Sesamum indicum Oleosin L improves oil packaging in Nicotiana benthamiana leaves

Abstract Plant oil production has been increasing continuously in the past decade. There has been significant investment in the production of high biomass plants with elevated oil content. We recently showed that the expression of Arabidopsis thaliana WRI1 and DGAT1 genes increase oil content by up to 15% in leaf dry weight tissue. However, triacylglycerols in leaf tissue are subject to degradation during senescence. In order to better package the oil, we expressed a series of lipid droplet proteins isolated from bacterial and plant sources in Nicotiana benthamiana leaf tissue. We observed further increases in leaf oil content of up to 2.3‐fold when we co‐expressed Sesamum indicum Oleosin L with AtWRI1 and AtDGAT1. Biochemical assays and lipid droplet visualization with confocal microscopy confirmed the increase in oil content and revealed a significant change in the size and abundance of lipid droplets.


| INTRODUCTION
Plant triacylglycerols (TAGs), more commonly known as vegetable oils, have a variety of uses ranging from human consumption to biodiesel production (Durrett et al., 2008). Studies have shown that the demand for plant oils for both food and non-food purposes have increased with the rising global population (Gunstone, 2011) and will continue to increase in the next decade (Chapman & Ohlrogge, 2012).
In oilseed plants like Sesamum indicum, storage oil and proteins accumulate during seed maturation within dynamic subcellular compartments called protein storage vacuoles (PSVs) and oil bodies (OBs), the latter also being referred to as oleosomes and lipid droplets (LDs) in plant seeds and non-plant organisms (Deruyffelaere et al., 2015).
OBs and LDs are approximately .5 to 2 μm in diameter and consist of a sphere-like core matrix of neutral lipids, such as TAGs and sterol esters, which are surrounded by a monolayer of amphipathic lipids, primarily phospholipids and sterols (Huang, 2018). The small size of OBs and LDs provides a large surface area-to-volume ratio of lipid monolayer per unit of TAG, and generally contain specific membrane-associated proteins such as oleosins to facilitate lipase binding and lipolysis during germination (Chapman & Ohlrogge, 2012;Deruyffelaere et al., 2015;Hsieh & Huang, 2004;Huang & Huang, 2016;Murphy, 2012).
Oleosins are a family of oil body-membrane proteins that function in proper OB and LD formation and stabilization for long-term lipid storage (Shimada & Hara-Nishimura, 2010). Oleosins are generally Suyan Yee and Vivien Rolland contributed equally to the work. made up of an N-terminal domain, a central hydrophobic domain, and a C-terminal domain (Hsiao & Tzen, 2011). Oleosins are typically not expressed in leaf tissue (Vanhercke et al., 2014), and are instead predominantly expressed in seeds (Jolivet et al., 2004). Lipid reserves are metabolized via the successive events of lipolysis, fatty acid (FA) transport to glyoxysomes, activation of acyl-CoA derivatives, β-oxidation, the glyoxylate cycle, the partial tricarboxylic acid cycle, and gluconeogenesis (Deruyffelaere et al., 2015).
Recent proteomics-and homology-based studies have also led to the identification of several new protein components involved in the formation, maintenance, function and turnover of LDs, specifically in non-seed plant tissue (Pyc et al., 2017;Reynolds et al., 2019;. These protein components include small rubber particle proteins (SRPPs) isolated from Persea americana (avocado) mesocarp LDs, caleosins, and LD-associated proteins (LDAPs) in Arabidopsis thaliana (Horn et al., 2013). There have also been a number of other proteins found to be frequently associated with LDs in the seeds of plants, including proteases, phospholipases, lipoxygenases, and lipases Rudolph et al., 2011).
Given the breakthroughs in elucidating and manipulating the lipid biosynthesis pathway in plants over the last 10 years, producing oil in the vegetative tissues of high biomass crops has never been a more attractive alternative to the conventional approach of producing oil in seeds. Several parts of the plant lipid biosynthesis pathway have been successfully targeted using various approaches. Strategies with a focus on down-regulating or overexpressing single genes or multiple gene combinations involved in FA and TAG metabolism have been investigated (Dong et al., 2014;Eastmond, 2006;Meyer et al., 2012;Reynolds et al., 2015;Vanhercke et al., 2013;Vanhercke et al., 2014;Vanhercke et al., 2019;Zhang et al., 2018). Examples of these approaches include the upregulation of lipid biosynthesis by overexpressing transcription factors found in seed tissue, like LEAFY COT-YLEDON2 (LEC2) and WRINKLED1 (WRI1), upregulating oil accumulation pathways via diacylglycerol acyltransferase (DGAT) overexpression, and minimizing lipase-mediated catabolism of seed oil through the silencing of TAG lipases, such as SUGAR-DEPENDENT LIPASE1 (SDP1) (Ohlrogge & Chapman, 2011;Eastmond, 2006;Kong & Ma, 2018;Baud et al., 2009;Kim et al., 2012;Vanhercke et al., 2019). Such metabolic engineering strategies rely on the simultaneous "push" to favor the production of FAs through the upregulation of WRI1, the "pull" towards conversion into TAGs via DGAT1, and the "protection" of newly synthesized TAGs from degradation or oxidation by downregulating breakdown pathways and/or expressing protective LD coat proteins, like OLEOSIN during storage (Vanhercke et al., 2014).
Several studies have focused their attention on WRI1 and DGAT1, as their roles in the biosynthetic pathway of plant oils have been well characterized. Similarly, these critical genes have been shown to impart a significant synergistic TAG accumulating effect when co-expressed in various plant models and crops Liu et al., 2017;Vanhercke et al., 2017;Vanhercke et al., 2019). The WRI1 transcription factor is a key transcriptional regulator of FA biosynthesis in both seed and non-seed tissue (Baud et al., 2009;Cernac & Benning, 2004;Deng et al., 2019;Kong & Ma, 2018;Ma et al., 2013), whereas DGAT1 acts as a gate keeper to the committed metabolic step towards TAG production by catalyzing the conversion of diacylglycerols and fatty acyl CoA substrates into TAGs Ståhl et al., 2004).
Various groups have also demonstrated the capacity to stably accumulate these TAGs in non-seed tissue of a variety of species, such as Nicotiana tabacum, Nicotiana benthamiana, Sorghum bicolor, and Solanum tuberosum (potato) tubers through the co-expression of AtWRI1, AtDGAT1 and an intron-interrupted oleosin from sesame Liu et al., 2017;Vanhercke et al., 2017;Vanhercke et al., 2019). However, few studies have investigated the effects of other S. indicum oleosin isoforms on the accumulation and stability of LDs in leaf tissue. Similarly, lipid droplet proteins from other high oil-producing plants, such as P. americana (avocado) and Vanilla planifolia (vanilla) have rarely been examined as potential protein targets to enhance TAG accumulation, as well as to improve LD stability in transient transgenic systems.
In this study, we hypothesized that screening a number of LD This resulting plasmid was designated pOIL380.
The genes coding for the S. indicum Oleosin isoform L (Tai et al., 2002) (GenBank: AF091840) and the S. indicum Oleosin isoform H1 LD proteins (Tai et al., 2002) (GenBank: AF302807) were also codon optimized for N. benthamiana expression and synthesized as NotI-SacI fragments through the GeneArt gene synthesis platform (ThermoFisher Scientific -AU). These synthesized gene fragments were subsequently cloned downstream of the 35S promoter in pJP3343 using the NotI-SacI restriction sites. The resulting plasmids were designated pOIL382 and pOIL383, respectively.
Finally, the genes encoding the V. planifolia leaf Oleosin U1 (Huang & Huang, 2016) (GenBank: SRX648194) and the P. americana mesocarp Oleosin M lipid droplet protein (Huang & Huang, 2016) (GenBank: SRX627420) were codon optimized for expression in N. benthamiana, synthesized as EcoRI-SpeI fragments via the GeneArt gene synthesis service (Thermo Fisher Scientific -AU), and subsequently cloned downstream of the 35S promoter using the respective restriction sites in the binary vector, pJP3343. These resulting plasmids were then designated pOIL386 and pOIL387, respectively.
Binary expression vectors containing AtWRI1 and AtDGAT1 were previously described (Vanhercke et al., 2013). Each of these expression constructs were transformed into Agrobacterium tumefaciens strain AGL1 for transient expression in plant tissue.
Assembled plasmid maps and sequences of pOIL380, pOIL382, pOIL383, pOIL386 and pOIL387 containing their respective codon-optimized oleosin genes are supplied as Supplementary Materials.

| Transient expression assays in N. benthamiana
Transient expression in N. benthamiana leaves was performed as previously described (Wood et al., 2009), with some minor modifications.
Specifically, A. tumefaciens AGL1 cultures containing the plasmids coding for the p19 viral suppressor protein and the LD gene(s) of interest were mixed such that the final OD 600 of each culture was equal to .125 prior to infiltration (Vanhercke et al., 2013). Samples being compared were randomly located on the same leaf. Infiltrations for each sample were repeated across three different leaves located on independent plants, acting as separate biological replicates. The infiltrated N. benthamiana plants were then grown for a further four days before leaf discs were harvested. For biochemical analyses, leaf discs were harvested and pooled from across the three infiltrated leaves, freeze-dried, weighed and stored at À80 C. For confocal microscopy analyses, fresh leaf discs from the same infiltration spot as those used for biochemical analyses were imaged within 30-45 min of harvesting (see below).

| Confocal imaging of lipid droplets
Lipid droplets located in the spongy mesophyll of fresh N. benthamiana leaf discs were imaged four days post-infiltration, as follows. The abaxial epidermis was peeled off in 50 mM PIPES pH 7.0 immediately after collection and discarded. One half of each peeled disc was stained for 10 min in 2 μg/ml BODIPY505/515 in 50 mM PIPES pH 7.0, followed by 2-3 washes in 50 mM PIPES pH 7.0.
During this time, the other half of the leaf disc was kept in 50 mM PIPES pH 7.0. Peeled leaf discs were then mounted in 50 mM PIPES pH 7.0 and imaged immediately, using a Leica SP8 Laser-Scanning Confocal Microscope (Leica Microsystems AG, Germany), a 20x objective (NA = .75), and the Leica LAS X software (Leica Microsystems AG, Germany). Lipid droplets and chloroplasts were imaged by exciting leaf discs with a 505 nm laser. BODIPY 505/515 signal was collected between 510 and 540 nm, whereas chloroplast signal was collected between 650 nm and 690 nm. Unstained half discs were imaged with the same settings to determine tissue autofluorescence.

| Statistical analyses
All statistical tests, including Student's t-tests, two-way ANOVAs and two-way MANOVAs with subsequent Dunnett's tests as post-hoc analyses were performed using R (version 3.6.1) loaded onto R Studio (version 1.2.5001).

| RESULTS
3.1 | Effect of transiently expressed AtWRI1, AtDGAT1, and lipid droplet proteins on TAG and FA composition To determine which lipid droplet protein best protects lipid accumulation, we used a N. benthamiana transient assay to express each one of the five selected LD protein constructs in a p19 + AtWRI1 + AtDGAT1 (PWD) background (Figure 1). A highly significant increase in TAG content in leaves transiently expressing SiOleosinL was observed. This was equivalent to a 2.3-fold increase compared with the p19 + AtWRI1 + AtDGAT1 (PWD) control, and a 122-fold increase compared with the p19 only control. In contrast, no significant increase in TAG levels was observed with the coexpression of SiOleosinH, VpOleosinU1, PaOleosinM, and RoTadA lipid droplet proteins compared with the PWD background ( Figure 1).
Interestingly, the co-expression of both S. indicum oleosin isoforms, SiOleosinL + SiOleosinH also did not result in an increase in TAG content ( Figure 1). The significant increase in oil content observed with SiOleosinL was accompanied by a modification of the FA profile ( Figure 2). The co-expression of SiOleosinL with AtWRI1 and AtDGAT1 significantly increased C18:1 levels (22.3 AE .70%, Figure 2c), while decreasing C16:0 content (23.2 AE .31%, Figure 2a) compared with the PWD control, which displayed C18:1 and C16:0 levels as 15.9 AE .64% and 27.6 AE 1.16% of the total FA profile, respectively (Figure 2c and a, respectively). Interestingly, VpOleosinU1 also significantly increased C18:1 levels and decreased C16:0 levels compared with the PWD control, albeit not significantly increasing overall TAG content compared with the PWD control ( Figure 1). All other gene combinations tested did not show a significant effect on TAG content or FA profiles compared with the PWD control.

| Modification of lipid droplet size and abundance
Next, we used confocal microscopy to test the effect of SiOleosinL and SiOleosinH on the size of accumulating lipid droplets in leaves.
SiOleosinL expressed in a PWD background showed an accumulation of smaller lipid droplets (Figure 3c and c 0 ) than in the PWD control ( Figure 3b and b 0 ). Using identical settings, LDs were barely detectable in the p19 control (Figure 3a and a 0 ), confirming the results described in Figure 1. In contrast, the lipid droplets in leaves expressing PWD + SiOleosinH (Figure 3d and d 0 ) were larger and looked similar to F I G U R E 1 Triacylglycerol content in Nicotiana benthamiana leaves transiently expressing AtWRI1, AtDGAT1, SiOleosinL, SiOleosinH, vanilla leaf VpOleosinU1, avocado mesocarp PaOleosinM, and Rhodococcus opacusTadA lipid droplet protein. Error bars denote standard error with n = 3. **Significantly different at p < .01 against the p19 control (mean TAG content of .02% DW). *Significantly different at p < .01 against the p19 + AtWRI1 + AtDGAT1 (PWD) control those observed in leaves expressing the PWD control. Finally, when SiOleosinH and SiOleosinL were co-expressed with PWD ( Figure 3e and e 0 ), the lipid droplets were small in size and looked similar to those observed in leaves expressing PWD + SiOleosinL (Figure 3c and c 0 ). In leaves expressing PWD + VpOleosinU1 (Figure 4b and b 0 ), the signal coming from lipid droplets appeared less punctate than in the PWD control leaves (Figure 4a and a 0 ).

| DISCUSSION
Earlier work focusing on the co-expression of AtWRI1 and AtDGAT1 has illustrated a significant increase in the levels of TAG produced transiently in N. benthamiana leaf tissue (Vanhercke et al., 2013). However, only a few studies have expanded on this system by examining opportunities to better package newly synthesized TAGs and "protect" them from breakdown. In this study, we have investigated several oleosins for their potential to improve TAG packaging in non-seed tissue using the existing AtWRI1 and AtDGAT1 system as a baseline for improvement.

| SiOleosinL and SiOleosinH have different effects on TAG accumulation and lipid droplet size
The main finding of this study is that the co-expression of SiOleosinL with AtWRI1 and AtDGAT1 genes significantly increased TAG content by 2.3-fold compared with the expression of AtWRI1 and AtDGAT1 alone (Figure 1). This increase in TAG content was also correlated with a decrease in the size of accumulating lipid droplets ( Figure 3c and c 0 ). This suggests that SiOleosinL has a dual effect: it promotes TAG accumulation, and it reduces the size of the lipid droplets. In contrast, SiOleosinH co-expressed with AtWRI1 and AtDGAT1 did not result in a significant increase in TAG content (Figure 1), nor F I G U R E 4 Visualization of lipid droplets in fresh Nicotiana benthamiana leaf transiently co-expressing p19 + AtWRI1 + AtDGAT1 (PWD) alone (a and a 0 ) or in combination with VpOleosinU1 (b and b 0 ). The abaxial epidermis was peeled off, and the spongy mesophyll cells were imaged. Chloroplasts are shown in magenta and lipid droplets were stained with the neutral lipid stain BODIPY505/515 and are shown in green. Each image is a maximum projection of several images taken along the z axis, at intervals of 2 μm. Scale bars: 50 μm did it reduce the size of the LDs produced transiently in N. benthamiana (Figure 3d and d 0 ) compared with the PWD control ( Figure 3b and b 0 ). This demonstrates that both S. indicum oleosin isoforms affect LDs differently.

Interestingly, the co-expression of both SiOleosinL and
SiOleosinH together lead to a hybrid effect: TAGs did not accumulate significantly (Figure 1), but the LDs were reduced in size compared with those in AtWRI1 and AtDGAT1 controls (Figure 3b and b 0 ). This suggests that the mechanisms controlling LD size and TAG accumulation are likely to be uncoupled, and that when SiOleosinL and SiOleosinH are co-expressed, the L isoform has a dominant effect on size, whereas the H isoform has a dominant effect on TAG accumulation. These results suggest that SiOleosinL and SiOleosinH may regulate LD stability and degradation differently.

| SiOleosinH may be affected by ubiquitination
Indeed, some oleosin isoforms contain C-terminal domain ubiquitination sites, which allow for the regulation of LDs via ubiquitin-dependent degradation pathways (Hsiao & Tzen, 2011;Tai et al., 2002). Ubiquitin-dependent degradation of oleosins or LDs containing oleosins has previously been reported in sesame seedlings (Hsiao & Tzen, 2011). Ubiquitination is a post-translational modification in which ubiquitin chains or single ubiquitin molecules are appended to target proteins, thereby affecting protein longevity, as well as protein activity and/or localization (Guerra & Callis, 2012). More specifically, ubiquitination was shown to control the fate of oleosins, as well as regulate lipid droplet dynamics in plants (Deruyffelaere et al., 2015).
Interestingly, SiOleosinH has three predicted ubiquitination sites at residues 130, 143, and 145 ( Figure S1), whereas SiOleosinL lacks any ubiquitination sites (Hsiao & Tzen, 2011). This observation is compatible with the results we have presented in this study, and we hypothesize that the ubiquitination sites on SiOleosinH may counteract the positive stabilizing effects that SiOleosinL imparts on TAG accumulation in N. benthamiana leaf tissue. Although protein ubiquitination is integral to many biological pathways, such as proteasomal degradation, stress responses, hormone biosynthesis and signaling, morphogenesis, and battling pathogens (Sorokin et al., 2009), LD instability is not desired when trying to increase lipid accumulation in vegetative tissue. Although beyond the scope of this study, it will be interesting to further investigate the functions of the SiOleosinH ubiquitination sites and the effects that variably ubiquitinated SiOleosinH mutants may have on TAG accumulation.
The limited amount of oil produced in leaves transiently expressing these constructs suggests that these experiments would be better suited to testing in stable transformants. and SiOleosinL were all co-expressed, there was no rapid degradation of radiolabeled TAG observed in these leaves (Figure 5d), albeit the initial amount of TAG present at 5 min was moderate compared with the expression of AtWRI1 and AtDGAT1 alone. This suggests that the co-expression of SiOleosinL in the AtWRI1 and AtDGAT1 background was able to prevent TAG degradation, demonstrating the oil packaging stability provided by this LD protein.
Further analyses at 3 h post-feeding indicate degradation products that mainly consisted of PC and PA in the leaves expressing SiOleosinL ( Figure 6). Interestingly, the amount of TAG present in these leaves remained largely unchanged over the course of the 3 h assay. As the amount of [ 14 C] acetate used was in excess, the constant amount of TAGs was most likely attributed to the stability of

| Effect of other oleosins on LDs
Another oleosin with interesting characteristics was VpOleosinU1. Here, vanilla leaf oleosin showed an increase in LD abundance, which was distributed in a different pattern compared with SiOleosinL (Figures 3 and   4). This is in coherence with published data on vanilla LD characteristics, where the epidermis of vanilla leaves and most other Asparagales species are known to have clustered LDs (< .5 μm), which is speculated to be needed for cuticle formation (Huang & Huang, 2016).

| CONCLUSION
In summary, each oleosin tested in this study showed a differentiated effect on several aspects of LD production and protection, such as F I G U R E 6 Radiolabeled triacylglycerol, phospholipid, and phosphatidic acid accumulation at 3 h of [ 14 C] feeding in Nicotiana benthamiana leaves transiently expressing p19 + AtWRI1 + AtDGAT1 (PWD) in combination with SiOleosinL, SiOleosinH, vanilla leaf VpOleosinU1, and avocado mesocarp PaOleosinM. Error bars denote standard error with n = 3 increasing total oil content, modifying oil composition and regulating lipid droplet size. We have demonstrated that out of the Oleosins tested in this study, SiOleosinL was best able to package oils produced by the overexpression of AtWRI1 and AtDGAT1 in leaf tissue. This new knowledge can inform future metabolic engineering approaches to meet increasing oil demands through high biomass plant genetics.

CONFLICT OF INTERESTS
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.