Introduction

Plant seeds need a source of fuel to germinate. Once embryogenesis begins, the required chemical energy is released by catabolising fuel stores, which generally consist of starch, proteins, and fats (Waschatko et al. 2016). Triacylglycerides are glycerol esters of fatty acids and are a key energy storage molecule (Murphy 1993). However, because they are insoluble in water, plants store triacylglycerides in oil bodies, which are specialised organelles that provide easy access to the energy rich fats during the germination process.

The membrane of oil bodies comprises a phospholipid monolayer embedded with proteins (Yatsu and Jacks 1972; Fang et al. 2014; Kanazawa et al. 2020), which together envelop the stored triacylglycerides (Fig. 1a). The main protein component of oil bodies are oleosins (Tzen et al. 1993). Although the 3D structure of oleosins is largely unknown, they are predicted to fold into unique structure (Fig. 1b) that contains a central hydrophobic domain flanked by two hydrophilic terminal domains (an N-terminal domain and a C-terminal domain) (Pons et al. 2005). The terminal domain rest on the oil body membrane atop the phospholipid heads, while the central hydrophobic domain (H-domain) extends beyond the membrane and is embedded into the triacylglycerides of the oil body (Jolivet et al. 2017). Within the H-domain is an unusual proline knot motif that is predicted to form a hairpin turn (Fig. 1b) that is conserved across all oleosin sequences (Huang 1996).

Fig. 1
figure 1

a Schematic representation of the oil body structure (triacylglyceride core, phospholipids and proteins of the interfacial membrane). b. Schematic representation of predicted oleosin structure with the proline knot highlighted. c. Cryo-SEM images of the outer endosperm tissues of coconut adjacent to the testa. Magnified to show one cell. Bulbous looking spheres are the oil bodies indicated by arrows. Reproduced from Dave et al. (2019), with permission from Elsevier. d. Cryo-SEM images of the outer endosperm tissues of coconut adjacent to the testa. Multiple cells are in view, arrows show cells that have no oil bodies. Reproduced from Dave et al. (2019), with permission from Elsevier

Full size image

First reported in the late 1980s (Murphy and Cummins 1989), oleosins are vital to the structural integrity of the oil body (Murphy 2001). Due to their amphiphilic nature, however, they are difficult to study and this has slowed research efforts. It has also resulted in conflicting results, especially in the context of the predicted structure (Li et al. 1992, Millichip et al. 1996). Overall, little is known about oleosin structure or function—knowledge gaps that if addressed may inform new applications for oleosins in industry. Here, we will focus on the current state of knowledge for oleosin structure and function with a brief overview of their biosynthesis and examples of how they are being developed for commercial use.

Oil bodies: composition and structure

Oil bodies are small (0.5–2 µm) lipid-based intracellular organelles in plants (Fig. 1c) (Tzen et al. 1993; Huang 1996; Shimada et al. 2008). They are found in high levels in certain tissues, such as in seeds, flowers, pollen, stamen, and fruits (Dave et al. 2019). Functioning as an energy reserve for seeds during germination and post-germinative growth, oil bodies are thought to provide an increased surface area for lipase action during triacylglyceride mobilisation after dormancy or during germination (Huang 1992). Not all cells contain the same number of oil bodies—some can be oil body rich, while others have none (Fig. 1d shows coconut cells, highlighted by arrows, without oil bodies).

Oil bodies consist of a lipid core surrounded by an interfacial membrane of phospholipids and proteins (Dave et al. 2019). The lipid core is largely composed of triacylglycerides, but some bioactive compounds can be found (e.g., vitamin E, carotenoids and phytosterols) (Acevedo-Fani et al. 2020). The phospholipid fraction of the interfacial membrane predominantly contains phosphatidylcholine, which accounts for ~ 50% of the total phospholipids. Minor fractions of other phospholipids can also be found, such as phosphatidylserine, phosphatidylethanolamine and phosphatidylinositol (Huang 1992; Payne et al. 2014). The protein fraction consists of membrane- specific proteins: oleosins, caleosins and steroleosins. The most abundant structural proteins of oil bodies are oleosins, covering most of the oil body’s surface (e.g., in Arabidopsis thaliana oleosins make up 79% of the oil body proteins (Jolivet et al. 2004)). Generally, oil body composition is 94–98% (w/w) triacylglycerides, 0.6–2.0% (w/w) phospholipids, and 0.6–3.0% (w/w) protein (Tzen et al. 1993; Nikiforidis et al. 2014).

The general arrangement of the interfacial membrane components (phospholipids and proteins) is largely known. At the interface, the hydrophobic tails of acyl moieties of phospholipids interact with the lipid core and the hydrophilic head groups face the cytosol (Huang 1992; Tzen and Huang 1992). Oleosins are oriented with their terminal domains atop the phospholipid monolayer. It is thought that positive residues on the N- and C-terminal domains interact with the negatively charged phosphate groups of the phospholipids to support the interfacial structure of oil bodies (Ratnayake and Huang 1996; White et al. 2008). Functionally, oleosins are thought to stabilise oil bodies against coalescence inside plant cells through electrostatic repulsion and steric effects (Frandsen et al. 2001; Maurer et al. 2013). The saturated nature of the fatty acids in the phospholipid fraction may also increase the physical stability of oil bodies; the lack of double bonds allows the fatty acids to be fully extended, and closely packed, promoting a firm anchorage of the oleosins and strengthening the oil body interface (Payne et al. 2014). Recent evidence suggests that the interfacial proteins of coconut oil bodies are disulfide-linked (Dave et al. 2019), which may further contribute to the stability of oil bodies.

Biosynthesis of oleosins and oil bodies

Oleosins are synthesised via the usual protein synthetic machinery, although there are several features to note. The ribosome synthesising the oleosin polypeptide chain is transported to the endoplasmic reticulum via the co-translation synthesis pathway (Fig. 2a) (Hills et al. 1993, Beaudoin et al. 2000, Huang and Huang 2017). During translation, the signal recognition particle (Fig. 2a, in green) binds to the signal region within the developing polypeptide chain of the polypeptide/ribosome complex. The signal recognition particle then binds the signal recognition particle receptor attached to the endoplasmic reticulum membrane (Fig. 2a in orange). This mode of translation embeds the oleosin into the endoplasmic reticulum membrane (Loer and Herman 1993). There are two regions within the oleosin polypeptide chain that may bind the signal recognition particle: the first being a section of the H-domain found close to the N-terminal and the second is the proline knot motif (van Rooijen and Moloney 1995; Abell et al. 1997; Abell et al. 2002; Beaudoin and Napier 2002; Huang and Huang 2017).)

Fig. 2
figure 2

a Schematic of oleosin synthesis. The signal recognition particle (green) binds to the polypeptide as it is being transcribed. The signal recognition particle carries the transcription machinery to the signal recognition particle receptor (orange) found on the endoplasmic reticulum membrane. Transcription of the oleosin then finishes with the H-domain being deposited into the membrane of the endoplasmic reticulum. The endoplasmic reticulum expands as more triacylglycerides are synthesised and more oleosins are added, finally forming an oil body ready to bud off the endoplasmic reticulum. b. Predicted schematic of how oleosin breakdown could occur. Oleosin is first phosphorylated or ubiquitinated, this allows a protease to recognise the oleosin and begin the process of digesting the oleosin. This then allows room for lipase to bind and begin metabolising triacylglycerides

Full size image

The mechanism through which the oil body forms is not fully understood, although the most promising hypothesis is the budding model. Following the budding model, oil bodies have three main steps in their formation: 1) fatty acid synthesis, 2) triacylglyceride assembly and 3) oil body budding (Song et al. 2017) (Fig. 2a). First the fatty acids are synthesised in a plastid using glycerol derived from photosynthesis. They are then moved to the endoplasmic reticulum and assembled into triacylglycerides. As triacylglycerides are being synthesised within the endoplasmic reticulum membrane, oleosins are being deposited into the same region creating a budding oil body that is then detached from the endoplasmic reticulum (Fig. 2a) (Wanner et al. 1981; Hsieh and Huang 2004). The mechanism of release is not understood and there is the question of how oleosins fold into their final structure in the oil body (Sarmiento et al. 1997, Song et al. 2017)—does this occur when deposited into the membrane or during the budding process? Whether oleosins play a role in forming the sharp curvature that facilitates detachment from the endoplasmic reticulum is an open question.

Although the budding model is more widely accepted, a second hypothesis for oil body formation is the post-encasement model. This model suggests that triacylglycerides build-up in the cytoplasm and are only encased in the oil body during the later stages of germination (Murphy 1993). This though does not seem likely as triacylglycerides are extremely hydrophobic and would not ‘linger’ in the hydrophilic environment of the cytoplasm. This theory also doesn’t explain how oleosins are deposited into the membrane, whereas in the budding model oleosins are targeted to the endoplasmic reticulum. These factors make the post-encasement model difficult to rationalise.

Decoding oleosin function

Stabilising oil bodies

It was established early on that oleosins stabilise oil bodies by preventing their coalescence or aggregation (Tzen and Huang 1992), which ensures a large surface area for lipase activity (Huang 1992). As noted, oleosins sit in the membrane of the oil body, with the two hydrophilic termini sitting atop of the phospholipid monolayer, and the H-domain nestled into the oil body. It is thought that both the N- and C-terminal domains and the H-domain play a role in stabilising the oil body.

Early studies demonstrated that oil bodies coalesce when the charge of the oleosins is neutralised, which could be due to the oleosin terminals dissociating from the phospholipids (Tzen et al. 1992). The removal of these domains results in the oil bodies coalescing and bursting, this leads to the idea that the two hydrophilic terminal domains brace the structure of the oil body membrane (Maurer et al. 2013). It is worth noting that most plant species have two different oleosin isoforms, which may have subtle differences in their function (Tzen et al. 1990). These isoforms have different molecular weights (e.g., in maize there is a 16 kDa lower, and an 18 kDa higher isoform), due to a longer C-terminus (Tzen et al. 1990). Both isoforms have been found in oil bodies together (Hsieh and Huang 2004) with the lower isoform being more effective in its function of stabilisation (Tzen et al. 1998).

The length of the H-domain, which is identical in both plant isoforms, is important for the size and stability of oil bodies. Peng et al. (2007) tested this hypothesis by reducing the length of the H-domain in an oleosin and examining how this affected the protein and oil body. The wild type H-domain is predicted to have 30 residues preceding the proline knot followed by a further 29 residues (30r-PK-29r). The study generated five truncated H-domain variants recombinantly: 18r-PK-29r, 18r-PK-17r,18r-PK-5r, 6r-PK-5r, and 0r-PK-0r (i.e., just the proline knot). The N- and C-terminal domains are retained. When compared to the wild type oleosin, the 18r-PK-29r and 18r-PK-17r variants are able to form normal sized, stable artificial oil bodies and prevented coalescence. This is consistent with another protein in the membrane of oil bodies, caleosin, which also has a very similar, but shorter H-domain (18r-PK-18r) (Peng et al. 2007). Oil bodies with the 18r-PK-5r, 6r-PK-5r, and 0r-PK-0r variants, however, show increasing susceptibility to coagulation, especially at elevated temperatures, leading to the conclusion that 18r-PK-17r is the shortest H-domain length required for oleosin stabilisation of oil bodies (Peng et al. 2007). Although this experiment would appear to show how unnecessarily long the oleosin H-domain is, finer truncations of the H-domain may reveal subtleties in the role of symmetry in the residues either side of the proline knot of the H-domain. Perhaps the longer tail and the proline knot play a role for the formation of the oil body in vivo.

Role in germination

Oil bodies have an important role in germination, particularly in the initial stages (Hsieh and Huang 2004; Purkrtova et al. 2008; Quettier and Eastmond 2009; Itabe 2010; Jolivet et al. 2013; Song et al. 2017). In general, oil bodies are metabolised by lipases and the glyoxysome, which is often found close to oil bodies (Hayashi et al. 2001). Lipases catalyse the stepwise hydrolysis of triacylglycerides to diacylglycerides, which is the first step in the gluconeogenic pathway (Lin et al. 1983; Wong and Schotz 2002) and the glyoxysome is a peroxisome and holds many of the enzymes that are involved in breaking down fatty acids to carbohydrates (Beevers 1979, 1980; Chapman and Trelease 1991). How lipase and the glyoxysome interact with the oil bodies is a mystery, however there is evedence that oleosins may be involved in these crucial interactions.

Oleosins are reported to be phosphorylated by a ‘serine, threonine, tyrosine protein kinase’ during germination (Fig. 2b) (Parthibane et al. 2012a, 2012b); Ramachandiran et al. 2018). In Arachis hypogaea (peanut), oleosin (OLE3) has been shown to be part of a complex of proteins that has duel monoacylglycerol acyltransferase and phospholipase A2 activities (Parthibane et al. 2012a, 2012b). The ‘serine, threonine, tyrosine protein kinase’ was shown to bind to peanut OLE3 and phosphorylated predominantly serine residues, particularly Ser18, which is not conserved across oleosins (Fig. 3a) (Parthibane et al. 2012a, 2012b). Similar studies in Arabidopsis thaliana OLE1 found that it was also phosphorylated by a ‘serine, threonine, tyrosine protein kinase’ on Thr166, which is again not conserved (Fig. 3a) (Ramachandiran et al. 2018). The role of phosphorylation may be to recruit proteins, such as proteases, to the oleosin, although these studies raise the possibility that the N- and C-terminal domains may have their own catalytic functions.

Fig. 3
figure 3

a Sequence alignment of oleosins. The green line signifies the approximate N-terminal, the orange line signifies the approximate H-domain, the purple line signifies the proline knot motif, and the pink line signifies the approximate C-terminal, Box one highlights Ser18, box two highlights Thr166. b. Oleosin models of OLE1 A. thaliana generated using AlphaFold, and models of almond, hemp, and sunflower oleosins generated using RaptorX. The proline knot motif of A. thaliana is highlighted with the conserved residues indicated

Full size image

Thiol-proteases are reported to degrade oleosins from the oil body, allowing lipases access to the oil body for triacylglyceride digestion (Fig. 2b) (Sadeghipour and Bhatla 2002; Vandana and Bhatla 2006). Tracking the abundance of oleosins throughout germination using SDS-page analysis demonstrated that the lowest molecular weight sunflower oleosin disappears and this coincides with the increasing activity of a 65-kDa thiol-protease (Vandana and Bhatla 2006). Zymographic analysis demonstrated that the protease interacts with the oil body and that the protease could degrade oleosins when it was isolated with oil bodies. When the concentration of the protease is increased, all oleosins in the oil body were removed regardless of the isoform (Vandana and Bhatla 2006).

Similarly, oleosins, along with other proteins, may be removed from the oil body via the ubiquitination pathway (Fig. 2b) (Hsiao and Tzen 2011; Deruyffelaere et al. 2015). Removal of oleosins via ubiquitination is thought to occur as the first step in the germination process. Oleosins isolated from seeds during germination, and analysed by protein mass spectrometry, were found to contain ubiquitin. This was supported by immunological detection with antibodies against both oleosin and ubiquitin (Hsiao and Tzen 2011; Deruyffelaere et al. 2015). Higher isoforms of oleosins and caleosin (from Sesamum indicum (sesame) and Arabidopsis thaliana) were found to be ubiquitinated at the C-terminal regions (Hsiao and Tzen 2011; Deruyffelaere et al. 2015). There are lysine residues on the C-terminal domain of the higher isoform that could serve as a site for ubiquitination

A limited structural understanding of oleosins

The domain structure of oleosins is well established based on their amino acid sequence (Fig. 3a). The N- and C-terminal domains are hydrophilic, whereas the middle H-domain is hydrophobic (Huang 1992). The long H-domain, which spans around 68–74 residues, is thought to be the longest hydrophobic stretch of residues found in any protein (Huang 1992; Hsieh and Huang 2004; Huang and Huang 2015). Beyond the primary structure, where oleosin sequences have been accumulating as plant genome sequences are reported, the only structural information for oleosins comes from Fourier transform infrared spectroscopy and circular dichroism spectroscopy, which reports on the proteins secondary structure in the far-UV range (180–230 nm) and on protein tertiary structure in the near-UV range (260–320 nm). However, protein structures can now be predicted from the primary sequence using protein folding software, such as AlphaFold (Jumper et al. 2021) or RaptorX (Xu et al. 2021) with surprising accuracy—here we have considered the AlphaFold model of OLE1 from A. thaliana (Fig. 3b) and used RaptorX to generate oleosin structures from almond, hemp, and sunflower.

The N-terminal domain is roughly 40 residues in length and is predicted to contain both α-helices and β-sheets (Li et al. 1992, Li et al. 1993, Lacey et al. 1998). Despite these findings, the protein folding software AlphaFold predicts the N-terminal to be disordered (Fig. 3b). Similarly, the RaptorX models (Fig. 3b) of the N-terminal domain from almond, hemp, and sunflower oleosins were predicted to be largely disordered, although a short helix is predicted. The C-terminal domain, which is ~ 65 residues in length, is α-helical based on circular dichroism spectroscopy and Fourier transform infrared spectroscopy experiments (Li et al. 1992, Lacey et al. 1998). The C-terminal appears to have positively charged residues spaced periodically throughout the primary sequence (Fig. 3a). These positively charged residues are thought to be on the underside of the helix and interact with the negatively charged phospholipid heads to hold the terminal end to the oil body membrane (Li et al. 1992, Tzen et al. 1992, Lacey et al. 1998). The models from AlphaFold and RaptorX suggest that the C-terminal domain contains an α-helix, but is largely disordered. It may be that the N- and C-terminal domains have a unique fold, or that they require the interactions with phospholipids to correctly fold.

The secondary structure of the H-domain is a bone of contention. Early research carried out on rapeseed oleosins proposed that the hydrophobic domain was made up of antiparallel β-strands based on circular dichroism and Fourier transform infrared spectroscopy data (Li et al. 1992). However, this was quickly contested with contrasting evidence from circular dichroism experiments demonstrating largely α-helical content in the H-domain of sunflower oleosins (Millichip et al. 1996). These studies used different methods of protein purification leading to debate on which method best represents the in vivo structure (Beisson et al. 1996). However, there have been further reports since of α-helix content in safflower and sunflower seed oleosins (Lacey et al. 1998, Alexander et al. 2002) and β-strand content in rapeseed oleosins (Li et al. 2002). Whether these differences are due to species or differences in protein preparation remains to be seen.

Those who support the β-strand hypothesis suggest there are two β-strands, one going down from the N-terminus and one coming back up from the hairpin loop to the C-terminus, in an antiparallel arrangement (Li et al. 1992; Tzen et al. 1992; Huang 1996; Li et al. 2002). Li et al. (2002), also predicted that the β-sheets of separate oleosins will interact with each other via hydrogen bonds between the β-sheets. Those who support the α-helix hypothesis propose there are two antiparallel α-helices (Alexander et al. 2002). The α-helical model has the advantage of ensuring the partial charges on the peptide backbone are not exposed to the hydrophobic environment and instead form hydrogen bonds through the helix and that any hydrophilic sidechains can hydrogen bond in the inter-helical space securing the two helices together (Alexander et al. 2002). They further propose that two conserved residues Thr67 and Thr97 are in this inter-helical space and are conserved due to their role in holding the helices together (Alexander et al. 2002). However, a protein sequence alignment (Fig. 3a) shows that these residues are not highly conserved but there are usually threonine residues within the domain. The AlphaFold model suggests that the H- domain is α-helical. Others have generated their own models of oleosins with similar results to AlphaFold (Huang and Huang 2017).

Despite the inconsistencies in the secondary structure of the H-domain, most researchers agree on the importance of the proline knot motif that creates that hairpin turn (Tzen et al. 1992; Hsieh and Huang 2004). All the polypeptide chains of oleosins currently under study have the same three proline residues and one serine residue in the same position (PX5SPX3P) in the middle of the central hydrophobic chain (Hsieh and Huang 2004) (Fig. 3b). The proline knot also looks to be essential in inserting oleosins into the oil body during oil-body formation in the ER (as mentioned above). The conformation of the proline knot is noted in AlphaFold to be difficult to predict; this is likely due to its unusual protein sequence. It is becoming clear that the only way to determine the structure of oleosins is experimentally.

The translation of oleosins in industry

Despite the many gaps in our understanding of oleosin structure and function, oleosins have found application in commercial settings and we highlight some recent examples.

Oleosins have the potential to aid drug delivery. Cancer medications have a reputation of being non- specific (Schilsky 2010; La Thangue and Kerr 2011) and can be hydrophobic which makes drug delivery challenging. Hydrophobic medicines are not easily delivered by oral or intravenous methods due to poor solubility, instability, and low membrane permeability (Porter et al. 2007; Savjani et al. 2012; Cho et al. 2018). Some have exploited oleosin stabilised oil bodies to create easier and more effective methods for delivering cancer medications to their intended site. Here, oleosins are a part of an artificial oil body which holds in its centre the hydrophobic medicine meant for treating the cancer (Chiang et al. 2018; Cho et al. 2018). Both Chiang et al. (2018) and Cho et al. (2018) also fused specific signalling proteins to the oleosins to target the oil body and drug to the cancer cells. Chiang et al. (2018) fused an epidermal growth factor receptor targeting motif to the N-terminal domain of the oleosin which targets cells with the epidermal growth factor receptor, commonly found in lung cancer cells. Cho et al. (2018) instead fused to the C-terminal domain of oleosins an immunoglobulin-binding protein, which binds antibodies that could target breast cancer cells. The artificial oil body contained carmustine, which is a hydrophobic cancer drug. Both methods exploited oleosins fused with ancillary proteins that target the oil body and its hydrophobic drug payload to specific cells.

Human fibroblast growth factor (hFGF) has been shown to aid in wound healing and hair growth (Jimenez and Rampy 1999, Braun et al. 2004, Jang 2005, Lin et al. 2015). However, it is difficult to express recombinantly and has poor thermal stability and poor transdermal absorption (Kovacs et al. 2006; Wang et al. 2007). The expression of oleosins fused with hFGF in plants has been suggested as a possible solution (Li et al. 2017, Cai et al. 2018). Studies on oil bodies with oleosins fused to hFGF9 and hFGF10 isolated from safflower seeds found that both proteins were still able to work effectively (Li et al. 2017, Cai et al. 2018). Mice treated with oil bodies containing oleosin-hFGF had improved wound healing and hair growth compared with just recombinant hFGF. The oleosin and oil body were able to stabilise the hFGF and therefore make its application more efficient. Similar studies have used human epidermal growth factor (hEGF), which has similar applications as hFGF (Mroczkowski and Ball 1990; Jahovic et al. 2004; Hee Na et al. 2006) and the harvested oil bodies can be directly given to the patient (Qiang et al. 2020).

Despite a limited understanding of the structure of oleosins, these examples clearly demonstrate an opportunity to revolutionise drug delivery systems. A deeper knowledge of the structure of oleosins will allow us to understand how oil bodies are formed and stabilised. This will inform future engineering efforts, to utilise oil body systems to their full potential.