Genetic Engineering of Lesquerella with Increased Ricinoleic Acid Content in Seed Oil

Seeds of castor (Ricinus communis) are enriched in oil with high levels of the industrially valuable fatty acid ricinoleic acid (18:1OH), but production of this plant is limited because of the cooccurrence of the ricin toxin in its seeds. Lesquerella (Physaria fendleri) is being developed as an alternative industrial oilseed because its seeds accumulate lesquerolic acid (20:1OH), an elongated form of 18:1OH in seed oil which lacks toxins. Synthesis of 20:1OH is through elongation of 18:1OH by a lesquerella elongase, PfKCS18. Oleic acid (18:1) is the substrate for 18:1OH synthesis, but it is also used by fatty acid desaturase 2 (FAD2) and FAD3 to sequentially produce linoleic and linolenic acids. To develop lesquerella that produces 18:1OH-rich seed oils such as castor, RNA interference sequences targeting KCS18, FAD2 and FAD3 were introduced to lesquerella to suppress the elongation and desaturation steps. Seeds from transgenic lines had increased 18:1OH to 1.1–26.6% compared with that of 0.4–0.6% in wild-type (WT) seeds. Multiple lines had reduced 18:1OH levels in the T2 generation, including a top line with 18:1OH reduced from 26.7% to 19%. Transgenic lines also accumulated more 18:1 than that of WT, indicating that 18:1 is not efficiently used for 18:1OH synthesis and accumulation. Factors limiting 18:1OH accumulation and new targets for further increasing 18:1OH production are discussed. Our results provide insights into complex mechanisms of oil biosynthesis in lesquerella and show the biotechnological potential to tailor lesquerella seeds to produce castor-like industrial oil functionality.


Introduction
The conventional source of hydroxy fatty acid (HFA) is castor (Ricinus communis), which contains 90% ricinoleic acid (18:1OH) in its seed oil. The fatty acid and its derivatives are used as feedstocks for numerous industrial products, such as lubricants, plastics and surfactants [1,2]. The production of castor oil is hampered by the presence of the toxin ricin [3,4] and hyper-allergenic 2S albumins [5][6][7] in its seeds. Lesquerella (Physaria fendleri, Brassicaceae) seed oil contains a major HFA, lesquerolic acid (20:1OH) at 55-60% [8][9][10][11], and seeds of this plant lack any known toxins. As such, efforts have been made through plant breeding to develop lesquerella as a new oilseed crop that is a safe source of HFA [12,13]. With the success of lesquerella biotechnology [14,15] and the deep knowledge of genes for fatty acid and seed oil biosynthesis [16][17][18], lesquerella oil can be improved through metabolic engineering [15].
One of our research goals is to generate a castor-oil producing lesquerella that is safe, cost-competitive, and widely accepted as an industrial feedstock. We have previously attempted to generate a castor oil-producing lesquerella through over-expressing of a castor RcLPAT2 involved in TAG assembly. In that study we demonstrated that seed oils of transgenic lesquerella showed increases in 18:1OH from 1% to 4%, and castor oil-like TAGs from 5% to 14% [15]. In this study, we aimed to further enhance the 18:1OH level by downregulating the expression of lesquerella PfFAD2, PfFAD3, and PfKCS18 genes using RNA interference (RNAi) technology [54,55]. A RNAi silencing approach was used to suppress an endogenous target gene expression through transgenic expression of a double-stranded RNA (dsRNA) that shares sequence homology with the target and leads to cleavage of the targeted transcripts, [54,55]. We hypothesize that suppressing FAD2 and FAD3 reduces polyunsaturated FAs (PUFAs) levels, including 18:2 and 18:3, and subsequently increases 18:1, the substrate of PfFAH12 (Figure 1). Suppressing KCS18 reduces 20:1OH and subsequently increases 18:1OH for TAG assembly. PfFAD2, PfFAD3, and PfKCS18 share high sequence homology with camelina CsFAD2, CsFAD3, and arabidopsis AtKCS18, showing 93.6%, 95.6%, and 82.3%, respectively [18]. RNAi constructs, CsFAD2 RNAi, CsFAD3 RNAi and AtFEA1 RNAi, are effective in silencing corresponding gene expression in camelina [56][57][58]. We therefore generated transgenic lesquerella expressing CsFAD2 RNAi, CsFAD3 RNAi and AtFEA1 RNAi. We have demonstrated here that high levels of 18:1OH can be achieved by blocking the desaturation and elongation steps. Our results not only provide tools for engineering castor oil-producing lesquerella, but also enhance our understanding of the mechanisms of HFA synthesis.

High Levels of 18:1OH Accumulate in Lesquerella by Blocking Elongation and Desaturation of Fatty Acids
In this study, the dsRNA fragments in AtFAD3 RNAi and CsFAE1 RNAi contain a 323 bp or 251 bp sequence sharing 91.7% and 82.8% identity with lesquerella PfFAD3-1 (BenBank ID: MF611845) [32] and PfKCS18 (GenBank ID: AF367052) [27], respectively. When the 2-dsRNAs (AtFAD3 RNAi and CsFAE1 RNAi) were introduced to lesquerella, we observed changes in FA composition ( that AtFAD3 RNAi and CsFAE1 RNAi are effective in silencing PfFDA3-1 and PfKCS18, respectively. To see the effect of AtFAD3 RNAi and CsFAE1 RNAi in the next generation, we examined FA composition in T 2 seeds from the top six T 1 lines. The line 1 seeds produced the highest 18:1OH content at 26.6% in the T 1 generation, however, the 18:1OH decreased from 19% (line 1-1) to 9% (line 1-9) in the T 2 generation (Figure 2). Similar significant reduction of 18:1OH also occurred in line 2 from 16.8% (T 1 ) to 7.3% (T 2 line 2-1) to 2.3% (T 2 line 2-4). The remaining top best T 2 off-springs from line 3 to line 6 also showed reductions in 18:1OH contents, but the reductions were not as large, varying between line 3 from 11.7% (T 1 ) to 10.57% (T 2 line 3-1) and line 6 from 8.5% (T 1 ) to 6.8% (T 2 line 6-1). The reduction of 18:1OH content occurring in all 6 top T 2 lines implies that the competence of CsFAE1 RNAi was not fully transmitted to the next generation. A similar phenomenon was reported for an RNAi in arabidopsis where the influence of the silencing faded through several selfed generations due to a generation-dependent decrease in transcription of the RNAi [59]. In maize, analysis of an RNAi effect over multiple generations also reveals that some lines display reduced transgene silencing, but the effect of the RNAi can be maintained by outcrossing rather than self-pollination [60]. Such phenomenon is explained based on the assumption that hemizygosity would reduce any potential trans-interactions between the transgenes on homologous chromosomes that could lead to transgene silencing [60]. The molecular basis of CsFAE1 RNAi stability in lesquerella remains to be investigated. Another possibility is that the 18:1OH contents in these transgenic lesquerella lines resulted from equilibrium of FA and TAG metabolism, and there could be a ceiling for 18:1OH accumulation in lesquerella. We generated 31 independent lines expressing CsFAE1 RNAi (Tables 1 and 2), only one line accumulated a high level of 18:1OH at 26.6% and it dropped to 19% in the next generation ( Table 1). The second highest three lines contain 18:1OH ranging from 15-17% (Tables 1 and 2 (Tables 1 and 2). The results indicate that the silencing effect of AtFAD3 RNAi is more stable than that of CsFAE1 RNAi, which could be attributed to the higher nucleotide identity of 91.7% displayed between AtFAD3 RNAi and PfFAD3-1 than the 82.8% identity demonstrated between CsFAE1 RNAi and PfKCS18.
In a separate experiment, we introduced a construct carrying three dsRNAs, At-FAD3 RNAi + CsFAE1 RNAi + CsFAD2 RNAi into lesquerella and generated 15 independent transgenic lines. We observed strong negative correlation between 18:2 and 18:3 (r = −0.67), and between 18:1OH and 20:1OH (r = −0.75), indicating the strong impacts of AtFAD3 RNAi and CsFAE1 RNAi shown again in transgenics expressing the 3-dsRNAs construct ( Table 2 and Table S4). Regarding the effect of CsFAD2 RNAi, we observed a weak negative correlation between 18:1 and 18:2 (r = −0.38), but this is opposite to the weak positive correlation between 18:1 and 18:2 (r = 0.38) observed in the 2-dsRNAs which did not contain CsFAD2 RNAi (Table S4). The result suggests that CsFAD2 RNAi exerts certain silencing effect which resulted in shifting the accumulation of 18:2 to 18:1 in lesquerella. Unlike PfFAD3-1 and PfKCS18 which were specifically targeted by AtFAD3 RNAi and CsFAE1 RNAi, respectively, the CsFAD2 RNAi may target two homologous lesquerella genes, PfFAD2 (GenBank ID: DQ518313) and PfFAH12 (GenBank ID: KC972619) that share 78.3% nucleotide identity. In fact, the dsRNA fragment in CsFAD2 RNAi contains 299 bp sequences which exhibit 55.7% and 88.9% identity with PfFAD2 and PfFAH12, respectively. The effect of CsFAD2 RNAi on silencing PfFAH12 can be inferred by the reduction of correlation strength between 18:1 and 18:1OH from very strong positive (r = 0.93) in 2-dsRNAs lines, which did not express CsFAD2 RNAi (Table 1 and Table S4), to very weak positive (r = 0.15) in 3-dsRNA lines due to CsFAD2 RNAi (Table 2 and Table S4). The shift Plants 2021, 10, 1093 8 of 14 of 18:1OH accumulation to 18:1 was also evident by lower accumulation of 18:1OH at an average of 4.4% in the 3-dsRNAs lines compared with that of 7.7% (average) in the 2-dsRNAs lines (Tables 1 and 2). The silencing effect of CsFAD2 RNAi on both PfFAD2 and PfFAH12 gene expression can be deduced by the increased accumulation of 18:1 content at an average of 27.8% in lines expressing 3-dsRNAs ( Table 2) compared with that of 20.5% in lines expressing 2-dsRNAs (Table 1). Our results support previous observations of using these RNAi sequences to generate high 18:1 content in camelina lines, including CsFAD2 RNAi [56], CsFAD2 RNAi + CsFAE1 RNAi [57], or CsFAD2 RNAi + AtFAD3 RNAi + CsFAE1 RNAi [58] (in preparation).

Constrains and Potential for Production of a High 18:1OH-Containing Oil in Lesquerella
Expression of CsFAE1 RNAi, CsFAD2 RNAi, and AtFAD3 RNAi in lesquerella resulted in significant increases not only in 18:1OH, but also in 18:1 (Tables 1 and 2). The results indicated that 18:1 was inefficiently used for synthesizing 18:1OH. This could be partially due to the nature of PfFAH12 which is a bifunctional oleate hydroxylase:desaturase [25], that may not efficiently convert 18:1 to 18:1OH. Seed oil from castor or Physaria lindheimeri contains 90% 18:1OH [61] or 85% 20:1OH [62], respectively. These species have distinct FAH12s, RcFAH12 in castor [24] and PlFAH12 in P. lindheimeri [62]. Replacement of PfFHA12 with RcFAH12 or PlFAH12 should allow more efficient 18:1OH synthesis in lesquerella. Alternatively, the resulted substantial accumulation of 18:1 could also be due to some lesquerella endogenous genes having substrate preference to 18:1 and efficiently incorporating 18:1 into TAG. During seed development, a lesquerella LPAT acts like a typical plant LPAT that has substrate preference for unsaturated FAs including18:1-CoA, resulting in efficient incorporation of 18:1-CoA into TAG through the Kennedy pathway. Castor RcLPAT2 is useful for increasing 18:1OH at the sn-2 of TAGs in lesquerella [15,37]. Additional isoforms, RcLPAT3B and RcLPATB, have also been shown to increase 18:1OH in arabidopsis seed TAGs [52]. Substituting the lesquerella endogenous PfLPAT with these specific castor RcLPATs may increase 18:1OH flux to TAG by RcLPATs. Besides the Kennedy pathway, PCderived DAG pathway may also channel 18:1 into TAG by a lesquerella PfPDCT (Figure 1). Once 18:1-PC is synthesized, e.g., by PfLPCAT, some of the 18:1-PC could be converted by PDCT to 18:1-DAG for TAG assembly (Figure 1). Lesquerella seed TAGs contain about 21% PUFAs (18:2 and 18:3) (Tables 1 and 2). There is strong evidence that plants enriched with PUFAs in seed TAG may use the PC-derived pathway [19]. Therefore, it is likely that PC-derived DAGs are utilized in TAG assembly in lesquerella. Castor gene RcPDCT was demonstrated to enhance flux from 18:1OH-PC to 18:1OH-DAG [41]. It would be favorable to over-express RcPDCT in lesquerella to increase 18:1OH incorporation to TAG through PC-derived DAG pathway.
It is anticipated that the increased 18:1OH is at the expense of 20:1OH in transgenic lesquerella lines expressing CsFAE1 RNAi (Tables 1 and 2), however, total HFA decreased and showed strong correlation with 20:1OH contents (0.96 < r > 0.75, Table S4). The results indicated that 18:1OH was not incorporated into TAG at the same efficiency as 20:1OH. Lesquerella PfKCS18 is evolved to specifically elongate 18:1OH-CoA to 20:1OH-CoA [27] (Figure 1). It is possible that other lesquerella enzymes, such as PfGPAT, PfDGAT and/or PfPDAT also co-evolved to adapt and utilize 20:1OH efficiently. Most plant GPATs have a broad acyl-CoA substrate specificity [19,63]. There is evidence that castor RcGPAT9 plays an important role in acylating HFAs at the sn-1 position of G3P, resulting in sn-1-HFA-LPA, which facilitates the subsequent incorporation of sn-2 and sn-3 HFA into seed TAG by LPAT and DGAT [53]. In plant seeds accumulating unusual FAs, members of DGAT2 family are essential enzymes in acylating unusual FAs to the sn-3 position of DAG. For example, castor RcDGAT2 prefers 18:1OH to common FAs [50,64]. Lesquerella seed transcriptome analysis reveals one PfGPAT9 and three PfDGATs [17]. It would be interesting to explore whether these genes have substrate selectivity for HFA-CoA or common FA-CoA. The role of PfKCS18 has been explored in camelina [65]. Transgenic camelina expressing RcFAH12 accumulates 15% HFA [66] but the resulted transgenic seeds reduce TAG content and seed germination ability [65]. When RcFAH12 with PfKCS18 are co-expressed, the transgenic camelina seeds increase HFA content to 21% and also restore TAG content and seed germination ability [65]. Camelina is not a native species for HFA synthesis, 18:1OH-PC generated by RcFAH12 in camelina may be subjected to β-oxidation [67], or represents a bottleneck [40], limiting HFA accumulation. The elongation step by PfKCS18 may ease the 18:1OH flux from PC to cytosol FA-CoA pool, thus relieve the bottleneck and facilitate the utilization of HFA-CoA by the Kennedy pathway [33] (Figure 1). PDAT transfer FA at the sn-2 position of PC to the sn-3 position of DAG, yielding TAG [44,68] (Figure 1). Castor has two PDAT1s, but only RcPDAT1-2 (or RcPDAT1A) selects 18:1OH-PC as substrate, and it participates in HFA-TAG synthesis [45,46]. There are three PfPDATs expressed in lesquerella seeds [17]. Whether these PfPDATs are involved in transferring HFA-PC to DAG remains to be investigated. To further enhance 18:1OH accumulation in lesquerella TAGs, coordinated expression of multiple genes, such as RcGPAT9, RcDGAT2 and RcPDAT1-2 (or RcPDAT1A) should promote 18:1OH accumulation in seed TAG.
In summary, to develop a castor oil-producing lesquerella crop, we designed genetic engineering schemes based on known pathways of fatty acid biosynthesis in lesquerella. As predicted, high levels of 18:1OH were accumulated by reducing the elongation of 18:1OH to 20:1OH through expression of CsFAE1 RNAi. Additionally, high levels of 18:1 and 18:2 were accumulated through suppression of desaturation steps by expressing CsFAD2 RNAi and/or AtFAD3 RNAi. Intriguingly, the accumulated 18:1 was not efficiently utilized to produce 18:1OH and instead, 18:1 was largely channeled to seed TAG. On the other hand as discussed, multiple mechanisms could limit the acylation of 18:1OH into TAG. Our results direct future research efforts in implementing genetic approach that targets not only enhancement of 18:1OH synthesis, but also on increased 18:1OH acylation to TAG. Nevertheless, we demonstrated for the first time that lesquerella can be engineered for large increases in 18:1OH levels from 0.4-0.5% in WT to a stable high level of 15-20% in transgenic seed oils.

Construction of pBinGlyBar1 + AtFAD3 RNAi + CsFAE1 RNAi and pBinGlyBar1 CsFAD2
RNAi + AtFAD3 RNAi + CsFAE1RNAi Constructs used for transformation experiments were prepared as follows. The FAD3 RNAi hairpin cassette was prepared by PCR amplification of a 323-bp fragment of the FAD3 gene from Arabidopsis thaliana Col-0 cDNA in both antisense using primers Arm1-5 NheI-F3 5 -AATAAGCTAGCACCGGACACACCACCAGAAC-3 and Arm1-3 EcoRI-F3 5 -TATTGAATTCCGTAGACTTTAAGAACCGCGAG-3 and sense orientations using primers Arm2-5 PstI-F3 5 -TAATACTGCAGCACCGGACACACCACCAGAAC-3 and Arm2-3 XhoI-F3 5 -ATTACTCGAGCCGTAGACTTTAAGAACCGCGAG-3 and cloned into plasmid pGEMT-Easy-HTM3 [69], replacing the existing antisense and sense arms. The resulting FAD3 hairpin sequence was excised at EcoRI/XhoI from that plasmid and inserted into pBinGlyBar1 [59] as an EcoRI/XhoI fragment. Flanking the FAD3 hairpin sequence in pBinGlyBar1 was the seed-specific promoter and the 3 UTR for the Glycine max glycinin-1 gene. The new construct was designated pBinGlyBar1 + AtFAD3 RNAi. The FAE1 RNAi suppression cassette was prepared by PCR amplification of a 251-bp portion of the camelina FAE1 gene from cDNA using the oligonucleotides: 5 -TAATTCTAGACTCGAGGGGAATACTTCGTCTAGCTC-3 and 5 -TATAAAGCTTACTAGT CCGACCGTTTTTTGACATGAGTC-3 . The PCR product was assembled sequentially in an inverted repeat orientation of either side of the Flaveria trinervia pyruvate orthophosphate dikinase (Pdk) intron [70]. The hairpin cassette was then cloned downstream of the seed-specific promoter for the Glycine max glycinin-1 gene and upstream of the 3 UTR for the glycinin-1 gene as a Not1 fragment. The resulting vector contained AscI restriction sites that flanked the glycinin-1 promoter and 3 UTR. Using this restriction site the entire cassette containing promoter, RNAi hairpin and 3 UTR were assembled into the AscI site of the binary vector pBinGlyBar1 + AtFAD3 RNAi. The resulting construct was designated pBinGlyBar1 + AtFAD3 RNAi + cFAE1 RNAi. The FAD2 RNAi hairpin cassette was pre-pared by PCR amplification of a 299-bp portion of the camelina FAD2 gene from cDNA using the oligonucleotides:5 -TAATTCTAGACTCGAGCGTCTTGATCACTTACTTGCAG-3 and 5 -TATAAAGCTTACTAGTCTACATAGATACACTCCTTTGCC-3 . The product was cloned sequentially in an inverted repeat orientation of either side of the Flaveria trinervia pyruvate orthophosphate dikinase (Pdk) intron. The hairpin cassette was then cloned downstream of the seed-specific promoter for the soybean oleosin gene and upstream of the oleosin 3 UTR as a Not1 fragment. The resulting vector contained AscI restriction sites that flanked the oleosin promoter and 3 UTR. Using this restriction site the entire cassette containing promoter, RNAi hairpin and 3 UTR were assembled into the MluI site of pBinGlyBar1 + AtFAD3 RNAi + cFAE1 RNAi to make pBinGlyBar1 + AtFAD3 RNAi + cFAE1 RNAi + cFAD2 RNAi which also contains a bar marker gene for Basta selection of transgenic plants.

Plant Transformation and Growth Condition
The lesquerella seeds, WCL-LY2 [71] 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 [72] carrying the binary vector pBinGlyBar1 + AtFAD3 RNAi + CsFAE1 RNAi and pBinGlyBar1 CsFAD2 RNAi + AtFAD3 RNAi + CsFAE1RNAi. Tissue culture and plant growth conditions were as described before [14] with the exception of using Basta (1 mg/L) as a transgenic selective agent. In brief, leaves harvested from plants in sterile condition were wounded by slightly scratching the underside of the leaf and then dipping the leaf in the half strength MS medium containing the Agrobacterium for 5 min. Following the inoculation, leaves were blotted on sterilized filter paper and transferred to Callus and Shoot Induction (CSI) medium composed of basal medium (BM, half strength MS medium plus 30 g L −1 sucrose and 6 g L −1 agar, pH 5.7) supplemented with 1 mg L −1 6-benzylaminopurine (BA) and 0.1 mg L −1 1-Naphthaleneacetic Acid (NAA). After incubating the infected leaves in the growth chamber for 2 days, the leaves were cut into 5 mm segments and cultured on CSI media plus 1 mg L −1 Basta for transgenic selection and 100 mg L −1 timentin for inhibiting the Agrobaterium growth. In 6-8 weeks, yellow-greenish Basta resistant calli started to appear on the leaf segments. To eliminate chimeras, each shoot was cut into small pieces (about 2 × 2 mm 2 ) and placed on the CSI medium for shoot regeneration. After 4 rounds of successive regenerations, shoots were sub-cultured on BM plus 1 mg L −1 BA, 1 mg L −1 Indole-3-Butyric Acid (IBA) and 1 mg L −1 Basta for multiplication. Shoots 10-15 mm in length were transferred to rooting medium (BM plus 1 mg L −1 IBA and 50 mg L −1 Basta). When a shoot developed 2-3 roots (usually in 3-5 weeks), it was then transferred to a Magenta box (Sigma, St. Louis, Mo) containing sterilized peat-vermiculite growth mixture (Sunshine mix #4, Planet Natural, Bozeman, MT) presoaked with 1 mg L −1 IBA water solution. After 8-10 weeks in the growth mixture, well-developed primary plants showing 8-12 normal leaves and 2-3 inch height were transferred to a 6-inch pot and placed under a transparent plastic cover for the first 2 weeks for acclimation in the greenhouse. T 1 selfed seeds were obtained by hand-pollination between different flowers from the same transgenic plant (T 0 ). To estimate the number of transgene locus for each line, T 1 seeds were germinated on germination medium containing Basta at 1 mg L −1 for 3 weeks. Healthy seedlings showing normally developed cotyledons and 2-4 true leaves were counted as resistance seedlings (R); sensitive seedlings (S) had arrested yellow cotyledons and no true leaves. Transgene locus numbers were based on the Mendelian ratio of R:S, 3:1 for one locus, 16:1 for two loci. T 1 seedlings were transplanted into soil for T 2 seed production.

Analysis of Fatty Acid Composition
Seeds homogenized in a gas chromatograph (GC) autosampler vial were subjected to direct transesterification to produce fatty acid methyl esters (FAMEs) using trimethylsulphonium hydroxide (TMSH) as described [59,73]. The resulting FAMES were analyzed by GC-flame ionization detection using previously described instrument conditions [74].