs, which was the aim of the study as polyunsaturation is anticipated to confer thermal instability (Browse et al., 1998;Durrett et al., 2008). However, this and a subsequent study (Bai et al., 2019) stopped short of confirming the predicted impact on thermal stability of rapeseed oil. An improvement in thermal stability improves the shelf life of edible oils (Kinney and Clemente, 2005) but would be transformative for industrial applications where greater thermal stability is required, such as feedstock for lubricants and hydraulic fluids, whilst minimizing environmental impact by retaining biodegradability (which mineral oil lacks).
In this study, we aimed primarily to characterize the impact on seed oil fatty acid composition of combining the low PUFA trait with the high erucic trait and thereby improving our understanding of the factors that limit erucic acid content. Having developed such rapeseed lines, we could address our secondary aim of assessing the impact of reducing PUFAs on the thermal stability of oil. Importantly, we used a strategy for development of the material that used genomics to accelerate 'traditional' mutation breeding, resulting in cultivars that are not considered genetically modified (GM) organisms.

HELP development
We combined low PUFA and high erucic acid seed oil traits by crossing the low PUFA donors K0472 and K0047 (which had been generated by chemical mutagenesis of the low erucic acid variety Cabriolet as reported by Wells et al. (2014)) onto the current high erucic acid rapeseed cultivar Maplus. In F 1 progeny (successful crosses assessed by Bna.FAD2 genotyping), two lines were randomly selected from each cross-combination, grown and self-pollinated to produce F 2 progeny. Marker-assisted selection was used in F 2 generation to identify plants inheriting appropriate alleles at the two loci in this cross conferring the low PUFA trait (the inactive allele Bna.fad2.A5 and mutated versions of Bna.FAD2.C5 denoted either Bna.fad2.C5 K0472 or Bna.-fad2.C5 K0047 , depending on donor) along with the two alleles conferring the high erucic acid trait (Bna.FAE1.A8 and Bna.-FAE1.C3). PCR primers and allele sequences are shown in Table S1, and sequence alignments of the amplicons of F 2 plants are shown in Appendix S1. We had been unable to amplify the third orthologue of FAD2 reported previously as being required in inactive form for manifestation of the low PUFA trait (Bna.-fad2.C1), and all individuals selected as inheriting low PUFA alleles Bna.fad2.A5 and Bna.fad2.C5 had inherited the low PUFA trait. From this, we deduce that variety Maplus already has the inactive (deleted) allele Bna.fad2.C1 as well as the inactive allele Bna.fad2.A1 that appears common in European rapeseed (Wells et al., 2014). Out of 192 F 2 plants, one HELP plant '2-91' from cross Maplus x K0472 (K0472-HE) and another HELP plant '4-87' from cross Maplus x K0047 (K0047-HE) were selected. These were self-pollinated up to F 5 progeny, and the seeds were collected for fatty acid analysis. The summary of the development of HELP lines is shown in Figure S1.

Seed fatty acid composition of HELP lines
Fatty acid profile was analysed from F 3 seeds of two HELP lines, 2-91 (K0472-HE) and 4-87 (K0047-HE), and high erucic acid control, Maplus as shown in Table 1. In comparison with Maplus, both HELP lines showed a slight increase in the very-long-chain fatty acids (sum of C20:1, C22:1 and C24:1) but showed a large decrease in the polyunsaturates (C18:2 and C18:3) percentages.

Figure 1 A simplified overview of the fatty acid biosynthesis pathway in
A. thaliana seeds. In the Brassicaceae, the predominant product of fatty acid biosynthesis in the plastid is oleic acid (denoted C18:1 to indicate an acyl chain comprising 18 carbon atoms and containing 1 double bond). Further elaboration of acyl chains is catalysed by microsomal enzymes that either extend the acyl chains, for example to produce the very-long-chain fatty acids (VLCFAs) C20:1 and C22:1, or increase the extent of desaturation, that is the number bonds, for example to produce the polyunsaturated fatty acids (PUFAs) C18:2 and C18:3.
As there were no biological replicates available for both HELP lines, F 3 progeny was multiplied to F 4 generation to assess the stability of fatty acid composition.
Fatty acids were analysed for F 4 seeds from 10 individual plants from each of the crosses, Maplus x K0472 and Maplus x K0047 (designated HELP lines K0472-HE and K0047-HE, respectively), along with parental lines as controls. The results are summarized in Table 2, and detailed fatty acid profiles for individual plants are provided in Appendix S2. Compared with Maplus, both HELP lines showed large increases in content of the main VLCFAs eicosenoic (C20:1), erucic (C22:1) and nervonic acid (C24:1) as shown in Figure 2. These increased from 55.7% of total fatty acids in Maplus to 63.3% in each of K0472-HE and K0047-HE HELP lines. Erucic acid increased from 46.1% in Maplus to 53.7% in K0047-HE and 54.4% in K0472-HE. Oneway ANOVA (analysis of variance) was calculated on five genotypes (having 10 replicates each)-Maplus, K0472, K0047, K0472-HE and K0047-HE. The P-values for each of VLCFAs, PUFAs, OA and SAFAs were found to be <2 9 10 À16 (highly significant), and thus, these fatty acid groups differ significantly from each other. To know the difference in the individual genotypes, post hoc Tukey's test was performed. The P-values for the genotype groups, Maplus-K0047 and Maplus-K0472 were found to be highly significant. Thus, HELP lines differ significantly from HEAR, Maplus for these fatty acid groups. The relative proportions of the two most abundant VLCFAs (C20:1 and C22:1) varied between the individual plants representing the HELP lines and showed a strong inverse correlation (R 2 = 0.795), as shown in Figure S2. It shows that in some instances, the elongation from oleic acid (C18:1) goes up to eicosenoic acid and is not elongated to erucic acid, but the overall VLCFA composition remains constant.
In addition to the anticipated increase in the content of VLCFAs, the introduction of the low PUFA trait affected the content of other fatty acids as shown in Figure 2. As expected, compared with Maplus, both HELP lines showed large decreases in content of the main PUFA linoleic (C18:2) and linolenic acid (C18:3). These decreased from 24.9% of total fatty acids in Maplus to 5.4% in K0472-HE and 7.4% in K0047-HE. The proportion of saturated fatty acids (SAFAs) also reduced, with the total of all SAFAs reducing from 6.5% in Maplus to 4.1% in K0472-HE and 4.2% in K0047-HE. Both PUFAs and SAFAs are lower in the HELP lines than they are in the low erucic acid donors of the PUFA trait (6.6% and 7.0% SAFAs in K0472 and K0047, respectively; 7.0% and 8.9% PUFAs in K0472 and K0047, respectively).
In order to assess the stability of the HELP profile in further generations, 169 F 4 plants were grown on, and selfed seed (F 5 generation) was collected and analysed for fatty acid composition. The detailed fatty acid profiles for individual plants are provided in Appendix S3. These showed that the HELP profile is stably inherited. The relative proportions of erucic and eicosenoic acid were analysed. As illustrated in Figure S3, this showed correlation between the F 5 and F 4 generations. This suggests a heritable basis for the variation, such as a minor effect locus (the same alleles are present for the FAE1 orthologues, so the difference is not attributable to allelic variation in these).

Pattern of very-long-chain fatty acid incorporation into triacylglycerol
The increased proportion of VLCFAs in the HELP lines (63.3%) approaches the hypothetical limit for TAG in Brassica species (66.7%) imposed by the inability of Brassica LPAAT to incorporate such moieties at the sn-2 position of the TAG molecule. We therefore tested whether, given the large proportion of VLCFAs that are synthesized and accumulated, incorporation had taken place at this position. Total lipids were extracted from HELP (K0472-HE, F 4 seeds) and its progenitors (Maplus and K0472) and TAGs analysed for fatty acids incorporated at the sn-2 position (van Erp et al., 2010). The results, illustrated in Table S2, show negligible incorporation at the sn-2 position, meaning that VLCFA occupation at the sn-1 and sn-3 positions can be calculated as 95% (i.e. 63.3% observed of the 66.7% available).

Thermal stability of low polyunsaturated rapeseed oils
The reduced PUFA content of HELP rapeseed oil is predicted to improve thermal stability. To test this, we produced oil from fieldgrown low PUFA lines K0472-HE and K0472. The crude oil was subjected to Rancimat stability testing (L€ aubli et al., 1988), along with field-grown conventional high and low erucic rapeseed oils (HEAR variety Maplus and LEAR variety Nikita, respectively). The fatty acid compositions of the crude oils are shown in Table S3, and the results of the stability test are shown in Table 3. As predicted, thermal stability improved greatly with the reduction in PUFA content. The stability of low erucic oil (variety Nikita) improved from 3.8 h for conventional LEAR rapeseed oil to 11.3 h for HOLP rapeseed oil. The stability of high erucic oil (variety Maplus) improved from 4.2 h for conventional HEAR to 16.4 h for HELP. The oxidative capacity of the oil types was assessed (shown in Table S4) and was not correlated with thermal stability.

Discussion
Despite the opportunities for exploiting high erucic rapeseed oil as a renewable feedstock, growing demand from the oleochemical industry and its widespread cultivation, there has been relatively little progress increasing the proportion of erucic acid in oil produced commercially (Iakovlieva et al., 2017;Johnson and Fritz, 1989;Knutsen et al., 2016;Meakin, 2007;Nieschlag and Wolff, 1971;R€ obbelen, 1991;Zanetti et al., 2012). Transgenic approaches have enabled exploration of one limitation in the accumulation of erucic acid in storage lipid, that imposed by the Brassica LPAAT enzyme, which cannot incorporate VLCFAs at the sn-2 position of TAG (Bernerth and Frentzen, 1990;Brockerhoff, 1971;Cao et al., 1990;Frentzen, 1993;Katavic et al., 2001;Nath, 2008;Nath et al., 2007;Nath et al., 2009;Sasongko and M€ ollers, 2005). The analysis we undertook of FA composition at TAG sn-2 positions (Table S2) confirmed expectations that even in the presence of additional substrate, VLCFAs are not incorporated at this position. Our studies do, however, provides insights into limitations imposed by the availability of monounsaturated 18carbon moieties for elongation to VLCFAs. Recapitulation of the combination of Bna.fad2 mutations shown previously to greatly restrict the conversion of oleic acid to PUFAs (Wells et al., 2014), but, in a high erucic acid background, increases the availability of C18:1 for extension by the Brassica elongases (controlled by the Bna.FAE1 loci). The observed impact is to increase the proportion of erucic acid in seed oil (by~17% of its original value), demonstrating that the availability of C18:1 is, indeed, an important limitation for erucic acid content. To increase erucic acid content further, the introduction of an LPAAT capable of incorporation of VLCFAs at the sn-2 position would be required. However, the pull through to VLCFAs already observed decreases the content of both saturated and polyunsaturated fatty acids (bỹ 38% and~23%, respectively, of their values in the low PUFA donor line K0472), resulting in the oil produced by K0472-HE containing~90% monounsaturated fatty acids.
The same enzyme complex extends oleic (C18:1) to eicosenoic (C20:1) and eicosenoic to erucic acid (C22:1) (Kondra and Stefansson, 1965;Lassner et al., 1996). We observed variability in the content of eicosenoic, along with a negative correlation between contents of eicosenoic and erucic acids ( Figure S2). Similar observations have been made previously in the progeny produced by crossing high erucic acid rapeseed with high oleic acid rapeseed (Sasongko and M€ ollers, 2005). It has been suggested that different alleles of B. napus orthologues of FAE1 have a different potential of producing erucic acid (Mahmood et al., 2003). This would not be unprecedented as, for example, the elongase encoded by the A. thaliana FAE1 locus uses oleic as a preferred substrate, resulting in the accumulation of eicosenoic rather than erucic acid (Katavic et al., 2001). In contrast, our results are consistent with the hypothesis that elongase activity is limiting the content of erucic acid in low PUFA lines, K0472-HE and K0047-HE. The observation of heritability to the F 5 generation, as shown by correlation of the relative proportions of erucic and eicosenoic in F 4 individuals and their progeny, indicates a genetic basis. As there is no variability for the active alleles of Bna.FAE1 in our experiments (these being inherited from Maplus), our observations can be attributed to the presence of 'modifier' loci for erucic acid content. An associative transcriptomics analysis of seed erucic acid content in a genetic diversity panel of B. napus indicated such a locus may be present on chromosome A5, with a gene encoding a fatty acid hydroxylase 1 identified as a candidate (Havlickova et al., 2018).
An increased content of erucic acid in HELP oil would undoubtedly be beneficial for industries producing compounds such as erucamide, for which rapeseed is the major source of erucic acid (Zanetti et al., 2012). HELP lines are expected to be superior to existing HEAR varieties because they contain a higher proportion of erucic acid in their oil (i.e. the yield of the primary product is greater) and due to anticipated reduction in processing costs. There is also an increasing global demand for more 'environmental-friendly' alternatives to mineral oil. The key drawback of rapeseed oil (and, indeed for vegetable oils in general) is the thermal instability that is a consequence of their high content of PUFAs. To test our prediction that low PUFA rapeseed oil (both HOLP and HELP types) would have substantially greater thermal stability than conventional rapeseed oil, we tested this using an industry standard test (Rancimat;L€ aubli et al., 1988). The results confirmed a dramatic improvement compared with conventional rapeseed oil. For HOLP, the oil is suitable for edible uses such as repeated high-temperature frying and retains the benefit of low saturate content. For HELP, the oil is a 'non-GM green feedstock' for the production of large volumes of low-cost, biodegradable products such as lubricants and hydraulic fluids, enabling more widespread use in sensitive environments. The development of non-GM varieties even for industrial crops is important as the cost of deregulation (a requirement due to the release of the organisms into the environment) is prohibitive and many countries do not permit the cultivation of GM crops at all.

Plant material and growth conditions
Donors of the low PUFA trait were selected from the ethyl methane sulphonate-mutagenized population JBnaCAB_E, derived from the winter oilseed rape variety Cabriolet (Wells et al., 2014). These lines carried mutations in Bna.FAD2.C5: K0472 and K0047. The donor of the high erucic acid trait was the current, open-pollinated high erucic acid winter rapeseed variety Maplus (breeder: NPZ-Lembke, Germany; https://www.npz.de/). Seeds were sown in the medium-grade compost (Scotts Levington F2 + S) and kept in the glasshouse under long-day conditions of 16-h photoperiod and temperatures of 20°C/ 14°C for day/night. At four-leaf stage (after~3 weeks of  PCR product purification, sequencing and analysis PCR products were purified using the 'Mag-Bindâ RXNPurePlus' by following manufacturer's protocol. Genewiz (https:// www.genewiz.com) was used for the Sanger sequencing of the purified PCR products. Mutation surveyorâ v5.0 software (SoftGenetics LLC, State College, PA) was used for the detection of the mutations from trace files.

Fatty acid composition analysis
Four milligrams of tripentadecanoin (Sigma-Aldrich, Poole, Dorset, UK) internal standard was added to 30 mg of seeds, followed by the addition of 400 µL of cold hexane: isopropanol (3:2 v/v) solution. The seeds were ground using a Tissuelyser and incubated for 1 h on ice. The samples were centrifuged at~9600 g for 5 min, and supernatant was transferred to a new tube. The pellet was washed another two times with 400 µL of cold hexane: isopropanol solution, vortexed, centrifuged and supernatants were pooled to the same tube. Then, 600 µL of 6.7% Na 2 SO 4 was added, vortexed and centrifuged for 30 s. The upper phase was removed into a new tube and dried in the Genevac TM (EZ-2.3 Elite model). These were reconstituted in 750 µL chloroform. 1/15 th part of the sample (2-3 technical replicates) was transferred to a 2-mL screw cap Supelcoâ glass vial. It was followed by the addition of 750 µL hexane and 500 µL 1N methanolic hydrochloric acid; incubation at 85°C for 3 h; and addition of 250 µL of 0.9% KCl. The vials were vortexed, and upper hexane layer containing fatty acid methyl esters (FAMEs) was used the gas chromatography (GC) analysis. FAMEs were analysed on Thermo Scientific's Trace GC Ultra-FID with GC column (specifications: BP 9 forte 10 m 9 0.1 mm ID 9 0.2 m film thicknesses) with a run time of~5 min per sample. Supelcoâ 37 component FAME mix (Sigma-Aldrich) was used as an external standard, and hexane was used as a blank. Data were analysed with the Thermo Scientific's ChromQuest TM software (version 4.2.34) platform.

Lipid extraction for TAG analysis
One millilitre of isopropanol containing butylated hydroxy toluene (BHT) was added to 4-6 seeds of each sample and heated in the oven at 85°C for 15 min. Heat-quenched samples were homogenized in a homogenizer; then, 2 mL of chloroform and 3 mL of methanol were added to each sample. It was followed by the addition of 1.6 mL of water, 2 mL of chloroform and 2 mL of 0.88% (w/v) KCl and gentle mixing. The tubes were centrifuged for 2 min at 400 g. The bottom phase was collected carefully in another tube and dried under N 2 . The lipid was resuspended in 500 µL of toluene containing 0.005% BHT for further analysis. FAMEs were derived using 10 µL of this sample, and total lipid content was calculated using GC results.

Positional distribution analysis of TAGs
Total lipid (1500 µL) was loaded on the thin-layer chromatography (TLC) plate (20 9 20 cm, silica), and the mixture was separated using hexane: diethylether: acetic acid (70:30:1, v/v). TAGs were eluted from TLC silica twice by washing with 5 mL of chloroform: methanol (4:1, v/v). A phase separation was induced by adding 2 mL of methanol and 4 mL of 0.88% (w/v) potassium chloride. The chloroform phase was collected, and the aqueous phase was back extracted with 5 mL of chloroform. The chloroform phase was dried under N 2 and re-suspended in 500 L of toluene containing 0.005% BHT. TAGs were derivatized to FAMEs by using 10 µ of this sample and analysed by GC. Samples were dried down using N 2 , and 1 mg of TAG was resuspended in: 1 mL of diethylether, 0.8 mL of buffer (50 mM NaBr, 5 mM CaCl 2 , pH 7.6) and 200 µ of lipase (Rhizomucor miehei lipase; Sigma-Aldrich). It was vortexed for 40 min, and the reaction was stopped by the addition of 2 mL of methanol was sourced from the breeders (location grown not specified). Oil pressing was carried out using a small capacity Komet screw press (Model CA 59 G; IBG Monforts, M€ onchengladbach, Germany), with a 6-mm press nozzle die and a screw speed of 20 rpm. The oxidative stability of the pressed oils was determined using a 'Metrohm Rancimat model 743', according to AOCS Official Method Cd 12b-92. Briefly, the induction times (n = 4) for portions of oil (3.0 g) was determined at 120°C with 10 L/h air throughput. The induction point, as a measure of the oil stability index (OSI), is the time point of maximal change in the rate of oxidation.

Antioxidant properties of oil samples
The total antioxidant capacity (TAC) of oils was determined using a commercial kit (Sigma-Aldrich, Catalogue number MAK187) to measure the concentration of the small molecule and protein antioxidants. This method is based on the detection of reduced Cu + ion chelates with colorimetric probe at absorbance 570 nm, which is proportional to the total antioxidant capacity in Trolox equivalents (a water-soluble vitamin E analogue used as an antioxidant standard). The absorbance at 570 nm was measured using spectrophotometric plate reader (Clariostar, BMG LAB-TECH Ltd. Oterberg, Germany). The concentrations of the samples were then calculated with the value obtained from trolox standards.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article.

Figure S1
Summary of the development of HELP lines, K0472-HE (2-91) and . Figure S2 Relationship between erucic and eicosenoic acid content of seeds of K0472-HE and K0047-HE. Figure S3 Relationship between erucic / eicosenoic acid ratios in F4 plants and their F5 progeny.