Towards oilcane: Engineering hyperaccumulation of triacylglycerol into sugarcane stems

Metabolic engineering to divert carbon flux from sucrose to oil in high biomass crop like sugarcane is an emerging strategy to boost lipid yields per hectare for biodiesel production. Sugarcane stems comprise more than 70% of the crops' biomass and can accumulate sucrose in excess of 20% of their extracted juice. The energy content of oils in the form of triacylglycerol (TAG) is more than twofold that of carbohydrates. Here, we report a step change in TAG accumulation in sugarcane stem tissues achieving an average of 4.3% of their dry weight (DW) in replicated greenhouse experiments by multigene engineering. The metabolic engineering included constitutive co‐expression of wrinkled1; diacylglycerol acyltransferase1‐2; cysteine‐oleosin; and ribonucleic acid interference‐suppression of sugar‐dependent1. The TAG content in leaf tissue was also elevated by more than 400‐fold compared to non‐engineered sugarcane to an average of 8.0% of the DW and the amount of total fatty acids reached about 13% of the DW. With increasing TAG accumulation an increase of 18:1 unsaturated fatty acids was observed at the expense of 16:0 and 18:0 saturated fatty acids. Total biomass accumulation, soluble lignin, Brix and juice content were significantly reduced in the TAG hyperaccumulating sugarcane lines. Overcoming this yield drag by engineering lipid accumulation into late stem development will be critical to exceed lipid yields of current oilseed crops.


| INTRODUCTION
Plant lipid consumption for food and fuel has increased the demand for lipids, which is expected to double in the next 15 years (Chapman & Ohlrogge, 2012). In the last decade, metabolic engineering efforts have advanced from single pathways to holistic approaches with overexpression and suppression of multiple genes and are expected to fuel the emerging bioeconomy (Haslam, Sayanova, Kim, Cahoon, & Napier, 2016;Yoon, Zhao, & Shanks, 2013). Production of lipids in vegetative tissues of high biomass crops like sugarcane has the potential to elevate oil yields per hectare compared to traditional oilseed crops (Ohlrogge & Chapman, 2011). Triacylglycerol (TAG) has more than twice the energy density of carbohydrates and is one of the most energy dense molecules on earth (Sanjaya et al., 2013). TAG represents the major lipid component of plant seeds where it provides energy to developing seedlings during germination.
Each TAG molecule has a glycerol backbone to which three fatty acids (FAs) are esterified (Thelen & Ohlrogge, 2002). The FA components of TAG are synthesized in the chloroplast and exported to the endoplasmic reticulum (ER) where FA is assembled into TAG (Bates, Ohlrogge, & Pollard, 2007;Petrie et al., 2012). The Kennedy pathway, comprises three sequential acylation reactions, glycerol-3-phosphate (G3P) acyltransferase catalyzes the first acylation of G3P by the incorporation of an acyl group from acyl-CoA to produce lysophosphatidic acid. The second acylation is catalyzed by lysophosphatidic acid acyltransferase by the addition of second acyl group to produce phosophatidic acid which is dephosphorylated into diacylglycerol (DAG). The final acylation is completed with the addition of the third acyl group by the diacylglycerol acyltransferase (DGAT) enzyme for the production of TAG (Chapman & Ohlrogge, 2012). WRINKLED1 (WRI1) is a transcription factor that belongs to the family of APETALA2/ethylene responsive element binding proteins. Expression of WRI1 is highest during seed development. A TAG reduction of 80% was observed in seeds of WRI1 mutants of Arabidopsis thaliana (Cernac & Benning, 2004;Focks & Benning, 1998). Ectopic expression of WRI1 in non-seed tissues upregulates the expression of genes involved in FA biosynthesis (Baud et al., 2007).
Diacylglycerol acyltransferase1-2 (DGAT1-2) catalyzes the addition of an acyl group to sn1-sn2-G3P, which can be limiting for the production of TAG from DAG. Overexpression of DGAT1-2 in Nicotiana benthamiana increased the TAG content in leaves by 18-fold while suppression of DGAT in A. thaliana decreased the lipid content in seed tissues by 9% to 49% (Petrie et al., 2012;Taylor et al., 2009). DGAT mutants display increased β-oxidation of FA thereby reducing total lipid content in seed tissues. For example, a phenylalanine insertion in DGAT1-2 at codon position 469 elevated the oil and oleic acid (OA) content of maize seeds (Zheng et al., 2008). Vanhercke et al. (2013) reported the synergistic effects of co-expressing WRI1 and DGAT1-2 with respect to increased TAG production in leaves.
Oleosin (OLE1) is a structural protein that covers the surface of oil bodies thereby protecting lipid droplets from coalescence and reducing lipid degradation (Capuano, Beaudoin, Napier, & Shewry, 2007;Parthibane, Rajakumari, Venkateshwari, Iyappan, & Rajasekharan, 2012). Cysteineoleosin (CYSOLE1) is a synthetic version of OLE1 into which six cysteine residues have been engineered. The Cys-residues contribute to disulfide crosslinking of OLE1s which stabilizes lipid droplets contributing to increased FA content in vegetative tissues like leaves and roots (Winichayakul et al., 2013).
SUGAR-DEPENDENT1 (SDP1) is a specific TAG lipase, responsible for initiating oil breakdown and directing FAs from the TAG pool in the ER towards the peroxisome for β-oxidation (Eastmond, 2006). Suppression of the SDP1 gene can increase the accumulation of TAG in vegetative tissues (Fan, Yan, Roston, Shanklin, & Xu, 2014;Kelly et al., 2013).
Trigalactosyl diacylglycerol1 (TGD1) protein is a permease-like component located in the inner chloroplast membrane and is a putative component of lipid transporter (Roston, Gao, Murcha, Whelan, & Benning, 2012). TGD1 is involved in lipid trafficking from the ER to the chloroplast. A TGD1 mutant showed increase in leaf TAG in A. thaliana, while impairing thylakoid lipid biosynthesis in the chloroplast (Xu, Fan, Froehlich, Awai, & Benning, 2005).
Stems contribute to 70%-80% of the sugarcane biomass and accumulate sucrose in mature internodes up to 20% of culm dry weight (DW) which could be diverted into lipid biosynthetic pathway for a step change in the production of TAG (Papini-terzi et al., 2009;Waclawovsky, Sato, Lembke, Moore, & Souza, 2010).
In a recent report from our laboratory, engineering of TAG pathway in sugarcane by co-expression of AtWRI1, ZmDGAT1-2, SiOLE1 and co-suppression of ADP-glucose pyrophosphorylase and peroxisomal ABC-transporter1 increased the TAG content in leaf and stem by 95-fold and 43fold to 1.9% and 0.9% of DW respectively (Zale et al., 2016). In this study, we report a refinement of our strategy in which we co-express and co-suppress alternative versions of genes including CYSOLE1 gene instead of SiOLE1, WRI1 gene from closely related Sorghum bicolor instead of A. thaliana (AtWRI1) and target ribonucleic acid interference (RNAi) suppression with respect to TGD1 and SDP1, which were previously not evaluated in sugarcane. The transgenes were delivered biolistically, either as multiple unlinked plasmids or as a single multigene construct flanked by insulators.  Figure S5) by conventional restriction enzyme digest of vector components and ligation. In vivo homologous recombination in Saccharomyces cerevisiae (Shao, Zhao, & Zhao, 2009) in combination with ligase cycling reaction (Yuan, Andersen, & Zhao, 2016) were used for the assembly of multiple expression/suppression cassettes into a single, large multigene construct ( Figure 2f).

| Tissue culture and biolistic gene transfer
Preparation of media for callus induction, direct embryogenesis, selection and regeneration of plants was carried out as described by Taparia, Fouad, Gallo, and Altpeter (2012). Co-precipitation of different plasmids for biolistic transfer was performed as described by Sandhu and Altpeter, (2008). For the co-delivery of unlinked constructs, the neomycin phosphotransferase II (nptII) containing plasmid was coprecipitated in 1:2 molar ratio with the plasmids carrying target genes. The backbone of the petross-2 plasmid was removed by using PacI restriction enzyme to linearize the plasmid prior gene transfer, followed by gel electrophoresis and gel purification to recover the linearized fragment as described earlier (Taparia et al., 2012).

| Greenhouse trial for evaluation of lipid accumulating sugarcane
Vegetative progenies of eight transgenic, lipid accumulating sugarcane lines were compared with non-transgenic sugarcane in a randomized block design with eight replicates in the greenhouse under natural photoperiod with 11-12 hr day length. The maximum daily light intensity above the canopy ranged between approximately 300 or 800 µmol m −2 s −1 during a cloudy or sunny day respectively. Plants were grown under controlled temperature (28°C/22°C, day/night) with drip fertigation system ( Figure 3a). Transgenic sugarcane lines were selected based on differences in TAG content in leaf tissues and delivered recombinant DNA constructs. Three of the transgenic lines (109; 316; 327) were generated from co-bombardment of unlinked cassettes. Four transgenic lines (1565; 1566; 1569; 1580) were generated from transformation with the single multigene construct. One transgenic sugarcane line (19B) from the earlier report (Zale et al., 2016) was also included as well as non-transgenic sugarcane of cultivar CP88-1762 (wild-type [WT]). The mature sugarcane stem of primary lines or WT was cut into individual node segments for initiation of vegetative progenies and one node per lines was planted in each pot. Nine months after vegetative propagation, an equal amount of immature, mid-mature and mature sugarcane stem segments from each line were collected for analysis of gene expression, TAG, total fatty acid and FA. During harvesting, total biomass, dry biomass, TAG content of juice, juice volume and Brix were determined as described earlier (Ewanick & Bura, 2011;Kannan, Jung, Moxley, Lee, & Altpeter, 2018;Zale et al., 2016).

| Analysis of gene expression by realtime quantitative reverse transcription PCR
Stem and leaf samples were collected for RNA extraction from plants 9 months after vegetative propagation. The immature stem samples and first dew lap leaf were selected and approximately 0.1-0.2 g of tissue was collected and flash frozen in liquid nitrogen followed by brief storage at −80°C. Total RNA from different leaf and stem tissue was extracted by using TRIzol reagent (Ambion Life Technologies). About 1 µg of total RNA was used for the synthesis of cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems).
Evaluation of (trans)gene expression was done by quantitative reverse transcription polymerase chain reaction (qRT-PCR) of cDNA by using gene-specific primers designed to amplify a 100-200 bp sized product (Table S3). Sugarcane glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene fragment was amplified as a reference gene for the normalization of transcripts (Iskandar et al., 2004).
Quantitative real time RT-PCR was carried out in Real Time PCR detection system (CFX connect; Bio-Rad) with SsoAdvanced Universal SYBR green supermix (Bio-Rad) under the following conditions: 95°C for 3 min for denaturation, 40 cycles at 95°C for 10 s and 58°C for 30 s and melt curve analysis at 95°C for 10 s and 65-95°C with 0.5°C increment every 5 s. The relative expression of each gene was calculated by using 2 {Ct (GAPDH)−Ct (transgene)} . The suppression of TGD1 and SDP1 WT gene were calculated by using the formula 2 −∆∆Ct (Livak & Schmittgen, 2001).

| Sampling of leaves and stem tissues for TAG and FA analysis
Transgenic and WT plants having 16-20 internodes 9 months after initiation of vegetative progenies were sampled for TAG and FA analysis. Two tillers per event were selected and three leaf and three stem samples were sampled from each tiller with a total of 12 samples (six leaf and six stem samples) from each biological replication/plant. For sampling of leaves, +1 (dewlap leaf), +3 and +5 leaves were sampled from each tiller. For stem sampling, second most mature internode, eighth to tenth internode and second youngest, visible internode were selected from two different tillers and ground in liquid nitrogen using pestle and mortar and freezedried in a lyohpilizer (Labcono). The lyophilized sample was used immediately for lipid analysis following shipment to Brookhaven National laboratory on dry ice.

| TAG and FA analysis
About 10 mg of freeze-dried leaf or stem tissue was taken for the analysis of total lipids. Seven hundred microlitres of extraction solvent (methanol/chloroform/formic acid) in the ratio of 2:1:0.1 by volume was used and vigorous mixing was carried out for 3 hr with the use of a vortex mixer. Total lipid extracts were separated by thin layer chromatography using a hexane/ diethyl ether/acetic acid (70:30:1, by volume) solvent system as described earlier (Zale et al., 2016). Lipids were visualized after spraying with 0.05% primuline (in 80% acetone). TAG fractions were identified, excised from the plate and transmethylated into FA methyl esters (FAMEs) by incubation in 1 ml of boron trichloride-methanol at 80-85°C for 40 min. For total FA analysis, total lipid extracts were transmethylated into FAMES by incubation in 1 ml of boron trichloride-methanol directly. For quantification, 5 μg of C17:0 was added as an internal standard before transmethylation. FAMES were extracted into hexane and down before dissolving in 100 μl hexane and analysed by gas chromatography-mass spectrometry with the use of an Agilent Technologies 7890A GC equipped with a 5975C mass-selective detector.

| Lipid staining
The first dewlap leaf tissues from both non-transformed and transgenic sugarcane were cut into 8 × 2 mm pieces. For BODIPY staining, the solution containing 100 µg/ml BODIPY 493/503 (Invitrogen) and 0.1% Triton X-100 (by dilution from a 10 mg/ml dimethyl sulfoxide stock solution) was used to stain fresh leaf tissues as described earlier (Zale et al., 2016). The BODIPY-stained lipid droplets were imaged using a Leica SP5 confocal laser scanning microscope (Leica) with excitation wavelength set at 488 nm. Lipid droplets were visualized at 63× magnification, with the gain set to 824 and 1,200 for BODIPY stain and chlorophyll respectively.

| Southern blotting
Total genomic DNA was extracted from leaf tissue using a modified cetyl trimethyl ammonium bromide protocol (Murray & Thompson, 1980). Twenty micrograms (20 µg) of genomic DNA was digested with PmeI (New England BioLabs). The digestion products were separated by electrophoresis overnight on 0.8% agarose gel with 1× tris-acetate-EDTA buffer at low voltage and transferred onto Hybond-N + nylon membranes (GE Healthcare Biosciences) overnight with 10× saline-sodium citrate (SSC) buffer. The membranes were rinsed with 6× SSC for 5 min, air-dried and exposed to UV light in a crosslinker (Select™XLE-Series, Spectroline ® ). Probes amplified by ZmUbiP-SB-F and DGAT-SB-R primers (Table S3) were labelled using 32 P-dCTP (Perkin Elmer Inc.) with a Prime-It II Random Primer Labeling Kit (Stratagene Inc.; Wu et al., 2015). Prehybridization, hybridization and washing were performed following the manufacturer's instructions. The membranes were exposed to X-ray film for 2 days and visualized by autoradiography.

| Determination of cell wall composition of oilcane bagasse
Dried biomass samples were ground in a Retsch cross beater mill (model SK100; Retsch GmbH & Co. KG) equipped with a 2 mm round whole screen. Ground samples were stored in sealed plastic bags in refrigerator at 4°C. Structural carbohydrates and lignin in the bagasse samples were determined using National Renewable Energy Laboratory (NREL) standard laboratory analytical procedures (Sluiter et al., 2011). Before carbohydrate analysis, the extractives (both water soluble and ethanol soluble) were removed from the ground biomass samples using a Soxhlet apparatus and a method adapted from NREL technical report number NREL/TP-510-42619 (Sluiter, Ruiz, Scarlata, Sluiter, & Templeton, 2008). Extractive-free biomass samples were analysed for carbohydrate and lignin composition using a two-step acid hydrolysis process. In the first stage, biomass samples were hydrolysed using 72% (w/w) sulphuric acid at 30°C for 1 hr. In the second stage, deionized water was mixed in the samples to dilute reduce acid concentration to 4% and samples further hydrolysis by at 121°C for 1 hr in an autoclave. Sugar concentrations in the liquid were determined using a high-performance liquid chromatography (Aminex HPX-87P column; Bio-Rad) equipped with refractive index detector. Acid-insoluble lignin was determined using gravimetric analysis of filtered solids by drying at 105°C for 4h, followed by determining ash at 575°C. The acid-soluble lignin was estimated by measuring absorbance of the filtered liquid at 105 nm in a UV-Vis spectroscope.

| Statistical analysis
The overexpression of genes and relative suppression of genes were analysed with descriptive statistics and ANOVA using excel and R programming respectively. Means of gene expression, relative gene suppression, TAG and total FA were compared using Fisher's least significant difference test. The correlation of expression/suppression of genes with the TAG content in leaves was analysed by using Pearson correlation coefficients in R programming. A minimum of three independent biological replicates were used for all statistical analyses.

| RESULTS
3.1 | Generation of transgenic sugarcane coexpressing WRI1, DGAT1-2 and OLE1 and/or suppressing SDP1 and/or TGD1 In order to elevate lipid accumulation in stem tissue of sugarcane ectopic expression/suppression of several genes was explored ( Figure 1). To obtain a rapid readout of the impact of alternative combinations and delivery strategies of target genes (WRI1, DGAT1-2, OLE1, CYSOLE1, SDP1 and TGD1; Figure 2) initial analysis focused on TAG accumulation in leaf tissue.
Biolistic co-delivery of two unlinked vectors carrying constitutive expression cassettes of nptII, DGAT1-2, OLE1 and RNAi suppression cassette of TGD1 with 15 shots resulted in 41 independent, transgenic, nptII-positive plants. Four of these lines co-expressed/suppressed all the transgenes resulting in TAG levels of up to 1.54% of DW in leaf tissue of line 109 (Table S1).

DGAT1-2, OLE1, WRI1
and RNAi suppression cassette of TGD1 with 20 shots resulted in 51 independent, transgenic, nptII-positive lines. Four of these lines co-expressed/suppressed all the transgenes resulting in TAG levels of 2.35% of DW in the leaf tissue of line 322 (Table S1).
Biolistic delivery of the linearized fragment of the same single large multigene construct with constitutive expression cassettes of CYSOLE1, DGAT1-2, WRI1 and RNAi suppression cassettes of TGD1 and SDP1 with 10 shots resulted in 31 independent, transgenic, nptII-positive lines. Five of those lines co-expressed CYSOLE1, DGAT1-2, WRI1 and suppressed SDP1 with highest TAG levels detected in line 1580 with 8.1% of DW in leaf tissue (Table S1).

| Pearson correlation between the levels of transgene expression/target gene suppression and TAG accumulation in stem or juice
Pearson correlation of TAG accumulation in stem tissues or juice with the level of gene expression/suppression was

| Analysis of sugar and biomass accumulation of transgenic sugarcane progenies under replicated greenhouse conditions
The sugar content of the extracted stem juice was estimated by measuring the Brix value of the juice. The Brix value of the transgenic lines was 11%-20% lower than the non-transgenic control (WT;

| Pearson correlation between stem or juice TAG content with biomass and sugar accumulation and juice volume
Triacylglycerol accumulation in stem tissues or extracted juice was evaluated for Pearson correlation with biomass and sugar accumulation and juice volume. Stem TAG content was positively correlated with TAG content of juice (.61) and negatively correlated with dry biomass (−.85) and stem juice volume (−.69). Dry biomass accumulation was negatively correlated with WRI1 (−.85) and DGAT1-2 expression (−.49) but positively correlated with SDP1 expression (.75). Stem juice volume was positively correlated with total dry biomass (.80) and SDP1 expression (.71) and negatively correlated with stem TAG (−.69) and juice TAG (−.48) content, WRI1 (−.73) and DGAT1-2 (−.48) expression (Table 2). Brix value was negatively correlated to the TAG content of the juice (−.73) and CYSOLE1 (−.6) expression (Table 2).

| Cell wall composition of oilcane bagasse
Cellulose and hemicellulose contents in the oilcane bagasse were calculated based on the monomeric sugars released after acid hydrolysis. Cellulose was the main component in the extractive free bagasse samples, followed by hemicellulose and lignin in all samples (Table 3). Cellulose content of the bagasse was lowest for non-transgenic sugarcane (WT, 42.85%) and ranged from 43.03% (line 327) to 46.10% (line 109) in oilcane lines, with the latter being significantly different from WT (Table 3). The hemicellulose content of bagasse was negatively correlated with stem TAG content (−.68; Table S2). The hemicellulose content of bagasse was up to 9% (line 109) higher in transgenic oilcane compared to non-transgenic sugarcane (WT). One line (line 1565) with highest stem TAG content (1.6% of DW) in this subset of lines displayed 5% reduced hemicellulose content (Table 3). The amount of insoluble lignin did not differ between transgenic oilcane plants ( Table 3). The soluble lignin content was significantly decreased by 44%-59% in transgenic lines (Table 3). The amount of insoluble lignin in transgenic lines was not significantly different from WT (Table 3).

| Visualization of lipid droplets
Large lipid droplets were visualized by boron-dipyrromethene (bodipy) dye staining in the leaf tissue of transgenic oilcane plants from both line 1580 and 1566 using confocal microscopy ( Figure 4). Lipid droplets of similar size were not observed in non-transformed sugarcane (WT; Figure 4).

| Southern blot analysis
Southern blotting was performed to evaluate the integration of the large multigene construct into the genomic DNA of sugarcane following delivery to transgenic lines as a linearized fragment. The genomic DNA from transgenic lines 1565, 1566, 1579 and 1580 displayed a hybridization product of more than 23 kb following restriction digestion with PmeI, gel electrophoresis,and blotting and hybridization with a probe from the DGAT1-2 ORF ( Figure S4). Plasmid DNA of the large multigene petross-2 construct displayed a hybridization signal of >20 kb following the same procedure ( Figure S4). The genomic DNA from non-transgenic sugarcane (WT) did not show any hybridization signal following the same procedure ( Figure S4).

| DISCUSSION
Metabolic engineering for elevating TAG content of vegetative tissues of high biomass crops like sugarcane has been proposed as a transformative concept to boost production of lipids (Ohlrogge & Chapman, 2011). A step change in TAG accumulation in stem tissue of sugarcane was demonstrated here, achieving an average of 4.3% and a maximum of 9.4% of DW in a replicated greenhouse experiment. This was achieved by multigene engineering targeting FA synthesis, TAG synthesis and TAG protection from hydrolysis. Accumulation of TAG in leaf tissues has been successful in model plants, including A. thaliana (Fan et al., 2013(Fan et al., , 2014Winichayakul et al., 2013), N. tabacum (Andrianov et al., 2010;Vanhercke, El Tahchy, et al., 2014;, N. benthamiana (Reynolds et al., 2015;Vanhercke et al., 2013) and B. distachyon (Yang et al., 2015) and high biomass crops including sugarcane (Zale et al., 2016), and sorghum (Vanhercke et al., 2019). However, leaf tissues represent only a minor fraction (typically less than 30%) of the biomass of tall C4 grasses. Therefore, accumulation of TAG in stem tissues is very desirable although reported stem TAG values were below 1% of DW in sugarcane (Zale et al., 2016) and Vanhercke et al. (2019) reported less than 3% of total lipids in sorghum stems. Pearson correlation of transgene expression and target gene suppression with TAG content in sugarcane stems revealed a significant effect of RNAi-suppression of the TAG lipase SDP1 which was not explored previously in sugarcane (Zale et al., 2016)  stacking of an RNAi construct of SDP1 into tobacco lines expressing WRI1, DGAT1 and OLE1 doubled the TAG content of tobacco leaves and increased the TAG content of tobacco stems more than sevenfold from less than 1%-7.4% by DW (Vanhercke et al., 2017;Vanhercke, El Tahchy, et al., 2014;. Earlier, Kelly et al. (2013) reported that disruption of TAG hydrolysis with a mutant of lipase SDP1 leads to a greater accumulation of TAG in stems than leaves. Introgression of the SDP1 mutation into a line with constitutive WRI1 and DGAT1 expression more than doubled the TAG content to almost 7% of stem DW in Arabidopsis (Kelly et al., 2013).
The level of transgene expression/target gene suppression likely influences TAG accumulation. Quantitative real-time PCR of transcripts revealed that we were able to suppress the SDP1 gene to 15% of the levels found in non-transgenic sugarcane stems. The expression level of WRI1 (up to 0.164 that of GAPDH) was almost twofold higher than that reported by Zale et al. (2016). The transgene performance in this study may have benefitted from our use of insulators in the construct design, protecting the transgene expression cassettes from position effects by shielding the construct from the negative influence of the neighbouring chromatin (Pikaart, Recillas-Targa, & Felsenfeld, 1998;Recillas-Targa, Bell, & Felsenfeld, 1999;Singer, Liu, & Cox, 2012). Complete knockout of SDP1 may have additional benefits for TAG accumulation and could be achieved by targeted mutagenesis as recently demonstrated for a high copy gene (caffeic acid O-methyltransferase) in sugarcane Kannan et al., 2018).
The source of the WRI1 gene may also have contributed to the elevation of TAG accumulation in sugarcane stems in the present study. The sorghum WRI1 gene used in this study displayed a higher Pearson correlation (.96) with TAG accumulation than the previously used WRI1 from A. thaliana (.65; Zale et al., 2016). Using the transcription factor WRI1 from a closely related species may result in stronger activation of target genes involved in FA biosynthesis in sugarcane.
Pearson correlation of sugarcane juice TAG content suggested that CYSOLE1 (.72), WRI1 (.66) expression and SDP1 suppression (−.49) but not OLE1 expression (−.31) have a strong impact on juice TAG content. Overexpression of the synthetic CYSOLE1 protein (Winichayakul et al., 2013) with additional cysteine residues instead of the OLE1 used by Zale et al. (2016) may produce stronger intermolecular disulphide bonds which may reduce TAG turnover by reducing access to lipases. Winichayakul et al. (2013) reported an increase of TAG content and stabilization of lipid droplets in mature leaves of A. thaliana by overexpressing CYSOLE1.
Biolistic delivery of the multigene construct as a linearized construct may have contributed to its integration into the sugarcane genome as functional unit. Southern blotting indicated the presence of the delivered fragment in the high molecular weight genomic DNA fraction in several transgenic lines with elevated TAG accumulation. Fan, Zhai, Yan, and Xu (2015) proposed that the suppression of TGD1 increases the extraplastidial FA pool available for TAG assembly by compromising re-entry of lipids to the chloroplast. We observed a wide range of TGD1 expression following co-introduction of a TGD1 RNAi construct with WRI1 and other expression cassettes. TGD1 expression showed a significant Pearson correlation (.50) to WRI1 expression. This may suggest that a highly elevated extraplastidial FA pool may further upregulate TGD1 expression requiring a more stringent RNAi approach. Design of alternative hairpin constructs or targeted mutagenesis of TGD1 may result in a stronger TAG accumulation than observed with the introduced TGD1 RNAi construct in this study.
Fatty acid composition in both leaf and stem tissues were also studied. Transgenic sugarcane stem showed significantly decreased level of FA palmitate (16:0), stearate (18:0) and α-linolate (18:3), while oleate (18:1) was increased significantly as compared to non-transformed sugarcane. The higher amount of unsaturated FA (particularly OA) in vegetative tissues of sugarcane may produce biodiesel with improved oxidative stability and cold flow (Graef et al., 2009). Plastidially localized beta-ketoacyl-ACP synthase II determines the proportion of 16 carbon to 18 carbon FAs (Pidkowich, Nguyen, Heilmann, Ischebeck, & Shanklin, 2007) and the proportion of saturated to unsaturated FAs (FatB acyl-acyl carrier protein [ACP] thioesterase; Salas & Ohlrogge, 2002 and stearoyl-ACP desaturase;Carlsson, Yilmaz, Green, Stymne, & Hofvander, 2011;Shanklin & Somerville, 1991). Similar increase in unsaturated FA by elevating TAG accumulation by overexpressing WRI1 and DGAT1-2 or overexpression of phospholipid: diacylglycerol acyltransferase in a TGD1 mutant background was observed in Arabidopsis (Fan et al., 2013(Fan et al., , 2014 and sugarcane (Zale et al., 2016). The phenylalanine at codon position 469 of maize DGAT1-2 was reported earlier to elevate both lipid and OA content in maize (Zheng et al., 2008) and probably contributed to the increased OA content in sugarcane. The increase in OA in transgenic sugarcane was at the expense of linolenic acid, similar to an earlier report for tobacco (Vanhercke, El Tahchy, et al., 2014;. In contrast, the level of palmitic acid increased in transgenic tobacco while it decreased in transgenic sugarcane. Differences in FA composition after transgene expression in sugarcane or tobacco could either result from differences caused by the level of expression or type and combinations of genes ectopically expressed which should allow to modulate FA composition for desired applications (Vanhercke, El Tahchy, et al., 2014;Vanhercke et al., 2013). the DOE Center for Advanced Bioenergy and Bioproducts Innovation (U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research) under award number DE-SC0018420. Award number DE-AR0000206 supported vector construction, generation of primary transgenic plants and their analysis for transgene integration, transgene expression and lipid accumulation. Award Number DE-SC0018420 supported the analysis of transgenic progeny plants in a replicated greenhouse trial, including transgene expression and lipid accumulation. Any opinions, findings and conclusions or recommendations expressed in this publication are those of the authors and do not necessarily reflect the views of the U.S. Department of Energy. This work was also supported by the USDA National Institute of Food and Agriculture, Hatch project 1020425. Confocal microscopy was performed at the Center for Functional Nanomaterials at Brookhaven National Laboratory. The authors would like to thank Dr. Hardev Sandhu, Everglades Research and Education Center, UF-IFAS for providing sugarcane tops and Sun Gro Horticulture, Apopka, FL for the donation of potting mix. The authors have no conflict of interest to declare.

AUTHORS' CONTRIBUTIONS
S.L., F.A., J.S., H.Z. and V.S. conceived and designed the experiments. F.A., J.S., G.S., H.Z. and E.G.R. designed and assembled the recombinant DNA constructs. S.P. generated transgenic sugarcane plants and their vegetative progenies and confirmed the transgenic nature of different lines with PCR and transgene expression with qRT-PCR. S.P., R.K. and B.K. evaluated the transgenic lines in a replicated greenhouse trial, including biomass and sugar accumulation, gene expression analysis, Southern blot analysis and sample preparation for lipid analysis. D.K. and V.S. performed cell