Omega-3 long-chain polyunsaturated fatty acids (LC-PUFAs), eicosapentaenoic acid (EPA; 20:5Δ^5,8,11,14,17) and docosahexaenoic acid (DHA; 22:6Δ^{4,7,10,13,16,19}) are now accepted as being essential components of a healthy, balanced diet (Napier et al., 2019; West et al., 2021). The wild capture fisheries that supply omega-3 fatty acids are at their maximum levels of sustainable production; therefore, attempts to meet the growing demands of an increasing population depend on alternative sources of fish oils (Tocher et al., 2019). Camelina sativa is an oilseed crop with high levels (>35%) of α-linolenic acid (ALA; 18:3Δ^9,12,15) and has reconstituted a biosynthetic pathway from ALA to synthesize both EPA and DHA in Camelina cv. Celine seeds by expressing heterologous desaturase and elongase genes, producing levels of EPA and DHA equivalent to those in marine fish oils, exemplified by the prototype line DHA2015.1 (abbreviated to DHA1) accumulating over 25% n-3 LC-PUFAs (Figures S1 and S2 (Petrie et al., 2014; Ruiz-Lopez et al., 2014). Field trials of DHA1 in the UK, USA and Canada demonstrated the omega-3 LC-PUFAs trait was stable in distinct geographical locations and agricultural environments (Han et al., 2020). In parallel, salmon feeding trials and human dietary studies using DHA1 seed oils both demonstrated that these transgenic plant-derived oils could serve as effective replacements for marine-derived fish oils (Betancor et al., 2018; West et al., 2021).

Based on our observation that ALA is the endogenous C18 precursor for seed omega-3 LC-PUFA production (Han et al., 2020), we hypothesized that increasing the ALA pool can further enhance EPA/DHA accumulation in DHA1 Camelina. The DHA1 construct already contains a Δ12 desaturase to drive the flux of fatty acids into PUFA biosynthesis (Figures S1 and S2). However, as a less obvious approach, we proposed using gene-edited fae1 mutants of Camelina. Camelina FAE1 functions in competition with endogenous FAD2 Δ12 desaturase (which desaturates OA to produce linoleic acid (LA; 18:2Δ^9,12)) for C18 substrates, sequentially elongating oleic acid (OA; 18:1Δ⁹) to gondoic acid 20:1Δ¹¹ and then erucic acid 22:1Δ¹³. Previously, Ozseyhan et al. (2018) used CRISPR/cas9 to disrupt all three homologues of FAE1 present in Camelina cv. Suneson (allohexaploid). Since the expression of FAE1 is restricted to developing seeds, there is no impact of this mutation on other tissues. Within the seeds, ALA was increased from 36.9% in wildtype (WT) to 47.3% in fae1 mutant—this is in addition to the complete ablation of erucic acid, a fatty acid, which is considered undesirable above a modest threshold (5% of total oil) in human foodstuffs. Pollen from the fae1 GE mutant (previously selfed to segregate away from the CRISPR-Cas9 transgene and associated DsRed marker) was crossed with the DHA1 line described in Han et al. (2020), and the resulting F1 hybrid seeds were sown in the greenhouse. Selected seeds with strong DsRed fluorescence (associated with DHA1; Figure S2) from individual F2 plants were dissected, and half the cotyledon tissue was used for fatty acid composition analysis by GC-FID, the other half germinated on 1/2MS nutrition media. Based on gas chromatographic analysis of fatty acid methyl esters (FAMEs) data, only seeds showing a low 20:1Δ¹¹ content (<1.0%; Ozseyhan et al., 2018) were identified as the DHA1 and fae1 homozygous line (designated DHA1xfae1). Seeds from F3 plants were then sown alongside parental lines and WT as replicated plots in the field at the Rothamsted Experimental Farm (Harpenden, UK) on 23 May 2019, as permitted by Consent 19/R8/01 from DEFRA. After harvesting on 18 September 2019, replicated seed FAMEs and TAG analysis were completed for the plots.

GC analysis (Figure 1a) confirmed showed that in field-grown DHA1xfae1 seeds, the accumulation of 20:1Δ¹¹ is as low as 0.4%, consistent with the mutant background. Compared with parental line DHA1, the levels of palmitic acid (16:0) and stearic acid (18:0) were broadly the same, with the OA and LA levels slightly decreased by 1% and 2.5%, respectively, whereas the ALA level was increased from 17.0% in DHA1 to 21.5% DHA1xfae1. This confirmed that targeted mutagenesis by gene editing of the FAE1 genes in the DHA1 genetic background did not affect the OA precursor metabolites but did alter the downstream desaturation pathway, as predicted. Analysis of the total C20+ n-3 LC-PUFAs in DHA1xfae1 seeds showed little change in EPA compared with DHA1 and 9.1% compared with 9.3%. However, the DHA level was modestly increased, from 9.7% in DHA1 to 12.6% in DHA1xfae1. Eicosatetraenoic acid (ETA; 20:4Δ^8,11,14,17) and docosapentaenoic acid (DPA; 22:5Δ^{7,10,13,16,19}) levels were similarly enhanced, 1.3% and 1.5%, respectively. In sum, total C20+ n-3 fatty acids contents (comprising ETA, EPA, DPA and DHA) were increased by 5.5% from 27.5% in DHA1 to 33.0% in DHA1xfae1. These data demonstrated that increasing the precursor ALA substrate can be an effective approach to boost n-3 LC-PUFAs production in the DHA1 line.

Health studies in humans have reported that diets with an increased omega-6 (n-6) to n-3 ratio are highly pro-thrombotic and pro-inflammatory, and contribute to the prevalence of atherosclerosis, obesity, diabetes and a wide range of inflammation disorders (West et al., 2021). Therefore, we calculated the different n-3 and n-6 parameters (Figure 1b, c; Figure S3). The total C20+ n-6 fatty acids including dihomo-γ-linolenic acid (DGLA; 20:3Δ^8,11,14) and arachidonic acid (ARA; 20:4Δ^5,8,11,14) remained similar, at 3.1% in DHA1 and 2.8% in DHA1xfae1, giving a ratio of C20+ (n-3/n-6) as 8.8 in DHA1 and 11.7 in DHA1xfae1. The total n-3 content was 47.5% in DHA1 and 58.2% in DHA1xfae1, whereas the total n-6 content was 23.6% and 20.2%, respectively. Therefore, the ratio of total (n-3/n-6) was 2.0 in DHA1 and 2.9 in DHA1xfae1, which is also a significant increase compared with 2.2 in the fae1 mutant and 1.7 in both WT Celine and WT Suneson lines, indicating that DHA1xfae1 seed fatty acids have a even better health benefits than those of DHA1, fae1 and WTs.

Triacylglycerol (TAG) is the major storage lipid for fatty acids in the seed, and it consists of a glycerol backbone onto which three fatty acids are esterified. The fatty acid composition of individual TAG species (described by number of carbons:number of desaturations) can vary, and WT camelina seed can have ~80 different molecular species of TAG (Usher et al., 2017). We have previously shown how transgene-derived accumulation of EPA, DPA and DHA expands the repertoire of TAG species (Han et al., 2020). A comparison (Figure S4) of the TAG species in the seeds of field-grown DHA1 and DHA1xfae1 compared to relevant parental lines (WT Celine, WT Suneson, fae1) illustrates a change in the TAG fingerprint of fae1, DHA1 and DHA1xfae1. The most abundant TAGs in fae1 are C54 species with 3 to 9 unsaturations—combinations of ALA (18:3) and LA (18:2). There is an almost complete absence of any C56 TAG species in fae1. In the case of DHA1 and DHA1xfae1, more novel TAG (C58 but inclusive of C56 to C66) species are present. These novel TAG species are absent from all the parental and WT backgrounds. Notable is the occurrence of some DHA-specific TAG molecular species that differ between DHA1 and DHA1xfae1 (e.g. 64:17 and 66:17) and a number of TAG species such as 58:8–58:12 that are more abundant, indicative of elevated accumulations of long-chain polyunsaturated fatty acids.

In conclusion, the combination of CRISPR-Cas9 gene editing to inactivate the FAE1 pathway clearly results in a beneficial increase in the levels of EPA, DHA and other omega-3 LC-PUFAs in transgenic Camelina harbouring the DHA2015.1 transgenes. In particular, the fae1 mutant not only is devoid of C20+ monounsaturated fatty acids (including the undesirable C22 erucic acid; Bach and Faure, 2010) but also has increased levels of omega-3 fatty acids such as ALA. Our previous studies have indicated that ALA is the primary endogenous precursor fatty acid used to make EPA and DHA, and our data here further confirm this. This contrasts with Canola, where recent attempts to engineer the accumulation of EPA and DHA result in the metabolism of the omega-6 precursor LA. In that respect, Canola is biased towards the synthesis of omega-6 fatty acids whereas Camelina is biased towards omega-3, therefore Canola requires an additional transgene-derived ‘push’ to direct the flux of fatty acids onto the omega-3 track (discussed in Napier et al., 2019). It is also noteworthy that Canola already lacks the FAE1 activity, selected by conventional plant breeding for the absence of erucic acid, which is present in parental varieties of Brassica napus seed oil. Collectively, these data suggest that Camelina is a superior host for the transgene-derived seed-specific synthesis of omega-3 LC-PUFAs such as EPA and DHA, and representing the first example of an environmental release of a GM + GE stack.

Omega-3 长链多不饱和脂肪酸 (LC-PUFA)、二十碳五烯酸 (EPA; 20:5Δ ^5,8,11,14,17 ) 和二十二碳六烯酸 (DHA; 22:6Δ ^{4,7,10,13, 16,19}）现在被认为是健康均衡饮食的重要组成部分（Napier等人， 2019 年；West等人， 2021 年）。提供 omega-3 脂肪酸的野生捕捞渔业处于可持续生产的最高水平；因此，满足不断增长的人口不断增长的需求的尝试取决于鱼油的替代来源（Tocher等人， 2019 年）。Camelina sativa是一种油料作物，具有高水平 (>35%) 的 α-亚麻酸 (ALA; 18:3Δ ^9,12,15) 并在 Camelina cv 中重建了从 ALA 合成 EPA 和 DHA 的生物合成途径。Celine 种子通过表达异源去饱和酶和延伸酶基因，产生与海洋鱼油中相当的 EPA 和 DHA 水平，例如原型线 DHA2015.1（缩写为 DHA1）累积超过 25% n-3 LC-PUFA（图 S1和 S2（Petrie等人， 2014 年；Ruiz-Lopez等人， 2014 年）。在英国、美国和加拿大进行的 DHA1 田间试验表明，omega-3 LC-PUFAs 性状在不同的地理位置和农业环境中是稳定的（韩等人， 2020）。同时，使用 DHA1 种子油的鲑鱼喂养试验和人类饮食研究均表明，这些转基因植物油可以作为海洋鱼油的有效替代品（Betancor等人， 2018 年；West等人， 2021 年）。

基于我们观察到 ALA 是种子 omega-3 LC-PUFA 生产的内源 C18 前体（Han等人， 2020 年），我们假设增加 ALA 池可以进一步增强 DHA1 Camelina 中 EPA/DHA 的积累。DHA1 构建体已经包含 Δ12 去饱和酶，以驱动脂肪酸流入 PUFA 生物合成（图 S1 和 S2）。然而，作为一种不太明显的方法，我们建议使用 Camelina 的基因编辑fae1突变体。Camelina FAE1与 C18 底物的内源性 FAD2 Δ12 去饱和酶（它使 OA 去饱和以产生亚油酸 (LA; 18:2Δ ^9,12 )) 竞争，依次将油酸 (OA; 18:1Δ ⁹ ) 延长为 gondoic acid 20： 1Δ ¹¹然后是芥酸 22:1Δ ¹³。此前，Ozseyhan等人。（2018 年）使用 CRISPR/cas9 破坏了 Camelina cv 中存在的所有三个FAE1同源物。Suneson（异六倍体）。由于 FAE1 的表达仅限于发育中的种子，因此这种突变对其他组织没有影响。在种子中，ALA 从野生型 (WT) 中的 36.9% 增加到fae1突变体中的 47.3%——这是在完全消除芥酸（一种脂肪酸，超过适度阈值（5%总油）在人类食品中。来自fae1的花粉GE突变体（以前自交以与CRISPR-Cas9转基因和相关的DsRed标记分离）与Han等人描述的DHA1系杂交。( 2020 )，并将由此产生的 F1 杂交种子播种在温室中。从单个 F2 植物中选择具有强 DsRed 荧光（与 DHA1 相关；图 S2）的种子被解剖，一半的子叶组织用于通过 GC-FID 进行脂肪酸组成分析，另一半在 1/2MS 营养培养基上发芽。基于脂肪酸甲酯 (FAME) 数据的气相色谱分析，只有显示低 20:1Δ11^含量（<1.0%；Ozseyhan等人， 2018 年）的种子被鉴定为 DHA1 和fae1纯合系（命名为 DHA1x fae1）。然后，根据 DEFRA 的第 19/R8/01 号同意书，于 2019 年 5 月 23 日，将 F3 植物的种子与亲本系和 WT 一起播种在 Rothamsted 实验农场（英国哈彭登）的田间复制地块。在 2019 年 9 月 18 日收获后，对地块完成了重复的种子 FAME 和 TAG 分析。

GC 分析（图 1a）证实，在田间种植的 DHA1x fae1种子中，20:1Δ ¹¹的积累低至 0.4%，与突变体背景一致。与亲本 DHA1 相比，棕榈酸（16:0）和硬脂酸（18:0）的含量基本相同，OA 和 LA 水平分别略微下降 1% 和 2.5%，而 ALA 水平从 DHA1 中的 17.0% 增加到 21.5% DHA1x fae1。这证实了通过在 DHA1 遗传背景中对FAE1基因进行基因编辑的靶向诱变不会影响 OA 前体代谢物，但确实改变了下游去饱和途径，正如预测的那样。DHA1x fae1中总 C20+ n-3 LC-PUFA 的分析与 DHA1 相比，种子的 EPA 变化不大，而 9.1% 与 9.3% 相比变化不大。然而，DHA 水平适度增加，从 DHA1 的 9.7% 到 DHA1x fae1 的 12.6 %。二十碳四烯酸 (ETA; 20:4Δ ^8,11,14,17 ) 和二十二碳五烯酸 (DPA; 22:5Δ ^{7,10,13,16,19} ) 水平同样提高，分别为 1.3% 和 1.5%。总之，C20+ n-3 脂肪酸总含量（包括 ETA、EPA、DPA 和 DHA）从 DHA1 中的 27.5% 增加到 DHA1x fae1 中的 33.0%，增加了 5.5 %。这些数据表明，增加前体 ALA 底物可能是提高 DHA1 线中 n-3 LC-PUFA 产量的有效方法。

人类健康研究报告称，增加 omega-6 (n-6) 与 n-3 比例的饮食具有高度的促血栓形成和促炎作用，并导致动脉粥样硬化、肥胖、糖尿病和多种疾病的流行。炎症性疾病（West等人， 2021 年）。因此，我们计算了不同的 n-3 和 n-6 参数（图 1b，c；图 S3）。包括二高-γ-亚麻酸 (DGLA; 20:3Δ ^8,11,14 ) 和花生四烯酸 (ARA; 20:4Δ ^5,8,11,14 ) 在内的总 C20+ n-6 脂肪酸保持相似，为 3.1%在 DHA1 中，在 DHA1x fae1中为 2.8%，C20+ (n-3/n-6) 的比率在 DHA1 中为 8.8，在 DHA1x fae1 中为11.7。DHA1 中的总 n-3 含量为 47.5%，DHA1x fae1 中的总 n-3 含量为 58.2 %，而总 n-6 含量分别为 23.6% 和 20.2%。因此，在 DHA1 和 DHA1x fae1 中，总比值 (n-3/n-6) 为 2.0，在 DHA1x fae1中为 2.9，这与fae1突变体中的 2.2 和 WT Celine 和 WT Suneson 系中的 1.7相比也有显着增加，表明DHA1x fae1种子脂肪酸比 DHA1、fae1和 WT 的健康益处更好。

三酰基甘油 (TAG) 是种子中脂肪酸的主要储存脂质，它由甘油主链组成，三个脂肪酸在甘油主链上酯化。单个 TAG 物种的脂肪酸组成（由碳数：去饱和次数描述）可能会有所不同，WT 亚麻荠种子可以有约 80 种不同的 TAG 分子物种（Usher等人， 2017 年）。我们之前已经展示了 EPA、DPA 和 DHA 的转基因衍生积累如何扩展 TAG 物种的库（Han等人， 2020 年）。田间种植的 DHA1 和 DHA1x fae1种子中的 TAG 物种与相关亲本系（WT Celine，WT Suneson，fae1 ）的比较（图 S4）说明了 TAG 指纹的变化fae1、 DHA1 和 DHA1x fae1。fae1 中最丰富的 TAG 是具有 3 到 9 个不饱和度的 C54 物种——ALA (18:3) 和 LA (18:2) 的组合。fae1中几乎完全没有任何 C56 TAG 物种。在 DHA1 和 DHA1x fae1的情况下，存在更多新的 TAG（C58，但包括 C56 至 C66）物种。这些新的 TAG 物种在所有亲本和 WT 背景中都不存在。值得注意的是在 DHA1 和 DHA1x fae1 （例如 64:17 和 66:17）之间出现了一些 DHA 特异性 TAG 分子种类以及一些更丰富的 TAG 种类，例如 58:8-58:12，表明增加长链多不饱和脂肪酸的积累。

总之，CRISPR-Cas9 基因编辑与 FAE1 通路失活的组合明显导致含有 DHA2015.1 转基因的转基因 Camelina 中 EPA、DHA 和其他 omega-3 LC-PUFA 水平的有益增加。特别是，fae1突变体不仅没有 C20+ 单不饱和脂肪酸（包括不受欢迎的 C22 芥酸；Bach 和 Faure，  2010) 但也增加了 omega-3 脂肪酸（如 ALA）的水平。我们之前的研究表明，ALA 是用于制造 EPA 和 DHA 的主要内源性前体脂肪酸，我们的数据进一步证实了这一点。这与 Canola 形成鲜明对比，后者最近尝试设计 EPA 和 DHA 的积累导致 omega-6 前体 LA 的新陈代谢。在这方面，Canola 偏向于 omega-6 脂肪酸的合成，而 Camelina 偏向于 omega-3，因此 Canola 需要额外的转基因衍生“推动”来引导脂肪酸流向 omega-3 轨道（ Napier等人在 2019 年讨论过）。还值得注意的是，油菜已经缺乏 FAE1 活性，这是通过常规植物育种选择的，因为没有芥酸，而芥酸存在于欧洲油菜种子油的亲本品种中。总的来说，这些数据表明 Camelina 是转基因衍生的种子特异性合成 omega-3 LC-PUFA（如 EPA 和 DHA）的优良宿主，并且代表了 GM + GE 堆栈的环境释放的第一个例子。