Expression of Fungal diacylglycerol acyltransferase2 Genes to Increase Kernel Oil in Maize [OA]

Maize ( Zea mays ) oil has high value but is only about 4% of the grain by weight. To increase kernel oil content, fungal diacylglycerol acyltransferase2 ( DGAT2 ) genes from Umbelopsis (formerly Mortierella ) ramanniana and Neurospora crassa were introduced into maize using an embryo-enhanced promoter. The protein encoded by the N. crassa gene was longer than that of U. ramanniana . It included 353 amino acids that aligned to the U. ramanniana DGAT2A protein and a 243-amino acid sequence at the amino terminus that was unique to the N. crassa DGAT2 protein. Two forms of N. crassa DGAT2 were tested: the predicted full-length protein (L-NcDGAT2) and a shorter form (S-NcDGAT2) that encoded just the sequences that share homology with the U. ramanniana protein. Expression of all three transgenes in maize resulted in small but statistically signiﬁcant increases in kernel oil. S-NcDGAT2 had the biggest impact on kernel oil, with a 26% (relative) increase in oil in kernels of the best events (inbred). Increases in kernel oil were also obtained in both conventional and high-oil hybrids, and grain yield was not affected by expression of these fungal DGAT2 transgenes. line advancement and ﬁeld trials. Statistical analysis Greenhouse support provided High-oil maize lines were

The demand for vegetable oil is increasing both for human consumption and more recently for use in the manufacture of biodiesel. One way to meet this demand without devoting additional land to agriculture is to increase the yield of oil per acre. Although oil makes up only a small proportion of maize (Zea mays) grain (approximately 4%), maize oil is the second in volume vegetable oil produced in the United States, after soybean (Glycine max) oil (Ash and Dohlman, 2006). Maize kernel oil is composed primarily of triacylglycerol (TAG) and is used as an energy source by the plant during seed germination (Tan and Morrison, 1979). Most of the oil in a maize seed is found in the embryo, which is about 33% oil by weight (Watson, 1987). Breeding approaches to increase maize kernel oil have had some success, but high kernel oil concentration has been linked to a decrease in grain yield (Dudley et al., 1977;Misevic and Alexander, 1989). Transgenic approaches to increasing oil in the maize embryo have shown promise in allowing increased oil production without yield drag (Zheng et al., 2008;Shen et al., 2010). For example, overexpression of the embryo development transcription factor LEAFY COT-YLEDON1 (ZmLEC1) in maize increased kernel oil by up to 48% (Shen et al., 2010). Unfortunately, this transcription factor also produced negative pleiotropic effects, including reduced seed germination and leaf growth. However, overexpression of a transcription factor downstream of ZmLEC1, maize WRINKLED1 (ZmWRI1), resulted in similar increases in kernel oil but without any detectable off types (Shen et al., 2010). ZmWRI1 appears to control genes involved in glycolysis and fatty acid biosynthesis but not oil biosynthesis, and it is hypothesized that additional oil increases may be obtained by combining overexpression of this transcription factor with genes that have a more direct effect on oil biosynthesis.
Diacylglycerol acyltransferase (DGAT) catalyzes the final step of TAG biosynthesis by transferring an acyl group from acyl-CoA to the sn-3 position of 1,2diacylglycerol (DAG; Kennedy, 1961). Two classes of DGAT genes have been identified in eukaryotes, DGAT1 and DGAT2 (Lardizabal et al., 2001;Lung and Weselake, 2006). DGAT1 is related to the acyl-CoA:cholesterol acyltransferase family (Cases et al., 1998). DGAT2 has little sequence similarity to DGAT1 and is part of a family of proteins that transfer acyl groups from CoA to neutral lipids (Lardizabal et al., 2001;Turkish and Sturley, 2007). Whether these two classes of DGATs have redundant or specific functions in TAG biosynthesis in plants is not clear. They may function at different times during development or provide different fatty acid specificities (Lung and Weselake, 2006;Li et al., 2010aLi et al., , 2010b. Ectopic expression of DGAT transgenes from various plant and fungal species has been shown to increase seed oil content in Arabidopsis (Arabidopsis thaliana; Jako et al., 2001), soybean (Lardizabal et al., 2008), and maize (Zheng et al., 2008).
In addition to DGAT, other enzymes have also been shown to be involved in TAG biosynthesis in plants, including phospholipid:diacylglycerol acyltransferase (PDAT) and DAG transacylase (Lung and Weselake, 2006).
Mutation of DGAT1 in Arabidopsis reduced seed oil content 25% to 35% (relative to the wild type), supporting a role for DGAT1 in oil deposition during seed development but also suggesting the participation of other acyltransferases or transacylases (Routaboul et al., 1999;Zou et al., 1999). Work with Arabidopsis mutants established that PDAT1 participates in the accumulation of TAG in Arabidopsis seeds and suggested that DGAT1 and PDAT1 activities may account for all TAG biosynthesis in Arabidopsis seeds ).
DGAT2 appears to play an important role in oil accumulation in seeds of other species. In both castor bean (Ricinus communis; Kroon et al., 2006) and tung tree (Vernicia fordii; Shockey et al., 2006), DGAT2 is more highly expressed in seeds than DGAT1. Furthermore, in tung tree, the DGAT2 form was shown to have greater selectivity for eleostearic acid, an unusual fatty acid found at high concentration in tung tree seeds (Kroon et al., 2006). A survey of the expression patterns of DGAT1, DGAT2, and PDAT genes in a variety of species with seeds containing epoxy and hydroxyl fatty acids found that while all three enzymes are important in oil accumulation in these species, DGAT2 and PDAT are more highly expressed in developing seeds of these species than in Arabidopsis or soybean, supporting the concept that PDAT and DGAT2 are involved in the accumulation of these special fatty acids (Li et al., 2010b).
In maize, an allele of a DGAT1 gene (DGAT1-2) was shown to be responsible for a high-oil quantitative trait locus on chromosome 6 (Zheng et al., 2008). Cloning of the high-oil DGAT1-2 allele revealed that a Phe is present at position 469 of the encoded protein that is absent in DGAT1-2 from maize lines with oil content typical of commercial lines (normal-oil content). The high-oil DGAT1-2 protein was shown to have greater enzymatic activity compared with the normal-oil form. Overexpression of both the high-oil and normal-oil DGAT1-2 genes in transgenic maize was performed using an embryo-enhanced promoter, and the transgenic lines had increased kernel oil and increased levels of kernel oleic acid, a trait also associated with the high-oil quantitative trait locus on chromosome 6 (Zheng et al., 2008). Maize plants with the high-oil allele had larger increases in oil content and oleic acid levels than lines that contained the normal-oil allele transgene. The line with the largest increase had a 41% (relative) increase in seed oil content and a 107% (relative) increase in oleic acid.
Overexpression in soybean of a codon-optimized version of the DGAT2A gene from the fungus Umbe-lopsis ramanniana using a seed-specific promoter resulted in an increase in seed oil content of 1.5% by weight (an approximately 7.5% relative increase) without any effect on grain yield (Lardizabal et al., 2008). We introduced this same UrDGAT2A gene into maize using an embryo-enhanced promoter and demonstrated a small but significant increase in kernel oil of up to 0.7% by weight (a relative increase of 19%). We also cloned and overexpressed a DGAT2 gene from Neurospora crassa and observed slightly higher DGAT enzyme activities and oil increases up to 0.9% by weight (a 26% relative increase). We observed no effect on hybrid grain yield from overexpression of these transgenes. Importantly, oil increases were obtained from overexpression of the fungal DGATs in high-oil hybrids, indicating the possibility of further enhancing oil biosynthesis in high-oil germplasm. An unexpected seedling phenotype was observed in some of the transgenic events and may provide insight into other roles of DGAT proteins during development.

Expression of UrDGAT2A or NcDGAT2 in Seed Increased DGAT Activity and Kernel Oil Content
The identification of fungal DGAT2 sequences was previously reported for U. ramanniana (Lardizabal et al., 2001) and N. crassa (Lardizabal et al., 2004). The U. ramanniana and N. crassa predicted protein sequences were similar in length and shared 45% amino acid identity and 66% amino acid similarity (Fig. 1). A subsequent BLAST search of public databases with the N. crassa amino acid sequence identified a sequence annotated as "hypothetical protein NCU02665 [Neurospora crassa OR74A]" (National Center for Biotechnology Information [NCBI] accession no. 85118420; Galagan et al., 2003) that was 100% identical along the length of the originally identified NcDGAT2 sequence. However, the hypothetical protein included an additional 243 amino acids at the N terminus ( Fig. 1).
To learn more about the additional N-terminal part of the hypothetical N. crassa protein, we conducted a BLAST search of GenBank with just this 243-amino acid fragment. The only sequences that were identified were annotated "hypothetical protein" or "related to diacylglycerol acyl transferase" from N. crassa. Analysis of the N-terminal sequences using the Protean program indicated that they are hydrophilic (data not shown). The portion of the NcDGAT2 predicted protein that aligns with the UrDGAT2A sequence begins with a small hydrophilic tail in front of a large hydrophobic domain that is conserved structurally in the DGAT2 family and probably anchors the protein to the membrane (data not shown; Shockey et al., 2006). Because it was not clear whether the predicted N-terminal portion of the NcDGAT sequence is really part of the native N. crassa protein, we decided to test both forms in maize. The longer NcDGAT2 coding sequence will be re-ferred to as L-NcDGAT2, and the shorter form will be referred to as S-NcDGAT2.
To determine whether expression of a fungal DGAT2 can increase maize kernel oil, we constructed plant transformation vectors with the UrDGAT2A coding sequence and the two different lengths of the NcDGAT2 protein-coding sequences. All constructs included an embryo-enhanced promoter, Per1, from barley (Hordeum vulgare; Stacy et al., 1996). Maize plants were transformed with these sequences, the regenerated transformed plants (R0 stage) were self-pollinated, and ears were harvested at maturity. Oil and fresh weight were determined for 36 kernels per ear. Kernels were then dissected into embryo and endosperm, and oil and weight were determined for the separate tissues. DNA was extracted from each embryo and analyzed by PCR to determine presence or absence of the transgene. Oil and weight were compared for kernels on the same ear with or without the transgene. For events transformed with the UrDGAT2A transgene, nine of 11 events had a statistically significant (P , 0.05) increase in kernel oil ranging from +0.4% to +0.7% (by weight) oil (Fig. 2), a relative change of 9% to 19%. Only four of seven events containing the long form of the NcDGAT2 transgene tested in this generation had an increase in kernel oil, from +0.3% to +0.7% by weight (relative change of 8%-16%). However, the short NcDGAT2 transgene produced increased kernel oil in 13 of 15 events, with a range of +0.4% to +0.9% (relative change of 13%-26%). A construct-level analysis of the dissected kernel data shows that each of the constructs increased oil predominantly in the embryo (Table I) from about 12% (relative) for UrDGAT2A to about 17% for S-NcDGAT2, as expected, since the transgene was expressed using an embryo-enhanced promoter. A small but significant increase in oil in the endosperm was observed for each construct and is probably due to activity of the Per1 promoter in the aleurone. None of the constructs produced a significant change in total kernel weight or in the weight of the embryo or endosperm. The differences in oil and weight values for controls of the different constructs are likely due to the fact that the plants from each construct were grown in the greenhouse at different times. Environmental conditions are known to have some influence on kernel growth and composition (Val et al., 2009).
To more directly compare effects of the three constructs, homozygous lines were identified from multiple events for each construct and were grown in a field in Hawaii. Plants were self-pollinated, and ears were harvested at 22 to 24 d after pollination (DAP) or at maturity. In this experiment, only two of the four UrDGAT2A events tested had more oil than the control at maturity, and the best event increased oil by 0.6% by weight, resulting in a relative increase of 11% (Fig. 3A). Sixteen of 18 events containing L-NcDGAT2 increased kernel oil, from +0.1% to +0.8%, giving a relative increase of 3% to 16%. All 10 of the events with the S-NcDGAT2 protein increased oil, from +0.5% to +0.9%, producing a relative increase of 9% to 19%.
DGAT enzyme activity was measured in embryos of kernels from a subset of events containing the NcDGAT2 proteins from ears harvested at 22 to 24 DAP. There was not a strong correlation between the level of oil increase and the amount of DGAT activity on an event basis within a construct (Fig. 3B). However, kernels containing the S-NcDGAT2 form had higher DGAT enzyme activity than the L-NcDGAT2 form, consistent with the trend to greater oil increases in events with this construct (Figs. 2 and 3A; Table I).
Comparison of UrDGAT2A and S-NcDGAT2 Transgene Efficacy in the Greenhouse Since S-NcDGAT2 consistently produced slightly higher oil increases and greater enzyme activity com-pared with L-NcDGAT2, we focused on a comparison of S-NcDGAT2 with UrDGAT2A for a more detailed analysis of the effects of these transgenes in maize kernels. Three representative events were selected from each construct, and inbred plants were grown in the greenhouse alongside nontransformed controls. Each plant was self-pollinated, and ears were harvested at 15, 30, and 45 DAP.
Kernels of all six transgenic events had more oil at maturity than the controls (Fig. 4A). Oil was increased in kernels expressing UrDGAT2A up to 0.7% (by weight) and in kernels expressing S-NcDGAT2 up to 1.1%. Oil was measured in embryos at several time points during kernel development (Fig. 4B). There was a trend for increased oil compared with nontransgenic controls at 15 DAP for several events, but most differences were not significant. However, embryos of all events had significantly more oil than controls by 30 DAP. At 45 DAP (maturity), embryos of all S-NcDGAT2 events had sig- Figure 2. Effects of DGAT2 constructs on oil content of maize kernels from R0 plants grown in the greenhouse. Change in kernel oil (percentage of kernel weight [wt], on a dry matter basis) is shown. Asterisks indicate differences from the control (P , 0.05).

Table I. Kernel oil content of primary transformants (R0 stage)
Construct-level analysis of oil and weight of kernels dissected into embryo (germ) and endosperm. Transgene-positive samples include both homozygous and hemizygous kernels. Boldface numbers indicate significant differences (P , 0.05). nificantly more oil than controls, while embryos from only one of the UrDGAT2A events had significantly more oil than controls. For the best S-NcDGAT2 event, embryo oil was increased 12% by weight compared with the control, a relative increase of 36%. Enzyme activity was compared in embryos of kernels from ears harvested at 30 DAP (Fig. 4C). Both UrDGAT2A and S-NcDGAT2 enzymes were active in maize kernels. The short NcDGAT2 protein appeared to have greater enzyme activity in two of the events compared with UrDGAT2A at this developmental time point.
To determine whether oil accumulation was limited by availability of the DGAT substrate, DAG, DAG levels were determined in kernels during development. The DAG levels of all samples decreased during development. At 15 DAP, two transgenic events expressing NcDGAT2 had significant decreases in DAG pool size compared with the control. DAG pool size was decreased in embryos from all six transgenic events at both 30 and 45 DAP compared with the control (Fig. 4D).
The composition of embryo oil from control and transgenic events was compared in mature kernels (Fig. 5). Levels of 16:0 were decreased in all events compared with controls, up to 2.9% of total fatty acids.
There was a small increase in 18:0 for all events, up to 1%. The greatest changes occurred with the 18:1 (up to 18% increase) and 18:2 (up to 15% decrease) fatty acids. There was also a small decrease in 18:3 fatty acids in the transgenic events compared with controls. These changes were more pronounced in the NcDGAT2 events than in the UrDGAT2A events. The longer chain length fatty acids were much less abundant, but significant changes in their concentrations were also evident (Fig.  5B). Interestingly, for 20 and 22 carbon fatty acids, the increases observed compared with controls were greater with UrDGAT2A events than with NcDGAT2 events.

Seedling Phenotype
An abnormal phenotype was noticed for some inbred seedlings of most of the events used in the greenhouse study. In the most strongly affected seedlings, the first leaves appeared stuck together and did not unfold (Fig.  6A). However, not all seedlings were affected, and there was a range of severity for the phenotype. To quantify this effect, seeds from plants hemizygous or homozygous for fungal DGAT2 transgenes were planted in pots in the greenhouse, seedling phenotypes were scored, and zygosity of each plant was determined by testing the copy number for the T-DNA using a Taqman assay for the selectable marker. The number of seedlings with either a mild or severe phenotype for each line and transgene zygosity class is shown in Figure 6B. Events with the UrDGAT2A transgene were more strongly affected than events with the S-NcDGAT2 transgene. All three of the UrDGAT2A events examined had a high rate of affected seedlings (65%-75% of plants homozygous for the UrDGAT2A transgene). However, two of the three NcDGAT2 events examined had only a minority of seedlings affected (6%-25% of homozygous seedlings; Fig. 6B), and half of those seedlings had a more mild phenotype than observed for seedlings with the . Kernel oil and embryo DGAT activity from field-grown homozygous inbred plants with different DGAT constructs. A, Kernel oil concentration determined by NMR. Change in kernel oil (percentage of kernel weight [wt], on a dry matter basis) is shown. B, DGAT activity in embryos from kernels harvested at 20 to 22 DAP from homozygous inbred lines grown in the field containing L-NcDGAT2 or S-NcDGAT2 constructs. Error bars indicate 95% confidence intervals.
UrDGAT2A transgene (data not shown). Only one of 16 seedlings of S-NCDGAT2 event M185538, which was a homozygous line, appeared abnormal. Homozygous plants tended to have a higher rate of phenotype appearance (Fig. 6B), suggesting that the level of transgene expression was related to the abnormality. The abnormal seedling phenotype was not observed for any of the events in hybrids grown in the field (data not shown).

DGAT2 Transgene Efficacy in Hybrid Kernels
To evaluate the potential of the fungal DGAT2 transgenes to increase oil in kernels of hybrid maize plants, selected events containing UrDGAT2A or S-NcDGAT2 transgenes and controls were crossed with an elite tester, and F1 hybrid seed was planted at multiple U.S. Midwest locations in a field layout designed to minimize cross-pollination of transgenic and control plants (see "Materials and Methods"). In addition, hybrid plants were grown in two different fields and hand pollinated. Kernel oil content was determined for both trials (Fig. 7). Open-and hand-pollinated values were similar, although oil concentrations in kernels from hand-pollinated ears were slightly higher, as expected, likely due to reduced cross-pollination with the nontransgenic controls. Larger error in the hand-pollinated measurement was due to the lower number of replications. In the open-pollinated trial, the UrDGAT2A transgene increased oil in four of five events compared with the control, from +0.2 to +0.5% by weight, a relative increase of 6% to 10%. The NcDGAT2 transgene increased oil in all six events in this trial, by +0.4 to +0.5% by weight (8%-11% relative increase). The greatest hand-pollinated value for UrDGAT2A was about +0.5% (12% relative) and that for NcDGAT2 was almost +0.6% (13% relative change).
The open-pollinated hybrid trials were harvested by a combine harvester and grain yield was determined. Yields of transgenic events were not significantly different from the control except for UrDGAT2A event 873, which had a 5.5% increase in yield (Fig. 7B).
To learn more about effects of the DGAT transgenes on kernel composition, kernel samples from the handpollinated trial were analyzed by near-infrared trans- mittance (NIT). Many of the events had significantly reduced starch content, and a plot of starch versus oil contents showed that increases in kernel oil were accompanied by decreases in kernel starch (Fig. 7C). No significant differences from controls were observed for protein content for any of the events, but there did appear to be a trend of increased protein in events expressing fungal DGAT transgenes (Fig. 7D).

DGAT2 Transgene Expression in High-Oil Hybrids
To determine whether a fungal DGAT2 could increase kernel oil above levels produced when the highoil allele of the maize DGAT1 gene is present, we crossed events containing the fungal DGAT2 genes to seven near-elite high-oil testers to produce high-oil hybrids. These high-oil lines were shown by PCR analysis to possess the high-oil allele of DGAT1 (data not shown). The lines were homozygous for the high-oil DGAT1 allele, except High Oil 1, which was heterozy-gous. An eighth hybrid was made with a conventional elite tester, which contained the normal-oil allele of DGAT1. The F1 plants were grown in a field in northern California. Ears were hand harvested at 20 to 22 DAP and at maturity, and kernels were tested for DGAT activity (Fig. 8A, immature kernels) or oil concentration (Fig. 8B, mature kernels).
All eight control lines without the fungal DGAT2 transgenes had similar levels of DGAT activity in developing kernels (Fig. 8A). DGAT activity was increased in kernels of all lines by the DGAT2 transgenes. Activity was greater with the S-NcDGAT2 transgene than with the UrDGAT2A transgene in most of the lines. The elevation in DGAT activity from the NcDGAT2 transgene was greater in some of the high-oil hybrids than in the conventional-oil hybrid.
Final oil concentration for the nontransgenic highoil hybrids ranged from 6.2% to 7% compared with 4.5% for the conventional hybrid (Fig. 8C). Small, but statistically significant, increases in kernel oil were measured for most of the hybrids for both UrDGAT2A  . Abnormal seedling phenotype with varying penetrance observed in some inbred lines. A, Inbred seedlings from control or transformed lines with severely abnormal phenotype 8 d after sowing. These seedlings are shown as examples; not all transgenic seedlings were affected. Some transgenic seedlings had a more mild abnormal phenotype, and others were normal. B, Graph showing the percentage of seedlings from different events that were homozygous (Hom), hemizygous (Hem), or null for the transgene that displayed any observable abnormal phenotype from the greenhouse experiment. The remainder of the seedlings had a normal phenotype. Sixty-eight to 76 seedlings were examined for the first five lines, which were hemizygous, and 16 seedlings were examined for event S-Nc 538, which was homozygous. and S-NcDGAT2 (Fig. 8B). Oil increases obtained with the UrDGAT2A transgene were similar between conventional-and high-oil lines and ranged from 0.1% to 0.4%, producing 2% to 8% more oil. The greatest oil increase was produced by the S-NcDGAT2 transgene in the conventional-oil hybrid, with an increase of 1.3% by weight, a relative increase of 30%. In the highoil hybrids, S-NcDGAT2 produced oil increases of 0.2% to 0.8% by weight, giving 4% to 12% more oil.

DGAT2 Genes in Maize
To begin to provide more understanding of the relative contributions of DGAT1 and DGAT2 proteins in the accumulation of oil in maize kernels, we conducted a search for maize DGAT2 sequences in Monsanto cDNA libraries using the castor bean DGAT2 amino acid sequence as the query (GenBank accession no. AY916129; Kroon et al., 2006). We identified two maize sequences that are closely related to other plant DGAT2 sequences. Subsequently, the maize B73 genome was sequenced (Schnable et al., 2009), and DGAT2-1 and DGAT2-2 were mapped to chromosome 9 (LOC100217122) and chromosome 4 (LOC100283803), respectively, by BLAST against maize at http://maizesequence.org. Homologs of these genes can be found in many other sequenced plant genomes, including rice (Oryza sativa) and soybean. Alignment of plant and fungal DGAT2 sequences showed expected clustering of taxonomic groups (plant versus fungal and dicot versus monocot; Fig. 9).

DISCUSSION
Small but statistically significant increases in maize kernel oil were achieved by expression of fungal DGAT2 transgenes using an embryo-enhanced promoter. A DGAT2 gene isolated from N. crassa encoded an enzyme with somewhat greater activity in maize kernels than the enzyme from U. ramanniana, and expression of the S-NcDGAT2 transgene resulted in greater increases in maize kernel oil. Two forms of the NcDGAT2 gene were tested. The full-length amino acid sequence includes 243 residues that are unique to this protein and located in the N-terminal half of the protein. Therefore, we tested both the full-length sequence (L-NcDGAT2) and the shorter sequence that aligns with the UrDGAT2A sequence throughout its length (S-NcDGAT2). We found that expression of S-NcDGAT2 resulted in higher DGAT enzyme activity Figure 7. Kernel oil and grain yield for a multilocation hybrid field trial. A, Oil content measured by NMR expressed as the difference from the control. OP, Grain from an open-pollinated hybrid agronomic trial planted at 10 locations; HP, grain from a two-location hand-pollinated trial. Event 871 was not planted in the hand-pollinated trial because sufficient seed was not available. Numbers on the x axis are abbreviated event names. Error bars represent 95% confidence intervals. wt, Weight. B, Grain yield expressed as the percentage difference from the control in a 10-location field trial. Yields of controls were 200 bushels per acre (UrDGAT2A) and 212 bushels per acre (NcDGAT2). Error bars represent 95% confidence intervals. C and D, Oil and starch (C) and protein and starch (D) were measured by NIT for grain from the two-location hand-pollinated trial. Values are percentage dry weight of the grain. Each point represents the mean of 15 to 40 samples. White diamonds, Controls; black diamonds, transgenic events. in maize embryos and also larger increases in kernel oil (Figs. 2 and 3). Since we did not measure the amount of N. crassa DGAT2 protein in the tissues, it is possible that the construct with the shorter protein allowed more accumulation of NcDGAT2 protein or that the short form has more inherent DGAT enzyme activity. Further study will be required to determine whether this unique 243-amino acid sequence is really part of the native NcDGAT2 protein and, if so, what its function may be.
Most of the increased oil from either fungal DGAT2 was located in the embryo, as expected because of the specificity of the promoter that was used. However, some increases were also observed in the endosperm, probably within the oil-rich aleurone. The increases in oil in the embryo of 4% to 4.8% (by weight; Table I) were greater than the increases obtained for soybean of 1% to 2% (Lardizabal et al., 2008). However, because the embryo is only a small part of the kernel, the overall gain in oil was only a maximum of approximately 0.9% (by weight) in grain of inbred plants (Fig. 3). This gain equates to an increase of about 23% (relative; average oil content of controls in the field was 3.9%).
The amount of oil increase due to the expression of fungal DGAT2 transgenes was modest and a somewhat lower percentage increase than levels produced by the high-oil DGAT1-2 allele of maize. With the highoil maize DGAT1-2 transgene, an oil increase of up to 41% was achieved in an inbred line (Zheng et al., 2008), while the best oil increases with the NcDGAT2 transgene were only about 26% in inbred plants (Fig. 2). In both experiments, embryo-preferred promoters were used to drive expression of the DGAT transgene, but they were different promoters that may differ in timing, strength of expression, or cell-type specificity, and these variations could affect transgene efficacy. In addition, different varieties of maize were used in the two experiments, and growing conditions were different. It would be interesting to directly compare the effects of expression of various DGAT transgenes in maize in one experiment to determine the relative efficacy of these different DGATs to increase kernel oil.
The specific roles of DGAT1 and DGAT2 in plants are not well understood (Li et al., 2010b), and little work has been done to understand the role of these enzymes in monocot seed development. In addition to the two previously characterized maize DGAT1 genes (Zheng et al., 2008), we have identified two DGAT2 sequences in maize (Fig. 9). Preliminary results using expression profiling arrays indicate that the maize DGAT1-2 gene is expressed early during embryo development and its expression level declines in mature kernels, while the maize DGAT2-1 gene is expressed later during embryo development (data not shown). The maize DGAT2-2 gene appears to be expressed in endosperm tissue during kernel development (data not shown). We also looked at the expression of the DGAT1-1 gene (referred to in Zheng et al., 2008, Fig. 4). This gene is located on Figure 8. DGAT enzyme activity and oil concentrations of kernels of high-oil and conventional transgenic and control hybrids. The construct averages were plotted, representing three events. A, DGAT enzyme activity in immature kernels (20-22 DAP). B, Change in kernel oil compared with controls without the DGAT2 transgene in mature kernels. C, Absolute oil concentration in mature kernels of the control lines. For both graphs, error bars represent 95% confidence intervals. wt, Weight. chromosome 6 (LOC100383910), as is the DGAT1-2 gene (LOC100283803), and is also expressed early in embryo development, although at lower levels than DGAT1-2. The expression of DGAT1-1 appears to be less specific than DGAT1-2, with expression in shoot apical meristem and tassels (based on analysis of the NCBI EST profile data by Pontius et al. [2003]). Thus, these DGAT genes are differentiated by timing and tissue specificity. Whether the enzymatic activities of the two DGAT forms are distinct in maize remains to be determined.
To better understand the potential benefit of additional DGAT activity on increasing kernel oil, we measured the level of the substrate for DGAT, DAG, in wild-type and transgenic maize embryos during kernel development. The DAG pool was reduced by expression of the fungal DGATs, and very little DAG was present in the embryos at maturity (Fig. 4D). However, it seems possible that sufficient DAG would be present at 30 DAP to support higher levels of oil biosynthesis if DGAT activity were further increased.
As with overexpression of DGAT1 in maize (Zheng et al., 2008), overexpression of the fungal DGAT2 genes resulted in changes in fatty acid composition. In particular, there were increases in 18:1 (up to 18%) and decreases in 18:2 (up to 15%). In agreement with its greater enzyme activity, larger changes in fatty acid composition were observed from overexpression of S-NcDGAT2 compared with UrDGAT2A (Fig. 5). The fatty acid composition changes that were observed in events expressing S-NcDGAT2 are similar in type and magnitude to those reported for overexpression of the high-oil DGAT1 gene in maize (Zheng et al., 2008), and these values are within the range reported for nontransgenic maize (Reynolds et al., 2005). This effect on fatty acid composition has been hypothesized to result from competition for DAG as a substrate by both choline phosphotransferase and by DGAT (Zheng et al., 2008). Higher levels of DGAT activity could result in reduced levels of oleoyl-phosphatidylcholine made by choline phosphotransferase, which is a substrate for the FAD2 desaturase that is responsible for the production of 18:2 fatty acids. Thus, increased competition at this juncture in the pathway could have an impact on the production of polyunsaturated fatty acids.
We observed a phenotype in seedlings of some plants overexpressing the fungal DGAT genes in which the first leaves appeared to be stuck together and failed to expand normally (Fig. 6). It is interesting that in Arabidopsis, the highest levels of expression of the DGAT1 gene are in germinating seeds and young seedlings (Zou et al., 1999), although the need for high levels of DGAT expression at this stage of development is not clear. In addition, study of Brassica napus plants with reduced DGAT1 expression by antisense suppression revealed the presence of some developmental abnormalities, including effects on seed development (Lock et al., 2009). These researchers concluded that DGAT1 may have a specific role in seed development of B. napus in addition to control of TAG accumulation. Mutants of wax production can have fused organs (Chen et al., 2003), showing the importance of cuticular waxes in organ development. It is possible that excess DGAT activity may lead to the production of extra or abnormal cuticular waxes and cause the fusion of seedling leaves. However, the abnormal phenotype does not appear to be directly correlated with the amount of transgenic DGAT2 activity, since the phenotype was more common in lines with UrDGAT2A than with NcDGAT2, even though NcDGAT2 lines tended to have higher DGAT activity (Fig. 3B). Perhaps differences in the enzyme activities of these two DGATs may include different specificities for substrates of wax biosynthesis.
UrDGAT2A and S-NcDGAT2 activities demonstrated some interesting differences that cannot be explained simply by the greater activity of S-NcDGAT2. The abnormal seedling phenotype was greater with UrDGAT2A, and the two DGATs had different effects on fatty acid composition (Fig. 5). S-NcDGAT2 expression resulted in larger increases in 18:1 and larger corresponding decreases in 18:2 and 18:3. But expression of UrDGAT2A had bigger effects on levels of longer chain fatty acids (20:0, 20:1, and 22:0). It would be valuable to test DGAT2 genes from additional sources to gain a better understanding of the fatty acid compositional changes and to have an opportunity to further optimize the increase in kernel oil.
We have shown that fungal DGAT2 genes have the potential to be useful in the development of maize with higher oil content. These transgenes were able to increase kernel oil in hybrids in the field and appeared not to have any negative effect on grain yield (Fig. 7). They were also able to increase total oil content in high-oil hybrids, which contain the high-oil allele of DGAT1 (Zheng et al., 2008;Fig. 8). Additional increases in kernel oil may be achievable with alternative DGAT genes or by varying the timing of gene expression during embryo development and maturation. In addition, engineering  upstream pathways to provide additional DAG substrate would likely be a productive approach to obtaining further oil increases. Overexpression of the transcription factor ZmWRI1 has been shown to up-regulate genes involved in fatty acid biosynthesis (Shen et al., 2010), so combining overexpression of fungal DGAT2 with ZmWRI1 would likely produce a synergistic effect on oil accumulation in maize kernels.

MATERIALS AND METHODS
Cloning L-NcDGAT2 cDNA The L-NcDGAT2 sequence was identified from a BLAST search of public databases using the previously cloned NcDGAT2 sequence (Lardizabal et al., 2004) as the query. A sequence identified as "hypothetical protein NCU02665 [Neurospora crassa OR74A]" (NCBI accession no. 85118420; Galagan et al., 2003) was 100% identical along the length of the originally identified NcDGAT2 sequence but also included an upstream sequence encoding an additional 243 amino acids not found in the UrDGAT2A protein (Fig. 1). To clone this sequence, RNA was isolated from N. crassa mating type A (Fungal Genetics Stock Center) mycelium using Tri-Reagent (Sigma) according to the manufacturer's protocol. First-strand cDNA synthesis was completed using the SMART cDNA Amplification kit (Clontech). Oligonucleotide primers were designed to amplify the full-length coding region of the NcDGAT2 sequence, and the PCR products were cloned into plasmid pCR2.1 according to the manufacturer's protocol (Invitrogen).

Vector Construction and Plant Transformation
Maize (Zea mays) transformation vectors were made using the barley (Hordeum vulgare) Per1 promoter and 5# untranslated region, which contained an intron from maize (ZmHsp70) to control the expression of each of the three fungal DGAT2 genes. The sequence for UrDGAT2A was the codon-optimized form described previously (Lardizabal et al., 2008), but the native N. crassa sequences were used for both NcDGAT2 cDNAs. Either the 3# nontranslated region of the potato (Solanum tuberosum) proteinase inhibitor II gene (Keil et al., 1986; UrDGAT2A and S-NcDGAT2 constructs) or the 3# nontranslated region from the wheat (Triticum aestivum) hsp17 gene (L-NcDGAT2 construct) was fused to the 3# end of the coding sequence to provide a signal for polyadenylation. The selectable marker cassette for these vectors included Agrobacterium tumefaciens CP4 5-enol-pyruvylshikimate-3-phosphate synthase driven by the rice (Oryza sativa) actin1 promoter and completed by the 3# end with the Agrobacterium nopaline synthase termination sequence. Both the gene cassettes were within the T-DNA borders.
A near-elite maize inbred line was transformed by the method described by Armstrong and Rout (2003). Transgenic maize lines containing one or two copies of the construct were selected for further study. R0 plants were selfpollinated and grown to maturity in the greenhouse. Subsequent inbred generations were grown in the field in Hawaii.

Greenhouse Experiment
The greenhouse was maintained at a day temperature of 30°C to 31°C and a night temperature of 20°C to 21°C. The photoperiod was maintained at 18 h minimum with a light intensity of 900 to 1,400 mE m 22 s 21 . The soil was Sunshine No. 1 with 14-14-14 NPK fertilizer and Micromax micronutrient incorporated at a moderate rate. Plants were fed by continuous liquid fertilization via irrigation with calcium nitrate, Epsom salts, and 0-10-30 fertilizer with micronutrients. All plants were self-pollinated. Sampling of ears was conducted at 15, 30, and 45 DAP. At 15 and 30 DAP, ears were harvested and immediately frozen in liquid nitrogen. Ears were stored at 270°C, and subsequent manipulations and dissections were done on dry ice. Dissected embryos were hand ground in liquid nitrogen with a mortar and pestle. Approximately 50 mL of material was subsampled for DGAT assays. The remaining ground embryo sample was freeze dried for lipid analysis. Mature ears (45 DAP) were maintained at room temperature until dissection. Kernels were softened for dissection by boiling in water for 10 min. Dissected embryos were then frozen at 270°C, hand ground in liquid nitrogen with a mortar and pestle, and freeze dried for lipid analysis.

Enzyme Assays
UrDGAT2A and NcDGAT2 enzyme activities were measured as described (Lardizabal et al., 2006), with the following changes. The assay buffer for plant samples substituted 5 mM Tricine, pH 7.5, for the 10 mM potassium phosphate, 280 mM NaCl for the 150 mM KCl, and 0.06% CHAPS for the 0.1% Triton X-100. Assays were performed at 25°C for 10 min and were immediately processed after assay termination. Twenty-six percent of the final organic phase was liquid scintillation counted, and thin-layer chromatography was not performed.

NMR
Seed oil and moisture content for both whole maize kernels and dissected embryos and endosperm were determined by pulsed NMR (Tiwari et al., 1974;Rubel, 1994) using a Maran Ultra-20 Benchtop NMR spectrometer (Resonance Instruments) operated at 23.4 MHz. The NMR spectrometer was calibrated using 100% maize oil, and results were reported on a dry matter basis.

NIT Spectroscopy
Bulk quantities of maize consisting of 30 to 50 mL of maize kernels (i.e. 100-150 kernels) were analyzed for average oil, starch, and protein contents with a NIT method using a Tecator Infratec 1221 scanning monochromator (Foss North America) with a 30-mm path-length sample cuvette. Absorbance values, as log (1/T) where T represents transmission, were recorded at 2-nm intervals between 850 and 1,048 nm. These absorbance values were translated to oil, starch, and protein content measurements using Infratec Corn Analysis model CO980811. The commercial Analysis model and standard operating procedure were validated relative to Accelerated Soxhlet Extraction results prior to this analysis; the SE of accuracy (absolute errors of differences) was shown to be 0.6% for moisture, 0.4% for oil, 0.3% for protein, 1.2% for starch, and 1.5% for extractible starch relative to dry sample weight.

Lipid Analysis
Oil fatty acyl composition was analyzed by gas-liquid chromatography of methyl esters (Browse et al., 1986). DAG and free fatty acid pool sizes were determined by a modification of a HPLC-evaporative light scattering detector method using a silica column instead of a cyanopropyl column (El-Hamdy and Christie, 1993;American Oil Chemists' Society, 1999).

Hybrid Field Testing
Yield trials with elite testers were conducted at multiple locations in Iowa and Illinois using two-row, 22-foot-long plots. Transgenic lines and their controls were separated by two plots of sterile hybrids to minimize crosspollination. Harvest and yield determination was done using a combine. In addition, the same hybrid entries were planted at two locations in the Midwest and one in California and hand pollinated. Ears were handharvested at maturity for analysis of oil content.