Using lipidomics to reveal details of lipid accumulation in developing Siberian apricot (Prunus sibirica L.) seed kernels

Siberian apricot seed kernel (SASK) contains a high of 50% oil with suitable fuel properties conformed to biodiesel standard. To date, Prunus sibirica is a novel non‐crop feedstock for biodiesel production in China. Here, oil contents and fatty acid (FA) compositions were identified in developing SASK from AS‐80 and AS‐84, at intervals of 1 week from 3 weeks after anthesis (WAA) to 9 weeks. The major differences in oil content between C18:1 and C18:2 levels were greater among the AS‐80 (32.69/15.48 g/100 g) than among the AS‐84 (25.78/13.15 g/100 g). Subsequently, the SASKs from 4, 6, and 8 WAA, respectively, representing early, middle, and late phases of oil accumulation, were selected as optimal samples for lipidomics analysis. It was notable that 18:1/18:1/18:2, 18:1/18:1/18:3, and 18:2/18:2/18:2 were the prominent compositions in triacylglycerol (TAG), and their higher content found among the AS‐80 was consistent with FA results. Although phosphatidic acid (PA) is directly connected with diacylglycerol (DAG) in Kennedy pathway, we found significant difference between PA and DAG compositions. The resulting molecular species differ in acyl composition depending on whether they were generated via phosphatidylcholine (PC) or Kennedy pathway. By qRT‐PCR analysis, the expression levels of FAD3, PDCT, and DAG‐CPT related to the biosynthesis of polyunsaturated fatty acids (PUFAs) showed a gradual decrease with SASK mature, explaining the drastic change of DAG‐18:3/18:3 content. Additionally, the lipidomics data coupled with qRT‐PCR analysis suggested that phospholipid:DAG acyltransferase may play a critical role in incorporation of PUFAs into sn‐3 of TAG. Our data contribute significantly to understand the underlying mechanisms of lipid accumulation in P. sibirica, and may also present strategies for engineering oil accumulation in oilseed plants.


| INTRODUCTION
Siberian apricot (Prunus sibirica) is a woody oil species that belongs to the Rosaceae family. As an endemic species in northern China, it is widely used for soil and water conservation owing to its superior adaptability to a wide range of environment (Wang, 2011). Almost 192,500 tonnes of seeds are harvested in autumn every year (Zhang, 2003). The oil content of Siberian apricot seed kernel (SASK) is generally as high as 50%, of which about 95% is unsaturated fatty acid (FA), including 56.23%-76.69% oleic acid and 16.44%-34.69% linoleic acid (Wang, 2012). Because the cost of feedstock account for the cost of biodiesel approximately over 75% (Guo, Fang, Tian, Long, & Jiang, 2011), non-crop feedstock to produce biodiesel has drawn governmental and popular attention. Previous evaluation of SASK oil has shown that its biodiesel fuel properties, such as cetane number, iodine number, and kinematic viscosity, were conformed to EN 14214 and GB/T 20828-2007 standards (Wang, 2012;Wang, Yu, He, & Liu, 2012). Thus, SASK oil is a potential non-crop feedstock for biodiesel production.
In plants, sucrose can be transported from photosynthetic tissues to heterotrophic sinks through the phloem. Once delivered into sinks, sucrose can be cleaved into two hexoses by either invertase or sucrose synthase (Ruan, 2014). In oil seeds, the generated hexoses are metabolized through the oxidative pentose phosphate pathway and the glycolytic pathway, providing acetyl-CoA as precursor for de novo FA synthesis (Baud & Lepiniec, 2009). After FA formation in plastid, a mixture of palmitic acid (C16:0), palmitoleic acid (C16:1), stearic acid (C18:0), and oleic acid (C18:1) is produced, and transported to cytosolic acyl-CoA pool (Bates, Stymne, & Ohlrogge, 2013). These long-chain acyl-CoAs are the substrate for glycerolipid assembly (Kennedy pathway). Notably, while the classic Kennedy pathway relies on the sequential acylation of a glycerol-3-phosphate (G3P) backbone, the membrane lipid phosphatidylcholine (PC) is also important for the assembly of unsaturated or otherwise modified lipid in endoplasmic reticulum . Three mechanisms allow the flux of FA through PC for eventual triacylglycerol (TAG) synthesis . First, acyl-CoA: lysophosphatidylcholine acyltransferase catalyzes the esterification of nascent FAs to PC via "Acyl editing" (Bates, Ohlrogge, & Pollard, 2007). Second, PC-derived diacylglycerol (DAG) is utilized as the substrate for TAG synthesis . Third, the FA from PC pool can be esterified to sn-3 position of TAG by phospholipid:DAG acyltransferase (PDAT; Dahlqvist et al., 2000). Overall, PC pool is a central intermediate in flux of acyl chains, implying that lipid assembly exerted more control than FA biosynthesis.
Previously, we have studied the regulation of oil accumulation in developing SASK using high-throughput sequencing, revealing the involvement of some specific genes encoding functional enzymes and transcription factors in SASK oil accumulation (Deng, Mai, Shui, & Niu, 2019;Niu et al., 2015Niu et al., , 2016. However, some detailed controls could not be delineated. For example, we could not specifically determine why phosphatidylcholine: diacylglycerol cholinephosphotransferase (PDCT) and diacylglycerol choline phosphotransferase (DAG-CPT) have higher expression levels in early development of SASK (Niu et al., 2015). Moreover, the relative flux of acyl chains onto the sn-3 position of TAG through either PDAT or diacylglycerol acyltransferase (DGAT) is unclear in most oilseed plants . A double gene silencing in Arabidopsis seeds resulted in 80% decrease of oil content (Zhang, Fan, Taylor, & Ohlrogge, 2009), showing key roles of DGAT and PDAT in TAG accumulation (Mhaske, Beldjilali, Ohlrogge, & Pollard, 2005). However, the transcriptional level and enzyme activity for DGAT were much higher than that for PDAT in oilseed rape (Tang et al., 2012;Troncoso-Ponce et al., 2011), implying that DGAT was a more important factor in TAG biosynthesis. Our previous studies have characterized the regulatory roles of DGAT1 and PDAT2 in TAG biosynthesis (Niu et al., 2015), but their relative importance is currently uncertain.
Lipidomics as a novel technique can provide important data about lipid metabolism during oil accumulation. Compared with other metabolites, the composition and content of lipids could reflect the response of plants to specific stimuli. Lipometabolism is the master pathway for energy homeostasis, membrane structure, cell signaling, transcription and translation regulation, and cell interactions. Recent studies have indicated the importance of lipids in response to biotic and abiotic stresses (Liu et al., 2017;Moradi, Mahdavi, Khoshkam, & Iriti, 2017). Moreover, lipidomics technique has been applied to various plant tissues (Horn & Chapman, 2012;Okazaki, Kamide, Hirai, & Saito, 2013;Woodfield et al., 2018). To identify the optimal period of SASK used for lipidomics analysis, the oil contents and FA compositions from two different germplasms (AS-80 and AS-84) were first detected. Subsequently, we applied liquid chromatography-tandem mass spectrometry (LC-MS/MS) to developing SASK. The lipidomics data could provide detailed information about the end products and the intermediates of lipid metabolism (Horn & Chapman, 2012;Okazaki et al., 2013;Woodfield et al., 2018). The expression levels of some key genes were assessed using qRT-PCR. This study provides a basis for improving our understanding of TAG synthesis in P. sibirica.

| Plant material
Based on the collection and identification of P. sibirica germplasm resources (Wang, 2012), two germplasms (AS-80 and AS-84) with significant differences in oil content were used plant material in this study. AS-80 and AS-84 locating in Changping District of Beijing (geographical coordinates approximately 116°23′E, 40°22′N) were selected and marked in April 2018. Fruits were collected at 3, 4, 5, 6, 7, 8, and 9 weeks after anthesis (WAA). After removing the pericarps, the seeds were immediately frozen in liquid nitrogen and stored at −80°C until use.

| Oil extraction and component analysis
The 10 g of SASK was dried and crushed, and the powders were used for oil extraction. Using Soxhlet extraction system, the SASK oils were extracted with n-hexane at 45°C for 8 hr. The difference value in weight of SASK powders is the content of SASK oil. The 5 wt% heptadecanoic acid (Shanghai, China) was added into the extracted oils, and then, the mixed samples were trans-esterified under standard conditions employing a 5.5:1 molar ratio of methanol to oil using 1 wt% potassium hydroxide as a catalyst at 60-65°C for 1 hr. After cooling and standing, the reaction mixture was separated into two layers. The upper layer, including the FA methyl esters (FAMEs), was isolated, dried with Na 2 SO 4 , and analyzed by gas chromatography-mass spectrometer (GC-MS).
According to our previous report, the FAMEs were detected using the Agilent 6890 equipped with a flame ionization detector (GB/T17377-1998; Deng et al., 2019). The following chromatographic conditions were applied: inlet temperature 250°C, detector temperature 280°C, and injection volume 1 µl. The carrier gas was high-purity hydrogen. The qualitative analysis of FAMEs was identified by comparing their retention times with ANPEL FAMEs mix. The quantitative analysis of FAMEs was identified by comparing their peak area with heptadecanoic methyl ester. Peak integration was performed by applying HP3398A software. The above experimentations are repeated three times.

| Lipid extraction
The SASKs from 4, 6, and 8 WAA, which represent early, middle, and late phases of oil accumulation, respectively, were selected as the optimal period for lipidomics analysis. Each biological repeat contained at least three separate SASKs. 20 mg of SASK sample was weighted to an EP tube, and homogenized with 400 μl water. 200 μl of the homogenate was transferred to a fresh EP tube, and add water to 400 μl. Then, 960 μl extract solution of methyl tert-butyl ether (MTBE): methanol = 5:1 containing internal standard was added. This internal standard comprises 5 μl 100 μg/ml of d7lysoglycerophosphatidylcholine (LPC) −18:1, 5 μl 100 μg/ml of d7-phosphatidyl-ethanolamine (PE) −15:0/18:1, and 11 μl 100 μg/ml of d7-TAG-15:0/18:1/15:0. All standards were purchased from Avanti Polar Lipids. After 30 s vortex, the samples were sonicated for 10 min in ice-water bath, followed by centrifuged at 4°C 3,000 rpm for 15 min. Five hundred microliters of the supernatant was transferred to a fresh tube. The residual was added 500 μl MTBE again, followed with vortex, sonication, and centrifugation, and another 500 μl of supernatant was taken out. This step was repeated once. The three supernatants were combined and dried in a vacuum concentrator at 37°C. The dried samples were reconstituted in 200 μl of 50% methanol in dichloromethane by sonication on ice for 10 min. After centrifugation at 4°C 1,300 rpm for 15 min, 75 μl of the supernatant was transferred to a fresh glass vial for LC-MS/MS analysis. The quality control sample was prepared pooling 20 μl of each sample together.
Triple TOF 6600 (AB Sciex) was used for its ability to acquire MS/MS spectra on an information-dependent basis during an LC/MS experiment. The acquisition software (Analyst TF 1.7, AB Sciex) continuously evaluates the full scan survey MS data as it collects and triggers the acquisition of MS/MS spectra depending on preselected criteria. In each cycle, the most intensive 12 precursor ions with intensity above 100 were chosen for MS/MS at collision energy of 45 eV (12 MS/MS events with accumulation time of 50 msec each). ESI source conditions were set as following: Gas 1 as 60 psi, Gas 2 as 60 psi, Curtain Gas as 30 psi, Source Temperature as 600°C, Declustering potential as 100 V, Ion Spray Voltage Floating as 5,000 V (pos) or −4500 V (neg).

| Lipidomics data processing
The data were analyzed by Biomarker as follows. The raw data files (.wiff format) were converted to files in mzXML format using the "msconvert" program from ProteoWizard (version 3.0.6150). Then, the mzxML files were loaded into LipidAnalyzer for data processing. Peak detection was first applied to the MS1 data. The CentWave algorithm in XCMS was used for peak detection. With the MS/MS spectrum, lipid identification was achieved through a spectral match using an in-house MS/MS spectral library (Biomarker). The relative quantitation of lipids can be achieved using the peak area, SIL-IS, and RF information.

| qRT-PCR analysis
Based on previous transcriptome results generated from SASK (Niu et al., 2015), the qRT-PCR primers were designed using PrimerQuest (www.idtdna.com/Prime rQues t/Home/Index) and are shown in Table S1. The qRT-PCR was performed using the SYBR Premix Ex Taq Kit (TaKaRa) according to the manufacturer's protocol. Also, the genes encoding cyclophilin and ubiquitin-conjugating enzyme were used as reference genes in this study (Niu et al., 2014). The relative expression levels were calculated as log 2 ((1 + E1) ΔCt1(control-sample) / (1 + E2) ΔCt2(control-sample) ), E1: PCR efficiency of target-gene primer; E2: PCR efficiency of reference-gene primer; ΔCt1: the difference of Ct value between control and sample in experimental group; ΔCt2: the difference of Ct value between control and sample in reference group. The PCR efficiency (E) was estimated by the equation of E = 10 slope (Ramakers, Ruijter, Deprez, & Moorman, 2003).

F I G U R E 1 Variation of oil content
in developing Siberian apricot seed kernel (SASK). The oils were extracted by Soxhlet extraction. Oil content was determined by comparing the SASK weight before and after extraction

AS-80 and AS-84
The dynamic pattern of oil content during SASK development showed a sigmoid, exhibiting a sharp rise during 4-8 WAA (Figure 1). Comparative analysis in developing SASK indicated that AS-80 (1.14 g/100 g to 53.48 g/100 g) accumulated oil more strongly than AS-84 (1.03 g/100 g to 44.85 g/100 g), achieving an enrichment of about 10% (Figure 1). GC-MS analysis showed that eight main FAs are identified in SASKs, including 18:2 and 18:3 (polyunsaturated fatty acids, PUFAs), 20:0 and 20:1 (very long-chain fatty acids, VLCFAs; C>20; Table 1). Among those FAs, C18:1 (oleic acid) and C18:2 (linoleic acid) represented the major compositions both in AS-80 and AS-84 (Table 1). The FA profiles were similar between AS-80 and AS-84, but the former AS-80 accumulated oleic acid and linoleic acid more strongly than the latter AS-84 (Table 1). Based on these results, the SASKs from 4, 6, and 8 WAA, which represented early, middle, and late phases of oil accumulation, respectively, were selected as optimal samples for lipidomics analysis.

| PA and DAG compositions in developing SASK from AS-80 and AS-84
In the Kennedy pathway, dephosphorylation of phosphatidic acid (PA) molecules can generate corresponding DAG molecules. The main DAG compositions in developing SASK were 18:3/18:3, 18:1/18:1, and 18:2/18:2 ( Figure 3a). The 18:3/18:3 content showed a notable decrease during oil accumulation, whereas 18:1/18:1 and 18:2/18:2 contents changed little (Figure 3a; Table S3). Compared with DAG, 18:1/18:1 and 16:0/18:2 were the main compositions in PA ( Figure 3b; Table S4). The 18:1/18:1 content showed a significant abundance at 6 WAA, while 16:0/18:2 decreased with SASK maturity (Figure 3b). Other significant molecular species of PA were those containing PUFAs At the key phase for oil accumulation (6 WAA), it should be noted that, in the delivery of 18:1/18:1 molecule for TAG biosynthesis, there was significant increase in PA-18:1/18:1, but no accumulation was observed in DAG-18:1/18:1 (Figure 3). Additionally, the PA-18:1/18:1 content of samples from AS-80 was significantly higher than that of samples from AS-84, but no significant difference in the DAG-18:1/18:1 content between AS-80 and AS-84 (Figure 3). These results suggested that PA concentration rather than DAG concentration would be required for TAG accumulation in SASK development. F I G U R E 2 Relative content of the major molecular species of triacylglycerol during oil accumulation. The further information on minor species listed in Table S2 F I G U R E 3 Relative content of the major molecular species of diacylglycerol (a) and phosphatidic acid (b) during oil accumulation. Further information on minor species is listed in Tables S3 and S4, respectively
Second, acyl chains on PC can also be incorporated into the sn-1 and sn-2 positions of DAG via two known enzymatic routes ( Figure 5, blue arrows). Our qRT-PCR results indicated the expression levels of PDCT and DAG-CPT both displayed a downtrend with SASK development (Figure 6a). Surprisingly, the main species of 18:3/18:3 DAG also showed a notable decrease during SASK development (Figure 3a). Considering a very low content of 18:3/18:3 species in PA (Figure 3b), it seems plausible that PDCT and DAG-CPT can generate DAG-18:3/18:3 from PC.
Third, the final step in TAG synthesis can also be catalyzed by PDAT ( Figure 5, purple arrows), using the sn-2 and sn-1 (at a quarter of the rate for the sn-2 position) acyl group from PC (Banaś, Garcia, Banaś, & Stymne, 2013). PDATs from Arabidopsis and other plants have been shown to have high activity with PC containing unsaturated FAs  (Kim et al., 2011;Van Erp, Bates, Burgal, & Shockey, 2011). With this knowledge, we sought to use the lipidomics data to evaluate the possible contribution of PDAT to TAG accumulation. If the oilseeds are in the period of high oil biosynthetic rate, the flow of FAs is mostly into oil accumulation, and is not unduly influenced by other lipid metabolism (Woodfield et al., 2018). Thus, these lipidomics data from 6 WAA (high biosynthetic rate) were chosen for evaluating the PDAT role in TAG biosynthesis. Using the sn-2 and sn-1 (a quarter of rate) acyl group of PC as the acyl donor, we assumed the ratios of C18:1/C18:2 and C18:2/C18:3. Also, the rates of C18:1/C18:2 and C18:2/C18:3 from the sn-1, sn-2, and sn-3 acyl group of TAG were evaluated. As expected, the rates of C18:1/C18:2 and C18:2/C18:3 from the sn-1 and sn-2 of TAG had enormous differences with those from sn-2 of PC in both AS-80 and AS-84 (Figure 6b). These data suggest that PDAT does not function in sn-1 and sn-2 acylation of TAG. Although the C18:1/C18:2 rate showed a substantial difference between sn-3 of TAG and sn-2 of PC, it was worth noting that the C18:2/C18:3 rate from sn-3 of TAG was close to this from sn-2 of PC in both AS-80 and AS-84 ( Figure 6b). Our data implicate involvement of PDAT in sn-3 acylation of TAG, and strong substrate selectivity for PUFAs by PDAT enzyme.
In summary, the above-mentioned results imply the importance of PC-mediated "PUFA trafficking" for TAG biosynthesis, which may determine the TAG compositions in developing SASK.

| DISCUSSION
Plant oil is composed of long-chain hydrocarbons, and thus can also replace petroleum in many applications, such as feedstock for chemical and biofuel. Owing to its important uses, high value and increased demand, oil biosynthesis in plant seed has been extensively studied. Numerous studies in plants have elucidated the biochemical pathways producing storage lipids, namely TAGs Baud & Lepiniec, 2009;Bourgis et al., 2011;Troncoso-Ponce et al., 2011). In Arabidopsis, 120 enzymatic reactions and more than 600 genes were divided into 12 sections based on the types of lipids produced and their subcellular localization (Li-Beisson et al., 2013). Although our recent studies in developing SASK have provided details on the very large number of genes involved in FA synthesis and TAG biosynthesis (Deng et al., 2019;Niu et al., 2015Niu et al., , 2016, some detailed metabolisms remain largely unknown such as the FA flux from the basic Kennedy pathway into and out of PC (Bates, 2016). To extend and complement our knowledge about the regulation of lipid accumulation, we conducted the detailed lipidomics examination of important lipid classes during SASK development.
The plastid FA synthesis pathway is catalyzed by various enzymes responsible for producing chain lengths from 6 to 18 carbons (Baud & Lepiniec, 2009). Some 18:0-ACP are efficiently desaturated by a stearoyl-ACP desaturase (SAD; Bates, 2016). Based on our previous transcriptomic analysis of P. sibirica, SAD6 was identified as the potential enzyme for oleic acid formation (Niu et al., 2015). In both AS-80 and AS-84, SAD6 gene showed a bell-shaped pattern of expression (upregulated expression during 5-7 WAA), but it was sharply upregulated in AS-80 compared with AS-84 (Figure 6a). This could explain why AS-80 had a higher content of oleic acid (Table 1). These data coupled with our previous transcriptome (Niu et al., 2015) and co-expression studies (Deng et al., 2019) clearly indicated that SAD6 expression is important for directing the acyl-ACP flux toward the oleoyl-ACP synthesis in developing SASK. Acyl groups esterified to PC are the site of extra-plastidic FA desaturation, namely PC acyl editing (Speerling & Heinz, 1993). This process involves reacylation of lyso-PC with an acyl-CoA from the acyl-CoA pool, and deacylation of PC which generates lyso-PC and releases the FA or acyl-CoA to the acyl-CoA pool ( Figure 5; Li-Beisson et al., 2013). In endoplasmic reticulum, FAD2 (Okuley et al., 1994) andFAD3 (McConn et al., 1993) can convert PC-oleate (C18:1) to PC-linoleate (C18:2) and then PC-linolenate (C18:3), respectively. The qRT-PCR analysis showed upregulated expression for FAD2 and downregulated expression for FAD3 during 5-7 WAA (Figure 6a), in agreement with our previous results (Niu et al., 2015). Also, the temporal expression of FAD2 closely matched the cumulative rate of linoleic acid in both AS-80 and AS-84 (Table 1), corroborating the role of FAD2 in PC-mediated biosynthesis of linoleic acid. The PC acyl editing exchanges the FA on PC with the acyl-CoA pool ( Figure 5), resulting in increasing PUFAs in acyl-CoA pool. Previous studies in developing soybean embryos revealed that about 60% of nascent acyl chains are incorporated directly into the sn-2 position of PC through acyl editing mechanism rather than through the sequential acylation converting to TAG (Bates, Durrett, Ohlrogge, & Pollard, 2009).
The de novo assembly of TAG is initiated with two acylations of G3P by glycerol-3-phosphate acyltransferase (GPAT) and lysophosphatidic acid acyltransferase (LPAAT; Barron & Stumpf, 1962). Subsequently, PA phosphatase functions in the removal of the phosphate group from PA to generate DAG (Barron & Stumpf, 1962). Despite PA is directly connected with DAG in Kennedy pathway, the composition profile of PA was distinct from those of DAG in developing SASK (Figure 3), which was consistent with the lipidomics results in developing seeds from oilseed rape (Woodfield et al., 2018). This may be caused by the complex outflux of PA. Indeed, PA also can be used for PG biosynthesis ( Figure 5; Li-Beisson et al., 2013). In developing SASK, the 16:0/18:1 and 16:0/18:2 as the main molecules in PA could also be found in PG (Figure 3b,c). Interestingly, 18:3/18:3 was the main composition in DAG, but its corresponding content in PA was relatively low. This finding may reflect the very significant flux of acyl chains between DAG and PC. The acyl chains on PC can also be reversibly incorporated into the sn-1 and sn-2 positions of DAG by PDCT (Lu, Xin, Ren, & Miquel, 2009) and DAG-CPT (Slack, Campbell, Browse, & Roughan, 1983; Figure 5). As mentioned above, FAD3 catalyzes the biosynthesis of linolenate (C18:3). During SASK development, the expression patterns of FAD3, PDCT, and DAG-CPT showed a higher level at early development of AS-80 and AS-84 (Figure 6a), which was exactly similar to the cumulative pattern of DAG-18:3/18:3 ( Figure 3a). Therefore, it is reasonable to presume that first PA-derived DAG is converted to PC, and then after PC acyl editing, these unsaturated acyl chains from PC are transferred to generate newly synthesized DAG ( Figure 5). In vivo metabolic flux analysis of soybeans strongly suggests that PA-derived DAG and PC-derived DAG are distinct pools that do not intermix (Bates et al., 2009). The mutation of PDCT in Arabidopsis reduces 18:2 and 18:3 accumulation by 40% in seed (Lu et al., 2009). These expression data of FAD3, PDCT, and DAG-CPT also could explain the gradual decrease of DAG-18:3/18:3 with SASK mature. Taken together, our data suggest that PC acyl editing is important for the flux of PUFAs into PC-derived DAG.
As remarked before, there is a close metabolic connection among PA, PG, PC, and PE ( Figure 5; Kennedy, 1961;Li-Beisson et al., 2013;Lu et al., 2009;Slack et al., 1983). Moreover, previous studies in soybean suggested that DAG-CPT and DAG-EPT have been demonstrated to have distinct characteristics, but an isolated gene codes for both activities (Harwood, 1976). PE can also be converted to PC by methylation (Williams & Harwood, 1994). However, all of they had notable differences from qualitative and quantitative aspects in developing SASK. For example, the acyl chains of PG were mainly composed of VLCFAs (Figure 4c), while only minor amounts of PA, PC, and PE contained VLCFAs (Figures 3b  and 4a,b). The 18:1/18:1 species of PA and PE both had a highest level at 6 WAA (Figures 3b and 4c), but the maximum values of PC were at 8 WAA (Figure 4a). Similarly, various molecules from PA, PG, PC, and PE showed differently temporal changes. These subtle distinctions in metabolism for phosphoglycerides may reflect their different functions in plant cells (Viola-Magni, Gahan, & Pacy, 1985).
In plants, at least three mechanisms differing in their acyl donor sources have been identified to be involved in the final acylation reaction (Li-Beisson et al., 2013). In its simplest form, DGAT catalyzes the last acylation of Kennedy pathway using an acyl-CoA molecule (Liu, Siloto, Lehner, Stone, & Weselake, 2012). Second, DAG can be acylated by PDAT using PC as the acyl donor (Dahlqvist et al., 2000). Third, studies in Carthamus tinctorius seeds implied the existence of diacylglycerol:diacylglycerol transacylase activity that acylate DAG to form TAG using DAG as the acyl donor (Stobart, Mancha, Lenman, Dahlqvist, & Stymne, 1997), but the enzymes catalyzing such activity are yet to be identified. Although there are multiple TAG biosynthetic pathways, DAG is the direct substrate for TAG biosynthesis. However, in sharp contrast to the compositions of acyl chains for PA molecules, the acyl chains for DAG were lack of palmitoyl (Figure 3a). These results suggest that besides of PA-derived DAG in Kennedy pathway, there are other biosynthesis or modification pathways for DAG production in developing SASK. Moreover, at the mid-point (6 WAA) of oil accumulation, 18:3/18:3 represented one of the most of all DAG molecules in AS-80 and AS-84 (Figure 3a), but TAG lacks the 18:3/18:3/* compositions ( Figure 2). These data suggest the substrate selectivity of DGAT and PDAT, and/or the presence of separate pools of DAG, as discussed above in metabolic flux analysis of soybeans (Bates et al., 2009).
DGAT and PDAT are the key enzymes involved in the final step of TAG biosynthesis in plants. However, their relative contribution to TAG accumulation has not been determined in most oilseed plants . Mutant studies in Arabidopsis have shown that the PDAT1 mutant has no effect on TAG accumulation (Mhaske et al., 2005), while the DGAT1 mutant only reduces oil content by 20%-30% (Zou et al., 1999). However, a double knockout of PDAT1 and DGAT1 is lethal, and RNAi suppression of PDAT1 gene in DGAT1 mutant results in severe defects in pollen and seed development, including 80% reduction of oil content (Zhang et al., 2009). From these results, the overlapping functions of DGAT and PDAT in oil accumulation can be concluded. The absence of DGAT is partially compensated by PDAT, but DGAT fully compensates for a lack of PDAT. Based on our previous transcriptomic analysis, it was identified that DGAT1 and PDAT2 may play a major role in TAG biosynthesis in developing SASK (Niu et al., 2015). Indeed, the genes encoding DGAT1 and PDAT2 displayed high expressions during the rapid period of oil accumulation (Figure 6a), suggesting the cooperative role of DGAT1 and PDAT2 in TAG biosynthesis. Although our preceding studies showed that the transcription level of PDAT2 was much lower than that of DGAT1 (Niu et al., 2015), the expression level cannot be used to elucidate the relative contribution of the two enzymes in TAG accumulation. As was reported in a recent review, the relative importance of DGAT versus PDAT is a major uncertainty in plant lipid metabolism (Bates, 2016). In this study, using the lipidomics data, we tried to evaluate the role of PDAT2 in TAG synthesis. Under this assumption that unsaturated FAs at the sn-2 and sn-1 (at a quarter of the rate for the sn-2 position) position of PC can be utilized by PDAT (Dahlqvist et al., 2000), the C18:1/C18:2 and C18:2/ C18:3 rates from sn-2 of PC and sn-1, sn-2, and sn-3 of TAG were calculated. One conclusion can be proposed that PDAT2 may have substrate selectivity for TAG biosynthesis in SASK development, resulting in that PDAT2 preferentially utilized PUFAs from PC. In sunflower and safflower seeds, the specialization toward 18:2 in sn-2 position of PC displayed by PDAT, and in case of safflower, PDAT had 3.5 times higher activity for 18:2 than 18:1 (Banaś et al., 2013). To a certain extent, these results indicated that PDAT may play a critical role in incorporation of PUFAs into sn-3 of TAG in SASK development.
In this study, two germplasms of P. sibirica (AS-80 and AS-84) with significant differences in oil content were used for plant material. According to the temporal patterns of oil and FAs in developing SASK, the optimal samples from 4, 6, and 8 WAA that, respectively, represent early, middle, and late phases of oil accumulation were identified for lipidomics analysis. To our knowledge, this is the first report on lipidomics analysis of major lipids involved in TAG biosynthesis during seed development in woody oilseed plants. The lipidomics results showed quite distinct molecular species distribution in PA, DAG, PC, PE, PG, and TAG, reflecting the complexity of lipid flux metabolism during oil accumulation. Moreover, the lipidomics data coupled with qRT-PCR analysis indicated that PC is a central intermediate in the flux of PUFAs or DAG, or both substrates destined to TAG synthesis in developing SASK. All of these findings will be conducive to understand the underlying mechanisms how the storage lipid is formed in developing SASK.