Dedicated Industrial Oilseed Crops as Metabolic Engineering Platforms for Sustainable Industrial Feedstock Production

Feedstocks for industrial applications ranging from polymers to lubricants are largely derived from petroleum, a non-renewable resource. Vegetable oils with fatty acid structures and storage forms tailored for specific industrial uses offer renewable and potentially sustainable sources of petrochemical-type functionalities. A wide array of industrial vegetable oils can be generated through biotechnology, but will likely require non-commodity oilseed platforms dedicated to specialty oil production for commercial acceptance. Here we show the feasibility of three Brassicaceae oilseeds crambe, camelina, and carinata, none of which are widely cultivated for food use, as hosts for complex metabolic engineering of wax esters for lubricant applications. Lines producing wax esters >20% of total seed oil were generated for each crop and further improved for high temperature oxidative stability by down-regulation of fatty acid polyunsaturation. Field cultivation of optimized wax ester-producing crambe demonstrated commercial utility of these engineered crops and a path for sustainable production of other industrial oils in dedicated specialty oilseeds.


Results
Production of Wax Esters in Dedicated Industrial Oilseed Crops. Although crambe, carinata, and camelina are all members of the Brassicaceae family, their seed oils have distinct fatty acid compositions that impact metabolic engineering strategies for production of industrial oils. As shown in Fig. 1, crambe seed oil is comprised of almost 70% very long-chain fatty acids (VLCFA; ≥ C20), including ~60% erucic acid (22:1). While wild-type carinata oil contains about 38% erucic acid (data not shown), in a previous study, genetic modification by silencing the carinata FAD2 and co-expression of the crambe FAE1 yielded stable transgenic lines containing up to 58% erucic acid line and almost 64% VLCFAs 25 . Camelina seed oil, by contrast, has only 20 to 25% VLCFA and < 2.5% erucic acid and is instead enriched in α -linolenic acid (18:3), which comprises ~35% of the seed fatty acids (Fig. 1).
To demonstrate the feasibility of these three oilseeds as platforms for the design and biotechnological production of industrial oils, the wax ester biosynthetic pathway from jojoba was overlaid on to the existing triacylglycerol biosynthetic pathways of seeds of the three crops. Wax ester biosynthesis requires the esterification of a fatty acid from an acyl-CoA substrate to a fatty alcohol, effectively bypassing the typical incorporation of fatty acids onto glycerol backbones to form triacylglycerols (Fig. 2). To accomplish this reaction, jojoba FAR (ScFAR) and WS (ScWS) cDNAs were co-expressed under control of strong seed-specific promoters in the target crops. The jojoba wax ester biosynthetic pathway functions primarily with C20 and C22 fatty acyl-CoA substrates 23,24 , which is ideally suited for the fatty acid profile of crambe and carinata seeds. Because camelina seeds have relatively low VLCFA content (Fig. 1), a cDNA for the jojoba fatty acid elongase 1 (ScFAE1)-like 3-ketoacyl-CoA synthase was also introduced to increase VLCFA production as substrates for FAR and WS. For comparison, studies were also conducted in crambe and carinata with and without the inclusion of the jojoba FAE1-like enzyme. In addition, the host germplasm for carinata studies was a line optimized for production of wax esters with monounsaturated acyl chains, generated by suppression of the FAD2 gene that controls the desaturation of oleic acid (18:1) to form linoleic acid (18:2) and overexpression of a crambe FAE1 25 . Transgene cassettes for ScWS, ScFAR, and ScFAE1 were assembled in a single binary vector for transformation of the crops to increase the likelihood of simple, linked insertion patterns of the transgenes to simplify future breeding efforts and biotechnological introduction of additional transgenes.
Analysis of lipid extracts by thin layer chromatography from seeds of ten to fifteen independent transgenic events for each crop indicated the successful production of wax esters (Fig. 3a-c). Total wax ester content of seeds measured by gas chromatography following hydrolysis of purified wax esters and triacylglycerol fractions revealed wax ester amounts of 15 to 30% of the total oil content (wax esters + triacylglycerols; Fig. 4a,b). Wax ester amounts of > 50% of the total oil were detected in single seeds of the top performing crambe lines (Supplementary Table S1). Wax esters in seeds of the three crops had distinct fatty acid and fatty alcohol compositions. In both crambe and carinata seeds, wax esters were composed predominantly of 20:1 and 22:1 fatty acids and 22:1 fatty alcohols (Fig. 3d). This was reflected in the molecular species of wax esters produced in these seeds, which were enriched in C42 and C44 species consisting of 22:1 fatty alcohols esterified to 20:1 and 22:1 fatty acids. The wax esters of camelina seeds contained a more complex mixture of fatty acids and fatty alcohols, including high levels of 20:1, 22:0, 22:1, 24:0, and 24:1 fatty acids and 20:1, 22:0, 22:1, 24:0, 24:1, 24:2, and 24:3 alcohols (Fig. 3d). Of the fatty alcohols, C24 alcohols comprised nearly 50% of the total fatty alcohols of camelina wax esters. These combinations of fatty acids and fatty alcohols resulted in wax esters of primarily C46 and C48 chain-lengths. Similarly, the inclusion of the ScFAE1 along with jojoba ScFAR and ScWS in crambe and carinata Figure 1. Comparison of the fatty acid compositions of seed oils from the genetic backgrounds of crambe (Crambe abyssinica), carinata (Brassica carinata), and camelina (Camelina sativa) that were used for metabolic engineering for wax ester production in this study . Data shown are for wild-type crambe and camelina seeds, while carinata data are for seeds engineered for FAD2-RNAi suppression and crambe FAE1 overexpression to increase erucic acid from 38% of total fatty acids in wild type seeds to 58% of total fatty acids in seeds of lines used for metabolic engineering 25 . Figure 2. Metabolic engineering strategies for production of wax esters in the three target crops. Wax esters were produced by seed-specific expression of jojoba cDNAs for fatty acyl-CoA reductase (ScFAR) that converts fatty acyl-CoAs into fatty alcohols and wax synthase (ScWS), an acyltransferase-like enzyme that esterifies fatty acyl-CoAs to fatty alcohols to form wax esters. Oleic acid (18:1) synthesized from fatty acid synthase (FAS) localized in the plastid is the primary substrate for wax ester production. Pools of very longchain fatty acids (VLCFAs; ≥ C20) were increased by seed-specific expression of cDNAs for the enzymes 3-ketoacyl-CoA synthase-related fatty acid elongases from jojoba (ScFAE1) or Lunaria annua (LaFAE1). Levels of monounsaturated C18 substrate (18:1-CoA) for wax esters were increased by seed-specific RNA interference (RNAi) suppression of the FAD2 gene (FAD2-RNAi) for the Δ12-oleic acid desaturase. [Lane 1 = standard WE from jojoba seed oil. Lanes 2, 5 and 8 show wildtype of the three respective crops and lanes 3, 4, 6, 7, 9 and 10 transformed crambe, carinata and camelina, respectively. 2-4 = Crambe abyssinica. 5-7 = Brassica carinata. 8-10 = Camelina sativa.] (b) Camelina oil from wild-type seeds (left, 1) and WE-producing seeds (right, 2) at 4 °C. The solid material in (2) is enriched in WE and demonstrates the higher melting point of WE versus TAG. (c) Crambe WE extracted from oil of engineered seeds. (d) Twenty most abundant WE molecular species in engineered seeds of the three oil crops. n = 3 to 5 single seed replicates (crambe, carinata) or replicates of 5 mg of seed (camelina) ± SD. resulted in a shift to include higher content of C24 fatty acids and fatty alcohols (Fig. 3d), namely an increase in wax content of C46 and C48.

Tailoring Wax Ester Structures By
Step-Wise Metabolic Engineering of Fatty Acid Composition. Experiments described above in carinata and crambe indicated an influence of ScFAE1 expression on wax ester composition. To explore this in more detail, step-wise metabolic engineering experiments were conducted in camelina, which is readily amenable to rapid gene testing due to its ability to be transformed with a simple Agrobacterium-based method using multiple selection markers and its short life cycle 16 . Initially, the ScFAE1 cDNA was replaced in the construct with the LaFAE1 gene from Lunaria annua. In contrast to the broad acyl-CoA substrate specificity of ScFAE1 26 , the LaFAE1 displays higher activity in vivo for monounsaturated acyl-CoA substrates and is also effective in fatty acid elongation to the C24 chain-length 27 . Consistent with this, camelina seeds engineered for co-expression of the LaFAE1 with the ScFAR and ScWS displayed a marked reduction in wax esters containing saturated and polyunsaturated fatty alcohols, relative to seeds expressing the ScFAE1 (Fig. 3d). Most striking was a 2-fold increase in relative content of C22 and C24 monounsaturated fatty acids and alcohols in wax esters compared to those generated with the ScFAE1 in the three-gene construct. This accumulation occurred in part at the expense of C20 monounsaturated fatty acids, which decreased by ~50% in seeds expressing the LaFAE1. Based on these findings, additional experiments were conducted to obtain further increases in wax esters containing monounsaturated fatty acids and alcohols for achieving wax esters with better oxidative stability at high temperatures. For these experiments, two of the camelina lines expressing LaFAE1 and ScFAR and ScWS with the highest wax ester content were re-transformed with a seed-specific FAD2-RNAi transgene to reduce polyunsaturated fatty acid content and channel oleic acid (18:1) to wax ester fatty acids and fatty alcohols. This resulted in substantial increases in wax esters with monounsaturated 20:1 and 22:1 fatty acids and 22:1 fatty alcohols, as well as additional decreases in saturated and polyunsaturated wax ester components (Fig. 5a). These differences were also observed in the wax ester molecular species. The major wax esters in seeds expressing the LaFAE1 were C48 species consisting of a 24:1 fatty alcohol paired with either a 24:0 or 24:1 fatty acid (Fig. 5b). With the additional suppression of FAD2, the major wax ester was instead a C42 species comprised of a 22:1 alcohol paired with a 20:1 fatty acid (Fig. 5b). The shift in relative content from 24:1 to 20:1 and 22:1 in wax ester fatty acids and fatty alcohols with FAD2 suppression likely results from limiting fatty acid elongation flux for conversion of the increased pools of 18:1-CoA to 24:1-CoA. This strategy was also applied in crambe by crossing the line expressing ScFAR and ScWS with the line expressing the CaFAD2-RNAi. The hybrid progenies showed an increased level of oleic acid (Fig. 5c), indicating the channeling of 18:1 to 18:2 was effectively reduced. Wax ester was detected in seed oils in the overwhelmingly majority of F 3 seeds with proportions up to over 40 weight % of total oil (Supplementary Table S2). Similar to the results achieved in camelina reported above, a considerable reduction in seed oil polyunsaturated fatty acids and alcohols was observed in the seed oil, but 20:1 fatty acid, 20:1 alcohol and 22:1 alcohol increased (Fig. 5d). Overall, these results demonstrate that it is possible to alter fatty acid unsaturation for tailoring wax ester profiles by regulating the target gene expression in camelina and crambe.

Agronomic Properties of Dedicated Industrial Oilseeds Engineered for Wax Ester
Production. Transgenic wax ester crambe lines (T 5 generation) were used for a confined field trial in Kristianstad, Sweden. Plants were grown under insect nets in the field to mitigate bee pollination as required by EC regulations. Transgenic lines with ScWS and ScFAR genes showed a slightly reduced seed yield, oil content and germination rate compared to the wild type (Table 1), while the transgenic line with the ScWS, ScFAR and ScFAE1 genes had a dramatically reduced seed yield and oil content (Table 1a). In contrast, the thousand seed weight of both transgenic lines was higher than that of wild type (Table 1a). Except for delayed seedling establishment relative to the wild type, both transgenic lines had normal vegetative growth in the field. Flowering and fruit set were clearly delayed and reduced for the transgenic line containing three jojoba genes. Evaluation of agronomic traits of transgenic crambe (T 6 generation) in greenhouses showed a similar tendency to that of the field trial, but the reduction in seed yield and oil content of the transgenic lines were much lower than that of the field trial (Table 1b). Normal seedling growth, seed-set and seed development were observed for the transgenic line with ScFAR and ScWS genes. Similar to the results from the field, the transgenic line with the ScWS, ScFAR and ScFAE1 genes had weak seedling growth, poor seed-set, and delayed seed maturation relative to the wild type under the greenhouse conditions. Camelina lines expressing the ScWS, ScFAR, and LaFAE1 + /− FAD2 RNAi suppression also displayed reduced germination under greenhouse conditions, with germination rates of seeds from the transgenic lines ranging from 25 to 80% versus 94 to 100% in seeds from the wild type ( Supplementary  Fig. 1). Notably, germination rates were higher in wax ester-producing seeds with FAD2 RNAi suppression versus wax ester-producing seeds in a wild type background.

Discussion
Here we have demonstrated the feasibility of crambe, camelina, and carinata as biotechnological platforms for the seed-specific production of wax esters. Seed oils of crambe and carinata have the highest endogenous levels of very long-chain fatty acids of the three oilseeds evaluated, and yielded wax ester profiles that most closely mimic the C42-rich wax ester composition of jojoba seeds by seed-specific expression of only the jojoba FAR and WS genes. We have also shown that an array of different wax ester compositions can be generated by co-expression of FAE1 genes from different sources and by channeling monounsaturated fatty acids to wax ester production by suppression of the FAD2 ∆ 12 oleic acid desaturase gene (Fig. 2). Using these approaches, wax ester levels of > 30% of the total oil content were achieved in seeds of the engineered crops.
Germination, seed yield per plant, and oil content of crambe seeds producing 22% to 25% wax esters from expression of jojoba FAR and WS showed no obvious effect in greenhouse and field studies (Table 1). However, seed germination was significantly compromised in crambe and camelina lines engineered for co-expression of an additional FAE1 gene (Table 1, Supplementary Fig. S1). The common feature of these seeds was the accumulation of C48 wax esters species composed of C24 fatty acids and alcohols, which were nearly undetectable in crambe lines lacking the ScFAE1 transgene. The striking difference in seed germination between lines lacking and containing the ScFAE1 transgene may be due to physical properties of C48 wax esters, e.g. melting points that Line 1000 seed weight (g) Seed yield Yield in relative to wild type (%)* Wax ester content (mg/g seed)**  Table 1. Agronomic traits of wax ester transgenic lines from field trial (T 5 generation) and greenhouse (T 6 generation) evaluations of crambe. Note: 2 gene lines: ScFAR/ScWS; 3 gene line: ScFAR/ScWS/ScFAE1. *Results are normalized to 100% for wild type. **Three-seed pooled sample and three biological replicates. ***20-seed pooled sample, three biological replicates for the field trial material and 4 biological replicates for the greenhouse material. 1 Different letters followed by the figures in the columns of 1000 seed weight, wax ester content and oil content both for field and greenhouse data and seed yield for greenhouse data (n = 4 plants) indicate significant differences at p = 0.05, analyzed using the Kyplot program with the Tukey test. n.d., not detected.
compromise seed performance. Alternatively, C48 wax esters or their C24 acyl components, such as fatty alcohols, may negatively affect metabolism (e.g. β -oxidation) during seed germination or maturation. Based on these studies, it appears that engineered seeds may be more tolerant towards the presence of shorter chain wax esters. This is consistent with data in Supplementary Fig. 1 showing that FAD2 RNAi suppression in wax ester-producing camelina lines nearly doubled seed germination rates. Seeds from these lines have a relatively higher content of wax esters with 20:1 and 22:1 fatty acids and alcohols and lesser relative amounts of C24 fatty acids and alcohols compared to wax ester-producing seeds lacking FAD2 RNAi suppression (Fig. 5A). In the case of camelina, a more effective metabolic engineering strategy may be to express a FAE1 from a species such as Arabidopsis, Teesdalia, Brassica napus or Crambe abyssinica that produce high levels of 20:1 or 22:1 in their seeds along with FAD2 suppression to generate shorter chain wax esters compared to that obtained from the LaFAE1 or ScFAE1 expression. It is notable that jojoba seed oil is deficient in TAG, but instead consists almost entirely of wax esters 19,21 . By comparison, crambe lines selected for field testing have ~25% of the oil as wax esters, with the remainder in the form of TAG. It remains to be determined if this level of wax ester accumulation results in oil quality with industrial value. Since sperm whale wax esters were used as additives in petroleum based lubricants, it can be anticipated that the blend of TAG and wax esters produced in crambe will have superior lubrication properties compared to a pure TAG based lubricant. Alternatively, wax esters and TAG can be partitioned through methods such as winterization or crystallization based on the higher melting point of wax esters (as shown for winterized camelina oil + /− wax esters in Fig. 3b). In this way, value can be obtained from both the wax ester and TAG components of the engineered oil. Possible future metabolic engineering strategies to enrich for wax ester production include blocking triacylglycerol production via RNAi suppression of DGAT1 and PDAT1, based on recent findings with Arabidopsis 29 .
Overall, the findings reported here demonstrate the feasibility of transferring multigenic biosynthetic pathways to dedicated non-food oilseed crops for biotechnological production of an industrial oil trait. In the case of wax ester production, crambe was particularly well-suited because of the high content of VLCFAs, mainly erucic acid, in its seeds. New transgenic crambe lines with increased erucic acid have since been developed 28 , which are likely to be a more suitable background for wax ester production. However, as shown with camelina and crambe, through the use of biotechnological techniques, the fatty acid compositions of seeds of these crops can be rationally tailored to provide pools of fatty acid substrates that are optimized for production of target industrial oil traits. For instance, as we have shown above, down-regulation of the FAD2 gene in combination with overexpression of jojoba FAR and WS resulted in the production of wax esters in crambe seed oil consisting mainly of 20:1 and 22:1 monounsaturated fatty acids and fatty alcohols, which confers a higher oxidative stability amenable to formulations of motor oil and transmission fluids. Production of wax esters containing higher amounts of monounsaturated molecular species also allows for easier separation of wax and TAG components of seed oil, which reduces costs associated with downstream purification. Engineering of seeds for production of high value industrial proteins and/or small molecule co-products that can be readily partitioned from oils could offer additional revenue streams for establishing crambe, carinata, and camelina as economically viable dedicated industrial oilseed crops.

Plant materials. Crambe abyssinica cv. Galactica (crambe), Camelica sativa cv. Suneson (camelina), and
Brassica carinata (carinata) were used in this study. For carinata, the XS18A 6-2 line engineered for high erucic acid and low polyunsaturated fatty acid seed oil content was used for transformation 25

RNA isolation from jojoba seeds and first strand cDNA synthesis. Jojoba (Simmondsia chinensis)
developing and mature seeds were collected in an established jojoba plantation grown by New South Wales Department of Agriculture at Condobolin, New South Wales, Australia. The developing seeds were sorted visually into three classes, early maturity (small and juicy seeds), middle maturity (solid tissue) and late maturity (close to ripe seed). The embryos were dissected out from the seeds and stored in RNA later (Qiagen) solution for RNA extraction. Total RNA was extracted with TRIzol solution (Invitrogen) from the middle and late stage groups of developing embryos and precipitated with 0.5 vol isopropanol and 0.5 vol 3 M sodium acetate pH 5.2. The precipitate was washed with 75% ethanol. Air dried RNA was resuspended in 20 μl water. First strand cDNA was produced with the RevertAid H Minus Reverse Transcriptase (Fermentas/Thermo) according to manufacturer's protocol.
Preparation of transformation vectors. The following vectors were used in this study:1) pFWS2-Kan transformation with kanamycin as selection; vectors 2, 4-5 were used for camelina transformation with DsRed as a marker for screening; and vectors 6 and 7 were used for carinata transformation with hygromycin as selection.
Preparation of pFWS2-Kan, pFWS3, and pFWS3-Kan. First strand cDNA obtained above was used as template for PCR amplification of ScFAR, ScWS, and ScFAE1 cDNAs from jojoba. PCR reactions were conducted with Phusion polymerase (New England Biolabs) using the 5′ and 3′ oligonucleotide pairs: (1) FAR 5′-TATATAGCGGCCGCAAAATGGAGGAAATGGGAAGC-3′ and 5′-TATATAGCGGCCGCTTTAGTTAAGAACGTGCTCTACGACACC-3′ , (2) WS 5′-ATAGAATTCA AAATGGAGGTGGAGAAGGAGCTAAAGACC-3′ and 5′-AATAAGTCGACTCACCACCCCAA C A A AC C C A AT T T C -3 ′ , a n d ( 3 ) FA E 1 5 ′ -TATATAG C G G C C G C A A A AT G A AG G C C A A AACAATCACAAAC-3′ and 5′-TATATAGCGGCCGCTTCTACGAAGCGATAGGTGCGATTTTAG-3′ (added restriction enzyme sites are underlined). The ScWS cDNA was ligated under control of the seed-specific soybean glycinin-1 promoter in the EcoRI/XhoI sites of the binary vector pBinGlyRed2 12 as an EcoRI/SalI-digested fragment to generate pBinGlyRed2-ScWS. The ScFAR and ScFAE1 PCR products were ligated as NotI restriction enzyme fragments downstream of the seed-specific soybean glycinin-1 promoter and upstream of the glycinin-1 3′ UTR into the corresponding site of the vector pKMS3 12 , resulting in FAR or FAE1 expression cassette. The FAR cassette was cloned as an AscI fragment into the MluI site of pBinGlyRed2-ScWS to generate pFWS2. An nptII kanamycin resistance gene under control of the nos promoter was amplified from pART27 30 and ligated as an AscI fragment into the corresponding site of pFWS2 to generate pFWS2-Kan. Alternatively, the ScFAE1 cassette was digested from pKMS3 as an AscI fragment and ligated into the corresponding site of pBinGlyRed2-ScWS to generate pFWS3. The nptII marker above was subsequently cloned as a KpnI fragment into the corresponding site of the DsRed marker of pFWS3 to generate pFWS3-Kan. The resulting pFWS2-Kan and pFWS3 vectors contain the DsRed gene under control of the cassava mosaic virus promoter for selection of transgenic seeds, and the pFWS2-Kan and pFWS3-Kan vectors contain nptII for kanamycin selection of transgenic plants.
Preparation of pFWS3-LaFAE1. The Lunaria annua FAE1 27 was amplified from plasmid pBluescript II SK+ /LaFAE1 using the oligonucleotides (added restriction sites are underlined): 5′-TATATT GCGGCCGCAAAATGACGTCTGTGAACGTAAAACTCC-3′ and 5′-TATATTGCGGCCGCTTAGGACC GACCGTTTTGGACAG-3′ . The product was digested with NotI and cloned downstream of the glycinin-1 promoter and upstream of the glycinin-1 3′ UTR. The resulting vector contained AscI restriction sites that flanked the glycinin-1 promoter and 3′ UTR. Using this restriction site the entire cassette containing the promoter, gene and 3′ UTR was used to replace the ScFAE1 cassette in pFWS3 to generate pFWS3-LaFAE1.
Preparation of pFWS2-Hyg and pFWS3-Hyg. The DsRed selectable marker was replaced with a hygromycin resistance cassette by first digesting plasmid pMDC32 31 with AseI, which released the hygromycin cassette, then fragments were incubated with T4 DNA polymerase and dNTPs. The blunt-ended hygomycin cassette was isolated by DNA gel electrophoresis and purified using the Geneclean kit (MP Biomedicals, Solon, OH). The three-gene wax vector pFWS3 was digested with AflII to release the DsRed marker, then fragments were blunted with T4 DNA polymerase and the vector fragment isolated and purified as described above. The hygromycin resistance cassette was ligated into the purified vector to produce pFWS3-Hyg, and orientation and copy number of the hygromycin cassette was confirmed using a combination of restriction mapping and DNA sequencing. The ScFAE1 gene was then removed by digesting plasmid pFWS3-Hyg with AscI, gel purifying the vector fragment, and self-ligation. The final plasmid (pFWS2-Hyg) was subjected to DNA sequencing to verify removal of the ScFAE1 gene.
Agrobacterium strains and plant transformation. The super-virulent Agrobacterium strain AGL1 32 was used for crambe transformation, while strain GV3101(pMP90) was used for camelina and carinata transformations. Crambe transformation was performed using either hypocotyls explants according to the method described 15 or the cotyledonary node method as developed for sugar beet transformation 33 with minor modifications, essentially as described 34 . When using cotyledonary nodes as explants for crambe transformation, seeds were surface-sterilized with 70% (v/v) ethanol for 30 sec, followed by 3% (w/v) sodium hypochlorite with shaking at 42.5 °C for 20 min, and rinsed thoroughly with sterile water. The sterilized seeds were then placed onto the germination medium consisting of MS salts 35 and vitamins, 2% (w/v) sucrose and 0.8% (w/v) Phytoblend (Caisson Labs) at pH 5.8 in plastic containers (50 seeds/container) and incubated at 7 °C overnight in the dark, followed by a dark period at 24 o C for 3 days, and a further culture at 24 °C in 16 h photoperiod with a light intensity of 33 μmol·m 2 ·s −1 for another 3 days prior to transformation. Camelina transformation was performed according to the floral Agrobacterium vacuum infiltration method described 12,16 . Carinata transformation was carried out using cotyledonary petiole explants, essentially as described 12 with selection using 5 mg/L hygromycin.

Screening of stable transgenic lines for seed wax ester content and composition. Total lipids
were extracted from wild type and engineered seeds from the three oil crops using variations on the method described by Bligh and Dyer 36 (crambe, camelina) or using hexane (carinata) followed by solid phase extraction or thin layer chromatographic (TLC) separation of wax ester (WE) and triacylglycerol (TAG) fractions. The total amount of each fraction and the fatty acid and fatty alcohol compositions of each were determined by gas chromatography after transesterification with added internal standards for quantification. Detailed analytical methods for seeds for each crop are provided below.
Scientific RepoRts | 6:22181 | DOI: 10.1038/srep22181 Crambe: For screening stable transgenic lines, the half-seed technique 28 was used in all generations. Seeds were removed from siliques and surface-sterilized using 15% calcium hypochlorite for 15-20 min and rinsed thoroughly with sterile water. The surface-sterilized seeds were then placed on the germination medium in a climate-controlled chamber as described 15 . After culture for ~15 h, seed coats were removed by dissection under microscope, and the larger outer cotyledon was excised for wax ester analysis, while the rest of the seeds were continuously maintained on the culture medium until planted in pots. Seedlings with the highest WE content were planted in the greenhouse or biotron depending on season. Alternatively, seeds were soaked in water for 2 h. Partial cotyledons were then cut off with a fine knife under the microscope. After cutting, the chipped fragment of the cotyledon was immediately used for oil composition analysis by GC as described below. The remaining part of the seed was sown directly into a pot with well-irrigated soil in the greenhouse where a cover of gas-permeable mesh on the top of each pot was used to protect the emerging seedlings from insect infestation.
For total lipid extraction, excised cotyledons were homogenized in a glass tube with 1.9 ml of methanol:chloroform (2:1) and 0.6 ml of 0.15 M acetic acid. To the homogenate, 0.6 ml of chloroform and 0.6 ml of water were added, mixed and centrifuged for 3 min at 3,000 rpm 28 . The lower phase was transferred into a screw cap tube and dried under N 2 . The residue was resuspended in 60 μl of chloroform for TLC and GC analyses. Approximately 40 μl of lipid extract was analyzed on TLC silica gel plates (15 cm × 20 cm or 20 cm × 20 cm, Merck) in a solvent system of hexane:ethyl ether:acetic acid (90:10:1 v/v/v) to separate WE and TAG. The areas corresponding to the reference sample (standard jojoba wax ester, visualized by exposure to iodine) containing WE and TAG were scraped and collected in a screw-cap tube with addition of 200 μl methanol to remove the water from the gel under nitrogen stream. Two ml methylation solution (2% sulphuric acid in dry methanol) was immediately added and the samples were transesterified at 95 °C for 45 min. After methylation, 0.5 ml hexane, 2 ml water and internal standard 17:0 alcohol were added. After brief vortexing, the samples were centrifuged at 3000 rpm for 3 min. The hexane phase containing fatty alcohol and FAME of the wax part or FAME of the TAG part was transferred into a GC vial for analysis using a Shimadzu GC-17A GC fitted with a CP-Wax 58 column as described 28 . Peaks were identified according to their relative retention time in comparison with external standards run on the same instrument. The WE content was calculated based on the peak areas. Total lipids in crambe seeds were determined as described 37 .
Camelina: DsRed fluorescent seeds from T 1 plants of ≥ 10 independent transgenic lines for each construct were screened by TLC to identify wax producing lines. Lines containing waxes were further analyzed by GC following direct transesterification of homogenized seeds to quantify fatty alcohol content (as described below). Three independent lines from transformants containing the highest seed WE content were selected for advancing to homozygosity and for more complete characterization of WE content and composition and agronomic properties. Two homozygous lines from the pFWS3-LaFAE1 transformants with the highest seed WE content were retransformed with the pBinGlyBar1-cFAD2 HP construct, and transgenic lines were selected by glufosinate resistance as described 12 . Seeds from T 1 plants from > 15 independent transformants were screened by GC following methylation of the seed oil to measure fatty acid and fatty alcohol content of seeds as described below. For initial screening of transformants, 3-4 seeds were homogenized with a glass rod in 200 μl of heptane:ethyl ether (50:50 v/v). Approximately 25 μl of the extract was analyzed on TLC silica plates in a solvent system of heptane:ethyl ether:acetic acid (60:40:1 v/v/v) with TAG and WE standards. Lipids were visualized by staining in iodine vapor. In addition, ~5 homogenized seeds were methylated in 2.5% (v/v) sulphuric acid in methanol as described 12 and analyzed using an Agilent 7890 GC fitted with an Agilent HP-INNOWax column (30 m length × 0.25 mm inner diameter, 0.25 μm film thickness), H 2 carrier gas, and flame ionization detection. The oven temperature was programmed from 185 °C (1 min hold) to 235 °C at 7 °C/min (5 min hold) and to 240 °C at 10 °C/min (5 min hold). Fatty acid methyl ester and fatty alcohol were identified by mobility relative to authentic standards and their structures confirmed by GC-MS using an Agilent 7890 gas chromatograph interfaced with an Agilent 5973 mass selective detector. For quantification of WE and TAG in seeds of homozygous lines, seeds (~25 mg) were weighed in glass screw cap tubes (13 × 100 mm), and 175 μl of triheptadecanoin (10 mg/ml in toluene) and 70 μl of myristitic acid-myristoyl alcohol WE (5 mg/ml in toluene), both purchased from NuChek Prep, were added as internal standards. Total lipids were then extracted from seeds in 3 ml of chloroform:methanol (1:2 v/v) and grinding with an Omni THQ tissue homogenizer. After 1 h incubation with shaking at 25 °C, 1 ml of chloroform and 1.8 ml of water were added to the extract. Following thorough shaking and centrifugation, the organic layer was recovered. The organic layer was dried under N 2 and re-suspended in 1 ml of heptane. This was applied to an activated silica solid phase extraction column (Supelco Supelclean LC-Si, 3 ml column; Sigma-Aldrich, Saint Louis, MO USA) equilibrated with heptane. One ml of heptane was subsequently applied to the column. The WE fraction was then eluted with 2.5 ml of heptane:ethyl ether (95:5 v/v), and followed by 2.5 ml of heptane:ethyl ether (80:20 v/v) to eluted TAGs. An aliquot of each fraction was analyzed by TLC (as described above) to confirm separation of these fractions. The WE and TAG eluents were then transesterified and analyzed by GC as described above and quantified relative to the added internal standards.
Carinata: The initial screening of transgenic T 1 lines for presence of WE was carried out using single seed TLC by lightly crushing individual seeds placed in the wells of a 96 well microtiter plate. Two drops of hexane were added to each well and extracted lipids were transferred to a silica gel TLC plate using a multichannel pipette set to a 3 μl volume. Plates were developed in hexane:ethyl ether:acetic acid (140:30:3 v/v/v), and lipids were visualized by exposure to iodine vapors. Seed lipids were extracted by grinding 10 pooled seeds in hexane with a 10 min incubation at 42 °C. Sediment was removed by centrifugation and the clear hexane phase was evaporated to dryness under N 2 . Lipids were dissolved in hexane and stored at − 20 o C until used for analysis. WE and TAG were first separated by TLC on the silica gel plates using the same solvent system as described above for single seed analysis. Areas containing lipids, visualized using iodine vapors, were scraped from the plates and directly methylated along with internal standards to generate fatty acid methyl esters and fatty alcohols using 1 M hydrochloric acid in methanol, with heating at 80 °C for 2 h. Extracted fatty acid methyl esters and fatty alcohols Scientific RepoRts | 6:22181 | DOI: 10.1038/srep22181 were analyzed simultaneously by GC on a DB23 column (30 m length × 0.25 mm inner diameter; Agilent) as described 25 . Identity of the fatty alcohols was verified by GC-MS of their trimethylsilyl-ether derivatives.
Wax ester profiling by nanoESI-MS/MS. Quantitative wax ester profiling of oil from total lipid extracts of engineered crambe, camelina, and carinata seeds was performed following a previously published direct infusion nanoESI-MS/MS method 38 with minor modifications. WE molecular species (484 total) carrying acyl chain combinations from 16:0 to 26:1 were monitored. Detector signals below 50 counts per second were defined as background noise. The WE profiles represent amounts of the 20 (Fig. 5d) or 10 (Fig. 3b) WE most abundant molecular species [mol %] relative to the sum of these molecular species.
Crossing of crambe wax ester lines with FAD2 down-regulated lines. Two previously described transgenic crambe lines with CaFAD2 knocked down by RNAi (CaFAD2-RNAi) driven by the napin promoter 28 were crossed with two transgenic lines transformed co-expressing the ScFAR and ScWS genes. For the crossing, only unopened floral buds were selected, while all open flowers and non-selected buds were removed from the inflorescences on the mother plants. The selected flower buds were manually opened and anthers were removed. Styles and stigmas were exposed and used in pollinations. After 24 h, from the father plants, open flowers having matured anthers with visible pollen shedding were chosen. Anthers were tapped on the stigmas of the mother plants to cover them with pollen grains. Subsequently, the artificially pollinated flowers on mother plants were marked. Each of the plants was used both as father and mother in reciprocal crosses. The F 1 seeds were sown directly in the soil for bringing to F 2 for oil analysis. For the F 2 seeds, the wax ester analysis was carried out using the half-seed technique as described above.
Crambe and camelina seed germination test. For crambe, 150 seeds without siliques were put onto the wet filter paper in Petri dishes which were placed in the dark in the climate-controlled chamber at a temperature of 25 °C. The germination results were recorded at 3, 5, and 7 days after sowing and the test was repeated three times. For camelina, ninety seeds were sown directly in the soil with three replicates. The germination results were recorded after one week.
Crambe field trial. A field trial on crambe was conducted in southern Sweden. The planting area was 75.6 m 2 for the transgenic crambe with the FAR and WS genes, 25.2 m 2 for the transgenic crambe with the FAR, FAE and WS genes, and 65.7 m 2 for the wild type. The crambe plants were grown under insect net to prevent bees from spreading GM-pollens as required by the EU court. The wax esters were extracted from the seeds and analyzed as described above.