Triacylglycerols (TAGs) are the major storage lipids which accumulate in developing seeds, flower petals, anthers, pollen grains, and fruit mesocarp of a number of plant species [1, 2]. TAGs are thought to be not only the major energy source for seed germination but also essential for pollen development and sexual reproduction in many plants [3, 4]. In oilseeds, TAG bioassembly is catalyzed by the membrane-bound enzymes of the Kennedy pathway that operate in the endoplasmic reticulum . The biosynthesis of TAGs is catalyzed by the sequential acylation of the glycerol backbone involving three acyltransferases: glycerol-3-phosphate acyltransferase (GPAT), lyso-phosphatidic acid acyltransferase (LPAAT) and diacylglycerol acyltransferase (DGAT). DGAT catalyses the final acylation of DAG to give TAG, which has been suggested to be the rate-limiting step in plant lipid accumulation.
In the traditional Kennedy pathway DGAT was thought to be the only enzyme that is exclusively committed to TAG biosynthesis using acyl-CoA as its acyl donor. The first DGAT gene was cloned from mouse and is a member of the DGAT1 family, which had high sequence similarity with sterol: acyl-CoA acyltransferase .
We had previously characterized an EMS-induced mutant of Arabidopsis, designated AS11, which displayed a decrease in stored seed TAG, delayed seed development, and an altered fatty acid composition . We analyzed WT vs. AS11 lipid pools and Kennedy pathway enzyme activities in fractions isolated from green mid-developing seed, and performed parallel labeling of intact seeds at this developmental stage, with [14 C] acetate. We found that compared to WT, there was an increase in all fatty acids in the DAG pool of AS11 seeds at mid-development, and, to a lesser extent, an associated backup of fatty acids in the PC pool. DAG was elevated from 1% in WT to 10-12% in AS11 and PC pools were elevated from about 2% in WT, to 8-12% in AS11. Cell-free fractions from WT and AS11 green seeds at mid-development were compared for their ability to incorporate [14 C]-18:1-CoA into glycerolipids in the presence of G-3-P. Proportions of labeled LPA and PA formed during the incubation period were similar in WT vs AS11, indicating that the activities of the Kennedy pathway enzymes GPAT and LPAAT (EC 18.104.22.168) were relatively unaffected in the AS11 mutant. However, the proportion of labeled TAG was much lower and that of DAG was much higher in AS11. The TAG/DAG ratio was therefore consistently 3-5-fold lower in AS11 compared to WT at all developmental stages (early-, mid- and late development). Furthermore, the ratio of 18:3/20:1 dramatically increased about 7-10 fold .
Cumulatively, this data suggested a lesion in DGAT1 which was subsequently demonstrated upon cloning the mutated gene from AS11. There is an 81 bp in-frame insertion consisting entirely of exon 2 in the transcript from AS11. The exon 2 in the repeat is properly spliced, thus the alteration of the transcript does not disturb the reading frame. However, this additional exon 2 sequence in the AS11 transcript would result in an altered DGAT protein with a 27 amino acid insertion (131SHAGLFNLCVVVLIAVNSRLIIENLMK157) . It is important to note that two other labs independently and simultaneously cloned the A. thaliana
DGAT1 [8, 9].
Studies manipulating the expression of DGAT1 followed: We demonstrated that expression of the Arabidopsis DGAT1 cDNA in a seed specific manner in the AS11 mutant restored wild type levels of TAG and VLCFA content. The acyl distribution, specifically, the sn-3 composition of the TAGs, was also restored to WT proportions. Furthermore, overexpression of the Arabidopsis DGAT1 in wild type plants led to an increase in seed oil content and seed weight . Over the past 10 years, DGAT1 expression has been genetically manipulated to produce Brassica napus prototypes containing increased seed oil [11, 12].
A second family of DGAT genes (DGAT2), first identified in the oleaginous fungus Morteriella ramanniana, has no sequence similarity with DGAT1 . A human DGAT2 and plant DGAT2s from tung and castor were subsequently identified by Cases et al. , Shockey et al.  and Kroon et al. , respectively. The putative DGAT2 from Arabidopsis has been studied by several labs including ours; functional expression in yeast has not been successful, and therefore whether it is a true functioning DGAT is still in question. Notably, an Arabidopsis dgat2 knockout mutant has a wild-type seed oil content and fatty acid composition .
TAG can also be formed by an acyl-CoA-independent enzyme, phosphatidylcholine: diacylglycerol acyltransferase (PDAT), in which the transfer of an acyl group from the sn-2 position of PC to the sn-3 position of DAG yields TAG and sn-1 lyso-PC [18, 19]. During the exponential growth phase in yeast, PDAT1 is a major determinant in TAG synthesis. In Arabidopsis, two close homologs to the yeast PDAT gene have been identified: PDAT1 At5g13640 and PDAT2 At3g44830 . Mhaske et al.  isolated and characterized a knockout mutant of Arabidopsis which has a T-DNA insertion in the PDAT1 locus At5g13640 (PDAT1, EC 22.214.171.124). Lipid analyses were conducted on this mutant to assess the contribution of PDAT1 to seed lipid biosynthesis; surprisingly, and in contrast to the situation in yeast, the oil content and composition in seeds did not show significant changes in the mutant. At the time, these results were interpreted to indicate that PDAT1 activity as encoded by At5g13640 is not a major determining factor for TAG synthesis in Arabidopsis seeds.
Nonetheless, because the Arabidopsis DGAT1 mutant AS11 shows only a 30-35% decrease in oil content [6, 9], it was apparent that other enzymes must contribute to oil synthesis in the developing seed . Thus, an examination of the contribution of DGAT2, PDAT2 or PDAT1 to oil deposition in an AS11 background was studied by performing double mutant crosses with AS11 . While the dgat2-ko line has no oil phenotype, homozygous double mutants from cross of AS11 with dgat2-ko mutant showed an oil fatty acid profile similar to AS11. The same pattern was observed with the pdat2-ko mutant alone and in crosses of the pdat2-ko mutant with AS11. In contrast, while the pdat1-ko has no oil or fatty acid composition phenotype, crosses of the pdat1-ko with AS11 were embryo-lethal in the double homozygous condition; only heterozygous lines produced by having expression of the pdat1 or dgat1 gene only partially inhibited using RNAi, allowed an examination of the double mutants. These detailed studies resulted in the finding that DGAT1 and PDAT1 have overlapping functions in both embryo development and TAG biosynthesis in the developing seed and in pollen. When DGAT1 is compromised in AS11, it is PDAT1, and not DGAT2 or PDAT2, that is responsible for the remaining 65-70% of TAG which is synthesized. This finding suggested a major, perhaps dominant role of PDAT1 in this process .
Recently, a castor bean-specific PDAT, PDAT1-2, was cloned and found to be highly expressed in developing seeds and localized in the ER, similar to the castor FAH12 hydroxylase. Transgenic Arabidopsis co-expressing the castor PDAT1-2 and FAH12 showed enhanced ricinoleate accumulation to up to 25% in TAGs (compared to 17% in FAH12 -only transgenics) [23, 24]. This study suggests that specialized PDATs may play a significant role in channeling PC-synthesized unusual fatty acids such as ricinoleic (from castor), or epoxy fatty acids (e.g. from Vernonia galamensis), into TAGs.
The discovery of the critical role PDAT1 has in TAG synthesis  suggested the importance of re-acylation of lysophospholipids and especially LPC, since it is produced as a result of PDAT1 activity. We hypothesized that in a situation where the responsibility for TAG synthesis is shifted to PDAT1 such as in the AS11 dgat1 mutant, profound changes in activity of re-acylation enzymes would be evident. As candidates of genes coding such enzymes, we proposed the two Arabidopsis genes LPLAT1 (At1g12640) and LPLAT2 (At1g63050), characterized by Ståhl et al.  as possessing broad specificity lysophospholipid acyltransferase activity. Although both of the genes to some extent could acylate a range of different lysophospholipids, they were shown to have a strong preference towards LPC. For this reason we chose to name the two genes LPCAT1 (At1g12640) and LPCAT2 (At1g63050), in the current study.
Here we report the further genetic and biochemical characterization of the AS11 dgat1 mutant. During the course of microarray and qRT-PCR studies of AS11 vs WT gene expression in mid-developing siliques, we found that LPCAT2, encoding acyl-CoA:lysophosphatidylcholine acyltransferase 2 (EC 126.96.36.199), was up-regulated while LPCAT1 was not affected. By a series of biochemical studies and key crosses of AS11 with either lpcat1 or lpcat2 knockout mutants, we determined that LPCAT2 (and much less so LPCAT1) is most critical for TAG synthesis in the AS11 mutant, primarily to maintain the PC pool for TAG assembly primarily catalyzed by PDAT1.