In Vivo and in Vitro Evidence for Biochemical Coupling of Reactions Catalyzed by Lysophosphatidylcholine Acyltransferase and Diacylglycerol Acyltransferase*

Background: Plant polyunsaturated fatty acids (PUFAs) are mainly synthesized on phosphatidylcholine (PC). Results: Diacylglycerol acyltransferase (DGAT) produced higher amount of PUFA-containing TAG in the presence of acyl-CoA:lysophosphatidylcholine acyltransferase (LPCAT). Conclusion: The LPCAT-catalyzed reverse reaction can be coupled to the DGAT reaction for PUFA accumulation. Significance: A mechanism for enhancing the transfer of PUFAs from PC into TAG has been confirmed. Seed oils of flax (Linum usitatissimum L.) and many other plant species contain substantial amounts of polyunsaturated fatty acids (PUFAs). Phosphatidylcholine (PC) is the major site for PUFA synthesis. The exact mechanisms of how these PUFAs are channeled from PC into triacylglycerol (TAG) needs to be further explored. By using in vivo and in vitro approaches, we demonstrated that the PC deacylation reaction catalyzed by the reverse action of acyl-CoA:lysophosphatidylcholine acyltransferase (LPCAT) can transfer PUFAs on PC directly into the acyl-CoA pool, making these PUFAs available for the diacylglycerol acyltransferase (DGAT)-catalyzed reaction for TAG production. Two types of yeast mutants were generated for in vivo and in vitro experiments, respectively. Both mutants provide a null background with no endogenous TAG forming capacity and an extremely low LPCAT activity. In vivo experiments showed that co-expressing flax DGAT1-1 and LPCAT1 in the yeast quintuple mutant significantly increased 18-carbon PUFAs in TAG with a concomitant decrease of 18-carbon PUFAs in phospholipid. We further showed that after incubation of sn-2-[14C]acyl-PC, formation of [14C]TAG was only possible with yeast microsomes containing both LPCAT1 and DGAT1-1. Moreover, the specific activity of overall LPCAT1 and DGAT1-1 coupling process exhibited a preference for transferring 14C-labeled linoleoyl or linolenoyl than oleoyl moieties from the sn-2 position of PC to TAG. Together, our data support the hypothesis of biochemical coupling of the LPCAT1-catalyzed reverse reaction with the DGAT1-1-catalyzed reaction for incorporating PUFAs into TAG. This process represents a potential route for enriching TAG in PUFA content during seed development in flax.

Seed oils of flax (Linum usitatissimum L.) and many other plant species contain substantial amounts of polyunsaturated fatty acids (PUFAs). Phosphatidylcholine (PC) is the major site for PUFA synthesis. The exact mechanisms of how these PUFAs are channeled from PC into triacylglycerol (TAG) needs to be further explored. By using in vivo and in vitro approaches, we demonstrated that the PC deacylation reaction catalyzed by the reverse action of acyl-CoA:lysophosphatidylcholine acyltransferase (LPCAT) can transfer PUFAs on PC directly into the acyl-CoA pool, making these PUFAs available for the diacylglycerol acyltransferase (DGAT)-catalyzed reaction for TAG production. Two types of yeast mutants were generated for in vivo and in vitro experiments, respectively. Both mutants provide a null background with no endogenous TAG forming capacity and an extremely low LPCAT activity. In vivo experiments showed that co-expressing flax DGAT1-1 and LPCAT1 in the yeast quintuple mutant significantly increased 18-carbon PUFAs in TAG with a concomitant decrease of 18-carbon PUFAs in phospholipid. We further showed that after incubation of sn-2-[ 14 C]acyl-PC, formation of [ 14 C]TAG was only possible with yeast microsomes containing both LPCAT1 and DGAT1-1. Moreover, the specific activity of overall LPCAT1 and DGAT1-1 coupling process exhibited a preference for transferring 14 C-labeled linoleoyl or linolenoyl than oleoyl moieties from the sn-2 position of PC to TAG. Together, our data support the hypothesis of biochemical coupling of the LPCAT1-catalyzed reverse reaction with the DGAT1-1-catalyzed reaction for incorporating PUFAs into TAG. This process represents a potential route for enriching TAG in PUFA content during seed development in flax.
Seed oils are major plant-based sources of essential polyunsaturated fatty acids (PUFAs) 3 for human diet and many industrial applications. Understanding the fundamental aspects of how plant fatty acids are synthesized in seed oils is essential for developing new strategies to alter fatty acid composition for improved nutritional quality and industrial uses of seed oils.
As shown in Fig. 1, various pathways have been proposed for transferring PUFAs from PC into triacylglycerol (TAG), which is the main component of seed oil. PUFAs may be directly transferred from PC into TAG via the catalytic action of phospholipid:diacylglycerol acyltransferase (PDAT) (4). Some oilseeds that are high in PUFAs like flax (5) or unusual fatty acids like castor (Ricinus communis) (6,7) appear to have unique PDAT(s) with high selectivity toward these fatty acids. Alternatively, the phosphocholine head group of PUFA-enriched PC may be removed, producing PUFA-enriched diacylglycerol (DAG), which may be further incorporated into TAG via either or both of PDAT and diacylglycerol acyltransferase (DGAT)catalyzed reactions. The conversion of PUFA-rich PC to PUFArich DAG may also involve the catalytic action of phospholipase C, phospholipase D along with phosphatidic acid phosphatase, and the recently discovered phosphatidylcholine: diacylglycerol cholinephosphotransferase (8). In addition, PUFAs may be removed from PC and enter the acyl-CoA pool via acyl-exchange predominately between the acyl groups at the sn-2 position of PC and the acyl-CoA pool mediated by the combined forward and reverse reactions catalyzed by acyl-CoA:lysophosphatidylcholine acyltransferase (LPCAT) (9). PUFA-enriched acyl-CoA may also be generated through liberation of free fatty acid from the sn-2 position of PC by phospholipase A (PLA) followed by activation of free fatty acid to acyl-CoA through acyl-CoA synthetase with LPCAT acting in the forward reaction to catalyze the re-acylation of lysophosphatidylcholine (LPC) to PC in a process known as the "Lands cycle" (10,11). The resulting PUFA-CoAs may be used by enzymes involved in the Kennedy pathway to form PUFA-enriched TAG. Recently, by using reverse genetics and metabolic labeling approaches, it was demonstrated that LPCAT-mediated acyl-editing together with phosphatidylcholine:diacylglycerol cholinephosphotransferase-based PC-sn-1,2-diacylglycerol interconversion control the major fluxes of PUFAs from PC to TAG (ϳ66%) in Arabidopsis seeds (12). In addition, in vitro evidence has shown that plant LPCAT does indeed operate in a reversible fashion, even though the reverse reaction is a thermodynamically unfavorable process (13). Despite recent progress, direct evidence of how LPCAT interacts with the enzymes involved in the Kennedy pathway is still lacking.
In this study we focused on the interaction between flax LPCAT1 and a type 1 DGAT. LPCAT can catalyze both forward and reverse reactions (Reaction 1, Fig. 2). The reverse reaction of LPCAT uses PC as substrate and forms acyl-CoA and LPC. DGAT catalyzes the final step in the Kennedy pathway of TAG synthesis from the substrates acyl-CoA and DAG (Reaction 2, Fig. 2). The reverse reaction of LPCAT and the DGAT reaction share a common intermediate, acyl-CoA.
Based on Reactions 1 and 2 ( Fig. 2), we proposed that the PUFA synthesized on PC can be removed and transferred to the acyl-CoA pool by the reverse reaction of LPCAT and then be further utilized by the DGAT-catalyzed reaction for TAG biosynthesis. As highlighted in Fig. 1, the overall reaction represents a route for transferring PUFAs from PC to the acyl-CoA pool and then to TAG. We chose to test this hypothesis through heterologous expression of flax genes in yeast mutant background devoid of LPCAT activity and TAG-forming capacity. Flax oil naturally contains substantial amount of PUFAs, especially ALA. To accumulate a large proportion of PUFAs in seed oils, flax may have more efficient mechanisms to channel PUFAs from PC into TAG as compared with other plant species. Many TAG synthesis-related genes from flax are closely related to genes found in plant species that produce high amounts of unusual fatty acids, such as castor and tung tree (Vernicia fordii) (14). The PUFA-enriching mechanism found in flax may also represent a route for enriching unusual fatty acids in these species.
Taken together with earlier reported in vitro data on plant LPCAT action (13), our data further confirm that plant LPCAT can act in a reversible fashion under in vivo condition. These data demonstrate for the first time in vivo evidence of the reversibility of the LPCAT. More interestingly, our in vivo and in vitro data support the hypothesis of biochemical coupling of the reverse reaction catalyzed by LPCAT with the DGAT-catalyzed reaction for the incorporation of PUFAs into TAG. These results contribute to our understanding of the mechanisms responsible for channeling PUFAs from PC into TAG in oleaginous plants that produce PUFA-enriched TAG during seed development.  (15). sn-1-16:0-sn-2-[ 14 C]acyl-PC were prepared according to the assay used for measuring the forward reaction of LPCAT as described by Lager et al. (13) (see the details under "in vitro enzyme assays" for LPCAT forward reaction).

Experimental Procedures
Identification and Isolation of LPCAT Genes-BLAST (16) searches were performed against the flax genome (17) and expressed sequence tag (EST) (14) databases by using AtLPCAT1 (At1G12640) and AtLPCAT2 (At1G63050) as the protein queries. The LPCAT genes were isolated according to gene isolation and cloning methods described in Pan et al. (5) and then subcloned into a pYES2.1/V5-His-TOPO vector using pYES2.1 TOPO kit (Invitrogen). The plasmids are referred to as pYESTOPO-LPCATs.
Dating the Segmental Duplication Events-To date the segmental duplication events, the K S values (synonymous substitution rates) of the duplicated genes, and 20 homologous gene pairs from their upstream and downstream flanking regions were used to calculate the average K S . The full-length cDNA sequences of the homologous gene pair were first aligned based on their corresponding amino acid translations using Translator X server (18). Then the resulting alignment was further subjected to PAL2NAL (19) to calculate the corresponding K S value. Only K S values between 0 and 1 were included. The approximate date of the duplication event (T) was calculated using the equation T ϭ K S /2 (20), where the synonymous mutation rate per base () of 1.5 ϫ 10 Ϫ8 or 8.1 ϫ 10 Ϫ9 is assumed (17) and the mean K S values from 21 homologous gene pairs is used.
Plasmid Construction-Plasmids used in our study are listed in Table 1. The integrity of all constructs was confirmed by sequencing.
A pair of duplicated DGAT1 genes was previously identified in flax genome by our group (5). We cloned one of the two DGAT1s, named as DGAT1-1, into the pYESBOP vector as described in our previous study (5). In this study we cloned the DGAT1-1 gene into the multiple cloning site 2 (MCS2) of pESC-URA for expression from the GAL1 promoter (P GAL1 ). The resulting plasmid is referred as pECS-DGAT1-1. Next, the region from CYC1 terminator (T CYC1 ) to ADH1 terminator of pESC-URA or pESC-DGAT1-1 was amplified by PCR, and the resulting products were referred as fragment 1. At the same time, the entire plasmid, pYESTOPO-LPCAT1 or pYESBOP, was also amplified by PCR, and the resulting PCR products were referred to as fragment 2. After the PCR amplification, fragment 1 was cloned into fragment 2 using the one-step, isothermal assembly method described by Gibson (21). Four different combinations of fragment 1 with fragment 2 resulted in four different plasmids, which are referred to as follows: 1) pLPCAT1ϩDGAT1-1 contains both LPCAT1 and DGAT1-1 genes, and both genes are under P GAL1 ; 2) pLPCAT1 contains only the LPCAT1 gene under P GAL1 ; 3) pDGAT1-1 contains only the DGAT1-1 gene under P GAL1 ; 4) pCTR is the control plasmid.
To construct the plasmid (pFAD2ϩFAD3ϩHPH) containing the replacement cassette for yeast mutant generation, previously isolated flax FAD2-1 and flax FAD3B genes (5) were first amplified using PCR with the gene-specific primer pairs and then inserted into MSC2 of the pESC-URA, yielding pESC-FAD2 and pESC-FAD3. Next, the cassettes P GAL1 -FAD2-T CYC1 of pESC-FAD2, P GAL1 -FAD3-T CYC1 of pESC-FAD3, and the hygromycin B resistance gene hph (hygromycin B phosphotransferase gene) from the plasmid pAG32 (22) were amplified by PCR and then inserted into the same pYESBOP plasmid using XhoI, EcoRI, and HindIII restriction sites, respectively. The resulting plasmid contained flax FAD2-1 and FAD3B genes under the GAL1 promoter and the selectable marker gene hph.
Yeast Strains-Yeast strains used in this study are listed in Table 2.
Two types of yeast mutant strains (H1246 ⌬ ale1 and H1246 ale1 ⌬::FAD2ϩFAD3) were constructed from the quadruple mutant strain Saccharomyces cerevisiae H1246, in which all four TAG biosynthesis genes (dga1, lro1, are1, and are2) are disrupted (23). H1246 ⌬ ale1 is a quintuple mutant strain. Besides the disrupted four genes in H1246, the YOR175c gene (also referred as ALE1) encoding LPCAT (24) was further deleted by using a disruption cassette containing a gene confer- ring resistance to hygromycin B (hph). Another yeast mutant strain, H1246 ale1 ⌬::FAD2ϩFAD3, was created by replacing the ale1 coding region with the gene replacement cassette containing genes encoding flax FAD2-1, flax FAD3B, and hygromycin B phosphotransferase .
A PCR-mediated gene disruption strategy (25) was used for yeast mutant construction. Using pAG32 and pFAD2ϩ FAD3ϩHPH as templates, the ale1 gene disruption and replacement cassettes were amplified by PCR, and the resulting PCR products were purified and then introduced into H1246 yeast cells using a standard lithium-acetate/PEG/single-strand DNA yeast transformation method (26) with slight modifications. Transformants were selected on YAPD plates containing hygromycin B (300 g/ml) (Invitrogen). The correct disruption or replacement of the ale1 gene in the mutants was verified by PCR with the ale1 gene-specific and disruption cassettespecific/replacement cassette-specific verification primers designed according to Hegemann and Heick (27). The disruption of the ale1 gene was further confirmed by performing the lyso-platelet-activating factor (lyso-PAF) sensitivity tests (the detailed procedure can be found under "Lyso-PAF sensitivity test" below).
Lyso-PAF Sensitivity Test-A Lyso-PAF sensitivity test was performed using the method described in Chen et al. (24) with slight modifications. Yeast mutant strain Y02431 (ale1⌬) transformed with flax LPCATs was first grown in liquid minimal medium (lacking uracil) supplemented with 2% glucose overnight at 30°C with shaking. After washing with sterile water 3 times, the overnight cultures were transferred to induction medium for 12 h. Cultures were then serial-diluted 1:10 from A 600 nm ϭ 2. A measure of 5 l from each dilution was inoculated on minimal medium plates (lacking uracil) containing 0, 5, 10, and 20 g/ml lyso-PAF. Yeast H1246 and Y02431 mutants transferred with empty vector pYES-LacZ were used as positive and negative controls, respectively. The plates were incubated at 28°C for 2 days.
The in Vivo Experiment-The yeast mutant strain (H1246 ale1 ⌬::FAD2ϩFAD3) was used for in vivo studies. In the heterologous yeast expression system, an increased production of unsaturated fatty acids at low growth temperatures has been reported for many higher plant desaturases (28 -30). For this reason, yeast mutant cells transformed with pLPCAT1ϩ DGAT1-1 or pDGAT1-1 were grown at 20°C in induction medium. After 3 days, the cells were harvested and used for yeast lipid analysis. Yeast lipid extraction and analysis was performed according to the method described in Pan et al. (5).
The in Vitro Experiments-Microsomal preparations were performed according to the method described in Pan et al. (5). The crude protein concentration in the microsomal preparations was measured using the Bio-Rad Bradford assay (Bio-Rad). Before the enzymatic assays, the effect of protein concentration, incubation time, and substrate concentrate on enzyme This study pAG32 pAG32-P TEF1 -hphMX (22)  activity was evaluated to optimize enzymatic reaction conditions for measuring substrate specificities of the DGAT-and LPCAT-catalyzed forward reactions.
DGAT Enzyme Assays-The DGAT enzyme assays were performed using the method described by Byers et al. (31), with some modifications. Microsomal preparations from H1246 cells transformed with pYES-DGAT1-1 were used in the assays. Control microsomes were prepared from H1246 yeast transformed with pYES-LacZ. The reaction mixture contained 5 g of microsomal protein, 200 mM HEPES-NaOH, pH 7.4, 3.2 mM MgCl 2 , 333 M sn-1,2-diolein, and 15 M [ 14 C]acyl-CoA in a total volume of 60 l. The reaction was incubated at 30°C for 6 min and then quenched by the addition of 10 l of sodium dodecyl sulfate 10% (w/v). The reaction was then spotted on a silica thin layer chromatography (TLC) plate (SIL G25, 0.25 mm, Macherey-Nagel, Germany). The plate was developed with hexane/diethyl ether/acetic acid (80:20:1, v/v/v). Corresponding TAG spots were scraped from the plate, and radioactivity was measured by a LS 6500 multi-purpose scintillation counter (Beckman-Coulter).
LPCAT Enzyme Assays-The LPCAT enzyme assays were performed according to the method previously described by Lager et al. (13) with slight modifications. Microsomes were prepared from the yeast mutant strain Y02431 (ale1⌬) transformed with pYESTOPO-LPCAT1. Microsomes prepared from yeast containing pYES-LacZ were used as a negative control.
For testing the forward activity, the reaction mixture contained 0.3 g of microsomal protein, 200 M [ 14 C]18:1-LPC, 100 M acyl-CoA, and 0.1 M potassium phosphate buffer, pH 7.2, in a total volume of 50 l. The reaction was incubated at 30°C for 5 min and then terminated by the addition of 170 l of 5% acetic acid and 500 l of chloroform/methanol (1:1, v/v). The mixture was extracted and then applied to a TLC plate. The TLC plate was developed with a solvent system of chloroform/ methanol/formic acid (75:25:10, v/v/v). Radioactivity of the PC spots was determined by scintillation counting.
For measuring the reverse activity, an aliquot of microsomal protein (43 g, which equals ϳ10 nmol of endogenous PC) was lyophilized overnight. The substrate, 9 nmol of sn-1-16:0-sn-2-[ 14 C]acyl-PC (2.5 nCi/nmol), was dissolved in 14 l of benzene and applied directly to the freeze-dried microsomes. The benzene was dried immediately under N 2 at 37°C, after which 100 l of reaction mixture containing 200 nmol of free CoA, 10 nmol of 18:1-CoA, and 1 mg of bovine serum albumin (BSA) in 0.1 M potassium phosphate buffer, pH 7.2, was added. The assay was incubated at 30°C for 60 min and then quenched with 100 l of 0.15 M acetic acid and 500 l of chloroform/methanol (1:2, v/v). The mixture was then extracted with the Bligh and Dyer method (32). The [ 14 C]acyl-CoAs in the upper phase was further hydrolyzed to free fatty acids with KOH. After neutralization with HCl, the free fatty acids were extracted into chloroform. Radioactivity of 14 C-labeled free fatty acids was determined by scintillation counting.
LPCAT1 and DGAT1-1 Coupling Enzyme Assays-The yeast quintuple mutant strain H1246 ⌬ ale1 was used to test the coupling hypothesis under in vitro conditions. Microsomes were prepared from the cells transformed with pLPCAT1ϩ DGAT1-1, pLPCAT1, or pDGAT1-1. Control microsomes were prepared from yeast cells transformed with pCTR. A 43-g portion of microsomal preparations was lyophilized overnight. The substrates of 9 nmol of sn-1-16:0-sn-2-[ 14 C]acyl-PC (2.5 nCi/nmol) and 32 nmol of sn-1,2-diolein were dissolved in 14 l of benzene and applied directly to the freeze-dried microsomes. The benzene was dried immediately under N 2 at 37°C. The assay was started by adding 100 l of reaction mixture containing 200 nmol of free CoA, 10 nmol of 18:1-CoA, and 1 mg of BSA in 0.1 M potassium phosphate buffer, pH 7.2. After incubation for 60 min at 30°C, the assay was terminated by the addition of 100 l of 0.15 M acetic acid and 500 l of chloroform/methanol (1:2, v/v). Lipids were extracted into chloroform and then applied to the TLC plate. The resulting TLC plate was first developed in a solvent system of chloroform/ methanol/acetic acid/H 2 0 (60:30:3:1, v/v/v/v) to half-plate. After drying in a N 2 -filled chamber, the plate was then developed with a solvent system of hexane/diethyl ether/acetic acid (80:20:1, v/v/v) to full plate. The corresponding TAG spots were scraped, and radioactivity was measured by scintillation counting. The coupling activities were determined as the amount of [ 14 C]TAG produced.

Results
As a prelude to investigating possible coupling of the LPCAT-catalyzed reaction to the DGAT1-catalyzed reaction, we first examined the properties of individual flax enzymes.
Identification and Isolation of LPCAT Genes from Flax-Through exploring flax genome and EST databases, we identified a pair of duplicated LPCAT genes in flax. Both LPCAT genes have a coding region of 1380 bp and encode proteins with 460-amino acid residues. These two LPCATs share 96.3% and 96.5% identity at the nucleotide and amino acid level, respectively. Both LPCAT genes contain eight exons and seven introns (Fig. 3A). Further genomic analysis revealed that the proteincoding genes flanking the LPCAT genes are highly conserved, suggesting that the duplicated LPCAT genes were derived from a segmental duplication event. In addition, the mean K S value of 21 homologous gene pairs (including the LPCAT gene pair) within the segmental duplication blocks was used as a proxy of time to estimate the date of the segmental duplication event. The result presented in Fig. 3B shows that a segmental duplication event occurred ϳ5.1-9.5 million years ago (MYA), which is consistent with the time of a recent (5-9 MYA) whole-genome duplication (WGD) in flax (17), suggesting that the segmental duplication is the result of a recent WGD event.
LPCAT cDNAs from Flax Encode Functionally Active LPCAT Enzymes-To determine whether LPCAT cDNAs from flax encode proteins possessing LPCAT activity, the cDNAs of both LPCATs were amplified from flax embryos and cloned into the yeast expression pYES2.1TOPO vector. The resulting recombinant plasmids were transformed into the yeast mutant strain Y02431 (ale1⌬), which has a very low LPCAT background activity (Chen et al.;Ref. 24). After transformation, a lyso-PAF sensitivity test was performed to confirm the proper expression of LPCAT genes in yeast mutant Y02431. Yeast H1246 and Y02431 mutants containing pYES-LacZ were used as positive and negative controls, respectively. The result (Fig. 3C) showed that there was a severe growth defect in yeast mutant Y02431 transformed with pYES-LacZ when the plate was supplemented with 10 g/ml lyso-PAF. No growth was observed for the negative control when 20 g/ml lyso-PAF were added into the plates. However, yeast cells expressing LPCAT1 or LPCAT2 were able to overcome the growth inhibition of lyso-PAF and exhibited a growth even better than the vector-only H1246 transformants in the presence of 5, 10, and 20 g/ml lyso-PAF. These results confirmed that these recombinant yeast mutant Y02431 can be further used for enzyme assays. Given our previous finding that the recently duplicated genes in flax, such as DGAT2-1/DGAT2-3 and PDAT1/PDAT5, exhibit similar gene expression patterns and encode proteins with similar substrate selectivity (5), only one of the duplicated LPCAT genes, LPCAT1, was used for subsequent analyses.
Flax DGAT1-1 Displayed Small but Significantly Enhanced Specificity for 18:3-CoA-We previously identified a pair of DGAT1 genes with 97.7% sequence identity in the flax genome (5). K S analysis showed that similar to LPCATs, the duplicated DGAT1s were derived from a recent whole-genome duplication (Fig. 3B). Our previous study (5,33) showed that flax DGAT1-1 was able to complement the TAG synthesis deficiency in yeast mutant strain H1246, in which all four TAG biosynthesis genes are disrupted. Experiments of in vivo fatty acid feeding indicated that DGAT1-1 produced a similar amount of TAG in the presence of diverse, exogenously provided fatty acids, suggesting that DGAT1-1 may have a broad substrate preference (5). In this study, we further examined fatty acyl-CoA substrate specificity for flax DGAT1-1 under in  vitro conditions. The same yeast mutant strain, H1246, was used because it provides a clean background to detect [ 14 C]TAG produced from the introduced flax DGAT1-1. The assays were carried out with microsomal preparations from the H1246 mutant expressing flax DGAT1-1. Different [ 14 C]acyl-CoAs were provided along with nonradiolabeled sn-1,2-dioleoylglycerol. As shown in Fig. 5, DGAT1-1 exhibited small but significant changes of substrate specificity toward the acyl-CoA substrates, and the highest specificity was observed with 18:3-CoA, which is in close agreement with substrate specificity data for overall microsomal DGAT activity from developing flax embryos (34).
In Vivo Evidence of Biochemical Coupling of LPCAT1 and DGAT1-1-catalyzed Reactions-After enzymatic characterization of individual flax LPCAT1 and DGAT1-1 enzymes, we started testing the coupling hypothesis in a yeast system under in vivo conditions. To avoid interference from endogenous TAG-synthesizing and LPCAT activities, we constructed a new yeast mutant strain from H1246. In the newly generated mutant strain, referred to as H1246 ale1 ⌬::FAD2ϩFAD3 mutant strain (Table 2), the endogenous ale1 gene was replaced by genes encoding FAD2-1 and FAD3B from flax; thus, no significant TAG-synthesizing and LPCAT activities were present in the mutant. The rationale behind inserting FAD2 and FAD3 into the yeast chromosome is that: 1) 18:2 and 18:3 do not naturally occur in the H1246 mutant, and 2) FAD2 and FAD3 catalyze the synthesis of 18:2 and 18:3, respectively, on the sn-2 position of PC, which is also the major substrate for the LPCAT reverse reaction. The newly created yeast strain containing flax FAD2-1 and FAD3B genes can synthesize 18:2 and 18:3 on the sn-2 position of PC, and in this way, we avoided exogenous fatty acid feeding, which usually gives a random distribution of fatty acids over all sn positions in all types of lipid classes. Constructs containing flax DGAT1-1 alone or both DGAT1-1 and LPCAT1 were then individually introduced into this newly generated yeast strain. We hypothesized that the reverse action of LPCAT can catalyze the transfer of PUFAs produced on PC directly into the acyl-CoA pool, making the PUFA-CoAs available for the DGAT1-catalyzed reaction for TAG production. Consistent with our hypothesis, whereas no significant differences in the total amount of TAG were found between yeast cells expressing flax DGAT1-1 and yeast cells co-expressing flax DGAT1-1 and LPCAT1 (Fig. 6A), co-expressing flax DGAT1-1 and LPCAT1 significantly increased 18-carbon PUFAs (18:2 and 18:3) in TAG with a concomitant decrease of 18-carbon PUFAs in phospholipid, as opposed to those expressing DGAT1-1 only (Fig. 6, B and C).
In Vitro Evidence of Biochemical Coupling of LPCAT1 and DGAT1-1 Reverse Reactions-Next, we tested the coupling hypothesis under in vitro conditions. For this purpose, we constructed another quintuple yeast mutant, referred as H1246 ⌬ ale1 (Table 2), in which the endogenous ale1 gene of H1246 was disrupted. The newly generated quintuple mutant was unable to synthesize TAG and had a low background LPCAT activity. The mutant was then transformed with pLPCAT1ϩDGAT1-1, pLPCAT1, pDGAT1-1, or pCTR, individually. After induction, microsomes of the recombinant yeast cells were isolated and fed with PC radiolabeled with 14 C at the sn-2 position. We hypothesized that the biochemical coupling can enhance the transfer of 14 C-labeled acyl groups from PC into TAG via the reverse reaction of LPCAT1 coupled to the DGAT1-1-catalyzed reaction. As shown in Fig. 7, A and B In contrast, no [ 14 C]TAG was detected after incubation of the radiolabeled PC with microsomes containing either flax LPCAT1 or flax DGAT1-1 alone. In addition, no [ 14 C]TAG could be detected when incubations were conducted with control microsomes. This result confirmed our hypothesis. It should be noted that, as shown in Fig. 7A, the assays also produced 14 C-free fatty acids, which may arise from the hydrolysis of [ 14 C]acyl-CoAs catalyzed by the endogenous acyl-CoA thioesterase (13).
We further tested the acyl specificities in overall LPCAT1 and DGAT1-1 coupling process by presenting sn-2-[ 14 C] 18:1-,  18:2-, or 18:3-PC as single substrates to the microsomes containing both flax LPCAT1 and flax DGAT1-1. As shown in Fig.  7C, yeast microsomes containing flax LPCAT1 and flax DGAT1-1 were able to transfer the 14 C-labeled 18:2 or 18:3 moiety at a higher rate than 18:1 moiety from the sn-2 position of PC to TAG, which was consistent with our in vivo data.

Discussion
Using acyl-CoAs as donors, two major biochemical pathways, the Kennedy pathway (de novo pathway) (35) and the Lands cycle (remodeling pathway) (10), contribute to PC biosynthesis (Fig. 1). Both pathways were proposed in 1950s. PC serves not only as a major component of cellular membrane but also as the primary site for PUFA or unusual fatty acid biosynthesis. The accumulation of a large proportion of PUFAs in flax seed oils necessitates efficient mechanisms to channel PUFAs from PC into TAG. Our previous studies have shown that PDAT-mediated TAG biosynthesis (5) and phosphatidylcholine:diacylglycerol cholinephosphotransferase-based PC-DAG interconversion (36) play critical roles in regulating PUFA content in flax seed oil. The Kennedy pathway and the Lands cycle are connected by LPCAT-catalyzed reactions. Here, we hypothesize that the reverse reaction of LPCAT can be coupled to the DGAT1 reaction to produce the PUFA-containing TAG.
Our hypothesis is supported by the following observations. 1) Acyl specificity tests showed that the reverse reaction of flax LPCAT1 has the highest preference toward PUFA-containing PC (Fig. 4B). In addition, the flax DGAT1-1 forward reaction showed a small but significant preference for 18:3-CoAs (Fig.  5). 2) Under in vivo conditions, DGAT1-1 produced a higher amount of PUFA-containing TAG and a lower amount of PUFA-containing phospholipid in the presence than in the absence of LPCAT1 (Fig. 6, B and C). 3) Under in vitro conditions, when sn-1-16:0-sn-2-[ 14 C]acyl-PC was included as the substrate, radiolabeled [ 14 C]TAG was formed only in the reaction with the yeast microsomes containing both flax LPCAT1 and flax DGAT1-1 (Fig. 7, A and B). In contrast, under the same assay conditions, no [ 14 C]TAG was detected in reactions with microsomes containing either flax LPCAT1 or flax DGAT1-1 alone. These results provided strong evidence supporting our hypothesis. 4) The acyl group specificity of overall LPCAT1 and DGAT1-1 coupling process under in vitro conditions showed that yeast microsomes containing DGAT1-1 and LPCAT1 were able to transfer the 14 C-labeled linoleoyl or linolenoyl moiety at a higher rate than the oleoyl moiety from the sn-2 position of PC to TAG (Fig. 7C).
Given that PUFA groups could be deacylated by the action of PLA on PC and previous work (37) indicated that the expression of PLA 2 can be affected by altered expression of LPCAT, there was the possibility that expression of LPCAT triggered the expression of PLA 2 , which could contribute to the increased PUFA content in TAG. It is important to note that PLA-catalyzed deacylation of PC produces free fatty acids. Before they can be used by the DGAT reaction, free fatty acids need to be activated to acyl-CoAs. The formation of the acyl-CoAs is catalyzed by acyl-CoA synthetase, which requires ATP for its action. However, the assay mixture used in this study did not contain ATP. More importantly, we failed to detect any [ 14 C]TAG products in the reaction with microsomes containing flax DGAT1-1 alone (Fig. 7, A and B) indicating that under our assay conditions, there was little or no interference from PLA. Interference may also arise from PC to DAG conversion catalyzed by the phospholipase C reaction or a combined reaction of phospholipase D/phosphatidic acid phosphatase. The resulting PUFA-containing DAG may be further incorporated into TAG by the DGAT-catalyzed reaction and lead to an increased content of PUFAs in TAG. For the same reason described above, this interference was negligible because we did not detect any [ 14 C]TAG when microsomes contained only recombinant flax DGAT1-1. Taken together, our data support the hypothesis that the transfer of PUFAs formed on PC to the acyl-CoA pool and further incorporated into TAG is at least to some extent catalyzed through coupling between the reverse reaction of flax LPCAT1 and the forward reaction of flax DGAT1-1.
It was proposed nearly 30 years ago that LPCAT can catalyze both forward and reverse reactions in plants (9). However, equilibrium of the LPCAT reaction favors the exothermic forward reaction. It was not until recently that Stymne and coworkers (13) proved that plant LPCAT enzyme does in fact operate in a reversible fashion under in vitro conditions. An intriguing question remains: How does the endothermic reverse reaction take place in vivo? It was previously suggested that for the LPCAT reverse reaction to take place in vivo, the PC content and free CoA has to be very high in compared with the LPC content and available acyl-CoA within the cell (13). Based on our data, we propose an alternative scenario as follows: under certain metabolic conditions, the endothermic LPCATcatalyzed reverse reaction can be pulled via coupling it to the DGAT1-catalyzed forward reaction, which is strongly exothermic and has a shared intermediate, acyl-CoA. The effective removal of acyl-CoAs from the acyl-CoA pool catalyzed by DGAT1 can lower the concentration of acyl-CoA and thus stimulate the LPCAT-catalyzed reaction to operate in the reverse direction. It should be noted that the LPCAT-catalyzed reaction is not a part of a linear biosynthetic pathway, but it is the combined forward and reverse reaction of the enzyme that catalyzes an exchange of acyl groups between PC and acyl-CoA pools where the acyl-CoA and LPC formed by the reverse reaction can be immediately re-acylated by the forward reaction. As shown in Fig. 4A, the forward reaction of flax LPCAT1 has highest specificities toward the PUFA-CoAs, indicating that once transferred to the acyl-CoA pool, PUFA groups can be quickly re-acylated into PC by the forward action of LPCAT. To maintain a sustained reverse reaction, the re-acylation of the formed LPC by the forward reaction of LPCAT has to occur simultaneously. In addition, the specific activity of flax LPCAT1-catalyzed forward reaction is ϳ1000 times higher than that of flax DGAT1-catalyzed reaction when the expressed recombinant enzymes were assessed in yeast microsomes under in vitro enzyme assays ( Fig. 4A and Fig. 5). This forward reaction of LPCAT has the potential to compete with the DGAT1-1-catalyzed reaction for PUFA-CoA substrates. Therefore, to produce PUFA-enriched TAG, especially ALAenriched TAG, the DGAT reaction has to be highly selective toward ALA-CoAs. It should be noted that even though flax DGAT1-1 showed enhanced substrate specificity toward ALA-CoAs, the differences in substrate specificity were fairly small (Fig. 5). Our previous in vivo fatty acid feeding experiments showed that recombinant flax DGAT2 appears to have a stronger preference than flax DGAT1-1 toward ALA-containing substrates (5). The activity of flax DGAT2 in the yeast system, however, was very low, which limits its potential use in testing the coupling hypothesis in this study. In the future it will be very interesting to find out whether using codon-optimized flax DGAT2 and flax LPCAT1 can further enhance the ALA content in TAG. The enhanced specificity of DGAT2 for ALA-CoA may provide a further operational advantage in the coupling process, and the extent of this coupling process could potentially be influenced by the fatty acyl composition of the acyl-CoA pool. Furthermore, LPCAT and DGAT enzymes might physically interact with each other in the membrane. If LPCAT is physically associated with DGAT1 in vivo, the products of the LPCAT-catalyzed reverse reaction, PUFA-CoAs, can be potentially channeled directly into DGAT, providing an enhanced efficiency of PUFA-CoA transfer and establishing a favorable operating condition for the reverse reaction of LPCAT. Thus, substrate selectivity and substrate channeling may work collectively to promote the incorporation of PUFAs into TAG while overcoming a formidable thermodynamic barrier.
It is well known that PC serves not only the primary site for PUFA biosynthesis but also as the major site for biosynthesis of a number of unusual fatty acids. Our results show that the reverse reaction of LPCAT1 can be coupled with the DGAT1 reaction to enhance PUFA content in TAG. The biochemical coupling mechanism found in flax might also come into play in other plant systems, such as castor, which accumulates oil enriched in the unusual fatty acid known as ricinoleic acid. It may be difficult, however, to test the hypothesis of biochemical coupling of castor LPCAT-catalyzed and DGAT-catalyzed reactions using a yeast system because of potential interference by endogenous thioesterase activity, which appears to readily catalyze the hydrolysis of hydroxyacyl-CoA (13). If the hydroxyacyl-CoA generated by the castor LPCAT-catalyzed reaction, however, is immediately used by the castor DGATcatalyzed reaction and castor DGAT exhibits a high affinity for hydroxyacyl-CoA, then the endogenous thioesterase activity may not be an issue. Indeed, as indicated in the previous study, castor DGAT2 exhibits enhanced preference for hydroxyacyl-CoAs (38). The high activity of castor DGAT2 on hydroxyacyl-CoA may keep the free concentration of hydroxyacyl-CoA low enough to effectively pull LPCAT in the reverse direction. Further experiments are needed to confirm the biochemical coupling of LPCAT and DGAT in other plant species.
In conclusion, seed oils producing enhanced levels of PUFAs or unusual fatty acids have a wide range of food, feed, and industrial applications. Increasing our insight into the incorporation of PC-modified fatty acids into TAG can provide new strategies to effectively modify seed oils for different uses. In this study we provide direct in vivo and in vitro evidence for biochemical coupling of the flax LPCAT1-catalyzed reverse reaction to the flax DGAT1-1-catalyzed reaction as a route for enhancing the amount of PUFA in TAG. A similar mechanism may apply to other plant species that accumulate PUFAs or unusual fatty acids in their oils. Our results provide an increased understanding of PC metabolism in relation to the Kennedy pathway for generating PUFA-enriched TAG and will benefit future initiatives aimed at enhancing the PUFA content of TAG in plants.
Author Contributions-X. P. planed and performed all experiments, carried out data analyses, and drafted the article. R. J. W. and X. P. designed the research. J. Z., G. C., M. K., M. S. G., and K. M. C. contributed valuable discussions during this study. All co-authors contributed to the revision of this work.