The goal of MFA is to quantify all the in vivo metabolic fluxes in a given organ or cell, here developing embryos, which results in a metabolic flux map. The determination of intermediary carbon fluxes requires ¹³C-labeling. Therefore, establishing culture conditions that mimic the development of embryos in planta is decisive to build carbon flux maps. For instance, it has been found for multiple species of the Brassicaceae family that developing embryos readily grow in liquid cultures (Schwender and Ohlrogge, 2002; Lonien and Schwender, 2009; Chen and Shachar-Hill, 2012; Tsogtbaatar et al., 2020). Substrates, furnished by the mother plant, are unloaded in the endosperm liquid and taken up by the embryo. Therefore, the best way to design a liquid growth medium that mimics the in planta liquid environment is to analyze the constituents of the endosperm or the vascular tissue (Alonso et al., 2007; Alonso et al., 2010; Cocuron et al., 2019; Tsogtbaatar et al., 2020). Substrate composition, total osmotic pressure of the medium, and light intensity are important factors influencing plant tissue development (Goffman et al., 2005; Allen et al., 2007; Alonso et al., 2010), and hence have to be optimized. To meet requirements for ¹³C-MFA, it is important to maintain homeostasis and metabolic steady state in culture embryos. Therefore, photoperiod is not commonly applied to embryos in culture for ¹³C-MFA studies. Ideal culture conditions for plant embryos are validated when the dry weight gain and the biomass composition of embryos grown in culture are not significantly different from the ones grown in planta (Alonso et al., 2007; Alonso et al., 2010; Alonso et al., 2011; Cocuron et al., 2019; Tsogtbaatar et al., 2020). Parallel labeling experiments, using different ¹³C-substrates, are usually conducted to have a better coverage of the metabolic network (Schwender et al., 2006; Alonso et al., 2007; Alonso et al., 2010; Antoniewicz, 2015; Crown et al., 2016; Antoniewicz, 2018; Acket et al., 2019; Tsogtbaatar et al., 2020). Labeled embryos are harvested after they have reached an isotopic steady-state, meaning that the labeling in intermediary metabolites and products have reached a constant pattern. The resultant labeling in a range of metabolites is then determined using nuclear magnetic resonance and/or mass spectrometry (MS) (Ratcliffe and Shachar-Hill, 2006; Dieuaide-Noubhani and Alonso, 2014). Several sensitive gas chromatography-mass spectrometry (GC-MS) and liquid chromatography tandem mass spectrometry (LC-MS/MS) methods have been recently developed to follow the labeling directly in key metabolic intermediaries such as sugars, phosphorylated compounds, free amino acids, and organic acids (Alonso et al., 2010; Koubaa et al., 2013; Cocuron and Alonso, 2014; Cocuron et al., 2017; Acket et al., 2019; Cocuron et al., 2020). Compartmentalization in plant cells is usually considered by following different labeling patterns of a few metabolites and hydrolyzed macromolecules whose biosynthesis occur in one compartment (Allen et al., 2007; Allen et al., 2012; Dieuaide-Noubhani and Alonso, 2014; Cocuron et al., 2020). The complete labeling information is entered into a mathematical model describing the metabolic network. This mathematical model includes equations expressing metabolic and isotopic steady states. Finally, the fluxes through the network that correspond to the observed label distribution are calculated using mathematical algorithms that can be computed using available software (Wiechert and De Graaf, 1997; Wiechert and De Graaf, 1997; Möllney et al., 1999; Wiechert et al., 1999; Wiechert et al., 2001). ¹³C-based MFA has been successfully applied to plant systems to characterize the in vivo carbon fluxes in important metabolic pathways during FAS in developing embryos (Goffman et al., 2005; Alonso et al., 2007; Allen et al., 2009; Lonien and Schwender, 2009; Alonso et al., 2010; Cocuron et al., 2019; Carey et al., 2020; Tsogtbaatar et al., 2020).

Determining the efficiency of the developing embryo in converting carbon sources into oil and other biomass components (e.g. proteins, and carbohydrates) known as carbon conversion efficiency (CCE), is important to evaluate the potential for improving seed oil quantity (Goffman et al., 2005; Alonso et al., 2007; Allen et al., 2009; Lonien and Schwender, 2009; Chen and Shachar-Hill, 2012; Cocuron et al., 2019; Carey et al., 2020; Tsogtbaatar et al., 2020). CCE results from the sum of all catabolic and anabolic metabolic processes, which varies in developing embryos from different oilseeds (Goffman et al., 2005; Allen et al., 2009; Lonien and Schwender, 2009; Carey et al., 2020; Tsogtbaatar et al., 2020). To assess the CCE, embryos are grown for several days culture media and conditions that mimic their development in plant, as explained in the above section. There are two options to determine the percentage of carbon uptaken that was stored into biomass components. The first option uses ¹⁴C-labeled substrates in sealed flasks, and quantifies the radiolabeling released as ¹⁴CO₂, and incorporated as ¹⁴C-oil, ¹⁴C-proteins, and ¹⁴C-carbohydrates (Goffman et al., 2005; Alonso et al., 2007; Allen et al., 2009; Alonso et al., 2010; Alonso et al., 2011; Carey et al., 2020). The second option is the quantify the substrate depletion from the media and the biomass stored during the incubation period (Cocuron et al., 2019; Tsogtbaatar et al., 2020). Table 1 summarizes the biomass composition and the CCE of different embryos from green and non-green oilseeds that have been studied and published so far. Under physiological conditions, the biomass composition of the developing embryos in culture differs among species. Oil content varied from 18% (w/w) for G. max to 56% for B. napus, proteins from 6% for Z. mays LH59 to 40% for G. max, and carbohydrates from 43% for G. max to 60% for Z. mays LH59. Under physiological conditions, T. arvense embryos were found to be the most efficient in converting carbon substrates into biomass (93%) while C. sativa were the least (32%). It is important to note that these differences in CCE reflect the amount of carbon loss as CO₂ during the synthesis of biomass components: the lower the CCE, the higher is the loss of carbon as CO₂, which has implications for metabolic engineering. For instance, improving oil content in C. sativa may be achieved by reducing the pathways producing CO₂. However, this strategy would not work in T. arvense which is already extremely efficient. Instead, for T. arvense, one would have to redirect carbon from another biomass component to increase FAS.

Except C. sativa, embryos from green seeds had higher CCEs than those seeds that cannot use light ( Table 1 ). The importance of light was further demonstrated by incubating developing embryos at different intensities. In general, higher light intensity significantly increased the CCE. This finding suggests that light has strong effects on the metabolism of developing green embryos—probably due to the additional production of NADPH and ATP via photosynthesis–resulting in faster growth and storage product accumulation, which improved the CCE (Goffman et al., 2005; Alonso et al., 2007; Allen et al., 2009; Lonien and Schwender, 2009; Chen and Shachar-Hill, 2012; Cocuron et al., 2019; Carey et al., 2020; Tsogtbaatar et al., 2020). Indeed, studies on these developing green embryos showed that all the major photosynthetic complexes (PSII, PSI and their antenna complexes, cytochrome b6f complex, and ATP synthase) are present at a necessary stoichiometric ratio, suggesting a high photochemical activity despite being partially blocked from light by the pod and seed coat (Asokanthan et al., 1997; Ruuska et al., 2004; Puthur et al., 2013; Allorent et al., 2015). The light reactions occurring in these green embryos have been associated with the rapid synthesis of ATP and NADPH needed for energetically-expensive FAS. It has also been reported that embryo photosynthesis contributes to a significant amount of oxygen, which fuels energy-generating biochemical pathways, including mitochondrial respiration (Ruuska et al., 2004; Borisjuk et al., 2005; Rolletschek et al., 2005; Tschiersch et al., 2011; Galili et al., 2014).

Differences in the flow of carbon through central metabolic pathways are responsible for the differences in biomass composition and CCE measured in developing embryos from various oilseed species ( Table 1 ). ¹³C-MFA has been the method of choice to measure in vivo rates of carbon flow, quantifying the metabolic pathways that are active during FAS.

¹³C-MFA was applied to developing embryos from various oilseed species to determine the sources of carbon and reductant for FAS, and to unravel the occurrence of non-conventional pathways that improved the CCE ( Table 2 ) (Schwender et al., 2004; Schwender et al., 2006; Alonso et al., 2007; Allen et al., 2009; Lonien and Schwender, 2009; Alonso et al., 2010; Hay et al., 2014; Cocuron et al., 2019; Acket et al., 2020; Carey et al., 2020; Tsogtbaatar et al., 2020). Glycolysis is a major source of pyruvate for de novo FAS ( Figure 1 ). ¹³C-labeling and metabolic flux analysis performed in non-photosynthetic and photosynthetic embryos revealed the contribution of the plastidic NADP-dependent malic enzyme (pNADP-ME) for the production of plastidic pyruvate (pPYR) in Z. mays, H. annuus, G. max, T. arvense, and C. sativa ( Table 2 ). The pNADP-ME provided up to 54% of pPYR in the high oil ALEX maize line (Cocuron et al., 2019). In plastids, pNADP-ME catalyzes the conversion of malate to pPYR with the production of CO₂ and NADPH ( Figure 1 ). In addition, pPDH further catalyzes the decarboxylation of pPYR to acetyl-CoA with the generation of CO₂ and NADH. Although the overall equimolar conversion of malate into plastidic acetyl-CoA results in the production of valuable reductant necessary for FAS (NADH and NADPH), it leads, in return, to the loss of two carbons as CO₂, which may affect the overall CCE. Interestingly, ¹³C-labeling in developing B. napus embryos demonstrated for the first time that Rubisco was fixing this plastidic ¹³CO₂ to produce phosphoglycerate (pPGA) through an unconventional “Rubisco shunt” ( Table 2 ) (Schwender et al., 2004). The fixation of pCO₂ by Rubisco contributes to additional sources of carbon for FAS: it compensates for the decarboxylation steps, and channels more substrates into FAS, improving the CCE (Eastmond et al., 1996; Ruuska et al., 2004; Schwender et al., 2004). Indeed, studies in developing B. napus embryos showed a CCE of 86% under physiological conditions due to CO₂ refixation by Rubisco into pPGA ( Tables 1 , 2 ) (Schwender et al., 2004). ¹³C-MFA demonstrated that up to 64% of pPGA was produced by Rubisco, contributing to the synthesis of acetyl-CoA for de novo FAS, and reducing by 40% the carbon lost as CO₂ ( Table 2 ) (Schwender et al., 2004). Similar processes have been described in developing embryos of G. max and T. arvense where Rubisco was found to contribute to 14% and 25% of the pPGA, respectively ( Table 2 ) (Allen et al., 2009; Tsogtbaatar et al., 2020).

In addition to the role of Rubisco, ¹³C-labeling and MFA also revealed the non-canonical function of isocitrate dehydrogenase (IDH) in photosynthetic embryos ( Table 2 ). This reaction, assumed to be thermodynamically irreversible, was reported to catalyze the carboxylation of α-ketoglutarate into isocitrate in vivo in developing B. napus embryos (Schwender et al., 2006). It is important to note that the CO₂ fixation by IDH may also improve the CCE. This phenomenon was explained by a high demand in citrate for FA elongation in B. napus ( Figure 1 ), and a high concentration of CO₂ (40 mM) available in developing oilseeds (Goffman et al., 2004). Similarly, reversibility of IDH was measured in developing green embryos from G. max, even when the concentration of CO₂ was lower (Allen et al., 2009), T. arvense (Tsogtbaatar et al., 2020), C. sativa (Schwender et al., 2006), and L. usitatissinum (Acket et al., 2019), but absent in heterotrophic embryos ( Table 2 ). Interestingly, developing embryos that have an active Rubisco and reversible IDH were found to be the more efficient in converting their substrates into biomass ( Tables 1 , 2 ).

Besides carbon, FAS requires reductant that must be synthesized in the plastid. For each mole of acetyl-CoA produced, the plastidic pyruvate dehydrogenase complex generated one mole of NADH, which can be directly used for FAS ( Figure 1 ). The plastidic production of NADPH necessary for de novo FAS may be ensured by the OPPP and/or pNADP-ME, and light reactions in the case of photosynthetic embryos ( Figure 1 ). Knowing that the OPPP and the pNADP-ME generate CO₂, their operation may affect negatively the CCE ( Figure 1 ). ¹³C-labeling and MFA were used to measure the relative contribution of the OPPP and pNADP-ME to the production of NADPH in developing embryos from different species ( Table 2 ). In general, the OPPP supported a higher production of NADPH than the pNADP-ME. For G. max and B. napus, these two combined pathways did not provide sufficient NADPH to support FAS in developing embryos. In those species, the remainder of NADPH may be supplied by the light reactions of photosynthesis and/or catabolism (Goffman et al., 2005; Schwender et al., 2006; Allen et al., 2009). For Z. mays, the production of NADPH by the OPPP and pNAPD-ME was just enough to support FAS in developing embryos ( Table 2 ). More specifically, the pNAPD-ME was found to work at maximal capacity in vivo; it was identified as the limiting step in the provision of pPYR and NADPH for FAS in maize (Alonso et al., 2010; Cocuron et al., 2019). The embryos from the other species, H. annuus, C. sativa, and L. usitatissinum, were producing NADPH via the OPPP and pNADP-ME in excess of the requirements for FAS ( Table 2 ), concomitantly generating CO₂, which resulted in the lowest CCE for these species ( Table 1 ).

Overall, the aforementioned studies have demonstrated that MFA is a valuable tool to quantify the carbon partitioning in vivo, and identify the sources of carbon skeletons and reductants necessary for FAS in the developing green and non-green embryos. Information gathered from MFA, combined with transcriptomics and metabolomics, will give insights on the genes that control oil synthesis and create novel approaches for the genetic engineering of oilseed crops. The following sections review some of the target genes and the common genetic engineering strategies for enhancing oil content and quality in seeds.

This strategy aims to enhance the flux of carbon through FAS to produce additional acyl chains in the plastids for subsequent assembly into TGs in the ER ( Figure 2 ). The “push” strategy is considered to be one of the most extensively used to manipulate oil content. Commonly, over-expression (or even down-regulation) of transcription factors (TFs) offers a great advantage of controlling a number of reactions simultaneously in the oil biosynthetic pathway ( Table 3 ). The best examples of such TFs are LEAFY COTYLEDON1 (LEC1), LEC1-LIKE (L1L), LEC2, and WRINKLED1 (WRI1) (Focks and Benning, 1998; Lotan et al., 1998; Stone et al., 2001; Kwong et al., 2003; Cernac and Benning, 2004; Mu et al., 2008). Other TFs, shown to regulate the expression of genes involved in FAS, have been tested, including dehydration-responsive element-binding (DREB) (Zhang et al., 2016), Dof-type transcription factor 1 (Dof1) (Su et al., 2017), FUSCA3 (FUS3) (Elahi et al, 2015), growth-regulating factor 2-like (GRF2) (Liu et al., 2012), GLABRA2 (GL2) (Shi et al., 2012), and MYB interaction factor 1 (MIF1) (Cheng et al., 2021). Among all the TFs, GRF2, LEC1, WRI1 were shown to be the most efficient in increasing seed oil content (40% to 48% higher compared to wild-type) ( Table 3 ). In parallel, specific enzymes in the first committed steps of de novo FAS have also been targeted, such as acetyl-CoA carboxylase (ACCase) (Dong et al., 2002; Cui et al., 2017), and malonyl CoA-ACP malonyltransferase (MCAMT) (Jung et al., 2019) ( Table 4 ). Finally, several studies demonstrated that overexpression of membrane-intrinsic proteins that mediate the export of FAs from plastids, such as FA export 1 (FAX1) (Li et al., 2015b; Tian et al., 2018; Li et al., 2020b; Cai et al., 2021), FAX2 and FAX4 (Li et al., 2015b; Li et al., 2016a; Li et al., 2020b), BILE ACID : SODIUM SYMPORTER FAMILY PROTEIN 2 (BASS2) (Lee et al., 2017), and fatty acyl-ACP thioesterases (FAT) (Bonaventure et al., 2003; Sinha et al., 2007; Parveez et al., 2010; Belide et al., 2012; Nguyen et al., 2015a; Ozseyhan et al., 2018; Ma et al., 2021; Liu et al., 2022) could also boost lipid accumulation in seeds ( Table 4 ). Using the “push” strategy, the highest TG levels recorded in seed so far were achieved by over-expressing ACCase in maize (+65%) (Dong et al., 2002) ( Table 4 ).

The pull strategy aims to ensure efficient assembly of FAs generated in the plastid into TG at the ER ( Figure 2 ). Most of the metabolic engineering studies that attempted to optimize this step in seeds have focused on the over-expression of diacylglycerol acyltransferase (DGAT) (He et al., 2004; Kroon et al., 2006; Lung and Weselake, 2006; Shockey et al., 2006; Yu et al., 2006; Burgal et al., 2008; Misra et al., 2013; Wang et al., 2014; Liu et al., 2015; Savadi et al., 2015; Zhou et al., 2020; Chen et al., 2022) and phospholipid: diacylglycerol acyltransferase 1 (PDAT1) (Zhang et al., 2010; van Erp et al., 2011; Guan et al., 2015; Zhou et al., 2020) which commit acyl chains to storage TGs. Manipulation of other genes encoding for “pull” proteins, such as glycerol-3-phosphate dehydrogenases (GPDH), glycerol-3-phosphate acyltransferase (GPAT), and lysophosphatidic acid acyltransferase (LPAT), have also been reported to increase TG levels in seeds (Liu et al., 2015) ( Table 5 ). Using the “pull” strategy, the combined over-expression of DGAT, GPAT, and LPAT was shown to be the most efficient to increase total seed oil content (+25%) (Shockey et al., 2019) ( Table 5 ).

As TGs accumulate within the ER phospholipid bilayer, the outer layer starts to expand into the cytosol, forming nascent LDs that bud off to form LDs ( Figure 2 ). The “package” strategy aims to facilitate LD biogenesis and maximize droplet stability by proper coating. Proteomic analyses of seed LDs indicated that they contain 20–30 coat proteins which were found to be involved in LD formation and/or stability (Jolivet et al., 2004; Jolivet et al., 2009). These coat proteins might also help blocking the accessibility of the TG within the LD core to lipases allowing more LD formation. Recent evidence shows that seed-specific overexpression or down-regulation of some genes encoding these proteins increased LD size and/or number. Such proteins include lipid droplet-associated protein (LDAP) (Gidda et al., 2016), LDAP-interacting protein (LDIP) (Pyc et al., 2017), oleosin (Lu et al., 2006, Lu et al., 2018; Siloto et al., 2006; Shimada et al., 2008; Hu et al., 2009; Liu et al., 2013; Zhang et al., 2019), and seipins (Lunn et al., 2018) ( Table 6 ). Interestingly, several studies also used mammalian “package” genes, such as fat-specific protein 27 (FSP27) and fat storage-inducing transmembrane protein 2 (FIT2) from mouse in Arabidopsis. These resulted not only in improvement of seed oil content but also stimulation of LD clustering and fusion (Cai et al., 2017; Price et al., 2020) ( Table 6 ). Among all the mentioned “package” genes, over-expression of seipin in Arabidopsis gave the highest increase in seed oil content (+62%), which enhanced hydroxy-fatty acid (HFA) levels too (Lunn et al., 2018) ( Table 6 ).

Coat proteins can be hydrolyzed by endogenous proteases and TGs can be degraded into free FAs by various lipases to produce acetyl‐CoA via β‐oxidation particularly during seed germination and seedling growth (Pracharoenwattana and Smith, 2008; Li-Beisson et al., 2013; Borek et al., 2015) ( Figure 2 ). Blocking the breakdowns of coat proteins and TGs (also known as “protect” strategy) is therefore an attractive strategy to increase oil content in seeds. The initial step in lipase-induced TG mobilization is the ubiquitination of the respective coat proteins, particularly oleosin and caleosin, and then subsequent digestion by the proteasome (Sorokin et al., 2009; Hsiao and Tzen, 2011; Deruyffelaere et al., 2015), which was extensively reviewed (Sorokin et al., 2009; Shao et al., 2019; Guzha et al., 2023). Important insights into the mechanism regulating the turnover of oleosins in plants has been recently published (Deruyffelaere et al., 2018; Kretzschmar et al., 2018). Two key components, PUX10 (a member of the plant ubiquitin regulatory X (UBX)-domain containing protein family) and CDC48A (the AAA ATPase, Cell Division Cycle 48) were identified. PUX10 localizes to LDs, binds to the ubiquitinated oleosins, and interacts with ubiquitin and CDC48A, respectively. As an adaptor, PUX10 recruits CDC48A to ubiquitinated oleosins, leading to dislocation of oleosins from LDs via the segregase activity of CDC48A (Deruyffelaere et al., 2018; Kretzschmar et al., 2018). In Arabidopsis pux10 mutant seeds, PUX10 deficiency impaired the degradation of ubiquitinated oleosins from LDs. However, this did not change the total FA content in seeds of mutant genotypes in comparison to wild-type (Deruyffelaere et al., 2018; Kretzschmar et al., 2018) ( Table 7 ). On the other hand, suppression of TG degradation to enhance FA content was achieved by down-regulating the genes that code for TG breakdown enzymes. Such genes are seed fatty acid reducer (SFAR) (Karunarathna et al., 2020), Gly-Asp-Ser-Leu (GDSL)-motif lipases (Ding et al., 2019), plastid lipase1 (PLIP1), and Sugar-dependent1 (SDP1) (Kelly et al., 2013; Kim et al., 2014; Kanai et al., 2019) ( Table 7 ). Interestingly, recent studies reported that the over-expression of several genes with major role in TG degradation have a positive effect on seed oil content, which may seem counterintuitive. Such genes include patatin‐related phospholipase (pPLAIIIδ) (Li et al., 2013; Li et al., 2015a), and nonspecific phospholipase C6 (NPC6) (Cai et al., 2020) ( Table 7 ). An intriguing question arising from these studies is how the over-expressed proteins increased lipid accumulation in seeds. One hypothesis is that phospholipase-mediated phospholipid turnover facilitates the movement of FAs from the plastid to the ER. Phosphatidylcholine (PC) is the most abundant class of phospholipids in plants and plays pivotal roles in TG production (Wang, 2005; Chapman and Ohlrogge, 2012; Karki et al., 2019). When hydrolyzed, PC can serve as a substrate for FA desaturation and also provides free FAs or DAG for TG synthesis (Mhaske et al., 2005; Lu et al., 2009; Zhang et al., 2010; Pokotylo et al., 2013). Therefore, over-expression of these phospholipases could enhance the acyl flux into ER and increase the overall levels of TGs in seeds. Among the genes tested as part of the “protect” strategy, it was shown that the down-regulation of PLIP1 in Arabidopsis gave the highest increase in seed oil content (+45%) (Wang et al., 2017) ( Table 7 ).

Metabolic engineering strategies have been used to produce novel high-value FAs or to improve their synthesis in seeds of established crops and other model species ( Tables 3 - 7 ). For instance, over-expression and down-regulation of desaturases, such as acyl-acyl carrier protein desaturase (AAD), fatty acid desaturases (FADs), and stearoyl-acyl carrier protein desaturase (SAD), and FAT, and/or down-regulation of fatty acid elongase (FAE) genes were attempted to divert the carbon flux towards the synthesis of beneficial FAs required for human health and nutrition, such as gamma-linolenic acid (C18:3), alpha-linolenic acid (C18:3), docosadienoic acid (C22:2), docosatrienoic acid (C22:3) (Hong et al., 2002; Anai et al., 2003; Eckert et al., 2006; Huai et al., 2015; Yin et al., 2018; Meesapyodsuk et al., 2021), and oils enriched in omega-7 monounsaturated FAs such as palmitoleic acid (C16:1) and its elongation products vaccenic acid (C18:1) and paullinic acid (C20:1) (Nguyen et al., 2015; Ettaki et al., 2018; Rodríguez-Rodríguez et al., 2021) ( Tables 4 - 5 ). Also, several oilseed plants produce industrially-relevant FAs, such as erucic acid (C22:1), lesquerolic acid (C20:1-OH), and nervonic acid (C24:1), providing renewable alternatives to petrochemicals for the manufacture of lubricants, coatings, or polymers. However, most plants producing these FAs usually have undesirable traits and are not economically viable crops. It is hence important to improve the production of these valuable FAs in other established crops via metabolic engineering technology. One example is the improvement of the production of erucic and nervonic acids. VLCFAs, such as erucic acid (C22:1) and nervonic acid (C24:1), are important in plastic, cosmetic, nylon, and lubricant industries (Mastebroek and Marvin, 2000). Nervonic acid has only been found in the seed oils of a few known plants, and among these, only Lunaria annua has been considered as a niche crop for future development. However, this plant is a biennial with highly variable seed yields (800–2,000 kg/ha) and has a major problem of seed shattering, making it an uneconomical source of nervonic acid (Mastebroek and Marvin, 2000). Consequently, there is a high demanded to improve the production of this valuable FA in other oilseed crops via metabolic engineering technology (Taylor et al., 2009). Over-expression of 3-ketoacyl-CoA synthase (KCS) and FAE genes were shown to increase the production of erucic and nervonic acids in Brassica napus, Camelina sativa, and Arabidopsis thaliana (Katavic et al., 2000; Mietkiewska et al., 2004; Taylor et al., 2009; Huai et al., 2015) ( Table 5 ). Another example is the production of HFAs in a toxin free oilseed crop to replace castor oil as a renewable source for numerous industrial applications. This was achieved through over-expression of fatty acid hydroxylase and elongase genes, and also acyltransferase genes from species producing HFAs, such as castor bean (Ricinus communis) and lesquerella (Physaria fendleri), in the seeds of the model species Arabidopsis and in the industrial oilseed crop Camelina (Smith et al., 2003; van Erp et al., 2011; Bates et al., 2014; Snapp et al., 2014; Shockey et al., 2019) ( Table 5 ).

To further increase the TG levels in seed oil, several studies have combined genes from different strategies. It is important to note that this combinatorial strategy has been mostly used to increase TG content in non-seed organs, such as leaves and roots (Slocombe et al., 2009; Fan et al., 2013; Kelly et al., 2013; Vanhercke et al., 2013; Winichayakul et al., 2013), and were extensively reviewed (Cahoon et al., 2007; Taylor et al., 2011; Xu et al., 2018; Vanhercke et al., 2019; Singh et al., 2021). However, it was shown that seed-specific overexpression of AtWRI1 (“push”) and AtDGAT1 (“pull”) combined with suppression of the triacylglycerol lipase SUGAR- DEPENDENT1 (AtSDP1) (“protect”) resulted in a higher seed oil content than manipulation of each gene individually (van Erp et al., 2014a) ( Table 5 ). Similarly, seed specific over-expression of Seipin (“package”) in transgenic Arabidopsis expressing R. communis fatty acid hydroxylase (FAH) (“pull”) not only increased the total oil content but also increased the HFA composition in seeds (Lunn et al., 2018) ( Table 6 ). However, this combinatorial strategy is not guaranteed to succeed: several attempts did not increase TG in seeds, such as the over-expression of WRI1 (“push”) and DGAT1(“pull”) from Arabidopsis in soybean, suggesting that some species put in place counteracting processes to keep oil content stable (Arias et al., 2022) ( Table 5 ). On the other hand, combining multiple strategies has also been successful to modify FA composition in seeds ( Tables 4 - 5 ). For instance, co-expression of Cuphea viscosissima FATB (“Push”) and LPAT (“pull”) with Cuphea avigera DGAT (“pull”) increased the accumulation of TGs rich in medium-chain FAs (C6-C14) in pennycress for industrial, jet fuel and improved biodiesel applications (Esfahanian et al., 2021) ( Table 4 ). Another example is the co-expression of Δ6 and Δ5 desaturases (“pull”) and FAE (“pull”) from a fungal species (Mortierella alpinia) in soybean, which caused the seed to accumulate novel FAs that are not naturally produced in soybean: γ-linolenic acid (GLA), eicosa-8, 11-dienoic acid (EDA), dihomo-γ-linolenic acid (DGLA), and arachidonic acid (C20:4), which are important for human health (Chen et al., 2006) ( Table 5 ).