Expression of ZmLEC1 and ZmWRI1 Increases Seed Oil Production in Maize

Increasing seed oil production is a major goal for global agriculture to meet the strong 3 demand for oil consumption by humans and for biodiesel production. Previous studies to 4 increase oil synthesis in plants have focused mainly on manipulation of oil pathway 5 genes. As an alternative to single enzyme approaches, transcription factors provide an 6 attractive solution for altering complex traits, with the caveat that transcription factors 7 may face the challenge of undesirable pleiotropic effects. Here we report that over- 8 expression of maize LEAFY COTYLEDON 1 ( ZmLEC1 ) increases seed oil by as much as 9 48%, but reduces seed germination and leaf growth in maize. To uncouple oil increase 10 from the undesirable agronomic traits, we identified a LEC1 down-stream transcription 11 factor, maize WRINKLED1 ( ZmWRI1 ). Over-expression of ZmWRI1 results in an oil 12 increase similar to over-expression of ZmLEC1 without affecting germination, seedling 13 growth, or grain yield. These results emphasize the importance of field testing for 14 developing a commercial high oil product and highlight ZmWRI1 as a promising target 15 for increasing oil production in crops. 16 The results presented highlight the potential application of

INTRODUCTION 1 2 Maize grain is the most important feedstock for meat, egg, milk, and fuel production in 3 the world. Approximately 65% of maize grain is used for feeding animals. High oil maize 4 shows a greater feed efficiency than normal oil maize in animal feed trials because the 5 caloric content of oil is 2.25 times greater than that of starch on a weight basis (Han et al., 6 1987;Perry, 1988). Maize oil is the most valuable co-product from industrial processing 7 of maize grain through wet milling or dry milling and is high quality oil for human 8 consumption. Compared to soybean oil which contains 6.8% linolenic acid (18:3) and is 9 susceptible to oxidation, maize oil is stable because it contains very little (<1.0%) 10 linolenic acid (Weber, 2003). With the rapid growth of human consumption and 11 industrial use for biodiesel production, the demanding for vegetable oil has increased 12 significantly. Therefore, high oil content is a desirable trait for the maize end-users and 13 becomes an important goal for genetic engineering. 14 15 Plant oil is synthesized from glycerol-3-phosphate and fatty acyl-CoA in the endoplasmic 16 reticulum as triacylglycerols (TAGs). Fatty acids are synthesized from acetyl-CoA 17 exclusively in the plastid, and then transported to the cytoplasm in the form of fatty acyl-18 CoA (Ohlrogge and Browse, 1995). In the endoplasmic reticulum, TAGs are synthesized 19 by the stepwise acylation of glycerol-3-phosphate, known as the Kennedy pathway. First, 20 fatty acyl moieties are added to the sn-1 and sn-2 positions of glycerol-3-phosphate by 21 glycerol-3-phosphate acyltransferase and lyso-phosphatidic acid acyltransferase, 22 respectively, to form phosphatidic acid. Phosphatidic acid is then hydrolyzed by 23 phosphatidate phosphatase to yield diacylglycerol (DAG). DAG can be used to form 24 TAGs, or it can be used as a substrate for membrane lipid biosynthesis. Diacylglycerol 25 acyltransferase (DGAT), the last enzyme for TAG synthesis, adds a third acyl chain to 26 DAG and yields TAGs. An alternative pathway for TAG formation may also exist in 27 plants. For example, phospholipid:diacylglycerol acyltransferase (PDAT) can transfer the 28 sn-2 acyl chain from phosphatidylcholine (PC) to DAG and form lyso-PC and TAG, and 29 has overlapping functions with DGAT for TAG synthesis in both seed and pollen in 30 Arabidopsis (Zhang et al., 2009). Finally, TAGs are stored in seeds in specialized 1 structures termed oil bodies. 2 3 Plant seed oil content is controlled by multiple steps in the oil biosynthetic pathway. 4 Manipulation of single steps in the pathway often shows a moderate effect on seed oil 5 content (Thelen and Ohlrogge, 2002;Durrett et al., 2008). For example, expression of a 6 fungal diacylglycerol acyltransferase in soybean results in an approximate 7.5% relative 7 increase in seed oil content (Lardizabal et al., 2008). Transcription factors regulate 8 multiple steps simultaneously, and they provide an attractive alternative to single enzyme 9 approaches for altering complex traits in crops (Broun, 2004;Broun, 2005;Grotewold, 10 2008). In Arabidopsis, LEAFY COTYLEDON 1 (LEC1) and WRINKLED 1 (WRI1) have 11 been identified as two key transcription factors involved in regulation of oil 12 accumulation. Mutations in both genes lead to reduced oil content in seeds. LEC1 13 encodes a HAP3 subunit of the CCAAT binding factor and plays an important role in 14 Arabidopsis embryo development. Ectopic expression of Arabidopsis LEC1 leads to the 15 formation of embryo-like structures containing oil and storage protein in leaves (Lotan et 16 al., 1998). WRI1 encodes a transcription factor containing two AP2 domains and may 17 play an important role in regulation of carbon metabolism. Over-expression of WRI1 in 18 Arabidopsis results in an increase in oil accumulation in seeds and leaves (Cernac and 19 Benning 2004). Expression profiling and genetic analyses also indicate that WRI1 20 functions downstream of LEC1 and is a key transcription factor controlling fatty acid 21 biosynthesis (Baud et al., 2007;Mu et al., 2008;Santos-Mendoza et al., 2008;Maeo et 22 al., 2009). 23 24 While these results are exciting, much of this has been done in Arabidopsis. The effect of 25 transcription factors on seed oil production in major crops has not been determined. 26 Furthermore, altered expression of transcription factors may show undesirable pleiotropic 27 effects on plant growth and development in addition to oil increase. Rigorous field trials 28 of transgenic genes in elite, high-yielding commercial varieties at various locations and 29 multiple environments are necessary to determine whether oil increase is associated with 30 yield penalty or poor agronomic performance. Here, we report that over-expression of 31 maize LEAFY COTYLEDON 1 (ZmLEC1) increases seed oil production but reduces seed 1 germination and plant growth. To uncouple oil increase from undesirable changes in 2 germination and growth, we identify maize WRINKLED1 (ZmWRI1) as a transcription 3 factor down-stream of LEC1 and demonstrate that over-expression of ZmWRI1 increases 4 seed oil content without the undesirable effects caused by ZmLEC1. The results presented 5 highlight the potential application of transcription factors for increasing oil production in 6 major crops. 7 8 RESULTS 9 10

Expression of ZmLEC1 in Transgenic Maize Plants 11
Previous studies indicated that over-expression of Arabidopsis LEC1 induced the 12 formation of embryo-like structures and increased expression of fatty acid pathway genes 13 in Arabidopsis leaves (Lotan et al., 1998). We have identified a maize homolog that 14 shares 41% identity to Arabidopsis LEC1 in amino acid sequence. ZmLEC1 is expressed 15 specifically in early embryo development and is not expressed in endosperm, leaf, and 16 root (Supplemental Fig. S1A). ZmLEC1 protein accumulated in embryos at 15 and 20 17 days after pollination (DAP) and diminished after 25 DAP (Supplemental Fig. S1B). 18 Expression of ZmLEC1 under a constitutive synthetic SCP1 promoter in Arabidopsis lec1 19 mutant plants can complement the lec1 mutant seed phenotype, indicating ZmLEC1 is 20 functionally equivalent to Arabidopsis LEC1 (Supplemental Fig. S2). To test whether 21 alteration of seed development in maize can increase seed oil production, ZmLEC1 was 22 expressed under two embryo-preferred promoters, a strong OLEOSIN promoter (OLE 23 pro) and a weaker EARLY EMBRYO PROTEIN promoter (EAP1 pro). Each construct 24 also contained a DS-RED2 marker gene driven by an aleurone-specific LIPID-25 TRANSFER PROTEIN 2 (LTP2) promoter to facilitate identification of transgenic and 26 null seeds for phenotypic analysis. Transgenic seeds with red fluorescence can be 27 separated from null seeds easily. Analysis of 15 transgenic maize lines expressing 28 ZmLEC1 under the EAP1 promoter revealed average increases in T 1 seed oil content by 29 35% and embryo oil concentration by 24% ( Figure 1A). Because maize seed oil content 30 is determined by the amount of oil in seed divided by seed weight and the amount of oil 31 in seed is determined by the oil concentration in embryo, embryo size and oil in 1 endosperm, we determined the effect of EAP1:ZmLEC1 on oil accumulation in 2 endosperm and on embryo size. Endosperm oil was extracted by hexane and determined 3 by the amount of oil divided by endosperm dry weight. ZmLEC1endosperm contained 4 0.55% oil which was not significantly different from null endosperm oil content (0.49%). 5 Transgenic lines showed an average increase of 14.4% in embryo size compared to null 6 but only line 103.1.12 showed a significant increase as determined by Student's t-test 7 (Supplemental Table S1). An increase in seed oil content by ZmLEC1 may be driven 8 primary by a higher embryo oil concentration and a small increase in embryo size. The T 1 9 seeds were propagated to obtain T 3 homozygous seeds. The high oil trait was stable 10 across three generations in different locations. T 3 homozygous transgenic seeds showed a 11 level of oil increase similar to that seen in the T 1 generation. The best transgenic ZmLEC1 12 line (line 103.1.12) showed as much as 48.7% increase in seed oil content relative to its 13 null ( Figure 1B). Detailed analysis, however, found that over-expression of ZmLEC1 14 reduced seed germination and leaf growth in addition to elevating oil content. The first 15 and second leaves of transgenic ZmLEC1 plants were 40-50% shorter than those of the 16 null plants and were narrow and dark green ( Figure 1D). In germination tests, root and 17 shoot growth of transgenic ZmLEC1 seedlings were slower than their corresponding nulls 18 ( Figure 1C), resulting in a poor early stand count and reduced plant height in the field. 19 Expression of ZmLEC1 by the OLE promoter increased seed oil content similar to that by 20 the EAP1 promoter, but these lines showed a more severe reduction in seed germination 21 and leaf growth than lines expressing ZmLEC1 under the EAP1 promoter (data not 22 shown). 23 24

Identification and Expression of ZmWRI1 in Transgenic Maize Plants 25 26
It is not surprising that alteration of a master switch transcription factor such as LEC1 27 may lead to pleiotropic effects on seed metabolism, development, and seedling growth. 28 We hypothesized that a transcription factor down-stream of LEC1 might uncouple the 29 high oil phenotype from the negative effects on germination and growth. In Arabidopsis, 30 WRI1 is another key transcription factor affecting seed oil accumulation (Cernac and 31 Benning 2004). We identified a maize WRI1 which showed 43% identity with AtWRI1 in 1 the amino acid sequence and was up-regulated by ~2 fold in ZmLEC1-expressing 2 embryos. ZmWRI1 showed an expression pattern similar to ZmLEC1 with a peak 3 expression in embryos at ~20 DAP and decreased expression after 25 DAP 4 (Supplemental Fig. S3). In contrast to ZmLEC1, which expressed specifically in embryos, 5 ZmWRI1 showed very weak expression in leaf, root and stalk. The up-regulation of 6 ZmWRI1 by ZmLEC1 was confirmed by co-expression of the ZmLEC1 protein and a 7 ZmWRI1 promoter:GUS reporter in maize culture BMS cell. Co-expression of ZmLEC1 8 protein increased GUS activity significantly (Figure 2), indicating that ZmLEC1 regulated 9 expression of ZmWRI1 directly or indirectly. Similar to ZmLEC1, expression of ZmWRI1 10 by the embryo-preferred OLE promoter increased T 1 seed oil content by an average of 11 30.6% across 15 transgenic lines analyzed ( Figure 3A). In contrast to ZmLEC1-12 expressing lines, the embryo size of transgenic ZmWRI1 seeds was not significantly 13 different from the nulls (Supplemental Table S2). Transgenic ZmWRI1 endosperm 14 contained 0.81% oil, which was significantly higher than 0.47% in null endosperm. The 15 increase of seed oil by ZmWRI1 may be primarily due to higher embryo and endosperm 16 oil concentration. In endosperm, oil bodies were found in aleurone cells but not in starchy 17 endosperm cells. To determine whether ZmWRI1 increases oil in starchy endosperm or in 18 aleurone cells, ZmWRI1 was expressed in starch endosperm under a maize 19 KD ZEIN 19 promoter (Lappegard and Martino-Catt, 2001). Expression of ZmWRI1 under the 19 KD 20 ZEIN promoter did not lead to an increase in seed oil content (Supplemental Fig. S4), 21 suggesting that higher oil content in endosperm expressing ZmWRI1 under the OLE 22 promoter could be due to expressions of ZmWRI1 by the OLE promoter in the aleurone 23 layer. Furthermore, we determined the protein and starch levels in the ZmWRI1 embryos 24 to understand the source of the additional carbon needed for the biosynthesis of the 25 increased embryo oil. Expression of ZmWRI1 did not affect protein content in the embryo 26 but did reduced starch content by ~60% compared to nulls (Figure 4), suggesting that 27 ZmWRI1 may enhance oil biosynthesis by reducing carbon flux to starch biosynthesis in 28 the embryo. The high oil trait was stable in three genetic backgrounds at three locations. 29 T 3 homozygous transgenic seeds showed an increase in oil similar to the T 1 generation, 30 with a 46% increase in the best line (line 25.2.1) ( Figure 3B). To determine whether oil 31 quality was affected by ZmWRI1 expression, we analyzed major fatty acid composition in 1 seed oil. There were no significant changes in fatty acid composition between mature 2 transgenic ZmWRI1 seeds and their corresponding null seeds (Supplemental Table S3). 3 In contrast to ZmLEC1-expressing lines, transgenic ZmWRI1 seeds germinated normally 4 compared to null seeds ( Figure 3C). Transgenic ZmWRI1 plants did not show any 5 significant growth differences from the null plants in the length of the first and second 6 leaves ( Figure 3D). Expression of ZmWRI1 by the weaker EAP1 promoter also increased 7 T 1 seed oil content, but to a lesser extent, averaging 16.9% increase in the top 15 8 transgenic lines (data not shown). construct was re-transformed into an inbred line, PHWWE, and was out-crossed to a 19 male tester line (PH1B5) to produce F 1 hybrid for field yield tests. Hybrids from 5 20 transgenic ZmWRI1 lines with grain oil increase from 10% to 22% and their 21 corresponding null lines were tested in 8 locations in the United States maize belt with 3 22 repeats for each line at each location. All 5 transgenic ZmWRI1 lines showed no 23 significant difference from their corresponding nulls in grain yield ( Figure 6). The 24 average yield of the 5 transgenic ZmWRI1 lines was 9.45 tonne /ha, which was not 25 significantly different from the average null yield of 9.55 tonne/ha ( Figure 6). In addition, 26 we did not observe any significant differences between transgenic lines and null lines in 27 early stand count, seedling vigor, flowering time, grain moisture, grain test weight, or 28 plant height. 29

DISCUSSION 31
Relative to a single enzyme approach, transcription factors provide an attractive solution 1 for increasing plant oil production (Broun, 2004;Grotewold, 2008). However, possible 2 pleiotropic effects of transcription factors are a key challenge for using them in a 3 commercial product (Century et al., 2008). For example, knockout of a homeobox gene, 4 GLABRA2, increased seed oil content in Arabidopsis but GLABRA2 also affected seed 5 coat, trichome, and root hair development (Shen et al., 2006). LEC1 encodes a CCAAT-6 binding transcription factor that is critical for seed development. Mutation of Arabidopsis 7 LEC1 resulted in desiccation-intolerant seeds with reduced oil content. Ectopic 8 expression of AtLEC1 in Arabidopsis led to formation of embryo-like structures which 9 accumulate oil and seed storage proteins (Lotan et al., 1998) and up-regulation of the 10 fatty acid biosynthetic pathway (Mu et al., 2008). We have identified a maize LEC1 gene 11 with 41% identity to the Arabidopsis LEC1 and have demonstrated that over-expression 12 of maize LEC1 increased seed oil content by as much as 48.7%. However, transgenic 13 seeds germinated poorly and plants showed stunted growth with dark green, narrow 14 leaves. Construct optimization with different promoters giving different expression levels 15 and tissue specificities reduced the undesirable phenotypes but was unable to eliminate 16 them. To uncouple the high oil phenotype from undesirable agronomic traits, such as 17 poor germination and plant growth, we have identified a down-stream transcription 18 factor, ZmWRI1, which appears to be more specific for oil biosynthesis. Expression of 19 ZmWRI1 increased seed oil content by as much as 46% but did not affect seed 20 germination and plant growth. Our work demonstrates that it is possible to uncouple a 21 desired trait from unwanted side effects by identifying a down-stream transcription factor 22 that is more specific for the trait. Assuming the grain yield of the transgenic line is equal 23 to current commercial hybrids, an average 25% increase in maize seed oil content will 24 add additional 87.5 kg oil or $70 per hectare based on a current yield of 10 tonne/ha, 25 3.5% kernel oil content, and an oil price of $0.80/Kg. If US farmers plant all their ~35 26 million hectares with high oil maize, then an additional ~3.0 million tonnes of oil will be 27 produced. Because starch endosperm accounts for 80-90% of seed mass, conversion of starch to oil 4 in starchy endosperm cells will increase seed oil content dramatically and make maize a 5 C4 oil crop potentially. We expressed ZmWRI1 under 19 KD ZEIN promoter to promote 6 oil biosynthesis in endosperm and did not detect a significant increase in seed oil content 7 (Supplemental Fig. S4). Overexpression of ZmWRI1 in embryo up-regulated multiple 8 genes in fatty acid biosynthesis, but not the genes involved in oil biosynthesis, such as 9 glycerol-3-phosphate acyltransferase, DGAT, and oleosin (Shen et al., unpublished data). 10 Failure of ZmWRI1 to increase oil in starchy endosperm could be due to lack of 11 expression of genes involved in oil biosynthesis and oil body formation. Interestingly, 12 long term recurrent selection for high oil in maize resulted in high oil lines with as much 13 as 22% kernel oil content, but did not increase oil content in starchy endosperm 14 (Lambert et al., 2004). High oil content resulted from higher oil concentration in embryo, 15 larger embryo, smaller seed, and more oil in aleurone cell. In contrast, recurrent selection 16 for high oil in oat led to oil accumulation in starchy endosperm (Peterson and Wood, 17 1997). It needs to be determined why oat and maize respond differently in endosperm to 18 recurrent selection for high oil. Expression of WRI1 may be able to increase oil in oat 19 endosperm while maize is selected for starch accumulation and is not competent for oil 20 biosynthesis in endosperm. 21

22
Development of high oil maize has been a breeding goal for many years. Alexander et al. 23 started a high oil breeding program using recurrent selection in synthetics in 1956 and has 24 developed the ASK high oil population with grain oil content as high as 22% (Lambert et 25 al., 2004). However, commercialization of high oil maize has not been successful, mainly 26 because of significant grain yield reduction and poor agronomic traits associated with 27 high oil germplasm. It is not known if grain yield reduction is caused by high oil content 28 directly due to the high energy input for oil biosynthesis or rather by genetic linkage drag 29 of old non-elite germplasm. Maize seed oil content is a complex trait affected by multiple 30 QTLs (Berke and Rocheford, 1995;Clark et al., 2006). One major oil QTL on 31 chromosome 6 (qHO6) affecting seed oil content and oleic acid content was cloned 1 recently (Zheng et al., 2008). qHO6 encodes a diacylglycerol acyltransferase (DGAT1-2) 2 which catalyzes the final step of oil biosynthesis. Over-expression of the unique high oil 3 DGAT1-2 allele increased seed oil content by up to 41%. With the qHO6 cloned, a 4 rigorous field yield test of DGAT1-2 transgenic plants should be able to answer whether 5 high oil content in grain affects grain yield or not. Interestingly, transgenic expression of 6 a fungal DGAT 2 resulted in a relative increase in oil of 7.5% in soybean with no 7 significant difference from control in seed yield and other agronomic traits at multiple 8 location field trials (Lardizabal et al., 2008). Over-expression of BnDGAT1 in canola 9 increased seed oil content by ~13% in the greenhouse but the increase in oil dropped to 10 ~3% under field conditions. The effect on seed yield and other agronomic traits was not Oil Analysis. Maize seed oil content was determined as described previously (Zheng et 6 al. 2008). For embryo oil and weight, seeds were soaked in water overnight at room 7 temperature. Embryos were then dissected from their endosperm and lyophilized. 8 Embryo oil content was determined by NMR. Because oil content in endosperm is too 9 low to be detected by NMR, the hexane extraction method was used to determine 10 endosperm oil content. Endosperm oil was extracted by hexane from 5-10 g endosperm 11 meal. Hexane supernatant was transferred to a pre-weighed aluminum boat after being 12 centrifuged at 20,000 rpm for 5 min. The hexane was evaporated in the hood and the boat 13 was baked at 100 o C for 5 min. After cooling down to room temperature, the boat was 14 The following materials are available in the online version of this article. 31 Supplemental Table S1. Effect of ZmLEC1 on embryo weight, seed weight, embryo oil 1 content, and seed oil content in maize. 2 Supplemental Table S2. Effect of ZmWRI1 on embryo weight, seed weight, embryo oil 3 content, and seed oil content in maize. This article is in memory of our great team leader, Dr. Mitchell Tarczynski, whose 4 scientific vision and strong leadership were critical for the success of the project. We 5 thank William Solawatz, Marjorie Rudert, and Shifu Zhen for growing maize plants in 6 the field and greenhouse; David Sevenich for protein measurement; Igor Oliveira for 7 Deciphering gene regulatory networks that control seed development and maturation in 8 Arabidopsis. Plant J. 54:608-620 9 10 Seebauer analyzed. Data were shown as mean ± SD. All 3 transgenic lines showed a significant 9 increase in seed oil content compared to their corresponding null segregates as 10 determined by Student's t-test (P<0.01). C, Warm germination test of transgenic seeds. 11 Transgenic and null seeds were placed between two sheets of filter paper and germinated     Figure 4. Protein and starch content of ZmWRI1 transgenic embryos. The data represent mean ± SD of 3 replicate samples. Each sample was run in triplets in starch and protein assays. All 3 transgenic lines showed no significant difference in embryo protein content (p>0.1 by Student's t-test) but did show a significant reduction in embryo starch content as determined by Student's t-test (p<0.05).

Figure 5.
Early stand count and plant height of transgenic ZmWRI1 plants in field. Six rows of transgenic and 3 rows of null were planted in field for each transgenic line. Early stand count % was calculated by number of plants divided by total seeds planted at 3 weeks after planting. Plant height of 10 plants in the middle of each row was measured. Data are shown as mean ± SD. All 5 transgenic lines showed no significant difference from null as determined by Student's t-test (p>0.1).