Critical metabolic pathways and SAD/FADs, WRI1s, and DGATs cooperate for high-oleic acid oil production in developing oil tea (Camellia oleifera) seeds

Abstract Oil tea trees produce high-quality edible oils with desirably high oleic acid (18:1) and low linoleic (18:2) and linolenic (18:3) fatty acid (FA) levels, but limited understanding of tea oil biosynthesis and regulation has become a significant obstacle for the breeding of high-yield and -quality oil tea varieties. By integrating metabolite and transcriptome analyses of developing oil tea seeds, we dissected the critical metabolic pathways, including glycolysis, fatty acid, and triacylglycerol (TAG) biosynthesis, as well as genes essential for tea seed oil production. Two plastidic stearoyl-acyl carrier protein desaturases (CoSAD1 and 2) and two endoplasmic reticulum-localized FA desaturases (CoFAD2 and 3) were functionally characterized as responsible for high 18:1 and low 18:2 and 18:3 proportions in tea oils. Two diacylglycerol O-acyltransferases (CoDGAT1 and 2) that may prefer to synthesize 18:1-TAG were functionally characterized and might be also important for high 18:1-TAG production. The highly expressed CoWRI1a and b were identified and characterized as activators of glycolysis and regulators of directing source carbon flux into FA biosynthesis in developing oil tea seeds. The upregulated CoSADs with downregulated CoFAD2 and CoFAD3 at the late seed developmental stages mainly accounted for high 18:1 levels. Two CoDGATs might be responsible for assembling TAGs with oleoyl acyl chains, whilst two CoWRI1s regulated carbons from parental sources, partitioning into oil production in oil tea embryo sinks. This study provides a deep understanding of the biosynthesis of tea seed oils and information on genes that may be used as molecular markers to breed oil tea varieties with higher oil yield and quality.


Introduction
Camellia oleifera Abel., commonly known as the oil tea or oil camellia, is a broadleaf perennial. It is a major woody oilseed tree mainly cultivated in the mountainous areas of southern China, and is also widely cultivated in other countries. The total oil tea tree plantation area is ∼5.3 million hectares, of which ∼3.1 million hectares are located in China [1]. Oil tea seeds are used for the extraction of edible oil, also called tea oil, which is composed of monounsaturated and polyunsaturated fatty acids (FAs) and minor saturated FAs [2]. Collectively, tea oil accounts for ∼25-40% of dry oil tea seeds, which is less than the proportion in other major oilseed crops, such as rapeseeds (40-50%), peanut (35-40%), or oil palm (30-50%) [3]. Therefore, there is still more room to increase the amount of oil in oil tea seeds. Moreover, ∼80% of tea oils is oleic acid (18:1) and ∼7% is linoleic acid (18:2), which gives what is regarded as the most suitable ratio of 18:1/18:2 for human health, comparable to that of olive oil. Oil tea seeds are also rich in saponins, vitamin E, polyphenols, phytosterols, and carotenoids [3][4][5][6][7]. Thus, tea oils are also used as herbal medicine for various health benefits, such as reducing serum triglycerides, improving the digestive system, reducing bad cholesterol, lowering blood pressure, increasing high-density lipoproteins, and strengthening the immune system [2,4,8]. Due to high levels of antioxidants, potential health benefits, and long shelf life, tea oil is recommended by the Food and Agriculture Organization (FAO) of the United Nations as a healthy vegetable oil [9,10]. In addition, tea oil is widely used in the cosmetics industry, to prepare margarines, lubricants, and rustproof oils because of its special physiochemical properties [2,3]. Therefore more tea oil production with enhanced components is the current demand of tea oil breeders and consumers as well [11,12].
Given the increasing market demand for high-quality tea oils and the huge vegetable oil supply deficit in China, the oil tea cultivation acreage has increased drastically in recent years. However, still only ∼3.4 million hectares of oil tea trees are cultivated in China, producing <2.3 million tons of tea seed oil annually, accounting for only 1.2% of the total consumption of edible oils in China [1,13]. The government of China has planned to double the oil tea tree planting area (6.8 million hectares) and increase the overall production of tea seed oil to 4.6 million tons to meet the challenge of increased demand for supplies of edible oils [1,13]. Thus, FA biosynthesis and triacylglycerol (TAG) production in developing oil tea seeds must be understood much better to facilitate the breeding of C. oleifera varieties with high oil yield.
Several studies on transcriptome profiling of developing oil tea seeds have been conducted [1,[14][15][16]. These preliminary transcriptome studies have generally described pathway gene transcripts and revealed the evolution of metabolic genes [1,14], by focusing on one or two metabolic genes without functional characterization [14], comparing the transcriptomes of high-and low-oil tea seeds to characterize differential gene expression [15], or combining proteomic and transcriptomic analyses to reveal key metabolic genes or enzymes for oil tea seed quality [16], as well as more recent genome sequencing of oil tea trees [17,18]. However, none of these studies has functionally characterized any of the key oil synthesis genes to support hypotheses. Many fundamental questions remain unanswered, such as what genetic factors primarily determine the diversity of oil yield or composition in oil tea seeds among the numerous available varieties. Despite the 2000year history of oil tea tree cultivation and tea seed oil utilization, extensive studies are needed to understand these quantitative traits and quality parameters of oil tea trees and tea seed oils. Here, metabolite profiling and transcriptome analysis of developing oil tea seeds at various developmental stages were performed to identify critical metabolic pathways and genes for the production of high-oleic-acid tea seed oils. We functionally characterized several genes involved in metabolic pathways from glycolysis, FA synthesis, to TAG assembly, including two stearoyl-acyl carrier protein (ACP) desaturases (CoSAD1 and 2), two FA desaturases (CoFAD2 and 3), and two diacylglycerol acyltransferases (CoDGAT1 and 2). Two oil tea WINKLED1 homologs (CoWRIs) were also characterized as oil biosynthesis regulators for oil production. Our study further employed transcriptome and FA profiling of developing seeds from more than 14 local C. oleifera varieties and verified the close correlation between the oil compositions and contents in C. oleifera seeds with expression levels of these genes.

Changes in storage substances in oil tea seeds at various development stages
We profiled and measured mature C. oleifera seed oil in comparison with other three common edible oils: peanut oil, rapeseed oil, and olive oil. Tea seed oil consisted of 6% palmitic acid (16:0), 1% stearic acid (18:0), 79% 18:1, 7% 18:2, and 0.3% linolenic acid (18:3). Thus, the FA composition of tea seed oil is similar to that of olive oil, but has much fewer polyunsaturated FAs (18:2 and 18:3) than rapeseed and soybean oils [14]. In total, the dry mass of the mature seeds of the 'Changlin #4' cultivar contains up to 35% oil (Fig. 1). To understand how these major valuable nutrients and basic energy substances, such as proteins and carbohydrates, are synthesized during the seed developmental stages, we first measured different types of nutrient substances in both seeds and fruit shells of a local oil tea tree cultivar (C. oleifera 'Changlin #4') at six developmental stages (Fig. 1A). The total FA contents in oil tea seeds at six development stages changed significantly from 2.2 at Stage 1 to 155.5 mg/g at Stage 6, and we observed a slight increase in the early period (Stage 1-Stage 4) and a considerable increase from Stage 5 to Stage 6 ( Fig. 1B). We sought to understand how carbohydrates derived from leaf photosynthesis and transported via phloem into the seeds are dynamically metabolized and partitioned into oils, proteins, and starches in oil tea seeds. For this, the accumulation patterns of seed storage substances such as starches, soluble sugars, and proteins at six developmental stages of oil tea seeds and fruit shells were measured (Fig. 1C). In developing seeds, the starch content decreased from 31.9 to 9.4 mg/g from Stage 1 to Stage 2 and then increased to 27.0 mg/g at Stage 6, and soluble sugar content increased from 117.7 to 166.7 mg/g from Stage 1 to Stage2 and then decreased to 67.8 mg/g at Stage 6 during the ripening process, suggesting partitioning of these carbohydrates into FAs and proteins in seed development (Fig. 1C). The protein contents of developing oil tea seeds decreased at Stage 4 and then increased to 2.9 mg/g at Stage 6 ( Fig. 1D). The increasing oil contents suggested a primary partitioning of these carbohydrates into the FA synthesis process during seed development. In fruit shells, soluble starch was slightly decreased. Soluble sugar continuously increased and was maintained at levels from 16.7 to 55.9 mg/g during the ripening process. Although the proteins of fruit shells remained similar at all developmental stages, the total FA contents in fruit shells were relatively low and decreased continuously in a manner opposite to the trend observed in the developing seeds (Fig. 1).

De novo assembly and annotation of the developing oil tea seed transcriptome
To dissect the biosynthesis of these seed storage substances, an RNA-Seq analysis of oil tea seeds of C. oleifera cv. 'Changlin #4' was conducted. RNA-Seq generated a range of 6-8 G clean reads for each sample, which were assembled and annotated into 104 287 transcripts assigned to biological processes, 97 270 transcripts assigned to cellular components, and 52 691 transcripts assigned to molecular functions (Supplementary Data Fig. S1). Approximately 29.25% of the entire homologous sequences had significant matches with genes from Vitis vinifera, followed by Coffea canephora Figure 1. Analyses of seed storage metabolites in developing oil tea seeds. (A) Oil tea fruits at various developmental stages (first six from left) and dry fruit (rightmost). The cultivar 'Changlin #4' is used as a representative to show developmental stages. The weight of fruit is given below each fruit. (B) FA contents in developing oil tea seeds (left panel) and fruit shells (right panel). (C) Starch and sugar contents in developing oil tea seeds (left panel) and fruit shells (right panel). (D) Protein contents in developing oil tea seeds (left) and fruit shells (right). (E) Appearance of seeds (left panel) and fruit shells (right panel) at various developmental stages. Data represent three independent experiments and are shown as means ± standard deviation (n = 3). Variance between numbers at various developmental stages was analyzed.

Changes in glycolysis pathway genes in developing oil tea seeds
Starch is an important storage form of carbohydrates in many seeds. Although starch accounted for a lower portion of storage substances compared with TAGs, the synthesis and degradation of starch in oil tea seeds is closely related to sucrose metabolism and carbon partitioning from the sources [19,20]. The glycolysis pathway is the major pathway involved in the oxidation of glucose to pyruvate for either FA or amino acid synthesis in all developing seeds. Using the Kyoto Encyclopedia of Genes and Genomes (KEGG), these metabolic pathways and corresponding metabolic genes qRT-PCR verification of expression patterns of the major genes involved in starch and sucrose synthesis and glycolysis pathways. All qRT-PCR data are expressed as means ± standard deviation (n = 3) from three independent experiments. STS, starch synthase; SUS, sucrose synthase; INVs, invertases; HXK, hexokinase; PGI, glucose phosphate isomerase; PPi-PFK, phosphofructokinase; ALDO, fructose-bisphosphate aldolase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; G3PDH, glycerol 3-phosphate dehydrogenase; GK, phosphoglycerate kinase; PGM, 2,3-bisphosphoglycerate-independent phosphoglycerate mutase; ENO, enolase; PK, pyruvate kinase; Amy, amylase; TPI, triose phosphate isomerase; SPS, sucrose phosphate synthase.

Expression of fatty acid and triacylglycerol synthesis genes in developing oil seeds
The major FA in oil tea seeds is 18:1, and the 18:1 FA contents exhibited the greatest increase at all oil tea seed developmental stages. The major increase in oils occurred at Stages 5 and 6. The proportions of 16:0, 18:0; 18:2, and 18:3 FAs increased (Fig. 3A). Stages 4-6 represent the seed developmental period when late seed development and seed filling actively occurred (Fig. 3A). During seed development, 18:1 TAGs increased continuously, whereas the proportions of 16:0, 18:0, 18:2, and 18:3 TAGs were reduced gradually (Fig. 3B). Regarding FA synthesis in the chloroplast, acetyl CoA carboxylase (ACCase) is the rate-limiting enzyme and converts acetyl-CoA to malonyl-CoA in the first step [26]. Among 49 ACCase transcripts that exhibited significantly different expression patterns at six developmental stages, CL1061.Contig4, CL5954.Contig3, and Unigene6929 were highly expressed ACCase genes ( Acyl-acyl carrier protein thioesterase (FATB) also plays an essential role in FA biosynthesis. As one of the major FATB genes in seeds, CL53.Contig4 (Unigene16664) was significantly upregulated at early developmental Stages 1 and 2 and decreased later on. Other acyl-ACP thioesterase (FAT) genes, such as CL7284.Contig2, also displayed the same expression patterns in developing seeds ( Fig. 3C; Supplementary Data Fig. S4; Supplementary Dataset S1).
The biosynthesis of photosynthetic-specific galactolipids in the chloroplast, including monogalactosyldiacylglycerol (MGDG), digalactosyldiacylglycerol (DGDG), and sulfoquinovosyldiacylglycerol (SQDG), essentially required MGDG synthase (MGD) and DGDG synthase (DGD) (Supplementary Data Fig. S5). [11] Highly expressed transcripts of these galactolipids linked with TAG production were also found in this study [27] (Fig. 3C). G3P is transported from the cytosol into the endoplasmic reticulum (ER) and acts as an acceptor of fatty acyl-CoAs for glycerolipid synthesis, and FAs are transported into the ER by ABCA transporters potentially with the help of long-chain acyl-CoA synthetases (LACSs). [12] Positioned at the cross point in the metabolic pathways involved in the synthesis of both phospholipids and TAGs,  S12). PA and DAG are also central metabolites for phospholipid synthesis, which occurs in parallel with TAG biosynthesis in the ER. Both PDAT and DGAT transfer the acyl group from phospholipids and acyl-CoAs, respectively, to DAG as a substrate to synthesize TAG. These genes were highly expressed in developing oil tea seeds; for example, the expression levels of CoDGAT1 (CL21487.Contig7) were increased by ∼2-fold at the late developmental Stages 5-6. CoDGAT2-1 (CL1666.Contig2) displayed increased expression levels at Stages 1-4, and then levels decreased. CoDGAT2-2 (Unigene7038) showed the highest expression at Stages 5-6. By comparison, CoPDAT1 (Unigene18970) and CoPDAT2 (Unigene6465) were expressed at lower levels in developing oil tea seeds compared with CoDGATs ( Fig. 3D; Supplementary Data Fig. S12). Phospholipase D (PLD) and A (PLA) genes were also highly expressed in the developing seeds ( Fig. 3D; Supplementary Data Figs S6-S10), and these PLDs modify phospholipids and seed storage TAGs and DAGs by affecting PA intermediate pools and acyl editing [27][28][29]. TAG can be successively hydrolyzed to FAs and glycerol by various lipases, such as those encoded by TAGLs (CL22819.Contig2, CL4510.Contig1, Unigene27904, and CL806.Contig7), DAGLs (CL11004.Contig2, CL14845.Contig3, and CL4510.Contig1), and MAGLs (CL14089.Contig2) ( Fig. 3D; Supplementary Data Figs S11-S13).
Several transporter genes, such as FA transporters (FAXs), LACSs, ATP-binding cassette type G transporter (ABCGs), and triglactosyldiacylglycerol transporters (TGDs) associated either with lipid precursor inter-organelle transport or their extracellular movement, were found to be highly expressed in the current study, indicating active FA movement and exchange among the chloroplast, cytosol, and ER. The genes involved in the biosynthesis of very long chain fatty acids (VLCFAs) and waxes, including the genes encoding ketoacyl-CoA synthase (KCS), hydroxyacyl-CoA dehydrase (HAD), ketoacyl-CoA reductase (KCR), enoyl-CoA reductase (ECR), and wax ester synthase (WSD), were found to be highly expressed in developing seeds ( Fig. 3D; Supplementary Data Figs S14-S17). In plants TAG is stored in special oil bodies surrounded by specific membrane proteins, including oleosins (OLEs), calcium-binding caleosins (CLOs), and sterol-binding dehydrogenases [steroleosins (SLOs)]. These proteins are abundant in seeds and significantly affect the assembly and stability of oil bodies and oil yield [11]. Based on RNA-Seq data of developing oil tea seeds, more than 30 highly expressed transcripts encoded OLEs, such as CL13885.Contig1, Unigene38830, Unigene25677, CL11045.Contig2, CL14158.Contig1, and Unigene13140 ( Fig. 3D; Supplementary Data Fig. S14, Supplementary Dataset S2). The expression levels of these genes were significantly upregulated during seed maturation; however, CLO-coding genes Uni-gene35942, CL16257.Contig2, and Unigene44817 were present at low levels. The ER-lipid droplet protein seipin is a homo-oligomeric integral membrane protein in the ER that concentrates at junctions with cytoplasmic lipid droplets ( Fig. 3D

Genes involved in high oleic acid accumulation in developing oil tea seeds
Generally, the high proportion of 18:1 in total FAs of oilseeds is mainly due to high activity of stearoyl-ACP desaturase (SAD) in the chloroplast and/or rather lower FA desaturase (FAD) activity in the ER [11,30]. The CoSAD genes include major transcripts CL23116.Contig1, CL9407.Contig 1, and CL17700.Contig2, which are named CoSAD1, CoSAD2, and CoSAD3, respectively.
The CoSAD sequences showed resemblance to SADs from various other oilseed species (Fig. 4). The transcripts of all CoSADs were found in abundance at all stages, hence a greater amount of 18:1 at each developing stage can be correlated with high expression of CoSADs. When CoSAD1-green fluorescence protein (GFP) fusion construct was infiltrated into tobacco epidermal cells, the CoSAD-GFP fusion signals were co-localized with chlorophyll autofluorescence, indicating their chloroplast localization (Fig. 4). To determine whether these enzymes are functionally active, we then overexpressed CoSAD1 and CoSAD2 in yeast cells that accumulated high levels of 16:0 and 18:0 FAs. We observed that increased levels of 16:1 and 18:1 FAs in CoSAD1/ 2-expressing yeast cells compared with the vector control, indicating that CoSADs indeed catalyzed 18:1 formation from 18:0-ACP or even non-specifically catalyzed 16:1 formation from 16:0-ACP (Fig. 4). The chloroplastderived FAs were transported into the cytosol by FAXs, which constitute a large gene family in C. oleifera. In addition, chloroplast-specific CoFADs, including CoFAD6, 7, and 8, convert galactolipids with the C18:1 acyl chain into the corresponding galactolipids with C18:2 or C18:3 acyl chains. [31] The ER-specific CoFADs, including FAD2 and 3, catalyze the desaturation of the fatty acyl chains in phosphatidylcholines (PCs), and their genes were also detected in the transcriptome of the developing oil tea seeds. All plastidic or ER-located CoFADs were clustered with their homologs from other plants (Fig. 5A) and may exhibit different functions of modification of fatty acyl chains in PCs and membrane glycerolipids.
The transcripts of two highly expressed genes in the developing seeds, namely CoFAD2 (CL17748.Contig2) and CoFAD3 (CL1435.Contig 3), were decreased by 5-fold and 10-fold, respectively, from Stage 2 to Stage 6 (Fig. 5B). These CoFAD gene expression patterns were consistent with the significant decreases in C18:2 and C18:3 FA levels in the oil tea seeds over different developmental stages (Fig. 5C). Other plastidic desaturases, including FAD6, 7, and 8, and ER-localized FAD2 or FAD3, encoded by CL5844.Contig2, CL18882.Contig3, CL9324.Contig2, and Unigene10772, displayed increasing trends throughout seed development, and these genes may be related to the synthesis of membrane phospholipids or galactolipids (Fig. 5D). We also predicted CoFAD3 localization in the ER membrane ( Supplementary Data Fig. S17). When we transiently expressed CoFAD2 in tobacco leaves and examined the lipid profile, increased 18:2 FA contents were observed (Fig. 5E). When CoFAD3 was transiently expressed in tobacco leaves, we observed greater 18:2 and 18:3 FA contents in the leaves compared with the vector control. These findings suggested that CoFAD2 and 3 are functionally active (Fig. 5E).

Functional characterization of CoDGATs in triacylglycerol biosynthesis
The deep analysis of the transcriptome of developing oil tea seeds showed four CoDGAT genes whose encoded proteins were clustered with soybean, olive, and Arabidopsis DGAT1, 2 and 3 (Fig. 6A). CL21487.Contig7, CL1666.Contig2, Unigene7038, and Unigene22423 were named CoDGAT1, CoDGAT2-1, CoDGAT2-2, and CoDGAT3, respectively (Fig. 6B). The highly expressed CoDGAT1 and CoDGAT2-1 (named CoDGAT2 hereafter) in developing seeds were further functionally characterized. The expression patterns of CoDGAT1 and CoDGAT2 at six development stages were examined via qRT-PCR, and the results were in accordance with the RNA-Seq outcomes. Both CoDGAT genes exhibited the highest expression at Stages 5 and 6, which is consistent with the stages of the highest oil contents (Fig. 6C). We next characterized the in vitro functions of CoDGATs in yeast cells. CoDGAT1 and CoDGAT2 were expressed in the TAG-deficient quadruple mutant yeast (Saccharomyces cerevisiae) strain H1246 individually, and large oil bodies were observed in CoDGAT1-and CoDGAT2-expressing H1246 cells, while only minimal oil bodies were found in control H1246 cells expressing empty vector after Nile Red staining (Fig. 7A). Separation of TAGs from the total lipid extracts in these yeast mutant cells by thin-layer chromatography (TLC) confirmed increased TAG levels in both CoDGAT1-and CoDGAT2-expressing H1246 cells compared with H1246 cells expressing empty vector ( Fig. 7B and C). Analysis of the FA composition of TAGs separated by TLC revealed that CoDGAT1 and CoDGAT2 expression facilitates the biosynthesis of TAGs with 16:1 and 18:1 fatty acyl chains but decreased    (Fig. 7D). Moreover, the functions of CoDGAT were further analyzed by the tobacco infiltration technique. According to TLC analysis of the infiltrated tobacco leaves, CoDGAT1 increased the TAG amount by 18% compared with control leaves (Fig. 7E). FA analyses showed a significant increase in the 18:1 proportion but a slight decrease in the 18:0 proportion in total TAGs of CoDGAT1-overexpressing leaves compared with these of the P19 control (Fig. 7F). We also examined the subcellular localization of a type I CoDGAT1. GFP-CoDGAT1 fusion driven by a cauliflower mosaic virus 35S promoter was transiently expressed in epidermis of tobacco leaves. Confocal microscopy showed that GFP-CoDGAT1 signals were primarily localized in the ER, and this finding was further supported by chloroplast autofluorescence (Fig. 7G). Due to the high similarity in protein sequences, we posited that both CoDGAT1 and 2 were primarily associated with the ER. Since TAG is synthesized and assembled in the ER, topological analysis of the CoDGAT1 protein sequence and its predicted localization in the ER are consistent with its function as a transmembrane protein (Supplementary Data Fig. S18).

Regulation of carbon partitioning into oil production by CoWRI1s in oil tea seeds
Oil biosynthesis in oilseed crops can be regulated at transcription level with regard to source-to-sink carbon partitioning into oils, proteins, or starches as seed storage substances [32,33]. WRI1 is the most important AP2 transcription factor regulating source carbohydrate glycolysis and carbon partitioning into plastid fatty acid biosynthesis, or even further to TAG production in the ER [32,33]. Phylogenetic analysis enabled us to identify two AtWRI homolog genes, CoWRI1a and b, both of which were more specifically expressed in fruit harboring the developing seeds among seven examined tissues and organs, with CoWRI1a as the major one as indicated by transcriptome data (Fig. 8A and B). qRT-PCR also verified that both CoWRI1a and CoWRI1b were highly expressed at Stages 3 and 4 of developing seeds when seed filling started, and actively participated in seed filling (Fig. 8C and D). Their expression patterns are highly consistent with late glycolysis genes (ALDO, G3PDH, G3PDH, PGK, ENO, and PK) and most FA synthesis genes (Fig. 3). To verify the functions of CoWRI1a and b, we further used tobacco leaves overexpressing both CoWRI1a or CoWRI1b with CoDGAT1. Overexpression of CoWRI1a or b or CoDGAT1 alone could clearly increase oil production in tobacco leaves (P < .05) ( Fig. 8E and F). However, when CoWRI1a or b was co-transformed and coexpressed together with CoDGAT1 in tobacco leaves, a significant increase in oil production was observed ( Fig. 8E and F). The additive or synergistic effects of CoWRI1a or CoWRI1b with CoDGAT1 indicated that they are metabolically connected but have different roles in oil production. As in most studies, overexpression of CoWRI1a or CoWRI1b enabled plants to produce more fatty acids and G3Ps for DGAT to synthesize TAGs.

Correlation between oil production with CoSAD/FADs, CoDGATs, CoWRI1s in 14 oil tea varieties
To verify the importance of these CoSADs, CoFADs, and CoDGATs in oil FA composition and production, Stage 5 seeds from 14 local oil tea trees were collected for an association study. These 14 local oil tea varieties were collected at the same time for seed oil and RNA analyses. The seed oil content and FA composition analyses of 14 oil tea varieties showed that the contents varied significantly but were largely consistent with the expression levels of CoSAD1 and CoSAD2, CoFAD2 and CoFAD3, and CoDGAT1 and CoDGAT2 in their seeds (Fig. 9). A close positive correlation between the 18:1 contents and expression levels of CoSAD1 and 2 was observed. In addition, a positive correlation was noted between 18:2 and 18:3 contents and CoFAD2 and 3 expression levels (Pearson correlation coefficient, P > .7) (Fig. 9). Furthermore, the total oil contents and expression levels of CoDGAT1 and 2 were clearly and positively correlated (Pearson correlation coefficient, P > .7) (Fig. 9).

Discussion
To meet the fast-increasing demand for edible oils from the growing population in China, expanding planting acreage, applying new breeding technology, and investing in basic research on oilseed crops, such as oil tea trees, have started to harvest fruits, such as the recent publication of oil tea tree genome sequences [17,18]. Still, how to increase oil productivity and meanwhile to improve nutritional quality of edible oils are the major concerns of the Chinese government for national food security. To gain an in-depth understanding and genetic improvement of oil tea trees in terms of oil yield and quality, this study integrated metabolite profiling, transcriptome analysis, and functional characterization to decipher the critical subcellular pathways and key metabolic and regulatory genes. The dissection of expression patterns of genes involved in glycolysis in the cytoplasm, FA biosynthesis in the chloroplast, and TAG synthesis and assembly in the ER of developing oil tea seeds unveiled different features of oil tea seeds, as compared with other wellknown oilseed crops, such as oil palm (Elaeis guineensis) and olive (Olea europaea) trees. Furthermore, we functionally characterized two major plastidic CoSADs, ER CoFAD2 and CoFAD3, and two ER CoDGATs for their coordinative up-or downregulation for cooperation in high 18:1 oil production in oil tea seeds. Two major CoWRI1s that could activate glycolysis and fatty acid biosynthesis in developing oil tea seeds were also functionally characterized.

Dynamic changes of seed storage substances and differential expression of glycolysis genes in developing seeds
Quantification of soluble sugars and starch indicated that soluble starch was maintained at relatively low levels in developing oil tea seeds, as most of the starch was converted into soluble sugars. A considerable number of soluble sugars was partitioned through glycolysis into oil biosynthesis in seeds of oilseed plants [32]. With the increasing contents of soluble proteins in small portions, oil accumulates continuously at seed filling stages and accounts for a significant portion of seed storage substances. While the total oil contents increased markedly throughout oil tea seed development, we observed significant increases in the 18:1 proportion at Stages 2-6 but reduced 18:2 and 18:3 levels in developing oil tea seeds. These findings inspired us to decipher the dynamic changes in these developing oil tea seeds using RNA-Seq analysis. Glycolysis is the central pathway of carbon metabolism in oilseeds because it converts source sugars into the precursors for TAG and protein biosynthesis at sites where ATP is also produced. The glycolytic pathway is the primary carbon source and produces power for lipid synthesis. In developing oil tea seeds, ALDO and GAPDH were each represented by at least three expressed transcripts, and these transcripts were highly expressed in developing seeds from the young to maturation stages, suggesting the importance of these two genes in oil production, similar to those in oil palm [34]. Increased ALDO levels combined with reduced TPI and GAPDH levels are potentially indicative of changes in carbon flux equilibrium in glycolysis, and amino acid metabolism also seemed to be tightly linked with oil production [34].
Reduction in TPI activity in a yeast knockdown mutant resulted in a 19% increase in lipid content. In contrast, yeast strains overexpressing oil palm ALDO and G3PDH showed 16 and 21%, more lipid content, respectively, compared with GAPDH [25,35]. Thus, highly expressed ALDO, TPI, and G3PDH genes are also required for oil production in oil tea seeds.
Some plant SUSs, such as CoSUS1, are involved in starch degradation for glycolysis and carbon flux into FA and amino acid synthesis pathways, whereas others, such as CoSUS2, are involved in sucrose synthesis. Two SUS genes, Unigene41968 (CoSUS1) and CL19004.Contig1 (CoSUS2), were highly expressed at the early seed developmental stages, where they displayed opposite patterns. We posited that the sucrose transported to oil tea seeds is mainly hydrolyzed via CoSUS1 rather than INV and channeled into glycolysis as a carbon source for lipid biosynthesis. The expression of an INV (CL170.Contig4) and amylase (CL3151.Contig4) was continuously increased over the seed developmental stage until maturation, potentially indicating a vital role of these genes in the maintenance of the low level of starch in oil tea seeds, thus ensuring major carbohydrate flux for oil synthesis and storage proteins in the seeds [21][22][23]. This finding is consistent with the downregulation of SUS2 in high-oil-palm lines compared with low-oil palms at late developmental stages [36]. The reduced CoSUS2 expression levels may lead to more lipid accumulation due to carbon channeling away from starch but towards oil biosynthesis. The downstream glycolysis pathway, with only one isoform of each of the PGK, PGM, ENO, and PK genes, was highly expressed; they may be also positively regulated by CoWRI1s, since WRI1 is the principal regulator activating late glycolysis and early FA synthesis in oilseed plants [12,32]. An increased quantity of the carbon flux channeled into FA synthesis via WRI-dependent upregulation of multiple source genes ensured the supply of precursors for increased oil production [12]. Indeed, the expression patterns of CoWRI1a and b in developing oil tea seeds were consistent with those of late glycolysis genes.

Active plastidic fatty acid and endoplasmic reticulum triacylglycerol biosynthesis in developing oil tea seeds
The plastidic FA synthesis pathway is a key pathway for oil production in oilseed plants. This pathway is vital for these high-oleic-acid oilseeds, given that oleic acid is mainly synthesized in the plastids before being transported into the ER and incorporated into TAGs. Most genes involved in this FA synthesis pathway were identified and differentially expressed during oil tea seed development. These genes included CoKAS I, CoKAS III, CoKAR, and ACCase. ACCase consists of three nuclearencoded subunits (biotin carboxyl carrier protein, biotin carboxylase, and carboxyl transferase α-subunit, and a plastid-encoded carboxyl transferase β-subunit) and catalyzes the first committed formation of malonyl-CoAs for FA synthesis [11]. Clearly, the transcript levels of multiple ACCases increased during seed development, indicating that these genes are critical for oil production. The oil content is increased 5-fold in plants via overexpression of ACCase in the plastid and ACCase expression levels are tightly linked with oil production in many oilseed plants [11,37]. At least one set of highly expressed FA synthetic genes were highly expressed in developing oil tea seeds: MAT, KASIII, KASI, KASII, SAD, FatA, FatB, and SAD in chloroplasts, and the galactolipid synthesis genes MGDs and DGDs in chloroplasts. Some of these early FA synthesis genes may be targeted by CoWRI1s [23]. GPATs, LPAATs, PAPs, PDATs, DGATs, and FADs as well as oil body protein genes OLEs, CLOs, and SEIPINs, and TAG lipase genes DAGL, TAGL, and MAGL, in the ER can be regulated by other transcription factors, such as LEC1 and 2 and FUS3 [33]. Surprisingly, at least 10 OLE gene transcripts were identified in the transcriptome, and most of them were highly expressed in oil tea seeds.

Differentially expressed SADs/FADs for high-oleic-acid tea oil production
The hallmark of oil tea seeds with high-quality edible oil is an extremely high content of 18:1, lower levels of saturated FAs (16:0 and 18:0), and ideally high levels of 18:2 and low levels of 18:3. In light of the current study it can be suggested that high plastidic CoSADs activity and reduced transcript levels of CoFAD2 and CoFAD3 in the ER during oil tea seed development likely account for the accumulation of a high 18:1 level.
In the plastid, stearoyl-ACP is desaturated to oleoyl-ACP (18:1-ACP) by SADs. Overexpression of SADs from oilseed plants resulted in high levels of 18:1. By contrast, mutation of SADs repressed 18:1 production in total oils from seeds or other organs of the plants [11]. Overexpression of CoSADs in yeast cells also confirmed the functions of these two highly expressed CoSADs in developing oil tea seeds. Here, 18:1 is exported from plastids to the cytosol, transported into the ER and incorporated into PA, DAG, and phospholipids. Phospholipids, such as PC and phosphatidylethanolamine (PE), are further desaturated to 18:2-or 18:3-phospholipids by the action of FAD2 and FAD3, which are transmembrane proteins that exert their enzymatic activity on these membrane glycerolipid substrates [11,38,39]. Overexpression of CoFAD3 in tobacco leaves also induced increases in 18:3 accumulation. Mutation of FAD2 and FAD3 in Arabidopsis, soybean, and rapeseeds resulted in significantly increased oleic acid levels in oilseeds [30,39,40]. We demonstrated that several highly expressed CoSADs in plastids enhanced 18:1 biosynthesis, whereas rapidly reduced expression levels of CoFAD2 and CoFAD3 genes in the ER suppressed the conversions of 18:1-PC/PEs into 18:2-or 18:3-phospholipids. The low proportion of 18:2 and even lower 18:3 content in tea seed oil could be caused by a reduced ratio of CoFAD2/CoFAD3 transcript levels, which is worthy of further investigation. The analysis of seed oil composition of 14 local oil tea tree varieties and transcript levels of CoSAD1 and 2 as well as CoFAD2 and 3 also suggested a close positive relation between 18:1 content and CoSAD1 and 2 expression levels, as well as between 18:2 and 18:3 contents and CoFAD2 and 3 expression levels (Fig. 9). These findings are similar to those noted in the olive tree, which produces oil with high oleic acid. In olive trees, SAD2 is highly expressed in the mesocarp and seed, and the major FAD2 and FAD3 genes exhibit low and decreasing expression levels throughout seed development. These characteristics are believed to be responsible for the composition of olive oil [41,42].

Triacylglycerol assembly via DGATs and weak acyl editing on triacylglycerols in oil tea seeds
The renewed synthesis and incorporation of new acyl-CoAs into PC occurs via PLA2 and acyl-CoA:lyso-PC acyltransferase (LPCAT), which play important roles in acyl editing on PCs or phospholipids. Generally, the sn-1 position of PC is esterified with saturated or monounsaturated FA, and sn-2 is esterified with polyunsaturated FAs (PUFAs). Acyl editing on PC is not only fulfilled through FAD2 and FAD3 but also achieved through deacylation and reacylation modification, which involves PLA2-LPCAT cycles. The PLA2-LPCAT cycle is very active in oilseeds containing high levels of PUFAs in TAGs, given that LPCATs prefer to use 18:2-or 18:3-acyl CoA for PC synthesis [11,40]. LPCAT transfers the newly synthesized acyl-CoAs from the plastids, mostly 18:1 by highly active SDAs, into the lyso-PCs generated by PLA2 to generate PCs with both oleoyl chains. However, in developing oil tea seeds, both PLA2 and LPCAT transcripts were highly expressed only at the early developmental stages (Stages 1 and 2) but repressed at the late developmental stages, when seed filling occurs.
TAGs assembled in the ER are mainly formed via two mechanisms: using fatty acyl-CoAs as acyl donors and DAGs as acyl acceptors via the functions of DGATs or using PCs as acyl donors and DAGs as acceptors via the functions of PDATs [11,19,[43][44][45][46]. Several highly expressed DGATs in developing oil tea seeds, including CoDGAT1 and CoDGAT2, may trigger increased 18:1 production. Both types of CoDGAT may prefer oleoyl CoA over other fatty acyl-CoAs as a substrate for TAG synthesis, which needs to be further verified. The expression levels of CoDGAT1 and 2 in seeds of 14 local oil tea varieties were clearly correlated with their total oil contents ( Fig. 9).
Extensive acyl editing on plant DAG and phospholipids via phosphatidylcholine diacylglycerol cholinephosphotransferase (PDCT) also enable the incorporation of more polyunsaturated FAs into TAGs through PDAT [11]. CDP-choline:DAG cholinephosphotransferase (DAG-CPT) transfers choline from CDP-choline to DAG to generate new PC to recycle DAG into PC; DAG-CPT-PDCT-PDAT forms an important PC-DAG exchange/conversion cycle to enforce the acyl editing on TAGs [11,45,47]. The reduced CoPDCT activity could be another reason for the increased 18:1 proportion in TAG in addition to low CoFAD2/CoFAD3 expression at late seed developmental stages (Supplementary Data Fig. S19). In developing oil tea seeds, CoPDCT was downregulated in general; thus, its activity of converting DAG back to the PC pool was reduced, thereby reducing the chance of desaturation of 18:1 on PC (Supplementary Data Fig. S19). The reduced number of PDCT, DAG-CPT, and PDAT transcripts at low expression levels over oil tea seed development indicated that PDCT-mediated acyl editing of TAG may play a minor role in 18:1 accumulation (Supplementary Data Fig. S19). The Arabidopsis PDCT mutant contains higher levels of oleic acid but reduced 18:2 and 18:3 levels [12,47]. Furthermore, compared with highly expressed CoDGATs, CoPDATs that encode the enzymes utilizing PC and DAG as substrates to create TAG were expressed at much lower levels and may contribute less to TAG production in oil tea seeds. By contrast, rapeseed, soybean, and flax seeds contain higher levels of PUFAs, such as 18:2 and 18:3; PDATs are regarded as major contributors to the biosynthesis of TAGs [11,45,48,49]. In general, the higher expression of DGATs and the lower expression of PDATs are likely responsible for the TAGs containing monounsaturated fatty acids [11,43,45]. Thus, the lower ratio of PDAT/DGAT gene expression in developing oil tea seeds may also determine the increased 18:1 percentage in TAGs. Ectopic expression of two CoWRI1s, individually or together with CoDGATs, which are assumed not to be directly regulated by WRI1, further confirms that CoDGATs are metabolically linked with CoWRI1activated genes/enzymes for precursors.

Comparison of mechanisms underlying high-oleate-oil biosynthesis in oilseeds
The genome sequence of high-oleate oil-producing olive revealed that the duplication and differential expression of OeFAD2 and OeSAD are responsible for the differential accumulation of 18:1 and 18:2 FA in olive seeds [50]. Several OeFAD2 genes in olive seeds are suppressed by an siRNA, resulting in reduced conversion of 18:1 into 18:2. Meanwhile, neofunctionalization of several OeSAD genes led to increased desaturation of 18:0 [50]. The decreased OeFAD2 and increased OeSAD expression explained the accumulation of high levels of 18:1 in olive oil [50]. This situation is quite similar to high-oleate production in oil tea seeds. The genomic bases of different expression patterns of CoSAD and CoFAD in oil tea seeds are still not clear, and they are not reported as regulation targets of WRIs [11,12]. However, the exceptional numbers of transcripts encoding oil body proteins, including 45 transcripts for OLEs, 11 for CLOs, 4 for SLOs, and 7 for seipins from individually assembled full-length transcriptome data (Supplementary Dataset S3), may indicate the multiple genetic loci for these oil body genes. Both our transcriptome and genome sequences of oil tea plants showed that there are more transcripts for oil body proteins in oil tea plants than in other oilseed plants, such as 17 OLE genes in diploid Arabidopsis and 13 in allopolyploid soybean (Supplementary Dataset S4) [18,51]. OLEs play an essential role in stabilization of the coat surface of oil bodies, and are positively associated with the high oil production in oilseeds, including in tea oil production [18,51]. Palm oils from mesocarps and kernels are compositionally different. Genome sequences revealed that palm oil biosynthesis is mainly controlled at transcriptional level [52,53]. The expression of FatB, FAD7/8, SAD, FatA, LACS, KASI, KASII, and KASIII genes in the chloroplast is critical for higher oil production in oil palm [52,53]. These genes are tightly regulated by a WRI, which is characterized as a key marker for high-oil-yield oil palm varieties, rather than in date palm [20,52,53]. Similar to oil palm WRI, CoWRI1a and b also play key roles in regulating oil biosynthesis in oil tea seeds. Sesame (Sesamum indicum) seeds contain a high level of oil, which constitutes ∼59% of the dry seed and consists of more unsaturated FA. In sesame seeds SiKASI and SiDGAT2 are two key genes responsible for the saturated/unsaturated fat ratio, whereas SiFAD2 and SiSAD contribute to the natural genetic variation in 18:1 FA [54]. All studies indicate that the expression levels of SAD, FAD, and DGAT/PDAT genes are critical for high oil production in oil seed crops and also critical for their oil composition [11,12,54]. Our study demonstrated that oil tea plant WRI1s may activate the late glycolysis genes, such as PK, and early FA biosynthesis genes, such as KAS, BCCP, ACP, and ACBP, in the chloroplast for oil production in oil tea seeds, which is worthy of further investigation.

Conclusions
Given that the breeding of C. oleifera to increase the production of tea seed oils could be seriously limited by poor understanding of the molecular and genetic basis for seed development and oil biosynthesis and regulation, we combined transcriptome and metabolic analyses of developing seeds to decipher the glycolysis and lipid synthetic pathways and genes. Furthermore, we characterized the key FA desaturase genes-CoSADs and CoFADsthat are responsible for the high 18:1, appropriate 18:2, and low 16:0, 18:0, and 18:3 levels in oil seed seeds. Two major CoDGATs that may prefer to synthesize TAGs with oleoyl acyl chains were characterized. Two key oil biosynthesis regulators CoWRI1s are also found in abundance in developing seed, and CoWRI1a and b activate late glycolysis and FA synthesis to direct the source carbon flux towards oil production. These genes are responsible for high-oleic-acid-oil production in oil tea seeds. These key metabolic and regulatory genes not only provide potential molecular tools for metabolic engineering of highquality edible oil production in oilseed crops or other engineered microorganisms, but also can be developed into useful markers for molecular-assisted breeding of new oil tea varieties with high oil yields and desirable FA compositions.

Plant materials
Developing fruits of Camellia oleifera Abel. cultivar 'Changlin #4' were used in this study. Twelve-yearold trees grown in Dechang Camellia tree farmer lands, Anhui Province, China, were used for sampling developing seeds of various stages. After the oil tea trees bloomed, seeds at different developmental stages were sampled regularly, and samples representing early embryo, growing embryo, early and late seed filing, and maturation stages were selected for FA, oil, RNA-Seq, and gene expression analysis. Six seed developmental stages, which were counted in weeks after flowering, were about 14,18,22,27,30, and 36 weeks after flowering, corresponding to the first weekends of May, June, July, the middle of August, early September, and late October in 2016, 2017, and 2018, respectively. For convenience, they were named Stage 1, 2, 3, 4, 5, and 6, respectively. The fruits at 36 week after pollination (WAP) of oil tea cultivar 'Changlin #4' were almost completely mature. For each time point, at least 50 young oil tea fruits or 20 older oil tea fruits were collected for the experiments. The oil tea fruits from each tree of the same cultivar in similar sizes from different bunches were collected and randomized to obtain repeat samples for metabolite and RNA analyses. The oil tea fruits and peeled seeds were stored at −80 • C for further studies. For the association study, the fruits of 14 other oil tea trees grown in the same garden were also picked (at about Stage 5) in September for the seed oil and RNA analyses.

RNA extraction, library construction, sequencing, and annotation
For the RNA-Seq analysis, we collected six developmental stages of C. oleifera seeds. The six cDNA libraries were sequenced using Illumina RNA-Seq technology, as described previously. Total RNA was extracted using an RNA Extraction Kit (BioTeke, Beijing, China). The Ultra™ RNA Library Prep Kit for Illumina (NEB) was used to synthesize double-stranded cDNAs. Further, cDNAs were prepared for Illumina HiSeq2500 paired-end sequencing. De novo assembly and functional analyses of transcripts were done as previously described [15,16]. For differentially expressed gene (DEG) analysis the assembled contigs with more than 10 reads mapped were chosen. For the expression analysis, the number of clean reads for each contig was calculated and then normalized to reads per kilobase per million reads (RPKM). The expression difference of each contig between different treatments was calculated based on the MARS model using the DEGseq package. Transcriptome annotation was performed using unigenes as query sequence against several databases, including NR, Swiss-Prot, and the Gene Ontology (GO) database. Unigenes were categorized into three broad categories: biological processes; cellular components; and molecular functions. Unigenes were also used to query the KEGG and Pfam databases for specific function identification.

Gene cloning and vector construction
The open reading frames (ORFs) for CoDGAT1, CoDGAT2, CoSAD1, CoSAD2, CoFAD2, CoFAD3, CoWRI1a, and CoWRI1b were cloned by PCR. For PCR amplification, genespecific primers were designed based on the RNA-Seq information of six developmental stages of oil tea seeds. Five micrograms of total RNA was used to synthesize the first-strand cDNA through the first-strand synthesis system Superscript III (Invitrogen, USA) for cloning CoSADs, CoFADs, CoWRI1s, and CoDGATs. The ORFs were cloned under PCR amplification conditions with pairs of gene-specific primers. The PCR products were cloned into the Gateway entry vector pDONR221 using BP Clonase (Invitrogen, USA). The resulting pDONR221 constructs were sequenced, and cloned into the destination vector using Gateway LR Clonase (Invitrogen, USA). All constructs were confirmed by sequencing and for transforming yeast strains or plants.

Yeast expression of CoDGAT proteins and functional assays
For the functional expression of CoDGAT1 or 2 and CoSAD1 and 2 in yeast, the quadruple mutant S. cerevisiae strain H1246 (W303; MATα are1-Δ::HIS3 are2-Δ::LEU2 dga1-Δ::KanMX4 lro1-Δ::TRP1 ADE2 ura3) was used as heterologous host. Stationary H126 cells were transformed using plasmid DNA pYESDEST52-CoDGATs and pYESDEST52-CoSADs through the PEG/lithium acetate method. Yeast cells harboring the empty pYESDEST52 vector were used as negative control. Transformants were carefully chosen on YNB medium lacking uracil and functional characterization of CoDGATs or CoSADs was carried out as described in Unver et al. [50]. The intracellular lipid bodies were stained using the Nile Red staining technique and visualized through fluorescence microscopy. Images of the stained cells were captured under a Nikon Eclipse 80I microscope using the emission wavelength of 538 nm and excitation wavelength of 485 nm. Alternatively, total lipids were extracted and separated by TLC and quantification was carried out by gas chromatography (GC).

Quantification of triacylglycerol and fatty acid composition
Extraction of all kind of lipids and seed TAG amount and composition were determined with little modification according to the methods described in previous studies [32,55]. Briefly, the three biological replicates of 30 mg oil tea seeds were ground in liquid nitrogen for total lipid extraction. The TAG from yeast cells and tobacco leaves was determined by TLC on a silica plate (SIL GF254, 0.25 mm). The plate was developed with hexane/diethyl ether/acetic acid (80:20:1, v/v/v), as described previously [56]. The amount of TAG and its composition in oil tea seeds was measured as explained in Chen et al. [55]. FAs in TAGs from seeds, hairy roots, and yeast cells were analyzed using a GC (Agilent 7890A) system having a flame ionization detector (FID). Triheptadecanoin in toluene (Nu-Chek Prep, Elysian, USA) was used as an internal standard during extraction, the transesterification reaction, and GC analysis. The oil content was calculated relative to 17:0 methyl ester (Sigma-Aldrich, CA, USA).

Analyses of seed sugar, starch and protein contents
The sugar and starch contents were analyzed via the anthrone assay as described previously [32,56]. Total seed protein was extracted following the method described previously [32].

Quantitative RT-PCR analysis of gene expression
Total RNA from six developmental stages of oil tea seeds was extracted and used to prepare first-strand cDNAs. All cDNA samples were further 20-fold diluted with sterilized water for quantitative real-time PCR (qRT-PCR). Gene-specific primers provided in Supplementary Data Table S1 were used for qRT-PCR in 96-well plates (iQ5 Real Time PCR System; Bio-Rad). The CoACTIN and CoETF genes were used as internal controls. All analyses were performed in three biological replicates with three technical replications.

Subcellular localization of CoFAD and CoDGATs
GFP-CoDGAT1 and GFP-CoSAD1 fusions were generated by using a Gateway recombination system into the pK7WGF2 vector in combination with GFP at the Nterminus of the genes [55]. Subcellular localization of these GFP fusions was achieved by using the tobacco (Nicotiana benthamiana) leaf infiltration system previously reported [55]. Images of GFP-CoDGAT1-or -CoSAD1 fusion proteins were captured with an Olympus confocal microscope using a 63× water-immersion objective with an excitation wavelength of 488 nm and emissions collected at 500 nm.

Transient expression of CoDGATs, CoFADs, and CoWRI1s in tobacco plants
Transient expression in tobacco leaves was carried out with some modifications as described previously [56]. Agrobacterium tumefaciens cultures harboring the gene coding for the P19 viral suppressor protein and the genes of interest were mixed together. The final concentration of both cultures was 0.125 at an OD of 600 prior to infiltration. For lipid analysis, 20 leaves from five plants were infiltrated with each of the genes individually or in different combinations. Samples for comparison were randomly located on the same leaf. After 5 days of infiltration, tobacco leaf disks (three from each plant) were harvested and stored at −80 • C. The TAG from tobacco leaves transiently expressing CoDGAT1, or CoFAD2 and 3, and CoWRI1, either individually or in different combinations, was separated by TLC and analyzed by GC.

Statistical analysis
All tests were conducted in triplicate. Statistical analysis was performed using Student's t-test, and P values <.05 were considered statistically significant.