Skip to main content
NCBI home page
Plant and Cell Physiology logoLink to Plant and Cell Physiology
. 2023 Nov 20;65(6):872–884. doi: 10.1093/pcp/pcad148

A Novel Soybean Diacylglycerol Acyltransferase 1b Variant with Three Amino Acid Substitutions Increases Seed Oil Content

Kayla S Flyckt 1, Keith Roesler 2, Kristin Haug Collet 3, Luciano Jaureguy 4, Russ Booth 5, Shawn R Thatcher 6, John D Everard 7, Kevin G Ripp 8, Zhan-Bin Liu 9, Bo Shen 10, Laura L Wayne 11,*
PMCID: PMC11209548  PMID: 37982755

Abstract

Improving soybean (Glycine max) seed composition by increasing the protein and oil components will add significant value to the crop and enhance environmental sustainability. Diacylglycerol acyltransferase (DGAT) catalyzes the final rate-limiting step in triacylglycerol biosynthesis and has a major impact on seed oil accumulation. We previously identified a soybean DGAT1b variant modified with 14 amino acid substitutions (GmDGAT1b-MOD) that increases total oil content by 3 percentage points when overexpressed in soybean seeds. In the present study, additional GmDGAT1b variants were generated to further increase oil with a reduced number of substitutions. Variants with one to four amino acid substitutions were screened in the model systems Saccharomyces cerevisiae and transient Nicotiana benthamiana leaf. Promising GmDGAT1b variants resulting in high oil accumulation in the model systems were selected for overexpression in soybeans. One GmDGAT1b variant with three novel amino acid substitutions (GmDGAT1b-3aa) increased total soybean oil to levels near the previously discovered GmDGAT1b-MOD variant. In a multiple location field trial, GmDGAT1b-3aa transgenic events had significantly increased oil and protein by up to 2.3 and 0.6 percentage points, respectively. The modeling of the GmDGAT1b-3aa protein structure provided insights into the potential function of the three substitutions. These findings will guide efforts to improve soybean oil content and overall seed composition by CRISPR editing.

Keywords: DGAT1, Glycine max, Lipid, Oil, Protein, Seed composition, Soybean

Introduction

Soybean (Glycine max) is a major crop grown for its oil and high-quality protein. Soybean seeds are composed of approximately 20% oil, 36% protein and 30% carbohydrates, with some variability depending on the growth environment (Assefa et al. 2019). While soy protein is primarily used for animal feed, the main use of soybean oil is for human consumption (Anderson et al. 2019). Soybean oil is the second most highly consumed cooking oil worldwide after palm oil (Thrane et al. 2017). Recently, the demand for soy oil as a feedstock for biofuel production has soared, with utilization expected to increase by 46% from 2022 to 2027 (International Energy Agency 2022).

Triacylglycerol (TAG), the primary component of seed oils, is a storage lipid comprising a glycerol backbone with three esterified fatty acids. In plants, fatty acids are synthesized from acetyl-CoA in the plastid, a process that requires acetyl-CoA carboxylase to form malonyl-CoA, and the fatty acid synthase complex to extend the acyl chain up to 16 or 18 carbons in length (Nikolau et al. 2003). TAG assembly is catalyzed by a series of integral acyltransferases within the ER (Cagliari et al. 2011). The enzymes glycerol-3-phosphate acyltransferase and lysophosphatidic acid acyltransferase subsequently add acyl-CoAs to glycerol-3-phosphate (G3P) at the sn-1 and sn-2 positions, respectively, resulting in phosphatidic acid (PA). PA is then hydrolyzed by PA phosphatase, forming diacylglycerol (DAG) (Nakamura et al. 2007). The final reaction, which acylates the sn-3 position of glycerol, is catalyzed by DAG acyltransferase (DGAT) (Cases et al. 1998, Liu et al. 2012). This terminal reaction is the final committed step in TAG biosynthesis and plays a key role in controlling the flux of carbon toward oil (Lung and Weselake 2006). Alternatively, DAG can be acylated independent of acyl-CoA by phospholipid:DGAT (Zhang et al. 2009).

DGAT enzymes are categorized into four distinct families that share little sequence similarity: DGAT1, DGAT2, DGAT3 and wax ester synthase (WS/DGAT) (Xue et al. 2022). Type-1 DGATs are the primary family contributing to seed oil accumulation in multiple plant species including maize, Arabidopsis and soybean (Chen et al. 2016). Of the three type I DGATs in soybean (i.e. GmDGAT1a, GmDGAT1b and GmDGAT1c), GmDGAT1a and GmDGAT1b are proposed to be most important for oil accumulation in seeds, as their expression is correlated with the rate of TAG biosynthesis (Li et al. 2013, Hatanaka et al. 2016). DGAT1s generally contain 8–10 transmembrane domains, with the characteristic conserved membrane-bound O-acyltransferase fold (McFie et al. 2010, Coupland et al. 2023). These proteins have been shown to exist as homodimers or homotetramers and contain a variable hydrophilic N-terminus, involved in regulation and oligomerization, and a more highly conserved C-terminus, which includes an invariable catalytic histidine at the active site (Weselake et al. 2006, Caldo et al. 2017). Additionally, the C-terminus contains a FYXDWWN motif that is conserved across DGAT1 and acyl-CoA cholesterol acyltransferase enzymes (Lopes et al. 2014, Chen et al. 2022). This motif has been suggested to coordinate acyl-CoA binding; however, its position in the human DGAT1 structure implies a role in dimerization via interaction with the N-terminus (Guo et al. 2001, Das et al. 2008, Wang et al. 2020).

Many studies have demonstrated an increase in seed oil content to various levels by overexpression of DGAT in species including Arabidopsis, rapeseed and cotton (Lardizabal et al. 2008, Weselake et al. 2008, Wang et al. 2014, Guo et al. 2017, Wu et al. 2021). Additionally, a high-oil quantitative trait locus (QTL) from maize has been mapped to a single amino acid insertion in ZmDGAT1-2 (Zheng et al. 2008), demonstrating that a modification within the gene can also lead to increased TAG. The heterologous expression of a highly active DGAT1 from Vernonia galamensis in soybean increased oil content by 4 percentage points (Hatanaka et al. 2016). Increasing oil in soybean by several percentage points with a native soybean DGAT1, however, is only possible using an engineered protein variant. We previously showed that overexpression of a kinetically improved soybean DGAT1b variant with 14 amino acid substitutions (GmDGAT1b-MOD) increases oil content when overexpressed in soybean (Roesler et al. 2016). Using this work to derive functional information on specific GmDGAT1b amino acid residues is challenging, as it is unknown which of the 14 substitutions are required to achieve the high-oil phenotype.

With the goals of increasing oil using a gene-editing approach and determining specific substitutions that can increase oil, we designed additional GmDGAT1b variants with one to four substitutions and evaluated them in the model systems yeast and transient Nicotiana benthamiana leaf. Four novel gene variants were selected for expression in soybean with the oleosin promoter. The most efficacious variant, GmDGAT1b C355S, N473S and I479S, was subsequently tested with the stronger promoter, β-conglycinin. In a multiple location field trial, this gene variant increased oil by 2.3 percentage points without decreasing protein. This work provides insight into the function of three amino acid substitutions that increase oil content and outlines a route to increasing oil by CRISPR editing.

Results

GmDGAT1b variants with one to four amino acid substitutions increased oil in yeast and N. benthamiana

GmDGAT1b variants were designed to contain four or fewer amino acid substitutions to simplify future CRISPR editing efforts and to examine specific residues that are important for increasing TAG production. In total, 16 amino acid positions were targeted primarily based on overexpression of DGAT1s and insights gained from screening the DNA-shuffled hazelnut (Corylus avellana) DGAT1 libraries (Roesler et al. 2016). These amino acid positions included some of the substitutions found in the GmDGAT1b-MOD variant (S58N, I470M, R206K, Y231F, D258E, S264T, V273L, I303V, I440M and R467K) and additional substitutions identified from the hazelnut DGAT1 variants. Soybean substitutions C355S, N473S, T216V and K328N were additional amino acids of interest due to the corresponding hazelnut substitutions identified in variants that resulted in high oil in yeast. The I479S substitution was also selected due to the presence of the corresponding hazelnut substitution in the five best variants from library D. Additionally, the A210V substitution was selected due to the presence of valine at the corresponding position in the highly active DGAT1 from V. galamensis (Hatanaka et al. 2016). Variants with diverse combinations of the selected substitutions were evaluated in the yeast DGA1Δ/LRO1Δ mutant strain by Nile red staining, which is highly correlated with lipid content (Zhao et al. 2019). Nile red analysis revealed nine GmDGAT1b variants with increased fluorescence intensities on average compared to yeast cells expressing GmDGAT1b wild type (WT) (Fig. 1A). One variant with a single amino acid substitution, I479S, exhibited the highest fluorescence, at 47% greater than GmDGAT1b WT.

Fig. 1.

Fig. 1

Open in a new tab

Evaluation of GmDGAT1b variants in model systems. (A) Nile red staining of GmDGAT1b variants expressed in the S. cerevisiae DGA1Δ/LRO1Δ double mutant strain. At least four biological replicates and six technical replicates were analyzed for each variant. Error bars denote standard deviation. (B) Total oil accumulation in N. benthamiana leaves expressing GmDGAT1b variants. Six biological replicates were analyzed for each variant. Individual data points are represented as dots, and error bars represent the standard error of the mean. One-way ANOVA revealed that statistically significant differences were present between at least two groups (F = 26.3, P < 0.0001). Tukey’s Honestly Significant Difference test was performed for pairwise comparisons between all gene variants, with non-overlapping letters signifying statistically significant differences (P < 0.05).

A subset of variants was then evaluated for the ability to promote oil accumulation when transiently expressed in N. benthamiana leaf. Six variants were selected in addition to the GmDGAT1b WT and empty vector (EV) controls based in part on the yeast Nile red results. As it was unknown if performance in the yeast model system directly correlated with oil accumulation in soybean, several variants that did not increase oil in yeast were also included. Additionally, the GmDGAT1b-MOD variant was included as a benchmark. Leaves expressing EV accumulated very small amounts of TAG at less than 0.01% of the dry weight. Expression of all DGAT1b variants increased leaf oil content compared to EV by at least 10-fold (Fig. 1B). Three of the variants tested, including GmDGAT1b-MOD, significantly increased oil content compared to GmDGAT1b WT. One novel DGAT1b variant with three amino acid substitutions, GmDGAT1b C355S, N473S and I479S (referred to as GmDGAT1b-3aa henceforth), accumulated oil to the same level as GmDGAT1b-MOD, which was significantly higher than WT and all other variants tested.

GmDGAT1b variants with one to four amino acid substitutions increased oil in soybean

Based on the results from the yeast and N. benthamiana model systems, four novel variants were selected for transgenic expression in soybean, which will provide proof of concept prior to CRISPR editing. These top GmDGAT1b candidates included a single I479S substitution, a double C355S, I479S, the triple GmDGAT1b-3aa and a quadruple T216V, D258E, S265T, I479S. Additionally, GmDGAT1b WT and GmDGAT1b-MOD were included as controls. These six genes, driven by the moderately high seed-specific oleosin promoter, were transformed into soybean for overexpression and grown to the T3 generation. Interestingly, overexpression of GmDGAT1b-MOD did not increase seed oil content significantly compared to respective sibling null segregants (Fig. 2A). This result was most likely due to the relatively lower seed-specific expression of the oleosin promoter, whereas the previous report used a strong seed-specific β-conglycinin promoter (Roesler et al. 2016). Three of the tested variants, (i) GmDGAT1b I479S; (ii) GmDGAT1b 3-aa and (iii) GmDGAT1b, T216V, D258E, S264T and I479S, showed a significant increase in seed oil content compared to the associated sibling nulls, with mean increases of 0.50, 0.69 and 0.55 percentage points, respectively. There were no significant differences in protein content (Fig. 2B). While significant, the measured oil increases were smaller than the changes previously recorded with expression of GmDGAT1b-MOD with the β-conglycinin promoter (Roesler et al. 2016). Thus, the variant with the highest oil increase, GmDGAT1b-3aa, was selected to be expressed in a new construct, driven by this stronger seed-specific promoter (Supplementary Table S1).

Fig. 2.

Fig. 2

Open in a new tab

T3 seed analysis of GmDGAT variants expressed using the oleosin promoter. Seed was analyzed by FT-NIR spectroscopy for oil (A) and protein (B), with n ≥ 6 representing at least two events per variant. Oil and protein values are reported on a wt%, at 13% moisture, basis. Statistical comparisons of seeds expressing each GmDGAT1b variant were made with respect to sibling null segregants by Student’s t-test, with asterisks representing significant differences (P < 0.05).

Overexpression of the GmDGAT1b-3aa substitution variant with a stronger promoter further increased oil

A GmDGAT1b-3aa construct with the β-conglycinin promoter driving expression was stably transformed into soybean. Analysis of T2 seed composition showed that DGAT overexpression conferred a greater increase in oil content compared with the previously analyzed oleosin promoter events (Supplementary Fig. S1A). Oil was significantly increased compared to segregating sibling nulls for the four events analyzed, with an average increase of 2.7 percentage points. No significant changes in protein were detected (Supplementary Fig. S1B). Three events were advanced to the T3 generation and were grown as short rows in the field along with a GmDGAT1b-MOD event expressed with the β-conglycinin promoter, named Soil 111. GmDGAT1b-3aa increased oil by 3.0 percentage points on average compared to sibling nulls (Fig. 3A). The oil produced by this variant was similar to the levels measured in GmDGAT1b-MOD-expressing seeds, which increased oil by 3.3 percentage points compared to nulls. These results indicate that only three amino acid substitutions are sufficient to increase oil to levels achieved with GmDGAT1b-MOD, which has 14 amino acid substitutions. There were no significant differences in protein content for any of the events (Fig. 3B). All GmDGAT1b-overexpressing events had increased C16:0, C18:0 and C18:1 fatty acids and decreased C18:2 and C18:3 fatty acids (Supplementary Table S2). Additionally, stachyose, sucrose and total carbohydrates were decreased (Supplementary Table S3). To further explore the effect of GmDGAT1b-3aa overexpression, 100-seed weight was determined on T3 seeds. Interestingly, two of the three GmDGAT1b-3aa events showed significant decreases of 5.7% and 5.9% in 100-seed weight compared to sibling null segregants (Supplementary Fig. S2). These results prompted additional investigation into the influence of GmDGAT1b-3aa on yield.

Fig. 3.

Fig. 3

Open in a new tab

T3 seed analysis of GmDGAT1b-3aa and GmDAT1b MOD (soil 111) expressed using the β-conglycinin promoter. Seed was analyzed by NIT spectroscopy for oil (A) and protein (B) (n ≥ 6). Oil and protein values are reported on a wt%, at 13% moisture, basis. Statistical comparisons for each event were made with respect to null segregants by Student’s t-test, with asterisks representing significant differences (P < 0.05).

Quantifying yield, protein and oil in a multiple location field trial

Two GmDGAT1b-3aa overexpression events with the highest oil content along with the corresponding null segregants were advanced to a multiple location field trial. Yield was measured across eight locations, and seed composition was determined on a representative selection of six locations. One of the two overexpression events resulted in a small but statistically significant decrease in yield compared to nulls (Table 1). There were small but significant increases in protein content of 0.48 and 0.55 percentage points for events 1 and 2, respectively. Oil content was significantly elevated in both GmDGAT1b-3aa events by 2.3 percentage points on average, with events 1 and 2 reaching 22.7% and 22.5% oil, respectively. These results were used to estimate kilograms of oil and protein per hectare (Supplementary Table S4). Oil was increased by an estimated 61.9 and 63.4 kg/ha for events 1 and 2, respectively. Protein was slightly decreased on a per-hectare basis by 9.6 kg ha−1 for event 1 and 33.2 kg ha−1 for event 2. Differences in fatty acid profile and carbohydrates were also noted. Positive seeds had elevated C16:0, C18:0 and C18:1 and decreased C18:2 and C18:3 fatty acids compared to null segregants (Table 2). Total soluble carbohydrates decreased, as did the individual sugars stachyose and sucrose (Table 3). Taken together, these results illustrate that the GmDGAT1b-3aa variant can significantly increase oil, while maintaining protein content across a variety of environmental conditions.

Table 1.

Overall seed composition and yield of GmDGAT1b-3aa grown in multiple location field trials. Oil and protein values are reported on a wt%, at 13% moisture, basis

Oil Protein Yield
Event Genotype % Δ % points % Δ % points kg ha−1 Δ kg ha−1
Event 1 Null 20.37 ± 0.29C 35.49 ± 0.25C 4,250.2 ± 168.1A
Positive 22.65 ± 0.29A 2.28 35.97 ± 0.25B 0.48 4,156.1 ± 168.1A −94.1
Event 2 Null 20.15 ± 0.29D 35.96 ± 0.25B 4,176.2 ± 174.9A
Positive 22.46 ± 0.29B 2.31 36.51 ± 0.25A 0.55 3,994.7 ± 174.9B −181.5
WT Control 20.07 ± 0.29D 35.63 ± 0.25C 4,250.2 ± 174.9A
Open in a new tab

Yield was determined based on eight locations, and compositional analysis was determined for samples from six locations by NIT spectroscopy. Indicated values represent mean ± standard error. Statistical comparisons across all groups were determined by BLUP analysis, with non-overlapping letters representing significant differences (P < 0.05).

Table 2.

Fatty acid profile for GmDGAT1b-3aa soybeans grown in the multiple location field trial. Fatty acid values are expressed on a relative percentage basis

Event Genotype C16:0 C18:0 C18:1 C18:2 C18:3
Event 1 Null 10.85 ± 0.13B 3.88 ± 0.08C 23.91 ± 0.49D 52.99 ± 0.54A 7.19 ± 0.23 CD
Positive 12.22 ± 0.13A 4.50 ± 0.08A 32.23 ± 0.49B 43.49 ± 0.54C 5.03 ± 0.23E
Event 2 Null 10.98 ± 0.13B 4.06 ± 0.08B 25.76 ± 0.49C 51.51 ± 0.54B 6.99 ± 0.23D
Positive 12.39 ± 0.13A 4.54 ± 0.08A 33.40 ± 0.49A 42.38 ± 0.54D 4.84 ± 0.23E
WT Control 11.00 ± 0.13B 3.92 ± 0.08BC 22.83 ± 0.49E 53.52 ± 0.54A 7.54 ± 0.23AB
Open in a new tab

Seed composition was determined from six locations determined by NIT spectroscopy. Indicated values represent mean ± standard error. Statistical comparisons across all groups were determined by BLUP analysis, with non-overlapping letters representing significant differences (P < 0.05).

Table 3.

Carbohydrate analysis for GmDGAT1b-3aa soybeans grown in the multiple location field trial. Values are reported on a wt%, at 13% moisture, basis

Event Genotype Raffinose Stachyose Sucrose Total carbohydrates
Event 1 Null 1.01 ± 0.05ABC 3.47 ± 0.08A 4.51 ± 0.22BC 9.46 ± 0.1AB
Positive 1.05 ± 0.05A 2.49 ± 0.08B 3.68 ± 0.22E 8.08 ± 0.1D
Event 2 Null 1.02 ± 0.05AB 3.50 ± 0.08A 4.58 ± 0.22AB 9.25 ± 0.1C
Positive 0.95 ± 0.05D 2.28 ± 0.08C 3.97 ± 0.22D 7.92 ± 0.1D
WT Control 0.98 ± 0.05BCD 3.53 ± 0.08A 4.74 ± 0.22A 9.62 ± 0.1A
Open in a new tab

Seed composition was determined from six locations, determined by NIT spectroscopy. Indicated values represent mean ± standard error. Statistical comparisons across all groups were determined by BLUP analysis, with non-overlapping letters representing significant differences (P < 0.05).

Analysis of the GmDGAT1b-3aa protein structure

Initially, the amino acid sequence was analyzed to determine if the three amino acid substitutions are located at positions of known functionality. A pairwise alignment was generated to compare GmDGAT1b-3aa to the human, maize and Arabidopsis DGAT1 sequences (Supplementary Fig. S3). The three amino acid substitutions were annotated in addition to predicted transmembrane domains and previously determined residues of interest (Wang et al. 2020). The C355 position is located within transmembrane domain 6, while N473 and I479 are near and within transmembrane domain 9, respectively. The N473 position is conserved between the four sequences, while the C355 and I479 positions are variable. These substitutions were not located at any known active site, acyl-CoA binding site, or within the conserved FYXDWWN motif.

To determine the function of the three amino acids within the three-dimensional structure, the GmDGAT1b WT and GmDGAT1b-3aa protein structures were predicted as homodimers. The overall structure of the GmDGAT1b protein is similar to that of the human DGAT1 (Sui et al. 2020, Wang et al. 2020), with a canoe-like shape (Fig. 4A). Most of the predicted structure was confident or highly confident apart from the N-terminus and two loops between alpha helices (Supplementary Fig. S4). The three amino acid residues of interest fell within confident or highly confident regions. No major differences in the overall structure were observed between GmDGAT1b WT and GmDGAT1b-3aa, and alignment of the structures yielded a very high similarity (Template Modeling (TM) score: 0.98633) (Supplementary Fig. S5). Residues described in Supplementary Fig. S3 were highlighted in the three-dimensional structure (Fig. 4B, C). As expected, the acyl-CoA binding residues faced inward toward the protein core, which functions as the reaction chamber. Notably, the N473 residue is in very close proximity to the conserved FYXDWWN motif. The N473 and I479 substitutions reside near the dimer interface and are within the same alpha helix as the acyl-CoA binding site Q485. The C355 position is not near any of the labeled functional residues. Next, we investigated the predicted hydrogen bonding between side chains at the three amino acid positions. In the WT protein, C355 and I479 did not form any hydrogen bonds (Supplementary Figs. S6A, S6E. However, in the improved variant, both S355 and S479 formed hydrogen bonds with nearby amino acid side chains (Supplementary Figs. S6B, S6F). Interestingly, in the WT protein, N473 formed hydrogen bonds with four amino acids: (i) W380 (within the FYXDWWN motif), (ii) A382, (iii) W477 and (iv) N113 of the neighboring protomer (Supplementary Fig. S6C). In the improved variant, only the W477 interaction remained (Supplementary Fig. S6D).

Fig. 4.

Fig. 4

Open in a new tab

Structural analysis of the GmDGAT1b-3aa variant modeled by AlphaFold. (A) Overall predicted structure represented as a dimer. (B and C) (two views) position of the three amino acid substitutions (yellow) in the GmDGAT1b-3aa variant, putative active site (red), acyl-CoA binding sites (blue) and the conserved FYXDWWN motif (green).

Considering these results, and that the individual C355S and N473S substitutions had not been previously evaluated in planta, an additional experiment was performed in transient N. benthamiana to dissect the impact of each of the three individual substitutions on oil accumulation. Of the three substitutions tested, N473S was the only single substitution that resulted in a statistically significant oil increase compared to GmDGAT1b WT (Supplementary Fig. S7). These results, taken together with the structural analysis, show that the N473S substitution is important for increasing oil.

Discussion

Since its discovery, DGAT1 has been the subject of multiple efforts to increase oil content in plants (Xu et al. 2008, Maravi et al. 2016, Chen et al. 2017), algae (Zhang et al. 2021) and yeast (Greer et al. 2015). As the final committed step in TAG biosynthesis, DGAT serves to ‘pull’ the flux of carbon toward oil. In soybean, there is generally a strong negative correlation between oil and protein. Given that both soy oil and protein are important commodities, it is essential to break this inverse relationship when increasing oil to generate a viable product. In a previous study, we showed that the improved GmDGAT1b-MOD variant with 14 amino acid substitutions increased oil in soybean while also slightly increasing protein (Roesler et al. 2016). The 14 substitutions were derived from DNA-shuffled variants of a DGAT1 from hazelnut, CaDGAT1, prior to applying the substitutions to GmDGAT1b. It was unknown which of the substitutions identified by the DNA shuffling experiment were required for the oil phenotype and if any of the substitutions had a negative impact on GmDGAT1b activity or stability. Also, some of the corresponding substitutions from CaDGAT1 may perform differently in GmDGAT1b. Therefore, we screened additional combinations of substitutions from the CaDGAT1 DNA shuffling effort in GmDGAT1b and discovered a novel variant with only three amino acid substitutions that increased total oil and protein when overexpressed in soybean and, for at least event 1, maintained yield across multiple environmental locations.

Model systems inform soybean oil accumulation

This study highlights the use of model systems to prioritize advancement to stable plant transformation. We initially screened variants in yeast, which confirmed functionality and provided preliminary insights into their efficacy (Fig. 1A). Transient expression in N. benthamiana leaf provided data on the performance of the gene variants in planta (Fig. 1B). Interestingly, oil accumulation between the model systems and stably transformed soybean did not correlate in all cases. For instance, the GmDGAT1b I479S variant performed well in yeast and significantly increased oil in soybean but did not increase oil in N. benthamiana leaf compared to GmDGAT1b WT (Figs. 1 and 2, and Supplementary Fig. S7). Conversely, the GmDGAT1b-3aa variant accumulated less oil than WT in yeast but was the best performing variant in soybean and N. benthamiana. Factors that may impact DGAT performance in different organisms and tissue types include the presence or absence of interacting proteins, impact on cell growth and viability and the available acyl-CoA and DAG substrates. For example, notable differences in DGAT activity measured in vitro have been reported depending on the fatty acid composition of both the acyl-CoA donor and the DAG acceptor provided (Lager et al. 2020). The available substrates in the model systems differ from soybean, with yeast rich in C16:1 and C18:1, and N. benthamiana leaf largely containing C18:3. In comparison, the predominant fatty acid in soybean seeds is C18:2, followed by C18:1.

One to three amino acid substitutions are sufficient to increase oil in soybean

Given our goal of increasing oil by CRISPR editing, we hoped to identify an efficacious GmDGAT1b variant with fewer substitutions compared to the previously discovered GmDGAT1b-MOD variant with 14 amino acid changes. Of the DGATs expressed in soybean with the oleosin promoter, three novel variants with one, three and four substitutions significantly increased oil content (Fig. 2). Previous studies overexpressing DGAT1 in seeds have used strong constitutive or seed-specific expression and have not explored the effect of promoter strength (Weselake et al. 2008, Hatanaka et al. 2016, Guo et al. 2017, Wu et al. 2021). We found that expressing GmDGAT1b-3aa with the stronger β-conglycinin promoter resulted in higher oil, with an increase of 2.3 percentage points (Table 1), compared to expression with the oleosin promoter, which increased oil by 0.69 percentage points (Fig. 2B). These results suggest that a strong seed-specific promoter driving the novel DGAT1b variant expression is likely required to attain high oil.

Soybeans expressing GmDGAT1b-MOD from our previous study did not significantly increase oil compared to nulls when expressed with the oleosin promoter, even though the MOD variant shows efficacy in soy when expressed under control of the β-conglycinin promoter (Roesler et al. 2016) (Fig. 3). These results indicate that the three novel variants are more effective than GmDGAT1b-MOD at lower expression levels. One hypothesis is that a stronger promoter is needed for GmDGAT1b-MOD to reach protein levels sufficient to increase oil accumulation due to lower protein stability. Additional work, such as quantifying DGAT protein levels in developing seeds and determining protein half-lives, is needed to characterize the stability of these engineered variants.

In addition to increased oil, overexpressing GmDGAT1b-3aa altered the fatty acid profile, with elevated C16:0, C18:0 and C18:1 and decreased C18:2 and C18:3 (Table 2). Similar changes were previously observed upon overexpression of both the DGAT1 from V. galamensis (VgDGAT1A) and GmDGAT1b-MOD in soybean (Hatanaka et al. 2016, Roesler et al. 2016). Given that C18:1 is further desaturated after incorporation into phosphatidylcholine (PC), the decrease in polyunsaturated fatty acids could be attributed to increased flux through DGAT, diverting C18:1-CoA away from PC (Roesler et al. 2016). In contrast, overexpressing a sesame DGAT1 (SiDGAT1) in a high-oil soybean cultivar resulted in increased C16:0 and C18:2 acid contents and decreases in C18:0 and C18:1 (Wang et al. 2019). These differences could be due to variation in substrate specificity between DGATs of different species. Taken together, these results demonstrate the complex factors influencing oil composition, which should be considered in the context of the end-use product.

GmDGAT1b-3aa-overexpressing soybeans maintain protein content with a minimal change in yield

The inverse correlation between protein and oil is a major challenge for the improvement of soybean (Clemente and Cahoon 2009). Protein and oil are produced simultaneously during development, so repartitioning between components is likely required due to limited carbon supply (Kambhampati et al. 2020). In fact, increased assimilate supply per seed results in higher protein content, but is associated with fewer seeds per plant in breeding populations (Rotundo et al. 2009). Thus, increasing oil while maintaining both protein content and yield presents an even greater challenge. In the field trial, we showed that protein content was slightly increased in both the tested events (Fig. 4B). This finding is consistent with previous studies, which have shown constant or slightly elevated protein when increasing oil by overexpression of a DGAT1 protein (Roesler et al. 2016, Al-Amery et al. 2019). These studies, however, have lacked robust yield data obtained across multiple locations. In 2022, we evaluated yield for two GmDGAT1b-3aa events across eight locations in the midwestern USA (Supplementary Table S5). A slight yield decrease was measured for both events, while only event 2 had a statistically significant difference (Table 1). Additionally, both events had significantly decreased 100-seed weight in T3 seeds (Supplementary Fig. S2). This finding is interesting, given that DGAT1 overexpression has been shown to increase seed weight in Arabidopsis and Camelina (Jako et al. 2001, Kim et al. 2016). Despite the yield reduction, both GmDGAT1b-3aa events resulted in increased oil on a land area basis by 61.9 and 63.4 kg ha−1 (Supplementary Table S4).

Protein structure prediction enables investigation into the function of GmDGAT1b-3aa substitutions

Recent advances have tremendously improved the accuracy and availability of three-dimensional protein structure predictions (Jumper et al. 2021, Varadi et al. 2021). Moreover, the experimentally determined structure of the human DGAT1 protein has helped to elucidate the role of specific amino acid positions (Wang et al. 2020). We therefore predicted the three-dimensional structure of GmDGAT1b-3aa and determined the location of the three substitutions, C355S, N473S and I479S, relative to the active site and other important residues and motifs (Fig. 4). This analysis revealed that GmDGAT1b N473 is predicted to form hydrogen bonds with multiple amino acids in WT GmDGAT1b, including interaction with the conserved FYXDWWN motif and the adjacent protomer, and that the N473S substitution disrupts these interactions (Supplementary Fig. S6). The N473S substitution is likely important for efficacy of the GmDGAT1b-3aa variant, as this individual substitution increased oil in N. benthamiana leaf (Supplementary Fig. S7). Thus, we hypothesize that the N473S substitution improves GmDGAT1b by disrupting a hydrogen bond between the FYXDWWN motif and transmembrane domain 9, which improves the position or flexibility of this essential motif. Both the N473S and I479S substitutions are within the same alpha helix as the acyl-CoA binding site at Q485, which is proposed to stabilize the acyl-CoA thioester (Sui et al. 2020, Wang et al. 2020). These substitutions, particularly I479S that is in close proximity, could possibly alter the positioning or stabilize a favorable conformation to promote the interaction of Q485 with the thioester. This substitution is likely also critical to the efficacy of GmDGAT1b-3aa, as the individual I479S substitution significantly increased oil in soybean when expressed with the oleosin promoter (Fig. 2). The I479 position corresponds to a high-oil QTL in the maize DGAT1, with the high-oil allele containing a phenylalanine insertion, further supporting the importance of this residue (Zheng et al. 2008). An explanation for the role of the C355S substitution is less obvious, given that it is not near any of the functional residues examined. We have not shown definitively that the C355S substitution is necessary for increasing oil in planta and the GmDGAT1b C355S and I479S variant did not significantly increase oil in soybean (Fig. 2A). Taken together, these results highlight the importance of the N473S and I479S substitutions for efficacy of the GmDGAT1b-3aa variant and provide potential mechanisms to explain the oil increase. Additional work, such as measuring enzyme kinetics and performing additional mutational studies at the three amino acid positions, is needed to confirm these hypotheses and to determine if all three substitutions are required to achieve high oil.

Given the similar effect on oil accumulation in soybean, it is interesting that none of the 14 amino acid substitutions in the GmDGAT1b-MOD variant overlap with the three substitutions in GmDGAT1b-3aa. The 14 amino acid substitutions are distributed throughout the protein and are not concentrated in the C-terminus as is seen with GmDGAT1b-3aa (Roesler et al. 2016). An effort to engineer the DGAT1 from Brassica napus, BnaDGAT1, by directed evolution also identified single amino acid substitutions throughout the protein that improved oil accumulation in yeast and N. benthamiana leaf, several of which were located in the C-terminal region (Chen et al. 2017). The importance of C-terminal substitutions for increasing DGAT efficacy is logical, as this region contains acyl-CoA binding and active site residues (Wang et al. 2020). The BnaDGAT1 N-terminus has been shown to contain an allosteric site that regulates activity based on the available acyl-CoA to CoA ratio (Caldo et al. 2017). N-terminal substitutions are also found in the GmDGAT1b-MOD variant and were identified in the BnaDGAT1-directed engineering effort. Taken together, these studies have shown that substitutions throughout the DGAT1 protein can have a positive effect on oil accumulation.

In conclusion, this work has identified three novel GmDGAT1b gene variants with one to four amino acid substitutions that increased oil in soybean. Additional investigation revealed that expression of GmDGAT1b-3aa with a stronger promoter further increased oil content without decreasing protein. Moreover, analysis of the predicted protein structure and individual substitutions suggest that the N473S and I479S are critical to achieving the high-oil phenotype. Our results demonstrate that producing high-oil soybean by CRISPR editing is possible by expressing GmDGAT1b-3aa with a strong seed-specific promoter such as β-conglycinin. With the advent of CRISPR-Cas genome editing technology, targeted modification of the endogenous GmDGAT1b gene has become a viable and appealing option. The native GmDGAT1b promoter can be replaced with a stronger soybean promoter, and targeted nucleic acid substitutions in the protein-coding sequence can also be achieved (Podevin et al. 2013, Svitashev et al. 2016, Shi et al. 2017). This approach provides an advantage over a transgenic solution by optimizing the trait gene at its native genomic location, thereby eliminating potential positional effects of the insertion and the possibility of endogenous gene disruption, resulting in no transgenic sequences inserted into the soybean genome (e.g. T-DNA borders). Taken together, we have provided insights into specific residues that are important for increasing oil and provide a promoter-gene replacement strategy for increasing soybean oil content by CRISPR editing.

Materials and Methods

Constructs

GmDGAT1b gene variants were generated by a combination of synthesis and site-directed mutagenesis (Genscript, Piscataway, New Jersey). A yeast vector that contained the high copy, 2µ, origin of replication and the URA3 gene for plasmid selection was used. Gene variants were cloned between the PGK1 promoter and the PGK1 terminator. Transient expression in N. benthamiana was accomplished with a binary vector. Expression of DGAT variants was driven by the GmUBQ promoter followed by the GmUBQ terminator. Transient expression constructs also included DsRed, which was used to confirm transfection efficiency.

For stable soybean transformation constructs, helper plasmid and binary vectors were used as previously described (Cho et al. 2022). The WT GmDGAT1b (glyma.17g053300) and GmDGAT1b substitution variants were expressed as non-codon-optimized sequences without introns. A 1090-bp genomic fragment containing a 1010-bp promoter and an 80-bp 5ʹUTR of the soybean oleosin gene (glyma.19g063400) was used to drive expression of the GmDGAT1b-coding sequences. The 527-bp soybean MYB gene (glyma.19g061600) terminator was used. For expression of GmDGAT1b-3aa with the β-conglycinin promoter, 610 bp of the soybean conglycinin alpha' subunit gene promoter (glyma.10g246300) and a 1163-bp phaseolin gene terminator (Phvul007.G059750) from Phaseolus vulgaris L. were used.

Yeast complementation and growth

A low-oil yeast (Saccharomyces cerevisiae) double mutant, DGA1Δ/LRO1Δ, strain was generated as a system for evaluating soybean DGAT variants by complementation. Briefly, single mutants DGA1Δ and LRO1Δ were obtained in the BY4741 and BY4742 backgrounds, respectively (Horizon Discovery, Cambridge, United Kingdom). After mating, heterozygous diploid cells were selected on synthetic defined (SD)-Lys-Met medium. Cells grown overnight in 1% yeast extract, 2% peptone, 2% potassium acetate media were transferred to sporulation media (1% potassium acetate, 0.005% zinc acetate, 0.05% Lys Met) and were shaken at 250 rpm at room temperature, for 7 d. Double mutants were identified by random spore analysis (Giaever and Nislow 2014). Plasmids were transformed into yeast using the Clontech Yeastmaker Yeast Transformation Kit (Takara Bio, San Jose, California), as described by the manufacturer. Liquid cultures in SD-Ura media were inoculated with single colonies and were grown for 18 h at 30°C under shaking at 250 rpm, to an approximate optical denity at 600 nm (OD600) of 1, prior to analysis.

Nile red staining

Oil content in yeast was estimated by staining with a fluorescence lipophilic dye, Nile red. An initial OD600 was measured, and cultures were diluted to correct for differences in cell density if needed. Nile red staining was performed in 96-well microtiter plates by adding 5 µl of a 0.02-mg per-ml stock of Nile red dissolved in 95% ethanol to 200 µl of a 1:10 dilution of the yeast culture in phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.8 mM KH2PO4, pH 7.4). Staining was performed for 5 min, followed by the determination of fluorescence intensity using an excitation wavelength of 489 nm and an emission wavelength of 581 nm. A second OD600 reading was taken, fluorescence intensity was divided by OD600 to further correct for differences in cell density and buffer blanks with no yeast were used to correct for background.

Transient expression and TAG analysis in N. benthamiana leaf

DGAT variants were evaluated by Agrobacterium-mediated transient expression in N. benthamiana leaves. Vectors were transformed into Agrobacterium strain AGL1 by electroporation. Liquid cultures were grown at 28°C overnight in a shaking incubator at 250 rpm. Cells were then pelleted and resuspended in infiltration buffer [5 mM MgSO4, 5 mM MES (pH 5.6) and 150 µM acetosyringone]. Cultures were then incubated at room temperature for 2–4 h and diluted to a final OD600 of 0.2 prior to infiltration. Plants were grown for approximately 5 weeks before infiltration in a growth chamber under a 16-h photoperiod, a light intensity of 180 μmol m−2 s−1, 24°C/20°C light/dark temperature and 65% relative humidity. The youngest fully expanded leaf was chosen for infiltration (one leaf per plant). Agrobacterium suspensions were injected into the underside of leaves using a syringe without a needle while applying counter pressure. Infiltrated leaves were harvested 3 d post-infiltration.

Analysis of leaf TAG

Approximately 10 mg of leaf tissue was sampled into pre-weighed 1.2 ml polypropylene tubes. The tissue was lyophilized, and the exact dry weight was obtained. An internal standard, 0.02 mg of tri-C17:0 TAG, was added prior to extraction. Neutral lipids were extracted four times in 100% hexane. Extracts from all fractions were combined and concentrated to a 0.5-ml volume. TAG was isolated by solid phase extraction (SPE) using 96-well HyperSep™ aminopropyl SPE plates (Thermo Fisher, Waltham, Massachusetts). Columns were preconditioned with 1 ml of hexane prior to loading samples as a 0.5 ml hexane fraction. Columns were washed with 1 ml of hexane:dichloromethane:chloroform (88:10:4 v/v), and then TAG was eluted with 1 ml of hexane:ethyl acetate (96:4 v/v). The TAG fraction was concentrated and resuspended in 180 µl of heptane. Fatty acids were derivatized by the addition of 20 µl of ∼0.25 m trimethylsulfonium hydroxide in methanol (Sigma-Aldrich, St. Louis, Missouri), followed by quantitation of fatty acid methyl esters by GC-FID (Agilent, Santa Clara, California). Separation was achieved with an OmegaWax 320 column (Supelco, Bellefonte, Pennsylvania), with the GC oven at a starting temperature of 190°C and then increased to 240°C at 5°C min−1, with a total run time of 10 min.

Expression of DGAT genes in soybean seed

Expression vectors were introduced into soybean by Ochrobactrum-mediated soybean embryonic axis transformation as previously described (Cho et al. 2022). Briefly, mature dry seeds of soybean cultivar 93Y21 were disinfected and imbibed on semi-solid medium containing 5 g l−1 sucrose and 6 g l−1 agar at room temperature in the dark. After an overnight incubation, embryonic axis explants were isolated and transferred to the deep plate with 15 ml of Ochrobactrum haywardense H1-8 further containing the appropriate construct and helper vector with suspension at an OD600 of 0.5 in infection medium containing 200 µM acetosyringone. After sonication for 30 s, embryonic axis explants were transferred to a single layer of autoclaved sterile filter paper. The plates are sealed with Micropore tape and incubated under dim light (5–10 µE m−2 s−1, cool white, fluorescent lamps) for 16 h at 21°C for 3 d. After co-cultivation, the embryonic axis explants were cultured on shoot induction medium solidified with 0.7% agar in the absence of selection. Shoot induction was carried out at 26°C with a photoperiod of 18 h and a light intensity of 40–70 µE m−2 s−1. Six to seven weeks after transformation, elongated shoots were isolated and transferred to rooting medium containing the selection agent. Marker-free transgenic plantlets were generated by excising the selection marker within the LoxP site using heat shock treatment described in Cho et al. (2022).

Transgenic plantlets were transferred to soil pots and grown in the greenhouse. T0 plants were evaluated for transgene copy number, the presence of desired genetic elements and full excision of marker genes by a series of quantitative polymerase chain reaction (qPCR) assays, as previously described (Lowe et al. 2016). For the oleosin promoter constructs, single-copy events with fully excised markers were planted in the field and qPCR was used to determine zygosity and confirm the molecular analysis from the previous generation. Homozygous positive and null plants were selected and harvested from the nursery field individually to preserve their identity. The resulting homozygous positive and null T2 seeds were analyzed by near-infrared spectroscopy (NIRS) to confirm changes in oil quantity and fatty acid profile across events, which is indicative of DGAT expression. Seed for the GmDGAT1b-3aa construct expressed with the β-conglycinin promoter was produced by the same process, except that the T2 seed was produced in the greenhouse. The T2 plants were grown in the field in replicated 3-m rows to generate T3 seeds. Overexpression of the DGAT1 protein was confirmed by western blot for the two GmDGAT1b-3aa events advanced to multiple location field trials (Supplementary Fig. S8).

Seed analysis by NIRS

T2 and T3 seeds were analyzed nondestructively by NIRS, as previously described (Roesler et al. 2016). Briefly, bulk T2 seeds from individual plants, approximately 20 g, were analyzed on a Tango Fourier-transform near-infrared (FT-NIR) spectrometer (Bruker, Billerica, Massachusetts) fitted with a robotic sampler (HTS-250; Pike Technologies, Fitchburg, Wisconsin) using a 50-mm spinning cup attachment and cells. T3 seed analysis was performed on bulk samples of 400–500 g of whole seed by near-infrared transmission (NIT) spectroscopy using an Infratec™ 1241 grain analyzer (Foss Analytical, Hillerød, Denmark). Captured spectra, from both platforms, were converted to predictions of moisture content, oil content, protein content and oleic acid content using Corteva™ derived models. Details on the model performance statistics are provided in Supplementary Table S5. The reference chemistry methods used for model development and validation for moisture, oil and protein were based on American Oil Chemists’ Society (AOCS) official methods in a laboratory certified by the AOCS for soybean reference chemistry. The reference chemistry used for the fatty acid calibrations was developed by gas chromatographic analysis of fatty acid methyl esters of oil extracts derived from soybean samples, after spectral capture. Reference chemistry was performed in a laboratory carrying Canadian Grain Commission Certification for fatty acid analysis. All compositional values are reported on a weight percent (wt%), at 13% moisture, basis. Fatty acid values are expressed on a relative percentage basis.

Multiple location field trials

The field trials experiment was conducted in 2022 and planted between May 11 and May 31 at nine locations (Supplementary Table S6). Locations were selected within the parental maturity group (3.2) and cover various longitudes across the midwestern USA, representing the main soybean growing region. The York, Nebraska site, was lost to a hailstorm on June 14, reducing the experiment to eight locations. Plots consisted of two 5.3-mr rows seeded with 120 or 160 seeds per row, depending on the location. Plots at each location were treated following standard agricultural practice for the region, with tillage and preemergent herbicide before planting. The fields were not irrigated. During the season, plots were actively monitored for weeds and insect pressure and sprayed as needed to maximize yield. At physiological maturity, the two rows were harvested with a plot combined for yield. A 0.45-kg seed sample per plot was retained at six of the eight locations for composition analysis. The six locations chosen for compositional analysis were selected based on geographical diversity. The calculated kilograms of oil and protein per hectare were estimated using the percent change in oil, protein and yield determined in the multiple location field trials applied to reported values for 2020 US commodity soybean (Naeve and Miller-Garvin 2020, USDA 2020).

Statistical analysis

The indicated analysis for N. benthamiana and single location seed composition data was performed in JMP Pro 16 (SAS Institute, Cary, North Carolina). Multiple location yield and composition were analyzed using Corteva proprietary software with a fixed effects statistical model to perform best linear unbiased prediction (BLUP) (Piepho et al. 2008). For all comparisons, P values < 0.05 were considered statistically significant.

Protein sequence and structural analysis

A multiple sequence alignment was generated with Clustal Omega (Madeira et al. 2022). AlphaFold version 2.3.1 was used to predict the structures of both WT GmDGAT and GmDGAT1b-3aa as dimers, with the following parameters: model_preset: multimer, db_preset: full_dbs, benchmark: false, use_precomputed_msas: true, run_relax: false and use_gpu_relax: false. Folded mutant protein structures were then visualized using PyMOL version 2.5.2. Amino acid substitutions relative to WT were manually color coded, and TM-align version 20220412 was then utilized to align the predicted WT and mutant structures to determine percent structural identity.

Supplementary Material

pcad148_Supp
pcad148_supp.zip (992KB, zip)

Acknowledgments

We would like to thank the plant transformation and controlled environments groups for generating the GmDGAT1b transgenic events, our compositional analysis and transgenic pipeline teams, and the field sciences team for planting, harvesting and managing the field trials. We thank Knut Meyer for the calculations of protein and oil per hectare, Masha Fedorova for providing advice on the discussion section and Ross Allen for developing and maintaining the NIRS models used throughout this study.

Contributor Information

Kayla S Flyckt, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Keith Roesler, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Kristin Haug Collet, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Luciano Jaureguy, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Russ Booth, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Shawn R Thatcher, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

John D Everard, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Kevin G Ripp, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Zhan-Bin Liu, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Bo Shen, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Laura L Wayne, Corteva Agriscience, 7300 NW 62nd Avenue, Johnston 50131, USA.

Supplementary Data

Supplementary Data are available at PCP online.

Data Availability

The data underlying this article are available in the article and in its online supplementary material.

Funding

Corteva Agriscience.

Author Contributions

B.S. and K.R. designed the study with input from all authors. K.H.C. designed the gene variants with input from K.R. K.H.C. and K.S.F. generated the model systems data. R.B., K.G.R. and Z.-B.L. facilitated transgenic plant generation. L.J. managed the field trials and analyzed the data. J.D.E. facilitated the compositional analysis. S.R.T. generated the protein structure. K.S.F. wrote the article. L.L.W. oversaw research and revised the manuscript with help from all authors. All authors read and contributed to the final article before submission.

Disclosures

All authors are employees of Corteva Agriscience.

References

  1. Al-Amery  M., Downie  B., DeBolt  S., Crocker  M., Urschel  K., Goff  B., et al. (2019) Proximate composition of enhanced DGAT high oil, high protein soybeans. Biocatal. Agric. Biotechnol.  21: 101303. [Google Scholar]
  2. Anderson  E.J., Ali  M.L., Beavis  W.D., Chen  P., Clemente  T.E., Diers  B.W., et al. (2019) Soybean [Glycine max (L.) Merr.] breeding: history, improvement, production and future opportunities. In  Advances in Plant Breeding Strategies: Legumes: Volume 7. Edited by Al-Khayri, J.M., Jain, S.M. and Johnson, D.V. pp. 431–516. Springer International Publishing, Cham. [Google Scholar]
  3. Assefa  Y., Purcell  L.C., Salmeron  M., Naeve  S., Casteel  S.N., Kovács  P., et al. (2019) Assessing variation in US soybean seed composition (protein and oil). Front. Plant Sci.  10: 298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cagliari  A., Margis  R., Dos Santos Maraschin  F., Turchetto-Zolet  A.C., Loss  G. and Margis-Pinheiro  M. (2011) Biosynthesis of triacylglycerols (TAGs) in plants and algae. Int. J. Plant Biol.  2: 40–52. [Google Scholar]
  5. Caldo  K.M.P., Acedo  J.Z., Panigrahi  R., Vederas  J.C., Weselake  R.J. and Lemieux  M.J. (2017) Diacylglycerol acyltransferase 1 is regulated by its N-terminal domain in response to allosteric effectors. Plant Physiol.  175: 667–680. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cases  S., Smith  S.J., Zheng  Y.W., Myers  H.M., Lear  S.R., Sande  E., et al. (1998) Identification of a gene encoding an acyl CoA:diacylglycerol acyltransferase, a key enzyme in triacylglycerol synthesis. Proc. Natl Acad. Sci. U.S.A.  95: 13018–13023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chen  G., Harwood  J.L., Lemieux  M.J., Stone  S.J. and Weselake  R.J. (2022) Acyl-CoA:diacylglycerol acyltransferase: properties, physiological roles, metabolic engineering and intentional control. Prog. Lipid Res.  88: 101181. [DOI] [PubMed] [Google Scholar]
  8. Chen  B., Wang  J., Zhang  G., Liu  J., Manan  S., Hu  H., et al. (2016) Two types of soybean diacylglycerol acyltransferases are differentially involved in triacylglycerol biosynthesis and response to environmental stresses and hormones. Sci. Rep.  6: 28541. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chen  G., Xu  Y., Siloto  R.M.P., Caldo  K.M.P., Vanhercke  T., Tahchy  A.E., et al. (2017) High-performance variants of plant diacylglycerol acyltransferase 1 generated by directed evolution provide insights into structure function. Plant J.  92: 167–177. [DOI] [PubMed] [Google Scholar]
  10. Cho  H.J., Moy  Y., Rudnick  N.A., Klein  T.M., Yin  J., Bolar  J., et al. (2022) Development of an efficient marker-free soybean transformation method using the novel bacterium Ochrobactrum haywardense H1. Plant Biotechnol. J.  20: 977–990. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Clemente  T.E. and Cahoon  E.B. (2009) Soybean oil: genetic approaches for modification of functionality and total content. Plant Physiol.  151: 1030–1040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Coupland  C.E., Ansell  T.B., Sansom  M.S.P. and Siebold  C. (2023) Rocking the MBOAT: structural insights into the membrane bound O-acyltransferase family. Curr. Opin. Struct. Biol.  80: 102589. [DOI] [PubMed] [Google Scholar]
  13. Das  A., Davis  M.A. and Rudel  L.L. (2008) Identification of putative active site residues of ACAT enzymes. J. Lipid. Res.  49: 1770–1781. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Giaever  G. and Nislow  C. (2014) The yeast deletion collection: a decade of functional genomics. Genetics  197: 451–465. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Greer  M.S., Truksa  M., Deng  W., Lung  S.-C., Chen  G. and Weselake  R.J. (2015) Engineering increased triacylglycerol accumulation in Saccharomyces cerevisiae using a modified type 1 plant diacylglycerol acyltransferase. Appl. Microbiol. Biotechnol.  99: 2243–2253. [DOI] [PubMed] [Google Scholar]
  16. Guo  Z., Cromley  D., Billheimer  J.T. and Sturley  S.L. (2001) Identification of potential substrate-binding sites in yeast and human acyl-CoA sterol acyltransferases by mutagenesis of conserved sequences. J. Lipid. Res.  42: 1282–1291. [PubMed] [Google Scholar]
  17. Guo  X., Fan  C., Chen  Y., Wang  J., Yin  W., Wang  R.R., et al. (2017) Identification and characterization of an efficient acyl-CoA: diacylglycerol acyltransferase 1 (DGAT1) gene from the microalga Chlorella ellipsoidea. BMC Plant Biol.  17: 48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hatanaka  T., Serson  W., Li  R., Armstrong  P., Yu  K., Pfeiffer  T., et al. (2016) A Vernonia diacylglycerol acyltransferase can increase renewable oil production. J. Agric. Food Chem.  64: 7188–7194. [DOI] [PubMed] [Google Scholar]
  19. International Energy Agency (2022) Renewables 2022. Edited by Publications, I. p. 142.
  20. Jako  C., Kumar  A., Wei  Y., Zou  J., Barton  D.L., Giblin  E.M., et al. (2001) Seed-specific over-expression of an Arabidopsis cDNA encoding a diacylglycerol acyltransferase enhances seed oil content and seed weight. Plant Physiol.  126: 861–874. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Jumper  J., Evans  R., Pritzel  A., Green  T., Figurnov  M., Ronneberger  O., et al. (2021) Highly accurate protein structure prediction with AlphaFold. Nature  596: 583–589. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kambhampati  S., Aznar-Moreno  J.A., Hostetler  C., Caso  T., Bailey  S.R., Hubbard  A.H., et al. (2020) On the inverse correlation of protein and oil: examining the effects of altered central carbon metabolism on seed composition using soybean fast neutron mutants. Metabolites  10: 18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kim  H., Park  J.H., Kim  D.J., Kim  A.Y. and Suh  M.C. (2016) Functional analysis of diacylglycerol acyltransferase1 genes from Camelina sativa and effects of CsDGAT1B overexpression on seed mass and storage oil content in C. sativa. Plant Biotechnol. Rep.  10: 141–153. [Google Scholar]
  24. Lager  I., Jeppson  S., Gippert  A.-L., Feussner  I., Stymne  S. and Marmon  S. (2020) Acyltransferases regulate oil quality in Camelina sativa through both acyl donor and acyl acceptor specificities. Front. Plant Sci.  11: 1144. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lardizabal  K., Effertz  R., Levering  C., Mai  J., Pedroso  M.C., Jury  T., et al. (2008) Expression of Umbelopsis ramanniana DGAT2A in seed increases oil in soybean. Plant Physiol.  148: 89–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Li  R., Hatanaka  T., Yu  K., Wu  Y., Fukushige  H. and Hildebrand  D. (2013) Soybean oil biosynthesis: role of diacylglycerol acyltransferases. Funct. Integr. Genomics.  13: 99–113. [DOI] [PubMed] [Google Scholar]
  27. Liu  Q., Siloto  R.M.P., Lehner  R., Stone  S.J. and Weselake  R.J. (2012) Acyl-CoA:diacylglycerol acyltransferase: molecular biology, biochemistry and biotechnology. Prog. Lipid Res.  51: 350–377. [DOI] [PubMed] [Google Scholar]
  28. Lopes  J.L.S., Nobre  T.M., Cilli  E.M., Beltramini  L.M., Araújo  A.P.U. and Wallace  B.A. (2014) Deconstructing the DGAT1 enzyme: binding sites and substrate interactions. Biochim. Biophys. Acta - Biomembr.  1838: 3145–3152. [DOI] [PubMed] [Google Scholar]
  29. Lowe  K., Wu  E., Wang  N., Hoerster  G., Hastings  C., Cho  M.-J., et al. (2016) Morphogenic regulators Baby boom and Wuschel improve monocot transformation. Plant Cell  28: 1998–2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lung  S.C. and Weselake  R.J. (2006) Diacylglycerol acyltransferase: a key mediator of plant triacylglycerol synthesis. Lipids  41: 1073–1088. [DOI] [PubMed] [Google Scholar]
  31. Madeira  F., Pearce  M., Tivey  A.R.N., Basutkar  P., Lee  J., Edbali  O., et al. (2022) Search and sequence analysis tools services from EMBL-EBI in 2022. Nucleic Acids Res.  50: W276–W279. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Maravi  D.K., Kumar  S., Sharma  P.K., Kobayashi  Y., Goud  V.V., Sakurai  N., et al. (2016) Ectopic expression of AtDGAT1, encoding diacylglycerol O-acyltransferase exclusively committed to TAG biosynthesis, enhances oil accumulation in seeds and leaves of Jatropha. Biotechnol. Biofuels  9: 226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. McFie  P.J., Stone  S.L., Banman  S.L. and Stone  S.J. (2010) Topological orientation of acyl-CoA:diacylglycerol acyltransferase-1 (DGAT1) and identification of a putative active site histidine and the role of the n terminus in dimer/tetramer formation. J. Biol. Chem.  285: 37377–37387. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Naeve  S.L. and Miller-Garvin  J. (2020) United States soybean quality 2020 Annual Report. https://ussec.org/wp-content/uploads/2021/06/2020-Quality-Report_US-Food-Soybean.pdf.
  35. Nakamura  Y., Tsuchiya  M. and Ohta  H. (2007) Plastidic phosphatidic acid phosphatases identified in a distinct subfamily of lipid phosphate phosphatases with prokaryotic origin. J. Biol. Chem.  282: 29013–29021. [DOI] [PubMed] [Google Scholar]
  36. Nikolau  B.J., Ohlrogge  J.B. and Wurtele  E.S. (2003) Plant biotin-containing carboxylases. Arch Biochem. Biophys.  414: 211–222. [DOI] [PubMed] [Google Scholar]
  37. Piepho  H.P., Möhring  J., Melchinger  A.E. and Büchse  A. (2008) BLUP for phenotypic selection in plant breeding and variety testing. Euphytica  161: 209–228. [Google Scholar]
  38. Podevin  N., Davies  H.V., Hartung  F., Nogué  F. and Casacuberta  J.M. (2013) Site-directed nucleases: a paradigm shift in predictable, knowledge-based plant breeding. Trends Biotechnol.  31: 375–383. [DOI] [PubMed] [Google Scholar]
  39. Roesler  K., Shen  B., Bermudez  E., Li  C., Hunt  J., Damude  H.G., et al. (2016) An improved variant of soybean type 1 diacylglycerol acyltransferase increases the oil content and decreases the soluble carbohydrate content of soybeans. Plant Physiol.  171: 878–893. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Rotundo  J.L., Borrás  L., Westgate  M.E. and Orf  J.H. (2009) Relationship between assimilate supply per seed during seed filling and soybean seed composition. Field Crops Res.  112: 90–96. [Google Scholar]
  41. Shi  J., Gao  H., Wang  H., Lafitte  H.R., Archibald  R.L., Yang  M., et al. (2017) ARGOS8 variants generated by CRISPR-Cas9 improve maize grain yield under field drought stress conditions. Plant Biotechnol. J.  15: 207–216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Sui  X., Wang  K., Gluchowski  N.L., Elliott  S.D., Liao  M., Walther  T.C., et al. (2020) Structure and catalytic mechanism of a human triacylglycerol-synthesis enzyme. Nature  581: 323–328. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Svitashev  S., Schwartz  C., Lenderts  B., Young  J.K. and Mark Cigan  A. (2016) Genome editing in maize directed by CRISPR–Cas9 ribonucleoprotein complexes. Nat. Commun.  7: 13274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Thrane  M., Paulsen  P.V., Orcutt  M.W., Krieger  T.M. (2017) Chapter 2 – soy protein: impacts, production, and applications. In  Sustainable Protein Sources. Edited by Nadathur, S.R., Wanasundara, J.P.D. and Scanlin, L. pp. 23–45. Academic Press, San Diego. [Google Scholar]
  45. USDA (2020) Crop Production 2020 Summary. https://downloads.usda.library.cornell.edu/usda-esmis/files/k3569432s/w3764081j/5712n018r/cropan21.pdf.
  46. Varadi  M., Anyango  S., Deshpande  M., Nair  S., Natassia  C., Yordanova  G., et al. (2021) AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res.  50: D439–D444. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Wang  Z., Huang  W., Chang  J., Sebastian  A., Li  Y., Li  H., et al. (2014) Overexpression of SiDGAT1, a gene encoding acyl-CoA:diacylglycerol acyltransferase from Sesamum indicum L. increases oil content in transgenic Arabidopsis and soybean. Plant Cell, Tissue Organ Cult.  119: 399–410. [Google Scholar]
  48. Wang  L., Qian  H., Nian  Y., Han  Y., Ren  Z., Zhang  H., et al. (2020) Structure and mechanism of human diacylglycerol O-acyltransferase 1. Nature  581: 329–332. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Wang  Z., Yang  M., Sun  Y., Yang  Q., Wei  L., Shao  Y., et al. (2019) Overexpressing Sesamum indicum L.’s DGAT1 increases the seed oil content of transgenic soybean. Mol. Breed.  39: 101. [Google Scholar]
  50. Weselake  R.J., Madhavji  M., Szarka  S.J., Patterson  N.A., Wiehler  W.B., Nykiforuk  C.L., et al. (2006) Acyl-CoA-binding and self-associating properties of a recombinant 13.3 kDa N-terminal fragment of diacylglycerol acyltransferase-1 from oilseed rape. BMC Biochem.  7: 24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Weselake  R.J., Shah  S., Tang  M., Quant  P.A., Snyder  C.L., Furukawa-Stoffer  T.L., et al. (2008) Metabolic control analysis is helpful for informed genetic manipulation of oilseed rape (Brassica napus) to increase seed oil content. J. Exp. Bot.  59: 3543–3549. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Wu  P., Xu  X., Li  J., Zhang  J., Chang  S., Yang  X., et al. (2021) Seed-specific overexpression of cotton GhDGAT1 gene leads to increased oil accumulation in cottonseed. Crop J.  9: 487–490. [Google Scholar]
  53. Xue  J., Gao  H., Xue  Y., Shi  R., Liu  M., Han  L., et al. (2022) Functional characterization of soybean diacylglycerol acyltransferase 3 in yeast and soybean. Front. Plant Sci.  13: 854103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Xu  J., Francis  T., Mietkiewska  E., Giblin  E.M., Barton  D.L., Zhang  Y., et al. (2008) Cloning and characterization of an acyl-CoA-dependent diacylglycerol acyltransferase 1 (DGAT1) gene from Tropaeolum majus, and a study of the functional motifs of the DGAT protein using site-directed mutagenesis to modify enzyme activity and oil content. Plant Biotechnol. J.  6: 799–818. [DOI] [PubMed] [Google Scholar]
  55. Zhang  M., Fan  J., Taylor  D.C. and Ohlrogge  J.B. (2009) DGAT1 and PDAT1 acyltransferases have overlapping functions in Arabidopsis triacylglycerol biosynthesis and are essential for normal pollen and seed development. Plant Cell  21: 3885–3901. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Zhang  Y., Pan  Y., Ding  W., Hu  H. and Liu  J. (2021) Lipid production is more than doubled by manipulating a diacylglycerol acyltransferase in algae. GCB Bioenergy  13: 185–200. [Google Scholar]
  57. Zhao  C., Luo  M.-T., Huang  C., Chen  X.-F., Xiong  L., Li  H.-L., et al. (2019) Determining intracellular lipid content of different oleaginous yeasts by one simple and accurate Nile Red fluorescent method. Prep. Biochem. Biotechnol.  49: 597–605. [DOI] [PubMed] [Google Scholar]
  58. Zheng  P., Allen  W.B., Roesler  K., Williams  M.E., Zhang  S., Li  J., et al. (2008) A phenylalanine in DGAT is a key determinant of oil content and composition in maize. Nat. Genet.  40: 367–372. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

pcad148_Supp
pcad148_supp.zip (992KB, zip)

Data Availability Statement

The data underlying this article are available in the article and in its online supplementary material.

Articles from Plant and Cell Physiology are provided here courtesy of Oxford University Press

RESOURCES