Altered Lipid Composition and Enhanced Nutritional Value ofArabidopsis Leaves following Introduction of an AlgalDiacylglycerol Acyltransferase 2C W

Sanjaya,a,b,1 Rachel Miller,c,d,1 Timothy P. Durrett,b,e,2 Dylan K. Kosma,e Todd A. Lydic,f Bagyalakshmi Muthan,a

Abraham J.K. Koo,a,d,3 Yury V. Bukhman,g Gavin E. Reid,a,f Gregg A. Howe,a,d John Ohlrogge,b,e

and Christoph Benninga,b,4

a Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, Michigan 48824bGreat Lakes Bioenergy Research Center, Michigan State University, East Lansing, Michigan 48824cCell and Molecular Biology Program, Michigan State University, East Lansing, Michigan 48824dDepartment of Energy–Plant Research Laboratory, Michigan State University, East Lansing, Michigan 48824eDepartment of Plant Biology, Michigan State University, East Lansing, Michigan 48824f Department of Chemistry, Michigan State University, East Lansing, Michigan 48824gGreat Lakes Bioenergy Research Center, University of Wisconsin, Madison, Michigan 53706

Enhancement of acyl-CoA–dependent triacylglycerol (TAG) synthesis in vegetative tissues is widely discussed as a potentialavenue to increase the energy density of crops. Here, we report the identification and characterization of Chlamydomonasreinhardtii diacylglycerol acyltransferase type two (DGTT) enzymes and use DGTT2 to alter acyl carbon partitioning in plantvegetative tissues. This enzyme can accept a broad range of acyl-CoA substrates, allowing us to interrogate different acylpools in transgenic plants. Expression of DGTT2 in Arabidopsis thaliana increased leaf TAG content, with some molecularspecies containing very-long-chain fatty acids. The acyl compositions of sphingolipids and surface waxes were altered, andcutin was decreased. The increased carbon partitioning into TAGs in the leaves of DGTT2-expressing lines had little effect ontranscripts of the sphingolipid/wax/cutin pathway, suggesting that the supply of acyl groups for the assembly of these lipids isnot transcriptionally adjusted. Caterpillars of the generalist herbivore Spodoptera exigua reared on transgenic plants gainedmore weight. Thus, the nutritional value and/or energy density of the transgenic lines was increased by ectopic expression ofDGTT2 and acyl groups were diverted from different pools into TAGs, demonstrating the interconnectivity of acyl metabolismin leaves.

INTRODUCTION

In angiosperms, deposition of triacylglycerols (TAGs) in lipiddroplets is most commonly associated with seed tissues. TheseTAGs represent a reduced form of carbon with high energy con-tent essential for seed germination and seedling establishment. Inaddition, many microalgae accumulate considerable amounts ofTAGs under stress conditions such as nitrogen deficiency, highlight, or high salt. The utility of this storage lipid has been intenselyexplored in both organism groups as a source of food and nu-traceuticals and as a feedstock for the petrochemical industry or

biodiesel fuel production (Durrett et al., 2008; Halim et al., 2012).The mechanisms of TAG synthesis and its regulation have beenmostly characterized in developing seed tissues. However, themolecular regulation and synthesis of TAGs in nonseed or vege-tative cells of plants and microalgae is still largely uncharacterizedbut is thought to involve different regulatory mechanisms thanthose observed for developing seeds (Fan et al., 2011; Chapmanand Ohlrogge, 2012).In plants and microalgae, the plastid and the endoplasmic

reticulum (ER) interact in the de novo formation of TAG, as thesynthesis of fatty acids (FAs) occurs in the plastid, followed by theexport of FAs and their assembly into TAGs at the ER. In bothorganism groups, the basic enzymatic mechanisms involved aresimilar for membrane glycerolipid and TAG synthesis. The carbonflux from the FA synthase in the plastid through the cytosolic acyl-CoA pool and the substrate specificities of the enzymes involved,especially that of the acyltransferases, likely affect the acylcomposition of the DAG intermediates and ultimately the TAGsformed at the ER. Thus, the manipulation of carbon fluxes be-tween acyl-CoA, TAG, and other acyl lipid pools are expected toprovide insight into the lipid metabolic network and its regulationin vegetative tissues.Different classes of diacylglycerol acyltransferases (DGATs) play

distinct roles in determining the quality and quantity of acyl-CoA

1 These authors contributed equally to this work.2 Current address: Department of Biochemistry, Kansas State University,Manhattan, KS 66506.3 Current address: Department of Biochemistry, University of Missouri,Columbia, MO 65211.4 Address correspondence to benning@cns.msu.edu.The author responsible for distribution of materials integral to the findingspresented in this article in accordance with the policy described inthe Instructions for Authors (www.plantcell.org) is: Christoph Benning(benning@cns.msu.edu).C Some figures in this article are displayed in color online but in black andwhite in the print edition.W Online version contains Web-only data.www.plantcell.org/cgi/doi/10.1105/tpc.112.104752

The Plant Cell, Vol. 25: 677–693, February 2013, www.plantcell.org ã 2013 American Society of Plant Biologists. All rights reserved.

flux into TAG synthesis. In eukaryotes, several classes of DGATshave been identified based on structure and activity. The two mostcommon are type 1 (DGAT1) and type 2 (DGAT2), which have bothbeen shown to be involved in the synthesis of TAG (Yen et al.,2008). DGAT1 proteins are predicted to possess six or moretransmembrane domains and belong to the membrane-boundO-acyltransferase family (Cases et al., 1998). Forward and reversegenetic approaches have confirmed that the Arabidopsis thalianaDGAT1 gene affects the TAG levels in seeds (Routaboul et al.,1999; Zou et al., 1999; Jako et al., 2001), and DGAT1 genes havebeen studied in detail in several species (Lung and Weselake,2006; Shockey et al., 2006; Banilas et al., 2011). Structurally,DGAT2 proteins differ from DGAT1 and possess only two to threepredicted transmembrane domains (Yen et al., 2008). Over-expression of fungal Umbelopsis ramannian DGAT2 increasedseed oil content in soybean (Glycine max) and maize (Zea mays;Lardizabal et al., 2008; Oakes et al., 2011). A third type of DGATincludes the bifunctional enzyme ADP1 from Acinetobacter cal-coaceticus, exhibiting both wax ester synthase (WS) and DGATactivity (Kalscheuer and Steinbüchel, 2003). ADP1 homologs havebeen identified in Arabidopsis and petunia (Petunia hybrida) andhave been shown to have WS and DGAT activity at different levels(King et al., 2007; Li et al., 2008). A fourth type, a soluble cytosolicDGAT enzyme (DGAT3), was identified in peanut (Arachis hypo-gaea), with homologs in Arabidopsis (Saha et al., 2006; Hernándezet al., 2012).

Like land plants, microalgae also possess DGAT-mediated TAGsynthesis. For instance, in the diatom Phaeodactylum tricornu-tum, a cDNA encoding a DGAT1-like protein, was isolated andshown to incorporate saturated FAs into TAGs (Guihéneuf et al.,2011). In Ostreococcus tauri, a gene encoding DGAT2 has beenidentified and the enzyme has been shown to possess broadsubstrate specificity (Wagner et al., 2010). Previously, we identi-fied five different genes encoding DGAT2 in Chlamydomonasreinhardtii and named them DGTT1 through DGTT5 (Miller et al.,2010). Recent studies have shown that of these five enzymes,DGTT1 and DGTT3 are active in TAG synthesis following nitrogendeprivation, but DGTT2, DGTT4, and DGTT5 are not (La Russaet al., 2012). Artificial microRNA silencing and overexpression ofDGTT2 in C. reinhardtii did not influence TAG levels despite al-tered transcript abundance (Deng et al., 2012). Recently, a genefor a DGAT1-type enzyme was identified in C. reinhardtii aftertranscript-based correction of gene models (Boyle et al., 2012).Although the different types of DGATs differ structurally, whetherthey have redundant or specific functions in TAG synthesis in C.reinhardtii is not clear. The distinct substrate specificities of thevarious DGAT enzymes may determine the FA composition ofTAG in microalgae and plants.

Here, we characterize DGTT2 of C. reinhardtii and use its ectopicproduction as a tool to affect carbon flux through the acyl-CoA poolin Arabidopsis, leading to the formation of TAGs in vegetativetissues. We focused on DGTT2 because it was the most activeisoform of C. reinhardtii in our hands. We also present in vitro dataon the acyl-CoA substrate specificity of DGTT2. Most notably, inDGTT2 transgenic plants, we observed a redirection of very-long-chain fatty acids (VLCFAs) from sphingolipids into TAGs. Theimplications of these results for plant lipid metabolism in vegetativetissues and for biotechnological applications are discussed.

RESULTS

Multiple Putative C. reinhardtii DGAT2 Isoforms

Amino acid sequence similarity to Arabidopsis DGAT2 was usedto identify potential DGAT orthologs, using BLAST against version3 of the C. reinhardtii genome. The BLAST search results returnedfive gene models with high sequence similarity to the Arabidopsistype 2 DGAT. These five candidates were named DiacylglycerolAcyltransferase Type Two (DGTT1 to DGTT5) and compared withtype 2 DGATs from Arabidopsis, O. tauri, yeast (Saccharomycescerevisiae), castor bean (Ricinus communis), and tung tree(Vernicia fordii). When analyzed with MEGA5 (Tamura et al., 2011),DGTT1 was the most closely related to Sc-DGA1. DGTT2, DGTT3,and DGTT5 formed their own clade, as did the three land plantDGATs. Ot-DGAT2B and DGTT4 were both more similar to the landplant DGATs than to the other C. reinhardtii DGTTs (Figure 1A; seeSupplemental Data Set 1 online).Structural analysis of the gene model–translated protein

sequences was performed in silico using the TMHMM server2.0 (http://www.cbs.dtu.dk/services/TMHMM/) to predict trans-membrane sequences. All five candidates had one to threetransmembrane domains in the N-terminal half of the protein,consistent with other type 2 DGATs. SignalP V3.0 (http://www.cbs.dtu.dk/services/SignalP/) and TargetP V1.1 (http://www.cbs.dtu.dk/services/TargetP/) failed to identify potential cellular localizationsignals. Searching against the National Center for BiotechnologyInformation Conserved Domain Database revealed a DAGAT do-main in the C-terminal half of all five candidates, although DGTT5has an apparent disruption in its domain (Figure 1B).

Expression of DGTT Constructs in Yeast

To test the function of the predicted DGTT genes, the coding se-quences of DGTT2-5 were isolated and expressed in yeast strainH1266, a triple knockout mutant for DiacylGlycerol Acyltransferase1(DGA1), Lecithin cholesterol acyl transferase Related Open readingframe1 (LRO1), and Acyl-coenzyme A: cholesterol acyl transferase-Related Enzyme2 (ARE2) (Sandager et al., 2002). This strain hasvery little native DGAT activity, providing a suitable backgroundfor testing the C. reinhardtii DGAT2 candidates. We were unableto isolate a DGTT1 cDNA based on the available gene model. Weconfirmed the presence of the DGTT2-5 proteins in yeast by im-munoblotting (see Supplemental Figure 1 online). To quantify theamount of TAG produced, electrospray ionization–mass spec-trometry (ESI-MS) was performed on lipid extracts from the trans-genic yeast (Figure 1C). Both DGTT2- and DGTT3-producing yeastaccumulated considerable levels of TAG (13.4 and 13.9% per dryweight [DW], respectively) compared with the empty vector control(0.5% per DW). Both DGTT4- and DGTT5-producing strains showedTAG levels equivalent to the empty vector (0.6 and 0.5% per DW,respectively), indicating that these may not have DGAT activity orlack appropriate cofactors or substrates in yeast necessary forproper DGAT activity.

In Vitro DGAT Activity of Recombinant Proteins

The substrate preference of the DGTT proteins was estimatedin an in vitro reaction with isolated microsomes. DGTT-containing

678 The Plant Cell

microsomes were incubated with radiolabeled palmitoyl-CoA(16:0) and oleoyl-CoA (18:1), along with dioleoyl DAG. The level ofDGAT activity was estimated from their incorporation of radiolabelinto isolated TAGs. Comparison of the results using the two dif-ferent substrate acyl-CoAs suggested an apparent difference insubstrate specificity, with DGTT2 and DGTT3 preferring 16:0-CoAand DGTT4 18:1-CoA (Figures 2A and 2B). These assays alsoconfirmed that DGTT5 has essentially no activity under these as-say conditions. This result, in conjunction with the apparent lack ofexpression of its gene in C. reinhardtii and disruption in the DAGATdomain, suggests that DGTT5 is a nonactive pseudogene.Although DGTT2 and DGTT3 belong to the same clade, following

nitrogen deprivation the level of DGTT2 transcripts remained un-changed, whereas DGTT3 transcript levels increased (Miller et al.,2010; Boyle et al., 2012). When expressed in yeast, DGTT2 activityin vitro was higher than that of the other isoforms. Based on theobserved activity levels, we focused on the characterization of thisenzyme as a potential tool for the manipulation of cytosolic acyl-CoAs in vegetative tissues of plants, the primary goal of this study.A microsome-based competition assay was used to further probethe substrate specificity of DGTT2. When incubated with equimolaramounts of 14C-labeled and unlabeled 16:0-CoA, the radioactivityin the TAG band was reduced to ;50% compared with thereactions with only 14C-labeled 16:0-CoA (Figure 2C). When in-cubated with equal moles of unlabeled 18:1-CoA, the decreasewas only ;40%. However, when incubated with equal moles ofunlabeled 22:1-CoA, the decrease was ;75%. Together, thesedata suggested that DGTT2 is able to incorporate varying acyl-CoAspecies into TAG, perhaps preferring very-long-chain acyl groups,which will become relevant for the interpretation of lipid data for thetransgenic plants described below.

Production of DGTT2 in Arabidopsis AffectsSeedling Growth

To explore DGTT2 as a tool to manipulate acyl-CoA pools in plantsand to engineer TAGs in vegetative tissues, we expressed the full-length DGTT2 coding sequence in Arabidopsis under the controlof the constitutive 35S cauliflower mosaic virus promoter, 35S:DGTT2. Five homozygous lines, 14, 22, 46, 52, and 57, werecarried forward to the T4 generation and used for detailed analysis.The hypocotyls of transgenic 10- to 12-d-old seedlings grown

on Murashige and Skoog agar plates supplemented with 1% Sucwere strikingly more elongated than those of control plants andslightly pale in color. Consistent with the pale-green phenotype,the total chlorophyll content in the transgenic lines was 28 to 30%lower. On soil, all transgenic lines followed wild-type growth anddevelopment patterns. To determine the abundance of DGTT2mRNA in the seedlings of overexpressors (15 d old), we usedquantitative RT-PCR. The expression of DGTT2 relative to ACTIN2ranged from 2 to 7 (ratio of DGTT2/ACTIN2) in the independenttransgenic lines tested (see Supplemental Figures 2A to 2D online),while no transcripts were detected in the wild type.

DGTT2 Production Causes Accumulation of TAGswith VLCFAs

To determine the effect of DGTT2 production on the accumulationof TAG, we analyzed 15-d-old whole seedlings by ESI-MS. The

Figure 1. Identification of C. reinhardtii Type 2 DGATs and Expression ofDGTT Constructs in Yeast.

(A) Phylogenetic tree of C. reinhardtii DGTT1-5 compared with DGAT2s fromother species: Arabidopsis (NP_566952.1), O. tauri (XP_003083539.1),S. cerevisiae (NP_014888.1), castor bean (DQ923084.1), and tung tree(DQ356682.1).(B) Predicted structure of DGTT1-5. Dark-gray boxes indicate predictedtransmembrane domains (TM), and light-gray boxes indicate the con-served enzymatic domain (DAGAT).(C) ESI-MS quantification of TAG levels extracted from transformedyeast. The total amount of TAG was normalized based on the DW of theyeast. Tritridecanoin (tri13:0) and tripentadecanoin (tri15:0) TAGs wereadded as internal standards (n = 4; average 6 SD). H1266 indicates theempty vector control.

Enhancing Nutritional Content of Leaves 679

TAG levels in 15-d-old seedlings of transgenic lines were in-creased up to 22- and 25-fold (Figure 3A). Lines 22 and 57contained 0.88 and 1.00% TAG per DW, respectively, comparedwith 0.04% TAG per DW in the wild type. The increase in TAGlevels correlated well with the relative abundance of DGTT2transcript in each line (see Supplemental Figure 2D online).Comparison of ESI-MS spectra of neutral lipid extracts from the

wild type and line 57 indicated an abundance of TAG molecularspecies containing VLCFAs in the latter (Figures 3B and 3C). Thepresence of long and VLCFA-containing TAG molecular species inthe seedling extracts of line 57 was further confirmed by electro-spray ionization–tandem mass spectrometry (ESI-MS/MS) (Figure3D), which produced product ions with masses consistent with theloss of VLCFAs. These spectra were consistent with each TAGmolecule containing only one VLCFA.We also used ESI-MS to analyze the TAG content in the leaves of

6-week-old soil-grown plants before bolting. As shown in Figure3A, TAG levels in transgenic lines 22 and 57 were 0.23 and 0.36%of DW, an 11- and 18-fold increase relative to wild-type plants(0.02% TAG per DW), respectively. These increases were con-siderably lower compared with those observed with young seed-lings and might reflect turnover of TAG in more mature leaves. TheESI-MS spectrum of neutral lipid extracts from 6-week-old leavesof line 57 confirmed the presence of TAG molecular speciescontaining VLCFAs, which were not detected in the wild type (seeSupplemental Figures 3A and 3B online). The presence of long andVLCFAs in neutral lipid extracts of soil-grown line 57 was alsoconfirmed by ESI-MS/MS (see Supplemental Figure 3C online).Interestingly, the level of TAGs with VLCFAs (e.g., C24:0) in leavesof soil-grown plants of line 57 was considerably higher (;8 to 10mol %) than the wild type (;2 mol %) (see Supplemental Figure 4online). Using an independent method, we also confirmed thelevels of TAG in transgenic soil-grown plants by transmethylationof thin layer chromatography (TLC)–separated TAG and gaschromatography–flame ionization detection analysis of the resultingFA methyl esters (FAMEs; see Supplemental Figure 5 online).

An Abundance of Oil Droplets in the Leaves ofTransgenic Lines

Leaves of 6-week-old TAG-accumulating line 57 and of the wildtype were compared by confocal microscopy following Nile Redstaining and by transmission electron microscopy (TEM). Oildroplets were abundant in line 57 and were distributed in theproximity of the chloroplasts (Figures 4A to 4C). By contrast, fewor no oil droplets were observed in the wild-type leaf sample(Figures 4D to 4F). We used TEM to analyze the location of oildroplets in leaf sections. In line 57, large and distinct electron-dense oil droplets were observed outside the chloroplast in me-sophyll cells, most likely associated with the ER (Figure 4H). No oildroplets were observed in cells of wild-type sections (Figure 4G).The shape of chloroplasts and distribution of starch granules weresimilar in both wild-type and transgenic lines.

Epidermal Surface Lipids Are Reduced in theTransgenic Lines

We suspected that a group of lipids specifically affected by theectopic accumulation of TAGs with VLCFAs in leaves might beepidermal surface lipids. To test this hypothesis, leaf epidermal cellsections of DGTT2 transgenic line 57 and wild-type plants wereexamined by TEM. Plants of line 57 had leaves with a noticeablyless osmium-dense cuticle than wild-type plants (Figures 5A and5B). Line 57 also contained very distinct and large numbers ofpresumed oil droplets in epidermal and mesophyll cells, including

Figure 2. DGTT Activity in Transgenic Yeast Microsomes.

(A) and (B) Autoradiograph of [1-14C]-16:0-acyl-CoA and [1-14C]-18:0-acyl-CoA radiolabeled lipids separated on a TLC plate. Signals repre-senting TAG, free fatty acids (FFA), and diacylglycerol (DAG) are indicated.H1266 indicates the empty vector control.(C) Competition assay. For each reaction, 1.725 nmol of [1-14C]-16:0-acyl-CoA was added. Competitors consisting of 1.725 nmol of unlabeled16:0-acyl-CoA, 18:1-acyl-CoA, or 22:1-acyl-CoA was added to the re-spective reactions and compared with a no-competitor control (0). Thenewly synthesized 14C-labeled TAGs were separated and quantified byscintillation counting. The results were normalized by subtracting thebackground (in dpm) and then dividing all the results by the no-competitorcontrol (n = 3; average 6 SD).

680 The Plant Cell

Figure 3. DGTT2 Leads to the Accumulation of TAG with VLCFAs in Arabidopsis Seedlings (15 d Old) and Soil-Grown Plants (6 Weeks Old).

Enhancing Nutritional Content of Leaves 681

guard cells. Similar oil droplet–like structures were not observed inepidermal cells of wild-type plants. Leaves of transgenic line 57and wild-type plants were further examined by scanning electronmicroscopy (Figures 5C and 5D). By comparison, line 57 leavesproduced relatively fewer rod-like wax crystals on the surface ofepidermal cells (Figure 5D). To observe quantitative and qualitativechanges in wax composition, the rosette leaf waxes of lines 14, 22,and 57 and the wild type (6 weeks old) were subjected to gaschromatography–mass spectrometry (GC-MS) analysis. We ob-served a considerable change in the wax composition, specificallya decrease in nonacosane (C29 alkane), which is the most abundantleaf wax constituent, and triacontanol (C30 1-alcohol) in the DGTT2transgenic lines (Figure 5E). However, no statistically significant

differences were detected between wild-type and DGTT2 trans-genic lines in the total wax amounts (Table 1).Furthermore, we observed changes in cutin and cutin-

associated features in the transgenic lines. Because alteration ofcutin content and composition is often associated with changesin the cuticular ledge ultrastructure of guard cells (Kosma andJenks, 2007; Li et al., 2007), we focused on the stomata ofleaves of line 57 and the wild type using scanning electron mi-croscopy and TEM. Scanning electron microscopy showed thatleaves of line 57 had an altered stomatal structure, with widerstomatal openings (Figures 6A and 6B). TEM images revealedsmaller and thinner cuticular ledges in guard cells of line 57(Figures 6C and 6D). These observations led us to investigateleaf cutin monomer composition. Ectopic production of DGTT2resulted in a 25 to 37% reduction in total identified cutinmonomer content (Table 1). In particular, the a,v-dicarboxylicacid (DCA) content of DGTT2 transgenic lines was lower, withslight reductions in 16:0 DCA and 18:1 DCA amounts (22 to 37%and 18 to 30% reductions, respectively, P < 0.001, n = 4) andparticularly large reductions in 18:2 DCA content (30 to 44%reduction, P < 0.001, n = 4; Figure 6E).

Composition of Sphingolipids Is Altered in DGTT2 Lines

Given that DGTT2 accepts a broad range of acyl-CoA substrates,we hypothesized that expression of DGTT2 could result in re-duced sphingolipid content in the leaves. We analyzed sphingo-lipid content in wild-type and transgenic lines using electrosprayionization–high-resolution/accurate mass spectrometry to identifyglycosyl inositolphosphoceramide (GIPC) lipids. Normalization ofGIPC lipid abundances against an internal standard (d18:1/12:0lactosyl ceramide) demonstrated that in wild-type plants, thet18:1/h24:0 and t18:1/h24:1 molecular species were the mostabundant (Figure 7A), constituting 46.2% 6 1.65% and 26.9% 60.05% of all GIPC lipids, respectively. Notably, line 57 (Figure 7A)exhibited 29.5 and 76.6% increases in levels of t18:1/h24:0 GIPC(P < 0.001, n = 3) and t18:1/h26:1 (P < 0.01, n = 3) relative to wild-type plants, respectively, and a concomitant 68.7% decrease inthe level of t18:1/h24:1 GIPC (P < 0.001, n = 3). These datasuggest that DGTT2 production in the leaves of transgenic plantsreduces 24:1 FA content in GIPC lipids, while enriching the GIPCcontent of 24:0.To determine whether DGTT2-dependent modulation of sphin-

golipid 24:1 and 24:0 FA levels was specific to GIPC lipids ordistributed across other sphingolipid classes, we also analyzed theless abundant glucosylceramide (GlcCer) and ceramide (Cer) lip-ids. As shown in Figure 7B, the presence of DGTT2 altered thecontent of t18:1 GlcCer lipids, as line 57 exhibited 39.8% (P < 0.05,

Figure 3. (continued).

(A) ESI-MS quantification of TAGs in neutral lipid extracts of seedlings and soil-grown plants of wild-type (WT) and homozygous transgenic plantsproducing DGTT2 (n = 4; average 6 SD).(B) and (C) Positive-ion electrospray ionization mass spectra of neutral lipid extracts from seedlings of the wild type and transgenic line 57. Tri-tridecanoin (tri13:0) and tripentadecanoin (tri15:0) TAGs were added as internal standards.(D) ESI-MS/MS analysis of neutral lipid extracts of line 57. Shown are the daughter fragment ions from TAGs with [M + NH4]+ adducts with mass-to-charge ratio values of 929, 957, and 983. The mass-to-charge values were rounded up to the nearest nominal mass.

Figure 4. Oil Droplets Are Abundant in Leaves of DGTT2 TransgenicLine 57.

(A) to (C) Confocal fluorescence image of leaf mesophyll cells of line 57,showing chloroplasts (red) and oil droplets (OD; arrows, green) stainedwith Nile red. Bar = 5 mm.(D) to (F) Confocal fluorescence image of leaf samples from the wild typeof the same age as line 57 (6 weeks old, soil-grown). Bar = 5 mm.(G) and (H) TEM analysis of the first leaf pair from 6-week-old wild type(G) and transgenic line 57 (H). Oil droplets (OD), chloroplasts (CH), andstarch granules (S) are indicated. Bars = 500 mm.

682 The Plant Cell

n = 3) and 46.1% (P < 0.001, n = 3) higher levels of h22:0 andh24:0 FAs, respectively, compared with the wild type. As observedfor the GIPC lipids, h24:1 species of t18:1 GlcCer were reduced inthe line 57 by 33.5% (P < 0.01, n = 3) relative to the wild type. Onlythree molecular species of d18:1 GlcCer were detected (see

Supplemental Figure 6D online), which contained h16:0, h24:0,and h24:1 FAs. Additionally, line 57 also contained 39.3% de-creased levels (P < 0.01, n = 3) of h24:1 FA in t18:1 Cer relative tothe wild type (see Supplemental Figure 6A online), while h26:0increased 66%. However, no statistically significant differenceswere detected between the wild type and line 57 in the totalabundances of these lipids (see Supplemental Figures 6A to 6Donline). Together, these results suggest that decrease in C24:1 ofsphingolipids reflects channeling of VLCFAs into TAG (Figure 3D;see Supplemental Figure 3C online).

Limited Changes in Global Expression of Genes inDGTT2 Lines

To evaluate possible compensatory mechanisms in the transgeniclines, we conducted microarray experiments with RNA extracts fromthe leaves of 6-week-old soil-grown plants. A two-factor mixed-model analysis of variance (ANOVA) analysis was employed toaccount for the effects of genotype and chip, and P values werecalculated for the wild type versus line 57. Since no meaningful hitscould be identified after a false discovery rate correction, 15 geneswere selected using the unadjusted P value threshold of <0.002 andfold change >2 or <22 (see Supplemental Table 1 online). Of the 15genes selected from the microarray study, 10 were upregulated andfive downregulated (Table 2). Two upregulated genes were predictedto encode enzymes with a known or possible involvement in TAGbiosynthesis, including a bifunctional enzyme WSD1 (wax estersynthase and diacylglycerol acyltransferase), At5g37300 (Li et al.,2008), and a gene encoding an HXXXD-type acyltransferase,At1g65450. We also identified the qua-quine starch (QQS) regu-latory gene, At3g30720, and cytosolic b-amylase encoding gene,At4g15210. Both proteins are involved in starch metabolism (Li et al.,2009). One of the downregulated genes with unknown function waspredicted to encode a mini zinc finger protein, At3g28917, and an-other a UDP-glycosyltransferase superfamily protein, At2g23210.To validate the microarray analysis, we quantified the transcript

levels of the four genes relevant to oil biosynthesis and starchmetabolism using quantitative RT-PCR. For this experiment, RNAwas isolated from an independent set of wild-type and line 57plants. When normalized to wild-type values, the transcript levelsfor the WSD1-encoding gene were 3.5-fold higher, for the HXXXD-type acyltransferase protein-encoding gene 4.4-fold, for the QQSgene 4.3-fold, and for the cytosolic b-amylase–encoding gene3.0-fold elevated in line 57, thus confirming the microarray results(see Supplemental Figure 7 online).

Growth, Starch, and Heating Value of DGTT2 Lines

The DW of the aerial parts of soil-grown transgenic line 57 andwild-type plants was measured on a weekly basis. The overall

Table 1. Total Amounts (nmol/mg DW) of Identified Leaf Wax and Cutin Constituents of DGTT2-Expressing Lines

Lipid Class Wild Type Line 14 Line 22 Line 57

Wax 2.91 6 0.29 2.13 6 0.41 2.34 6 0.30 2.28 6 0.40Cutin 5.24 6 0.17 3.25 6 0.26*** 3.89 6 0.31*** 2.31 6 0.21******Significant at P <0.001 compared with the wild type by one-way ANOVA with post-hoc Dunnett’s test (n = 4; average 6 SD).

Figure 5. Effect of DGTT2 on Cuticular Wax Production in Arabidopsis.

(A) and (B) Transmission electron microscopy of leaf epidermal sections ofthe wild type (A) and DGTT2 transgenic line 57 (B). Oil droplets (OD) wereonly present in the transgenic line. The arrows point at the cuticle (CT),showing a layer of osmium dense material. The cell wall (CW) is indicated.Bars = 200 nm.(C) and (D) Scanning electron microscopy of leaves of the wild type (C)and transgenic line 57 (D). Rod-like wax crystals are observed on thesurface of wild-type plants near the base of trichomes but are much lessabundant on the surface of cells of the transgenic line. Bars = 5 mm.(E) Leaf epicuticular wax profile of wild-type (WT) plants and transgenic linesas indicated. The number of carbons in the respective compound class isindicated. Significant at *P < 0.05 or **P < 0.01 compared with the wild typeby one-way ANOVA with post-hoc Dunnett’s test (n = 3; average 6 SD).

Enhancing Nutritional Content of Leaves 683

morphology of the leaves and shoots from the DGTT2 trans-genic line was unchanged. Transgenic line 57 gained a consid-erably higher biomass than the wild type (105.6 and 87.6 mg DWper seedling, P < 0.001, n = 3) during the first 5 weeks aftergermination, and DW values continued to be higher until the endof the experiment (Figure 8A). The growth rate of transgenic line57 and wild-type plants slowed after the 8th week.

The increased transcript levels of the QQS gene and cytosolicb-amylase–encoding gene in transgenic line 57 from the mi-croarray experiments led us to investigate the starch content ofDGTT2 transgenic plants. Contrary to expectation, given theincrease in the expression of the genes encoding QQS and

b-amylase, 6-week-old plants of line 57 accumulated 18.7%more starch than wild-type plants (58.0 and 47.1 µg Glu per mgDW [P < 0.01, n = 5], respectively) (Figure 8B). Heating valueswere calculated for 6-week-old line 57 and wild-type plantsbased on an elemental analysis of dry biomass in the presenceof oxygen. Consistent with the increase in FA content, a slightincrease was observed for line 57 (14.1 6 0.2 mJ/kg DW,compared with the wild type 13.9 6 0.4 mJ/kg DW; seeSupplemental Table 2 online), but this difference was not sta-tistically significant.

Caterpillars Feeding on DGTT2 Lines Gained More Weight

Increased TAG accumulation or decreased wax and cutin in veg-etative tissue may result in an unusual diet for phytophagous

Figure 6. Effect of DGTT2 on Cutin and Related Phenotypes in Arabidopsis.

(A) and (B) Scanning electron microscopy images of stomata from theadaxial surface of leaves of the wild type (A) and line 57 (B). Bars = 5 mm.(C) and (D) TEM images of sections of leaf guard cells. Arrows indicatethe cuticular ledge of the guard cells of the wild type (C) and line 57 (D).Bars = 2 mm.(E) Leaf cutin monomer profile of the wild type (WT) and DGTT2 transgeniclines as indicated. Significant at ***P < 0.001 compared with the wild typeby one-way ANOVA with post-hoc Dunnett’s test (n = 4; average 6 SD).

Figure 7. Analysis of Arabidopsis Sphingolipid Molecular Species byHigh-Resolution/Accurate Mass Spectrometry.

(A) Relative abundances of individual GIPC molecular species containingt18:1 long-chain bases in the wild type (WT) and line 57. t test significantat **P < 0.01, or ***P < 0.001 versus the wild type (n = 3; average 6 SD).(B) Relative abundances of individual GlcCer molecular species containingt18:1 long-chain bases in the wild type and line 57. t test significant at *P <0.05, **P < 0.01, or ***P < 0.001 versus the wild type (n = 3; average 6 SD).

684 The Plant Cell

organisms and may perhaps affect the nutritional value of theplants. To test this hypothesis, we conducted a “no choice” insectfeeding assay, in which the generalist lepidopteran herbivoreSpodoptera exigua was reared on the wild type and transgeniclines 22 and 57 (Figures 9A and 9B). Newly hatched larvae werecaged in pots containing 6-week-old plants of two transgenic lines(22 and 57) and the wild type and were allowed to feed and growfor 14 d. At the end of the feeding trial, larvae were recovered andtheir body mass was determined. No visible difference in plantconsumption was observed between DGTT2-producing lines andthe wild type during the 14-d trial (Figure 9A). However, insectsgrown on DGTT2-producing lines were heavier (9.75 6 3.42 mglarval weight) than larvae reared on wild-type plants (6.57 6 1.39mg larval weight) (P < 0.01, n = 95; Figures 9B and 9C).

Active plant defense responses to insect herbivory are con-trolled by wound-induced production of the 18:3-derived phy-tohormone, jasmonoyl-L-isoleucine (JA-Ile) (Koo and Howe,2012). The altered lipid content of DGTT2-producing plantsraised the possibility that increased larvae performance wascaused by reduced levels of JA-Ile in the transgenic lines. Totest this idea, we quantified JA-Ile levels in leaf tissue from insect-fed and control undamaged plants (Figure 9D). Insect feedingincreased the JA-Ile content in the damaged leaves of the wildtype and two transgenic lines. In both control and damagedleaves, the amount of JA-Ile in DGTT2-producing lines wascomparable to that observed in the wild type.

DISCUSSION

Different Roles for Different DGATs

High-throughput transcript profiling of C. reinhardtii undernitrogen deprivation showed that the expression levels andpatterns of the five putative DGAT2-encoding genes variedconsiderably (Miller et al., 2010; Boyle et al., 2012). DGTT1 wasthe most highly regulated, being strongly upregulated followingnitrogen deprivation. Based on the sequence analysis, DGTT2and DGTT3 are the most similar. Additionally, they showed near-identical results in yeast assays. Although DGTT3 did show achange in expression during nitrogen deprivation, the overall

change was less than twofold and statistically insignificant. Forthese reasons, it was assumed that their activity in Arabidopsiswould also be similar and we focused only on one of the twoisoforms.Structural analysis of the DGAT2 candidates indicates that they

are similar to previously identified DGAT2 proteins. The absenceof predicted signal or targeting sequences may be explained bythe fact that the programs were trained on land plant signalsequences, which could hinder the detection of such sequencesin C. reinhardtii. Given the absence of obvious organellar targeting

Table 2. Changes in the Gene Expression in the Leaves of DGTT2 Plants

Locus Gene Name Annotation Functional Category Fold Change

UpregulatedAt5g37300 WSD1 WS and diacylglycerol acyltransferase Lipid metabolism 3.86At3g30720 Qua-Quine Starch (QQS) Starch metabolism 3.69AT4g15210 ATBETA-AMY Cytosolic b-amylase expressed in rosette leaves

and inducible by sugarStarch metabolism 2.86

At1g65450 HXXXD-type acyltransferase family protein Lipid metabolism 2.27AT2g37390 NSKR2 Sodium potassium root defective 2 Metal ion binding 2.21AT5g58390 Peroxidase superfamily protein Peroxidase activity 2.05AT5g52390 PAR1 protein Unknown 2.97AT4g08300 Nodulin Mt-N21/EamA-like transporter

family proteinUnknown 2.66

DownregulatedAt3g28917 MIF2 Mini zinc finger 2 DNA binding, biological process unknown 22.34At2g23210 UDP-glycosyltransferase superfamily protein Transferase activity 22.26

Figure 8. Growth and Starch Content of the Wild Type and DGTT2Transgenic Line 57.

(A) The DW of the aerial parts of Arabidopsis plants was measured ona weekly basis for a period of 8 weeks. Each time point represents theaverage value of at least six individual plants. Black symbols representthe wild type (WT) and gray symbols transgenic line 57. t test significantat *P < 0.05, **P < 0.01, or ***P < 0.001 versus the wild type (n = 6;average 6 SD).(B) Starch content of leaves of the wild type and line 57 (6 weeks old). ttest significant at **P < 0.01 versus the wild type (n = 5; average 6 SD).

Enhancing Nutritional Content of Leaves 685

sequences, the most likely location is the ER membrane, which isthe main site of TAG synthesis in plants (Lung and Weselake,2006). Recently, however, the existence of a separate TAG as-sembly pathway in chloroplasts has been suggested (Fan et al.,2011; Goodson et al., 2011). It seems possible that some of theDGAT2 proteins are preferentially associated with the ER,whereas others are targeted preferentially to the chloroplast en-velope by mechanisms that would not require classic targetingsequences.

Biochemically, DGTT2 possesses broad substrate specificityand is able to incorporate varying acyl-CoA species into TAGin vitro. However, it is unclear what its natural substrate in C.reinhardtii is. Both O. tauri DGAT2B and yeast Dga1 also havebroad substrate specificity, in contrast with the more substratespecific type 2 DGATs found in land plants (Wagner et al., 2010).These, along with the relative abundance of type 2 DGATs inother green algae, suggest a more distinct role for these en-zymes in algae, compared with plants. This difference may berelated to the different roles and regulation of TAG synthesis,as microalgae produce oil in response to environmentalstresses, whereas land plants produce oil primarily during seeddevelopment.

Ectopic Production of DGTT2 Results in TAG Accumulationin Arabidopsis

Most of the DGTT2-producing Arabidopsis lines examined de-veloped distinct seedling phenotypes, such as accelerated hy-pocotyl elongation and pale-green cotyledons. The increasedhypocotyl length was presumed to be due to excess storagecompounds, such as TAG in the seeds that may accelerate cellelongation compared with the wild type. The reduction in chlo-rophyll content in the DGTT2 seedlings could be related to thereduction in the major plastid lipids, limiting the extent of thephotosynthetic membrane. The fact that this phenotype is onlyvisible at the seedling stages, when leaves need to rapidly expandand demands on membrane lipid biosynthesis are high, supportsthis hypothesis. Thus, it seems likely that the acyl-group supply islimited and TAG accumulation in leaves diverts acyl groups awayfrom other pathways, such as plastid glycerolipid biosynthesis,without a compensatory increase in acyl synthesis. This hypoth-esis is supported by the microarray results, which revealed nosignificant induction of genes encoding enzymes necessary forFA biosynthesis (Table 2; see Supplemental Table 1 online).In general, plants accumulate TAG in seeds as stored energy

that can be subsequently used to fuel germination and seedlingestablishment. We demonstrated that accumulation of TAG withVLCFA, such as C20:0 and longer, in green seedlings and leavesof soil-grown plants can be achieved by expression of DGTT2,even though VLCFA-containing glycerolipids do not normallyaccumulate in green tissues of Arabidopsis. This observation wasunexpected because C. reinhardtii TAG primarily contains C16:0and C18:1 FAs, but no TAG species containing VLCFAs (Fanet al., 2011). Following the ectopic production of DGTT2 in theleaves of Arabidopsis, most likely the availability of particular acyl-CoA species as well as the substrate preference of DGTT2 forVLCFAs leads to the observed composition of the TAG producedin the transgenic lines. It is interesting to note that the amount ofTAG accumulated in DGTT2 transgenic seedlings was higher thanthat of soil-grown plants (Figure 3A), while the relative abundanceof major chloroplast lipids, such as monogalactosyldiacylglycerol(MGDG) and digalactosyldiacylglycerol (DGDG), was reduced (seeSupplemental Figure 8 online). These seedlings had increasedlevels of 16:3-containing molecular species of MGDG but fewer18:3-containing species, similar to mutants affected in ER-to-plastid lipid trafficking (Xu et al., 2003), suggesting that FA inter-mediates normally transported from the ER to the chloroplast aspart of the eukaryotic pathway of thylakoid lipid biosynthesis areinstead channeled into TAG. Compared with seedlings, the lowerlevels of TAG that accumulated in the leaves of soil-grown DGTT2transgenic plants may perhaps be explained by FA catabolism inolder leaves (Yang and Ohlrogge, 2009).

Acyl Group Diversion from Sphingolipids and SurfaceLipids to TAG

In plants, FAs are synthesized in plastids, and following elonga-tion and modification at the ER they provide the precursors for thesynthesis of sphingolipids and surface lipids, such as cutin andwaxes (Chen et al., 2009; Kunst and Samuels, 2009). Sphingo-lipids are composed of a Cer backbone that consists of a long-chain base (C18 long) to which a FA (C16-C26) is bound by an

Figure 9. S. exigua Feeding Assay and JA-Ile Measurements.

(A) and (B) Photographs of representative wild-type (WT) and DGTT2transgenic plants (A) and larvae grown on each genotype (B) at the endof feeding trial.(C) Average fresh weight of insects. t test significant at **P < 0.01 versuswild-type hosts (n = 95; average 6 SD).(D) JA-Ile accumulation in leaf tissue collected from undamaged controlplants and those damaged during 14 d of insect feeding (n = 5; aver-age 6 SD). FW, fresh weight.

686 The Plant Cell

amide linkage. Sphingolipids are an important lipid component ofthe plasma membranes of plant cells. They are known to play rolesin membrane structure, signal transduction, and programmed celldeath (Chen et al., 2009). GlcCer and GIPC lipids are the two majorclasses of sphingolipids in plants (Chen et al., 2008; Li-Beissonet al., 2010; Chao et al., 2011). In Arabidopsis, the cutin polymer iscomprised of DCAs, v-hydroxy FAs, and glycerol monomers,which are embedded with and covered by waxes (Bonaventureet al., 2004; Franke et al., 2005). The increase in TAG molecularspecies containing VLCFA in the leaves of DGTT2 transgenic linesis consistent with a decreased carbon flux into the sphingolipidand surface lipid biosynthetic pathways. More specifically, theincreased levels of 18:2 and 24:1 in TAG of DGTT2 lines andconcomitant decrease in 18:2 DCA content of cutin, t18:1/h24:1 ofsphingolipids and 18:3 of MGDG and DGDG (see SupplementalFigure 9 online), and the observed decrease in very-long-chainalkane content of surface waxes suggests a redirection of satu-rated and unsaturated acyl-chains from sphingolipids, surfacelipids, and galactolipids into TAG, mediated by DGTT2 (Figure 10).Surface lipid metabolism is restricted to epidermal cells. The TEManalysis clearly demonstrated an alteration in epidermal cell lipidmetabolism in DGTT2 lines, apparent by the accumulation of oildroplets in both pavement and guard cells, and altered cuticle andguard cell cuticular ledge ultrastructure.

The microarray results suggested no change in the expression ofany known sphingolipid or surface lipid biosynthetic pathway genesother than WSD1. WSD1 is a bifunctional enzyme with both waxsynthase and DGAT activities. However, WSD1 is known for itsrole in wax ester synthesis in inflorescence stems of Arabidopsis

(Li et al., 2008). Wax esters are generally not present in the leaves ofArabidopsis, and our leaf wax analysis did not reveal the presenceof wax esters in leaf waxes of DGTT2 transgenic lines. Whether ornot WSD1 plays a role in the observed TAG accumulation ofDGTT2 transgenic lines via its DGAT activity remains to bedetermined.In agreement with Wagner et al. (2010), in vitro assays dem-

onstrated that DGTT2 was capable of using both saturated andunsaturated acyl-CoAs of several different carbon chain lengths.Interestingly, the ESI-MS/MS spectra of the VLCFA-containingTAGs from the transgenic lines are consistent with these mo-lecular species only containing one VLCFA (see SupplementalFigure 3C online; Figure 3D). This observation is consistent withDGTT2 adding a VLCFA to a DAG acceptor molecule. As wild-type Arabidopsis leaves typically do not possess VLCFAs in theTAG fraction (Li-Beisson et al., 2010), we can assume that theDAG-derived portion of TAG in DGTT2 lines did not contributethe VLCFA. This is further supported by the lack of VLCFAs inDAG and phosphatidic acid pools of DGTT2 transgenic plants(see Supplemental Figure 10 online). Thus, it is unlikely that thesupply of the acyl acceptor for TAG biosynthesis by the Ken-nedy pathway is limiting. Rather, the observed diversion of acylgroups from other pathways (Figure 10), such as sphingolipid,surface lipid, or thylakoid lipid biosynthesis, suggests that thesupply of acyl donors (acyl-CoAs) is limiting and that the broadsubstrate specificity of DGTT2 is responsible for the observedTAG molecular compositions.However, the increased transcript level of WSD1 in the trans-

genic lines opens the possibility that this enzyme is involved in the

Figure 10. A Schematic Overview of Altered Diversion of Acyl-CoA Groups from Other Pathways by DGTT2 in Arabidopsis Leaves.

The hatched arrow represents the activity of DGTT2 in channeling acyl-CoAs from wax/cutin/sphingolipids to the synthesis of TAGs with VLCFAs in thetransgenic lines. CW, cell wall; DAG, diacylglycerol; FAE, fatty acid elongation; FAS, fatty acid synthesis; G3P, glycerol-3-phosphate; LPA, lyso-phosphatidic acid; LPC, lysophosphatidylcholine; PC, phosphatidylcholine; PA, phosphatidic acid; PM, plasma membrane.[See online article for color version of this figure.]

Enhancing Nutritional Content of Leaves 687

elevated TAG levels observed in DGTT2-producing leaves as well.It is similar to a unique type of long-chain acyltransferase ex-hibiting both WS and DGAT activities (ADP1), which was identifiedin A. calcoaceticus (Kalscheuer and Steinbüchel, 2003). It hasbeen shown that the WS reaction of ADP1 accepts a wide rangeof different chain length acyl-CoAs as substrates. However, theDGAT reaction preferred acyl-CoAs of greater chain lengths(Kalscheuer and Steinbüchel, 2003). Heterologous expression ofADP1 in a TAG synthesis–deficient quadruple mutant of yeastrestored TAG synthesis (Kalscheuer et al., 2004). The petuniahomolog of ADP1 failed to restore TAG synthesis in yeast due tolack of DGAT activity (King et al., 2007), but the Arabidopsis ho-molog, WSD1, had WS and DGAT activities (Li et al., 2008). Whythe WSD1 gene is induced in the transgenic lines and whetherWSD1 plays a role in TAG accumulation in the DGTT2 transgeniclines remains to be seen, especially since it is unclear whetherWSD1 has an acyl chain length preference for its DGAT activity.Because DGTT2 showed affinity for C22:1 acyl-CoA in the in vitroassays, it is reasonable to assume that the activity of DGTT2contributes to channeling long-chain CoAs from sphingolipids/surface lipids to the synthesis of TAGs with VLCFAs in thetransgenic lines (Figure 10).

Enhancing the Nutritional Content of Vegetative Biomass

The microarray results suggested that in the DGTT2 transgeniclines, nonlipid pathways, especially starch biosynthesis, may beaffected, which was confirmed by observing an increase in starchcontent (Figure 8B). The two starch-related genes upregulated intransgenic lines encode the regulatory protein QQS and a cyto-solic b-amylase. Transgenic lines in which QQS levels are re-duced have been shown to accumulate more leaf starch content(Li et al., 2009), seemingly in contradiction with the phenotypesobserved for the DGTT2 transgenic lines. Similarly, in vitro studiessuggested that b-amylase could play a role in starch degradation,even though a mutation in the respective gene did not alter leafstarch content (Laby et al., 2001). One explanation is that thesetwo genes could be induced to compensate for the increase instarch content, rather than being responsible for the increase.Alternatively, the increased accumulation of these two transcriptsmay suggest a connection between lipid and carbohydrate me-tabolism. Whether the observed effects on starch metabolism andrelated genes have to be considered a pleiotropic effect or needto be interpreted in the context of altered carbon flow in theDGTT2 transgenic lines remains to be determined.

In principle, the enhanced TAG content should lead to increasedheating values reflecting an increase in energy density, assumingthe gains are not negated by other compositional changes. Onlyslight differences in growth rates of transgenic lines and the wildtype implied that the ectopic introduction of DGTT2 activity inArabidopsis had no detrimental effects under the employed labo-ratory conditions. The observed subtle increase in total heatingvalue in the DGTT2 transgenic line is consistent with the expectedcontribution of total FAs/TAG toward total heating values in theleaves, which we calculated to be;0.7%. Our results also indicatethat metabolic changes caused by DGTT2 activity do not nega-tively affect the growth and development of the insect herbivoreS. exigua. The growth of S. exigua larvae was positively correlated

with the increased levels of TAG and starch in the transgenic lines.A variety of factors may contribute to increased herbivore perfor-mance, including increased energy content (e.g., essential dietaryFAs), increased digestibility of the biomass due to altered surfacelipids, or decreased production of anti-insect compounds, such asglucosinolates. Either way, this study demonstrates the potentialbenefits of engineering TAG content in transgenic crops to in-crease energy density for biofuel production or enhanced nutri-tional value for animal feed.

METHODS

Phylogenetic Analysis

The protein sequences for the Chlamydomonas reinhardtii DGTT1-5genes were retrieved by sequence comparison with Arabidopsis thalianaDGAT2 using BLAST (Altschul et al., 1997). Sequences of DGTT1-5 andother DGAT2 proteins were aligned using the ClustalW software inMEGA5 (Tamura et al., 2011). The alignment was then used to constructa phylogenetic tree using the neighbor-joining method in MEGA5, with thetree being tested by bootstrapping with 1000 replicates.

Plasmid Construction

C. reinhardtii strain dw15.1 (cw15, nit1, mt+), provided by Arthur Grossman,was grown under continuous light (;80 µm/m2/s) and at 22°C in liquid Tris-Acetate Phosphate (TAP) media (Harris, 1989) until mid-log phase and thenpelleted. Total RNA was extracted from the cells using a Qiagen RNeasyPlant Mini kit. cDNA was synthesized with Invitrogen SuperScript III andoligo(dT) primer and used as a template for PCR. The primers used are listedin Supplemental Table 3 online. The amplified regions were digested withHindIII and SphI, for DGTT2 and DGTT4, and HindIII and XhoI, for DGTT3and DGTT5. The gene sequences were then subcloned into the InvitrogenpYES2 vector to form pYES2-DGTT2-5 for expression in yeast Saccha-romyces cerevisiae.

DGTT2 cDNA was amplified from pYES2-DGTT2 by PCR using gene-specific primers (see Supplemental Table 3 online). A fragment of 975 bpcontaining the complete open reading frame was digested with BamHIand EcoRI. This fragment was then placed between the cauliflowermosaic virus 35S promoter and OCS terminator of vector p9-35S-OCS(DNA Cloning Service) to form 35S:DGTT2.

Yeast Expression

Yeast strain H1266 (are2D lro1D dga1D) (Sandager et al., 2002) was grownto mid-log phase in yeast extract peptone dextrose (YPD) media (Sherman,2002) and transformed with the DGTT constructs, along with an emptypYES2 vector as a negative control, according to Gietz et al. (1995). Thetransformants were selected on Synthetic Complete (SC) medium(Sherman, 2002) with 2%Glc and the uracil omitted (SC-U). Colonies werepicked and grown overnight in SC-U + 2%Glc, before being transferred toSC-U + 2% Gal and 1% raffinose. After 48 h, 30 mL of the cultures werecollected and pelleted by centrifugation for 5 min at 3000g. Lipids wereextracted and analyzed as described below.

Microsome DGTT2 Assays

Transformed yeast colonies were picked and grown in SC-U + 2% Glcovernight and then transferred to SC-U + 2% Gal and 1% raffinose. Thecells were harvested after 12 h, and the microsomes were prepared asdescribed (Milcamps et al., 2005). The total protein concentration wasmeasured using Bio-Rad Protein Assay Dye Reagent by adding 900 mL ofthe reagent to 20 mL of BSA standards or microsome samples. Fifty

688 The Plant Cell

nanograms of microsome were added to a mix containing 100 mM Tris,8 mM MgCl2, 1 mg/mL BSA, 20% glycerol, 0.25 mg/mL DAG, and 1.725nmol [1-14C]-16:0-acyl-CoA or [1-14C]-18:1-acyl-CoA (Moravek Bio-chemicals). The reaction was incubated at room temperature for 1 h. Thelipids were extracted with chloroform:methanol (1:1 [v/v]) and phaseseparated with 0.2 MH3PO4 and 1MKCl. The organic layer was extractedand separated on a silica TLC plate using 80:20:1 (v/v/v) petroleum ether:ethyl ether:acetic acid as the solvent. The TLC plate was exposed to filmfor 72 h to visualize the radiolabeled lipids.

For the competition assay, 1.725 nmol unlabeled 16:0-acyl-CoA, 18:1-acyl-CoA, or 22:1-acyl-CoA was added. The separated TAG bands werescraped from the plate and counted in a scintillation counter to quantifythe amount of radioactivity incorporated into the lipid.

Plant Material and Generation of Arabidopsis Transgenic Plants

Arabidopsis wild-type (Columbia 2) seeds or transgenic seeds weresurface sterilized and grown on half-strength Murashige and Skoog(Murashige and Skoog, 1962) agar plates containing 1% Suc in a growthchamber adjusted to 16 h light/8 h dark (100 µm/m2/s) at 22°C after 3 d ofstratification at 4°C. Fifteen-day-old wild-type and transgenic plants weretransferred onto soil and grown in a growth chamber at 16 h light/8 h dark(100 µm/m2/s) and 22°C. These plants were harvested at 6 weeks for lipidand metabolic assays. Starch analysis was performed using a Megazymekit as per the manufacturer’s instructions (Megazyme). For insect assays,transgenic lines and the wild type were grown in a short-day growthchamber at 12 h light/12 h dark (100 µm/m2/s) at 22°C. The binary vectorcontaining DGTT2 was introduced into Agrobacterium tumefaciens strainGV3101 by electroporation.Arabidopsiswas transformed using the flowerdip method (Clough and Bent, 1998). Transgenic plants (T1) were selectedon half-strength Murashige and Skoog agar plates containing 1% Suc and100 mg/L kanamycin. Wild-type and homozygous transgenic seedlingswere grown on the same shelf of the growth chamber when used for lipid,metabolite, and insect assays.

Lipid Isolation and Quantification

Soil-grown 6-week-old and 15-d-old (grown on half-strength Murashigeand Skoog agar plates containing 1% Suc) wild-type and transgenicArabidopsis plants were freeze dried. Neutral lipids were extracted fromdried samples using chloroform:methanol (1:1 [v/v]) with 100 mM internalstandard tri15:0 TAG and separated on a silica TLC plate using a mixtureof solvents consisting of petroleum ether:ethyl ether:acetic acid (80:20:1,by volume). TAG bands were isolated from the TLC plate after separation,dissolved in toluene with 10 mM tri13:0 TAG internal standard, and as-sayed using ESI-MS as previously described (Durrett et al., 2010). Toquantify the amount of TAG accumulating in yeast expressing the DGTTconstructs, neutral lipids were extracted from the yeast pellets and sub-mitted for ESI-MS, following the method described by Durrett et al. (2010).

Distinct lipid and TAG bands were scraped from the TLC plates andused to prepare FAMEs by acid-catalyzed transmethylation. Identificationand quantification of FAMEs was performed as previously described (Xuet al., 2003). The amounts of lipids were calculated based on the contentof FAs derived from GC using C15:0 as an internal standard.

Microscopy

For oil droplet visualization and TEM, the leaf samples from 6-week-old soil-grown transgenic and wild-type plants were used. Whole leaf samples forTEM were fixed in a mixture of 2.5% glutaraldehyde and 2.5% para-formaldehyde in 0.1 M cacodylate buffer at 4°C for 24 h, postfixed in 1%osmium tetroxide, and dehydrated in a graded acetone series. Sampleswere infiltrated and embedded in Spurr resin (Polysciences). Thin sectionswere imaged using a JEOL 100CX transmission electron microscope at

a 100-kV accelerating voltage. Freshly harvested leaf sampleswere used foroil droplet visualization by confocal microscopy as previously described(Sanjaya et al., 2011). For scanning electron microscopy, leaves from6-week-old transgenic line 57 and wild-type plants were plunge frozen inliquid nitrogen, freeze dried with an EMS750X Turbo Freeze Drier (ElectronMicroscopy Sciences), andmounted on aluminum stubs using carbon tape(Ted Pella). These samples were coated with gold (;20-nm thickness) for3.5 min in an Emscope Sputter Coater model SC 500 (Ashford) purged withargon gas and coated with osmium (;10-nm thickness) in an NEOC-ATosmiumcoater (Meiwafosis) for 35 s. The sampleswere examined in a JEOLJSM-6400V (Lanthanum Hexaboride electron emitter) scanning electronmicroscope. Digital images were acquired using Analysis Pro softwareversion 3.2 (Olympus Soft Imaging Solution) at the Center for AdvancedMicroscopy, Michigan State University.

Analysis and Quantification of Sphingolipid Classes

Leaves of soil-grown (6-week-old) plants were used for the extraction ofsphingolipids, as previously described (Markham et al., 2006). Sphingolipididentification and quantification from wild-type and line 57 samples wasperformed by electrospray ionization–high-resolution/accurate mass spec-trometry operating in positive ion mode. All solvents used were HPLC gradeor the highest grade available. Isopropanol, methanol, and chloroform werepurchased from Macron Chemical. Hexane and HPLC water were fromFisher. Ammonium formatewaspurchased fromAlfa Aesar. The sphingolipidinternal standard mixture used for mass spectrometry analysis was AvantiSphingolipid Mix II (Avanti Polar Lipids). Around 30-mL aliquots of total lipidextract were dried under nitrogen and subjected tomild alkaline hydrolysis ofglycerolipids by the addition of 500 mL of methanol, 250 mL of chloroform,and 75 mL of 1 M KOH. Samples were incubated for 2 h at 37°C in a shakingwater bath and then neutralized with 6 mL of glacial acetic acid (Merrill et al.,2005). For Cer andGlcCer analysis, samples were reextracted by amodifiedFolchmethod (Busik et al., 2009) and resuspended in 300mL of isopropanol/methanol/chloroform (4:2:1 [v/v/v]). For GIPC lipid analysis, dried sampleswere extracted into 300 mL of isopropanol/hexane/water (3:1:1 [v/v/v])(Markham et al., 2006), centrifuged to remove particulates, and the super-natants saved for analysis. Samples were further diluted 30-fold in iso-propanol/methanol/chloroform (4:2:1 [v/v/v], for Cer and GlcCer analysis) orisopropanol/hexane/water (3:1:1 [v/v/v], for GIPC analysis) containing a finalconcentration of 20 mM ammonium formate and the sphingolipid internalstandard mixture diluted to 100 nM of each sphingolipid species. Sampleswere centrifuged and loaded intoWhatmanMultichem 96-well plates (SigmaAldrich) sealed with Teflon Ultra Thin Sealing Tape (Analytical Sales andServices).

Samples were directly infused into a Thermo Scientific model LTQOrbitrap Velos high-resolution/accurate mass spectrometer using an AdvionTriversaNanomate nano-electrospray ionization source (Advion) with a sprayvoltage of 1.4 kV and a gas pressure of 0.3 p.s.i. The ion source interfacesettings (inlet temperature of 100°C and S-Lens value of 50%) were opti-mized to maximize the sensitivity for precursor ions while minimizing in-source fragmentation. High-resolutionmass spectra were acquired using theFT analyzer operating at 100,000 resolving power, across the range ofmass-to-charge ratio from 400 to 2000, and were signal averaged for 2 min.All mass spectra were recalibrated offline using XCalibur software (ThermoScientific) and the exactmasses of the sphingolipid internal standardsd18:1/12:0 Cer, d18:1/12:0 glucosyl/galactosyl ceramide, d18:1/12:0 sphingo-myelin, and d18:1/12:0 lactosyl ceramide. Automated peak finding, cor-rection for 13C isotope effects, and quantitation of lipid molecular specieswere performed using Lipid Mass Spectrum Analysis software version 1.0(Haimi et al., 2006) linear fit algorithm. Sphingolipids were quantitated as thesum of their [M+H]+ and [M+NH4]

+ (when present) ion abundances againstthe d18:1/12:0 lactosyl ceramide internal standard. As no attempts weremade to correct for differences in ionization efficiency among individualmolecular species of Cer, glucosyl ceramide, and GIPC lipids, as well as the

Enhancing Nutritional Content of Leaves 689

lack of an ideal internal standard for quantitation of GIPC lipids, sphingolipidmolecular species are presented only as a fraction of the total normalizedabundance of each lipid class.

Surface Wax and Cutin Analysis

Rosette leaves of 6-week-old soil-grown transgenic and wild-type plantswere used for cutin and wax analysis, using slight modifications ofmethods previously described (Molina et al., 2006; Kosma et al., 2009).Briefly, surface waxes were extracted from leaves by a 30-s submersionin hexane. Internal standards pentadecanoic acid (C15:0), tricosanol(C23:0), and octacosane (C28:0) were added to the hexane extracts, andthe solvent was removed by evaporation under nitrogen gas. Driedextracts were derivatized with 100 mL each of pyridine and N,O-bis(trimethylsilyl trifluoroacetamide) (BSTFA). Excess pyridine and BSTFAwere removed by evaporation with N2, and samples were dissolved inheptane:toluene (1:1 [v/v]) for GC-MS analysis.

Thoroughly delipidated ground leaf tissues were used for cutinmonomeranalysis. Methyl heptadecanoate and pentadecalactone were used asinternal standards. Base catalyzed transmethylation reactions (NaOMe/methanol) consisted of 4.5% sodium methoxide (NaOMe) and 7.5%methyl acetate in methanol in a total volume of 6 mL.

Reaction mixtures were heated overnight (;16 h) at 60°C and thenallowed to cool to room temperature. Reaction mixtures were acidified withglacial acetic acid to pH 4 to 5, 2 to 3 mL saline solution was added (0.5 MNaCl), andFAMEswere extractedwith 7mLCH2Cl2. The organic phasewaswashed twice with dilute saline solution (0.9% NaCl [w/v]) and dried overanhydrous Na2SO4. Extracts were evaporated to dryness under N2 and theproduct silylated to convert hydroxyl groups to their trimethylsilyl ethersusing BSTFA and pyridine. Excess pyridine and BSTFA were evaporatedunder N2, and samples were dissolved in heptane:toluene (1:1 [v/v]) for GC-MS analysis. GC-MS temperature programs were as follows: wax, inlet350°C, detector 320°C, oven 130°C for 3 min then increased at 5°C/min to325°C, and 325°C for 10 min; cutin, inlet 330°C, detector 320°C, oven 140°Cfor 3min and then increased at 5°C/min to 310°C andheld at 310°C for 10min.

Elemental Analysis

Rosette leaves of 6-week-old soil-grown transgenic line 57 and wild-typeplants were harvested at the end of the day and thoroughly freeze-dried,ground into fine powder in a Retsch Mill at a frequency of 25 for 2 min, andused to measure carbon, hydrogen, sulfur, nitrogen, and ash contentemploying a commercial service (Elemental Analysis). Protocols are listed onthe company website (http://www.elementalanalysis.com/). Calculation ofheating value (mJ/kgDW) = 0.3491XC+1.1783XH+0.1005XS20.0151XN2

0.1034XO 2 0.0211Xash; Xs are entered as mass percentages (Gaur andReed, 1995).

Insect Feeding Assay and JA-Ile Quantification UsingMass Spectrometry

Insect feeding trials and JA-Ile analysis by liquid chromatography–tandemmass spectrometry were performed as previously described (Koo et al.,2011). Spodoptera exigua eggs (Benzon Research) were hatched at 30°Con an artificial insect diet (Southland Products) for 2 d and used for feedingexperiments. Newly hatched neonates were allowed to grow on artificialdiet for another 2 d before being transferred to 6-week-old plants. Tenlarvae were reared per single pot, and each pot contained two plants ofthe same genotype. A total of 14 pots were used for each genotype.Larvae were caged in each pot using an inverted clear plastic cup with anopening on the top covered with Miracloth for air exchange. Plants withinsects were grown in a growth chamber maintained at 21°C under a 12-hlight (100 µm/m2/s) and 12-h dark cycle. Fresh weight of individual larvaewas determined 14 d after the start of the feeding trial. For JA-Ile

measurements,;350mgof leaf tissuewas harvested from insect damagedand undamaged control plants at the end of the trial. Jasmonate extractionand JA-Ile quantification by liquid chromatography–tandem mass spec-trometry were performed as previously described (Koo et al., 2011).

Microarray and Data Analysis

Leaves from 6-week-old plants grown on soil at 22°C under 16 h light/8 h dark(100 µm/m2/s) were used for isolating total RNAswith theRNeasy plantmini kit(Qiagen). Subsequently, the total RNA samples were pretreated with RNase-free DNase I and cleaned with a Plant Total RNA isolation kit (Qiagen). Fourbiologically independent RNA samples from DGTT2 line 57 and the wild typeeach were used for the microarray experiments. Probe preparation, hybrid-ization to the4-plexGeneChipArabidopsisATH1GenomeArrays (NimbleGen),and subsequent processing steps were performed according to the manu-facturer’s instructions (Gene Expression Center, University of Wisconsin).Arrays were scanned at 5 mm on an Axon4000B scanner (Molecular Dy-namics).Microarray datawere analyzed using PartekGenomics Suite software(2009). RMA normalization (Irizarry et al., 2003a, 2003b) was performed byPartek upon data import, producing log2-transformed intensity values. Probeset log intensities had a bimodal distribution with a valley at ;9 (data notshown). Such bimodality appears to be a common feature of oligonucleotidearrays, and the lower mode is thought to correspond to unexpressed genes(Irizarry et al., 2003b; https://stat.ethz.ch/pipermail/bioconductor/2006-June/013255.html). Therefore, probe sets whose maximal values across all arrayswere <9 were taken to target genes that were not expressed at detectablelevels in any of the samples and were therefore removed from further analysis.This decreased the number of probe sets in the analysis from30,361 to 18,221.

The PCA plot of the filtered data set did not show clear separation be-tween wild-type and transgenic plants (data not shown), indicating that thetransgene expression did not have a strong effect on the global gene ex-pression profile of the plants. Since the experiment was performed on twoNimbleGen chips, with four arrays per chip, twowild type and two transgeniceach, two-factor mixed-model ANOVA analysis was employed to accountfor the effects of genotype and chip, the latter being a randomeffect. P valueswere calculated for the wild type versus line 57 contrast. Multiple testingcorrection using the step-up false discovery rate method identified only onegene with adjusted P value < 0.05, encoding a transposable element.Therefore, putative hitswere selected using the unadjusted P value thresholdof 0.002 and fold change threshold of >2 or <22. This identified 15 genes(see Supplemental Table 1 online), several of which could be expected to befalse positives.

Quantitative RT-PCR

Total RNA was extracted from 15-d-old and 6-week-old DGTT2 trans-genic and wild-type plants. cDNA synthesis, quantitative RT-PCR withgene-specific primers (see Supplemental Table 3 online), and dataanalysis were performed as previously described (Sanjaya et al., 2011).

Accession Numbers

The protein sequences used in the phylogenetic analysis were as follows:DGTT1, XP_001702848.1; DGTT2, XP_001694904.1; DGTT3, XP_001691447.1;DGTT4, XP_001693189.1; DGTT5, XP_001701667.1; Arabidopsis DGAT2,NP_566952.1; VfDGAT2, DQ356682.1; RcDGAT2, DQ923084.1; ScDGA1,NP_014888.1; and OtDGAT2B, XP_003083539.1. The microarray data setis deposited at the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) under accession number GSE38898.

Supplemental Data

The following materials are available in the online version of this article.

690 The Plant Cell

Supplemental Figure 1. Confirmation of DGTT2-5 Protein Expressionin Yeast by Immunoblotting Analysis.

Supplemental Figure 2. Overexpression of DGTT2 Affects theSeedling Phenotype.

Supplemental Figure 3. Analysis of Long and Very-Long-Chain FattyAcids in Soil-Grown Plants (6 Weeks Old) by ESI-MS/MS.

Supplemental Figure 4. Confirmation of Long and Very-Long-ChainFatty Acids in 6-Week-Old Soil-Grown DGTT2 Line 57 and Wild-TypePlants by ESI-MS/MS.

Supplemental Figure 5. Analysis of TAG in Soil-Grown 6-Week-OldLeaves of Wild Type and Homozygous Transgenic Plants ExpressingDGTT2 by GC-FID.

Supplemental Figure 6. Abundance of Sphingolipids in the Wild Typeand Line 57.

Supplemental Figure 7. qRT-PCR Quantification of Genes Expressedin 6-Week-Old Wild Type and Line 57.

Supplemental Figure 8. Glycerolipid Composition of DGTT2 Trans-genic Lines 22 and 57 and Wild-Type Plants (15-d-Old Seedlings).

Supplemental Figure 9. Glycerolipid Composition of DGTT2 Trans-genic Lines 22 and 57 and Wild-Type Plants (6-Week-Old Plants).

Supplemental Figure 10. Diacylglycerol and Phosphatidic AcidComposition of DGTT2 and Wild-Type Plants.

Supplemental Table 1. Differentially Expressed Genes Identifiedin DGTT2 Lines and Wild-Type Plants Using NimbleGen MicroarrayAnalysis at a P Value Threshold of 0.002 and Fold Change >2 or <22.

Supplemental Table 2. Elemental Analysis of the Wild Type andTransgenic Line 57.

Supplemental Table 3. Primers Used in This Work.

Supplemental Data Set 1. Multiple Sequence Alignment of Type 2DGATs from Algae and Plants.

ACKNOWLEDGMENTS

We thank Azrin Jamalruddin and Jeremy Letchford (Department of Bio-chemistry and Molecular Biology, Michigan State University) for technicalassistance. We also thank Alicia Pastor for TEM sections, Melinda Framefor assistance with confocal microscopy and Carol Flegler for scanningelectron microscopy (Center for Advanced Microscopy, Michigan StateUniversity), Kathy Richmond, Nick Santoro, and Shane Cantu (EnablingTechnologies) for microarrays, starch, and sugars analysis at the GreatLakes Bioenergy Research Center, Michigan State University. We thankChristopher Saffron and Jonathan Bovee (Department of Biosystems andAgricultural Engineering, Michigan State University) for helpful discussion onelemental analysis and calculation of heating values. This work was fundedby the Department of Energy Great Lakes Bioenergy Research Centerunder the Cooperative Agreement DE-FC02-07ER64494 and by a grantfrom the U.S. Air Force Office of Scientific Research (Grant FA9550-11-1-0264) to C.B. Insect feeding assays and jasmonate measurements weresupported by a grant from the Chemical Sciences, Geosciences, andBiosciences Division, Office of Basic Energy Sciences, Office of Science,U.S. Department of Energy (Grant DE-FG02-91ER20021).

AUTHOR CONTRIBUTIONS

Sanjaya, R.M., G.A.H., G.E.R., J.O., and C.B. designed research. Sanjaya,R.M., T.P.D., D.K.K., T.A.L., B.M., and A.J.K.K. performed research. Sanjaya,

R.M., T.P.D., D.K.K., T.A.L., B.M., A.J.K.K., and Y.V.B. analyzed data.Sanjaya, R.M., and C.B. wrote the article.

Received September 12, 2012; revised January 4, 2013; acceptedJanuary 23, 2013; published February 15, 2013.

REFERENCES

Altschul, S.F., Madden, T.L., Schäffer, A.A., Zhang, J., Zhang, Z.,Miller, W., and Lipman, D.J. (1997). Gapped BLAST and PSI-BLAST: A new generation of protein database search programs.Nucleic Acids Res. 25: 3389–3402.

Banilas, G., Karampelias, M., Makariti, I., Kourti, A., andHatzopoulos, P. (2011). The olive DGAT2 gene is developmentallyregulated and shares overlapping but distinct expression patternswith DGAT1. J. Exp. Bot. 62: 521–532.

Bonaventure, G., Beisson, F., Ohlrogge, J., and Pollard, M. (2004).Analysis of the aliphatic monomer composition of polyesters associatedwith Arabidopsis epidermis: Occurrence of octadeca-cis-6, cis-9-diene-1,18-dioate as the major component. Plant J. 40: 920–930.

Boyle, N.R., et al. (2012). Three acyltransferases and nitrogen-responsive regulator are implicated in nitrogen starvation-inducedtriacylglycerol accumulation in Chlamydomonas. J. Biol. Chem. 287:15811–15825.

Busik, J.V., Reid, G.E., and Lydic, T.A. (2009). Global analysis ofretina lipids by complementary precursor ion and neutral loss modetandem mass spectrometry. Methods Mol. Biol. 579: 33–70.

Cases, S., Smith, S.J., Zheng, Y.W., Myers, H.M., Lear, S.R., Sande, E.,Novak, S., Collins, C., Welch, C.B., Lusis, A.J., Erickson, S.K., andFarese, R.V. Jr.. (1998). Identification of a gene encoding an acyl CoA:diacylglycerol acyltransferase, a key enzyme in triacylglycerol synthesis.Proc. Natl. Acad. Sci. USA 95: 13018–13023.

Chao, D.Y., et al. (2011). Sphingolipids in the root play an importantrole in regulating the leaf ionome in Arabidopsis thaliana. Plant Cell23: 1061–1081.

Chapman, K.D., and Ohlrogge, J.B. (2012). Compartmentation of tri-acylglycerol accumulation in plants. J. Biol. Chem. 287: 2288–2294.

Chen, M., Cahoon, E.B., Saucedo-Garcia, M., Plasencia, J., andGavilanes-Ruiz, M. (2009). Plant sphingolipids: Structure, syn-thesis and function. In Lipids in Photosynthesis: Essential andRegulatory Functions, H. Wada and N. Murata, eds (Dordrecht, TheNetherlands: Springer), pp. 77–115.

Chen, M., Markham, J.E., Dietrich, C.R., Jaworski, J.G., andCahoon, E.B. (2008). Sphingolipid long-chain base hydroxylation isimportant for growth and regulation of sphingolipid content andcomposition in Arabidopsis. Plant Cell 20: 1862–1878.

Clough, S.J., and Bent, A.F. (1998). Floral dip: a simplified methodfor Agrobacterium-mediated transformation of Arabidopsis thaliana.Plant J. 16: 735–743.

Deng, X.D., Gu, B., Li, Y.J., Hu, X.W., Guo, J.C., and Fei, X.W.(2012). The roles of acyl-CoA: Diacylglycerol Acyltransferase 2Genes in the biosynthesis of triacylglycerols by the green algaeChlamydomonas reinhardtii. Mol. Plant 5: 945–947.

Durrett, T.P., Benning, C., and Ohlrogge, J. (2008). Plant tri-acylglycerols as feedstocks for the production of biofuels. Plant J.54: 593–607.

Durrett, T.P., McClosky, D.D., Tumaney, A.W., Elzinga, D.A.,Ohlrogge, J., and Pollard, M. (2010). A distinct DGAT with sn-3acetyltransferase activity that synthesizes unusual, reduced-viscosityoils in Euonymus and transgenic seeds. Proc. Natl. Acad. Sci. USA 107:9464–9469.

Enhancing Nutritional Content of Leaves 691

Fan, J., Andre, C., and Xu, C. (2011). A chloroplast pathway for the denovo biosynthesis of triacylglycerol in Chlamydomonas reinhardtii.FEBS Lett. 585: 1985–1991.

Franke, R., Briesen, I., Wojciechowski, T., Faust, A., Yephremov,A., Nawrath, C., and Schreiber, L. (2005). Apoplastic polyesters inArabidopsis surface tissues—A typical suberin and a particularcutin. Phytochemistry 66: 2643–2658.

Gietz, R.D., Schiestl, R.H., Willems, A.R., and Woods, R.A. (1995).Studies on the transformation of intact yeast cells by the LiAc/SS-DNA/PEG procedure. Yeast 11: 355–360.

Gaur, S., and Reed, T. (1995). An Atlas of Thermal Data for Biomassand Other Fuels. (Golden, CO: National Renewable Laboratory,).

Goodson, C., Roth, R., Wang, Z.T., and Goodenough, U. (2011).Structural correlates of cytoplasmic and chloroplast lipid bodysynthesis in Chlamydomonas reinhardtii and stimulation of lipidbody production with acetate boost. Eukaryot. Cell 10: 1592–1606.

Guihéneuf, F., Leu, S., Zarka, A., Khozin-Goldberg, I., Khalilov, I., andBoussiba, S. (2011). Cloning and molecular characterization of a novelacyl-CoA:diacylglycerol acyltransferase 1-like gene (PtDGAT1) from thediatom Phaeodactylum tricornutum. FEBS J. 278: 3651–3666.

Haimi, P., Uphoff, A., Hermansson, M., and Somerharju, P. (2006).Software tools for analysis of mass spectrometric lipidome data.Anal. Chem. 78: 8324–8331.

Halim, R., Danquah, M.K., and Webley, P.A. (2012). Extraction of oilfrom microalgae for biodiesel production: A review. Biotechnol.Adv. 30: 709–732.

Harris, E.H. (1989). The Chlamydomonas Source Book. (New York:Academic Press).

Hernández, M.L., Whitehead, L., He, Z., Gazda, V., Gilday, A.,Kozhevnikova, E., Vaistij, F.E., Larson, T.R., and Graham, I.A.(2012). A cytosolic acyltransferase contributes to triacylglycerolsynthesis in sucrose-rescued Arabidopsis seed oil catabolism mu-tants. Plant Physiol. 160: 215–225.

Irizarry, R.A., Bolstad, B.M., Collin, F., Cope, L.M., Hobbs, B., andSpeed, T.P. (2003a). Summaries of Affymetrix GeneChip probelevel data. Nucleic Acids Res. 31: e15.

Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D.,Antonellis, K.J., Scherf, U., and Speed, T.P. (2003b). Exploration,normalization, and summaries of high density oligonucleotide arrayprobe level data. Biostatistics 4: 249–264.

Jako, C., Kumar, A., Wei, Y., Zou, J., Barton, D.L., Giblin, E.M.,Covello, P.S., and Taylor, D.C. (2001). Seed-specific over-expression of an Arabidopsis cDNA encoding a diacylglycerol acyl-transferase enhances seed oil content and seed weight. Plant Physiol.126: 861–874.

Kalscheuer, R., Luftmann, H., and Steinbüchel, A. (2004). Synthesisof novel lipids in Saccharomyces cerevisiae by heterologous ex-pression of an unspecific bacterial acyltransferase. Appl. Environ.Microbiol. 70: 7119–7125.

Kalscheuer, R., and Steinbüchel, A. (2003). A novel bifunctional waxester synthase/acyl-CoA:diacylglycerol acyltransferase mediateswax ester and triacylglycerol biosynthesis in Acinetobacter cal-coaceticus ADP1. J. Biol. Chem. 278: 8075–8082.

King, A., Nam, J.W., Han, J., Hilliard, J., and Jaworski, J.G. (2007).Cuticular wax biosynthesis in petunia petals: Cloning and characteri-zation of an alcohol-acyltransferase that synthesizes wax-esters. Planta226: 381–394.

Koo, A.J., Cooke, T.F., and Howe, G.A. (2011). Cytochrome P450CYP94B3 mediates catabolism and inactivation of the plant hor-mone jasmonoyl-L-isoleucine. Proc. Natl. Acad. Sci. USA 108:9298–9303.

Koo, A.J., and Howe, G.A. (2012). Catabolism and deactivation of thelipid-derived hormone jasmonoyl-isoleucine. Front. Plant Sci. 3: 19.

Kosma, D.K., Bourdenx, B., Bernard, A., Parsons, E.P., Lü, S.,Joubès, J., and Jenks, M.A. (2009). The impact of water deficiencyon leaf cuticle lipids of Arabidopsis. Plant Physiol. 151: 1918–1929.

Kosma, D.K., and Jenks, M.A. (2007). Eco-physiological and mo-lecular-genetic determinants of plant cuticle function in drought andsalt tolerant crops. In Advances in Molecular Breeding TowardDrought and Salt Tolerant Crops, M.A. Jenks, P.M. Hasegawa, andS.M. Jain, eds (Dordrecht, The Netherlands: Springer), pp. 91–120.

Kunst, L., and Samuels, L. (2009). Plant cuticles shine: Advances inwax biosynthesis and export. Curr. Opin. Plant Biol. 12: 721–727.

Laby, R.J., Kim, D., and Gibson, S.I. (2001). The ram1 mutant ofArabidopsis exhibits severely decreased beta-amylase activity.Plant Physiol. 127: 1798–1807.

Lardizabal, K., Effertz, R., Levering, C., Mai, J., Pedroso, M.C.,Jury, T., Aasen, E., Gruys, K., and Bennett, K. (2008). Expressionof Umbelopsis ramanniana DGAT2A in seed increases oil in soy-bean. Plant Physiol. 148: 89–96.

La Russa, M., Bogen, C., Uhmeyer, A., Doebbe, A., Filippone, E.,Kruse, O., and Mussgnug, J.H. (2012). Functional analysis of threetype-2 DGAT homologue genes for triacylglycerol production in thegreen microalga Chlamydomonas reinhardtii. J. Biotechnol. 162: 13–20.

Li, F., Wu, X., Lam, P., Bird, D., Zheng, H., Samuels, L., Jetter, R.,and Kunst, L. (2008). Identification of the wax ester synthase/acyl-coenzyme A: diacylglycerol acyltransferase WSD1 required for stemwax ester biosynthesis in Arabidopsis. Plant Physiol. 148: 97–107.

Li, L., Foster, C.M., Gan, Q., Nettleton, D., James, M.G., Myers,A.M., and Wurtele, E.S. (2009). Identification of the novel proteinQQS as a component of the starch metabolic network in Arabi-dopsis leaves. Plant J. 58: 485–498.

Li, Y., Beisson, F., Koo, A.J., Molina, I., Pollard, M., and Ohlrogge,J. (2007). Identification of acyltransferases required for cutin bio-synthesis and production of cutin with suberin-like monomers.Proc. Natl. Acad. Sci. USA 104: 18339–18344.

Li-Beisson, Y., et al. (2010). Acyl-lipid metabolism. The ArabidopsisBook 8: e0133, doi/10.1199/tab.0133.

Lung, S.C., and Weselake, R.J. (2006). Diacylglycerol acyltransfer-ase: A key mediator of plant triacylglycerol synthesis. Lipids 41:1073–1088.

Markham, J.E., Li, J., Cahoon, E.B., and Jaworski, J.G. (2006).Separation and identification of major plant sphingolipid classesfrom leaves. J. Biol. Chem. 281: 22684–22694.

Merrill, A.H., Jr., Sullards, M.C., Allegood, J.C., Kelly, S., andWang, E. (2005). Sphingolipidomics: high-throughput, structure-specific, and quantitative analysis of sphingolipids by liquid chro-matography tandem mass spectrometry. Methods 36: 207–224.

Milcamps, A., Tumaney, A.W., Paddock, T., Pan, D.A., Ohlrogge, J.,and Pollard, M. (2005). Isolation of a gene encoding a 1,2-diacyl-glycerol-sn-acetyl-CoA acetyltransferase from developing seeds ofEuonymus alatus. J. Biol. Chem. 280: 5370–5377.

Miller, R., et al. (2010). Changes in transcript abundance in Chlamy-domonas reinhardtii following nitrogen deprivation predict diversionof metabolism. Plant Physiol. 154: 1737–1752.

Molina, I., Bonaventure, G., Ohlrogge, J., and Pollard, M. (2006).The lipid polyester composition of Arabidopsis thaliana and Bras-sica napus seeds. Phytochemistry 67: 2597–2610.

Murashige, T., and Skoog, F. (1962). A revised medium for rapidgrowth and bio assays with tobacco tissue cultures. Physiol. Plant.15: 473–497.

Oakes, J., Brackenridge, D., Colletti, R., Daley, M., Hawkins, D.J.,Xiong, H., Mai, J., Screen, S.E., Val, D., Lardizabal, K., Gruys, K.,and Deikman, J. (2011). Expression of fungal diacylglycerol acyl-transferase2 genes to increase kernel oil in maize. Plant Physiol.155: 1146–1157.

692 The Plant Cell

Routaboul, J.M., Benning, C., Bechtold, N., Caboche, M., andLepiniec, L. (1999). The TAG1 locus of Arabidopsis encodes fora diacylglycerol acyltransferase. Plant Physiol. Biochem. 37: 831–840.

Saha, S., Enugutti, B., Rajakumari, S., and Rajasekharan, R. (2006).Cytosolic triacylglycerol biosynthetic pathway in oilseeds. Molecularcloning and expression of peanut cytosolic diacylglycerol acyltransferase.Plant Physiol. 141: 1533–1543.

Sandager, L., Gustavsson, M.H., Ståhl, U., Dahlqvist, A., Wiberg,E., Banas, A., Lenman, M., Ronne, H., and Stymne, S. (2002).Storage lipid synthesis is non-essential in yeast. J. Biol. Chem. 277:6478–6482.

Sanjaya, Durrett, T.P., Weise, S.E., and Benning, C. (2011). In-creasing the energy density of vegetative tissues by diverting car-bon from starch to oil biosynthesis in transgenic Arabidopsis. PlantBiotechnol. J. 9: 874–883.

Sherman, F. (2002). Getting started with yeast. Methods Enzymol.350: 3–41.

Shockey, J.M., Gidda, S.K., Chapital, D.C., Kuan, J.C., Dhanoa,P.K., Bland, J.M., Rothstein, S.J., Mullen, R.T., and Dyer, J.M.(2006). Tung tree DGAT1 and DGAT2 have nonredundant functionsin triacylglycerol biosynthesis and are localized to different sub-domains of the endoplasmic reticulum. Plant Cell 18: 2294–2313.

Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., andKumar, S. (2011). MEGA5: Molecular evolutionary genetics analysisusing maximum likelihood, evolutionary distance, and maximumparsimony methods. Mol. Biol. Evol. 28: 2731–2739.

Wagner, M., Hoppe, K., Czabany, T., Heilmann, M., Daum, G.,Feussner, I., and Fulda, M. (2010). Identification and character-ization of an acyl-CoA:diacylglycerol acyltransferase 2 (DGAT2)gene from the microalga O. tauri. Plant Physiol. Biochem. 48:407–416.

Xu, C., Fan, J., Riekhof, W., Froehlich, J.E., and Benning, C. (2003).A permease-like protein involved in ER to thylakoid lipid transfer inArabidopsis. EMBO J. 22: 2370–2379.

Yang, Z., and Ohlrogge, J.B. (2009). Turnover of fatty acids duringnatural senescence of Arabidopsis, Brachypodium, and switchgrassand in Arabidopsis beta-oxidation mutants. Plant Physiol. 150:1981–1989.

Yen, C.L., Stone, S.J., Koliwad, S., Harris, C., and Farese, R.V. Jr..(2008). Thematic review series: Glycerolipids. DGAT enzymes andtriacylglycerol biosynthesis. J. Lipid Res. 49: 2283–2301.

Zou, J., Wei, Y., Jako, C., Kumar, A., Selvaraj, G., and Taylor, D.C.(1999). The Arabidopsis thaliana TAG1 mutant has a mutation ina diacylglycerol acyltransferase gene. Plant J. 19: 645–653.

Enhancing Nutritional Content of Leaves 693

DOI 10.1105/tpc.112.104752; originally published online February 15, 2013; 2013;25;677-693Plant Cell

BenningAbraham J.K. Koo, Yury V. Bukhman, Gavin E. Reid, Gregg A. Howe, John Ohlrogge and Christoph Sanjaya, Rachel Miller, Timothy P. Durrett, Dylan K. Kosma, Todd A. Lydic, Bagyalakshmi Muthan,

Introduction of an Algal Diacylglycerol Acyltransferase 2 Leaves followingArabidopsisAltered Lipid Composition and Enhanced Nutritional Value of

 This information is current as of October 14, 2020

 

Supplemental Data /content/suppl/2013/01/29/tpc.112.104752.DC1.html

References /content/25/2/677.full.html#ref-list-1

This article cites 56 articles, 27 of which can be accessed free at:

Permissions https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X

eTOCs http://www.plantcell.org/cgi/alerts/ctmain

Sign up for eTOCs at:

CiteTrack Alerts http://www.plantcell.org/cgi/alerts/ctmain

Sign up for CiteTrack Alerts at:

Subscription Information http://www.aspb.org/publications/subscriptions.cfm

is available at:Plant Physiology and The Plant CellSubscription Information for

ADVANCING THE SCIENCE OF PLANT BIOLOGY © American Society of Plant Biologists