The production of wax esters in transgenic plants: towards a sustainable source of bio-lubricants

We review the most recent developments in metabolic engineering of plants towards the production of wax esters for industrial applications.


Introduction
Wax esters (WEs) are a class of neutral lipids composed of fatty acids esterified with fatty alcohols. They are widely used in different industrial sectors to produce surface coatings, polishes, printing inks, and candles, as well as for cosmetic and pharmaceutical applications (Fig. 1). In particular, WEs with a low melting point and a high oxidation stability are highly in demand as components of lubricants (Carlsson et al., 2011). In the past, WEs for lubrication purposes were obtained from the sperm whale (Physeter macrocephalus), which can accumulate up to 4 tons of this valuable oil in its skull (Clarke, 1979). As a consequence, the sperm whale became almost extinct and was classified as an endangered species in the early 1970s. The ensuing ban on sperm whale hunting and the import of whale products resulted in an active search for suitable resources for replacing spermaceti oil (Nieschlag et al., 1977). The desert plant jojoba (Simmondsia chinensis), which accumulates WEs instead of triacylglycerols (TAGs) as storage lipids (Miwa, 1971), was considered as a potential substitute. Nevertheless, jojoba is a low yield plant that can only be cultivated on a limited scale in hot and dry climates. Thus, jojoba oil remains expensive and its use is restricted to the cosmetic sector (Carlsson et al., 2011). Nowadays, most WEs are produced through complex chemical processes relying on high energy consumption and fossil fuel resources. In order to reduce dependence on petroleum and to decrease the environmental impact of WE production, the development of so-called green factories for the production of bio-based lubricants represents a very promising alternative that would combine both sustainability and improved biodegradability. This review describes the different strategies that have been developed to produce WEs in plants, the first promising results, as well as possible ways to improve the quantity and quality of the desired end-products.

Wax esters: chemical structure and functions
Although WEs are simply composed of two acyl-chains, a great degree of WE structural diversity exists in nature (see Box 1 for WE nomenclature). While most WEs consist of even-chain saturated and unsaturated fatty acyl and alcohols moieties ranging from C12 to C30 carbons in length (Patel et al., 2001), WEs containing odd-chain or branched components, isoprenoid alcohols or diols have also been described (Holtzapple and Schmidt-Dannert, 2007;Chinta et al., 2016). WEs have been found in organisms from all kingdoms of life (except Archae and fungi) and shown to fulfill several important biological functions. Most commonly, WEs are found in surface lipid layers: as constituents of the cuticle of plants (Li et al., 2008) and the exoskeleton of insects , and as secretions produced by sebaceous glands in mammalian skin (Cheng and Russell, 2004a), they protect from water loss, pathogen attack, and ultraviolet light. In birds, WEs are produced by the uropygial gland for the lubrication, waterproofing, and maintenance of the plumage (Biester et al., 2012b). WEs also serve as energy storage in bacteria belonging to the genera Acinetobacter, Marinobacter, Rhodococcus, and Mycobacterium, in some zooplankton organisms (Lee et al., 2006), and in the phytoflagellate Euglena gracilis, which has the ability to synthesize WEs under anaerobic conditions (Inui et al., 1982;. Interestingly, the plant kingdom also contains a WE-storing species, S. chinensis (jojoba), a desert shrub native to North America, which accumulates WEs in its seeds to sustain post-germinative growth (Miwa, 1971;Sturtevant et al., 2020). Other functions of WEs include buoyancy regulation in marine organisms, such as the sperm whale and copepods (Clarke, 1978;Pond and Tarling, 2011), and chemical communication in insects and birds (Chinta et al., 2016;Grieves et al., 2019).
Among naturally abundant WE sources, the spermaceti oil is a mixture of TAGs and C24-42 WEs, the latter accounting for more than 70% of the total (Nevenzel, 1970). Highly heterogeneous WE compositions within the spermaceti organ have been reported (Morris, 1975) and as many as 240 different WEs were detected by Spencer (1979). The exact WE composition seems to be highly dependent on the specimen Fig. 1. Wax ester production methods and industrial applications. Wax esters are important industrial lipids used as ingredients for the formulation of lubricants, cosmetics, pharmaceuticals, paints, emulsions, and printing inks. They can be produced by chemical synthesis or lipase-catalysed esterification of petroleum products or plant oils. Nowadays, natural wax esters are mainly obtained from jojoba oil, carnauba wax and beeswax.
(its age and diet) and may not be systematically dominated by cetyl palmitate as often believed (Horiguchi et al., 1999). As mentioned above, this source of WEs is no longer available in order to protect sperm whales. The well-known jojoba oil, widely used in the cosmetic and pharmaceutical industries (Sánchez et al., 2016), is mainly composed of very long-chain monounsaturated (20:1, 22:1, and 24:1) fatty acids and alcohols (Miwa, 1971). Different studies have shown that performance of jojoba oil as a lubricant is similar or even superior to that of spermaceti oil (Gisser et al., 1975;Miwa et al., 1979). However, due to the rather high melting point (around 9 ºC), its usage as a lubricant in cold climates is limited (Carlsson et al., 2011). WEs are also obtained from the epicuticular waxes covering the leaves of the Brazilian palm tree Copernicia cerifera (carnauba wax), or the stems of the bush Euphorbia cerifera (candelilla wax). According to Doan et al. (2017), carnauba wax consists of 62% wax esters, which are composed of C16-C24 fatty acids and C18, C30, and C32 fatty alcohols, while candelilla wax contains only 16% WEs with C16 fatty acids and C18 and C30 fatty alcohols as predominant constituents. WEs are also a major component of beeswax (58%), which is extracted from honey combs. Beeswax mainly comprises C16 fatty acids and C24-C32 fatty alcohols (Doan et al., 2017;Moreau et al., 2018). Other sources of WEs include sheep wool max (lanolin), sorghum kernels, sunflower oil, and rice ban oil. Natural WEs are used as ingredients in cosmetics, food products, polishes, and coating agents (Harron et al., 2017;Moreau et al., 2018).

Properties of wax esters and industrial applications
The properties of WEs depend on the chemical structure of their fatty acid and alcohol components (Patel et al., 2001). The two key factors influencing WE utility for industrial applications are the melting point and resistance to oxidation, but the thermal and pressure stability are also important. The melting temperature (T m ) is determined by the carbon chain length, the degree of unsaturation, and the position(s) of the double bond(s). It was shown that for synthetic saturated WEs, the T m increased by 1-2 °C per additional carbon unit, from approximately 38 °C for a C26 WE (dodecyl myristate, 12:0-14:0) to 75 °C for a C48 WE (tetracosanyl tetracosanate, 24:0-24:0). For WE isomers, the T m is affected by the position of the ester linkage, with symmetrical WEs having the highest melting point. Moving the ester bond towards either end of the molecule results in decreasing the T m by 1-5 °C (Patel et al., 2001). Saturated, monoenoic, and dienoic WEs of the same length have a very different T m (Iyengar and Schlenk, 1969;Patel et al., 2001). For example, the presence of one or two double bonds in WEs composed of C18 moieties decreases the T m by approximately 30 °C and 60 °C, respectively (Patel et al., 2001). Moreover, Iyengar and Schlenk (1969) observed that the T m of various monounsaturated WEs with the double bond in the fatty acid chain is 10 °C lower in comparison with their isomers with the double bond in the fatty alcohol moiety. Similarly, the T m determined by Russell and Volkman (1980) for stearyl palmitoleate (18:0-16:1) was lower than that of oleyl palmitate (18:1-16:0), but with a difference of only 1.5 °C. Oxidation stability is mainly correlated with the degree of unsaturation of a lipid molecule. WEs with a higher number of double bonds are therefore more susceptible to oxidation, which can lead to their polymerization and degradation (Kodali, 2002;Fox and Stachowiak, 2007). In contrast, the presence of branched chains lowers the melting point of WEs (Patel et al., 2001) without negatively affecting oxidative stability.
WEs have been used for decades in many different sectors (Fig. 1). Whereas solid WEs are common ingredients of candles and polishes, liquid WEs are often found in printing inks, paints, surface coatings, leather waterproofing treatments, plasticizers, Box 1. Wax ester nomenclature The names of WEs are formed in the same way as the names of other esters. The first part of the name specifies the alkyl residue derived from the alcohol (alcohol residue), and the second part is derived from the acid residue. In literature, both common and IUPAC fatty acid and fatty alcohol names are used. The shorthand notation used in different reports can cause some confusion. Usually, it reflects the WE name: the alcohol residue precedes the acyl residue and they are connected with a hyphen. In the alternative shorthand notation the acyl residue is given first, but then a slash is used instead of a hyphen (Chen et al., 2015). However, in some studies these two notations were mixed: in the traditional notation a hyphen was replaced by a slash (e.g. 20:0/18:2 was used for arachidyl linoleate). Examples of WE names with the shorthand notation format used in this review are given below.

WE common name
Alcohol residue Acyl residue Like other lipids, WEs can be classified according to the length of their acyl-chains. Most common WEs are long-chain WEs, which are composed of C16 and C18 acyl moieties, and very long-chain WEs with >C18 acyl chains. WE containing <C8 and C8-C14 acyl-chains are referred as to short and medium, respectively. and oil solutions for lamps. WEs have also many applications in the food, pharmaceutical, and cosmetic sectors. For example, cetyl octanoate (16:0-8:0) is used in the formulation of different cosmetics, including cleansing products, hair conditioners and makeup removers due to its ability to retain moisture ('Final report on the safety assessment of cetearyl octanoate Kuo et al., 2012b). Nevertheless, in these application sectors, natural sources of WEs are preferred as feedstocks. The very high costs and low yields associated with culturing S. chinensis resulted in using jojoba oil only for dermatological formulations, health care products, and cosmetics. Finally, WEs with high oxidative stability and resistance to hydrolysis have outstanding lubrication properties. Such WEs, therefore, represent excellent components of high-performance factory machine lubricants and automobile transmission fluids (Carlsson, 2006). The sulfurized form of spermaceti oil was indeed considered as an ideal additive in many lubricant applications until sperm whale hunting was banned (Nieschlag et al., 1977).

Current large-scale wax ester production methods
Nowadays, large-scale production of WEs for industrial applications is based on chemical processes using petroleum or plant resources as feedstocks. Alternatively, WEs can be generated by enzymatic synthesis using lipases (Fig. 1). The chemical synthesis of WEs first requires the reduction of a fatty acid to an alcohol, followed by esterification with a fatty acid. Although seemingly simple, this multi-step process is rather expensive, generates waste, and requires high temperatures and pressure, catalysts such as sulfuric acid, tin or titanium, and a complex downstream purification (Lokotsch et al., 1996;Nguyen et al., 2017). Nevertheless, chemical esterification of fatty alcohols with fatty acids enables the production of synthetic WEs with properties similar to spermaceti oil or other natural WEs (Bell et al., 1977;Nieschlag et al., 1977). Lipase-catalysed synthesis, which can be carried out under moderate temperature and pressure conditions in solvent-free systems, has received increasing attention as an attractive alternative to chemical synthesis. The process has lower energy consumption and generates less waste compared with chemical methods (Petersson et al., 2005). In the presence of saturating concentrations of fatty alcohols and under low water content conditions to avoid the reverse reaction, lipases can catalyse alcoholysis/transesterification of oil (TAGs) or esterification of free fatty acids (Vilas Bôas and Castro, 2022). Among a large number of tested WE-synthesizing lipases, enzymes from Candida, Rhizopus, and Rhizomucor species, including the commercially available Lipozyme RMIM (lipase from R. miehei) and Novozym 435 (lipase from C. antarctica), efficiently produced different long-chain esters resembling natural WEs (Steinke et al., 2001;Guncheva and Zhiryakova, 2008;Lopes et al., 2011;Kuo et al., 2012a;Ungcharoenwiwat et al., 2016). Immobilization of lipases allows the reuse of the enzymes and improves their activity and stability (Kuo et al., 2012a). However, despite many advantages, lipase-based synthesis of WEs is still not productive enough to compete with chemical synthesis (Nguyen et al., 2017). It should also be noted that even though lipase-catalysed production of WEs is more environmentally friendly than the conventional chemical process, it still requires fatty alcohols, which are most commonly produced by hydrogenation of plant and animal oils or from petrochemical feedstocks using the Ziegler process or oxo synthesis (Noweck and Grafahrend, 2006). Both methods suffer from harsh conditions, hazardous reagents, and production of chemical waste (Hagström et al., 2013;Munkajohnpong et al., 2020). As pointed out above, there is no readily available source of WEs since the ban on the hunting of spermaceti whales, and current methods for obtaining large amounts of WEs for industrial purposes are not only harmful for the environment but also dependent on diminishing fossil reserves. In coming years, WE production may be limited by growing costs, and thus insufficient to meet the increasing demand. Therefore, there is a strong need for alternative bio-based methods for sustainable production of WEs. Metabolic engineering has enabled the establishment of WE synthesis in bacteria (Kalscheuer et al., 2006), yeast (Kalscheuer et al., 2004;Wenning et al., 2019), and plants (Heilmann et al., 2012;Iven et al., 2016;Zhu et al., 2016;Ruiz-Lopez et al., 2017). Among these WE-producing platforms, plants represent an attractive strategy for renewable, sustainable, and environmentally friendly synthesis of WEs tailored to industrial applications.

Enzymes involved in synthesis of wax esters
The biosynthesis of WEs from acyl-chains is straightforward as it relies on only two consecutive enzymatic activities. First, a fatty acyl reductase (FAR) reduces an acyl-chain to the corresponding fatty alcohol. Next, a wax ester synthase (WS) transfers the acyl group of an acyl-CoA onto the hydroxyl group of the fatty alcohol, yielding a WE (Fig. 2). This pathway was first described in the early 2000s in jojoba (Lardizabal et al., 2000;Metz et al., 2000), and afterwards in many species from bacteria to protists, birds, and mammals.

Fatty acyl reductases
FARs are acyl-thioester reductases (EC 1.2.1.50). They use NAD(P)H as reducing equivalents, and usually generate fatty alcohols from activated acyl-chains . However, some FARs from bacteria and cyanobacteria also produce fatty aldehydes (Reiser and Somerville, 1997;Lin et al., 2013). Pioneering biochemical studies on cell free preparations of etiolated Euglena gracilis cells showed that the fatty alcohol production occurred via an aldehyde intermediate that was not released by the FAR enzyme. Several early studies showed that this activity was associated with the microsomal fraction, and used acyl-CoAs as substrates (Kolattukudy, 1970;Pollard et al., 1979;Vioque and Kolattukudy, 1997). It was later shown that plants possess two types of FARs, which differ in their subcellular localizations and preferred thioester-linked acyl substrate (acyl-CoA or acyl-ACP). Whereas the first type are classic microsomal enzymes residing in the endoplasmic reticulum (Rowland et al., 2006), the second type are soluble plastid-localized proteins that use acyl-ACPs as substrates (Chen et al., 2011;Shi et al., 2011;Doan et al., 2012). These differences have allowed for the engineering of WE production in plants in either seeds or the plastids of leaves (see below).
The protein structure of FARs displays two conserved domains, an N-terminal Rossmann-fold NAD(P)H binding domain, and a C-terminal fatty acyl-CoA reductase (FAR_C) domain. Their N-terminal extremity usually harbors the motif (I/Y/F)-L-(I/V)-(T/V)-G-X-X-T-G-F-L-A, a canonical ADP binding domain most probably involved NAD(P)H binding (Aarts et al., 1997), and the classic YXXXK active site motif of short-chain dehydrogenase/reductase (Kavanagh et al., 2008). Their C-term FAR_C domain is often annotated 'sterile' or 'male sterile' in databases because the Arabidopsis MALE STERILITY2 (MS2) protein was the first publicly available FAR sequence, but at that time with unknown function (Aarts et al., 1997). FARs are usually about 500 amino acid polypeptides, with the plastidial isoforms containing an additional N-terminal targeting signal for chloroplast import (Chen et al., 2011;Shi et al., 2011;Doan et al., 2012). In contrast, mammalian FARs contain an extra C-terminal transmembrane domain that allows for their anchoring to the peroxisomal membrane (Heilmann et al., 2012). A similar hydrophobic region on the C-terminus was identified in AmFAR1 from honey bee , CfFAR2 and CfFAR3 from copepod Calanus finmarchicus (Teerawanichpan and Qiu, 2012), and avian FARs (Hellenbrand et al., 2011).
Interestingly, FARs with both wide and strict substrate specificities have been described. Among bacterial FARs, two FARs from Marinobacter aquaeolei VT8 (referred to as Maqu_2220 and Maqu_2507) have been intensively studied biochemically. Both displayed a broad specificity producing in vitro C10 to C20 fatty alcohols, while Maqu_2220 was additionally shown to reduce fatty acyl-CoAs, fatty acyl-ACPs, and fatty aldehydes to corresponding fatty alcohols (Wahlen et al., 2009;Hofvander et al., 2011;Willis et al., 2011;Liu et al., 2013). Despite the apparent contrasting results obtained in different heterologous host systems (reviewed in Rowland and Domergue, 2012), biochemical studies have generally shown that plant FARs have distinct substrate specificities with clear substrate chain-length preferences. For example, Arabidopsis FAR1, FAR4, and FAR5 mostly use saturated C22:0, C20:0, and C18:0 as substrates, respectively (Domergue et al., 2010). FAR3/CER4 seems to be specific for the production of saturated C24:0 to C28:0 fatty alcohols (Rowland et al., 2006), but the recent characterization of CER17 showed that FAR3/ CER4 is also active on ω6 monounsaturated fatty acids of similar chain length (Yang et al., 2017). Whereas most Arabidopsis FARs preferentially use saturated substrates, the seed-localized jojoba ScFAR mainly produces C20:1 and C22:1 fatty alcohols Miklaszewska and Banaś, 2016). Recently, two Brassica napus CER4 homologs were shown to be specific for reducing branched (anteiso) fatty acids (Liu et al., 2021). Similarly, some insect FARs were shown to prefer monounsaturated substrates and to have a strong specificity toward the configuration (cis/trans) of the double bond (Lassance et al., 2010). Altogether, these studies suggest that FARs with unique substrate specificities in terms of chain length, degree of saturation, and branching may exist in nature to produce tailored WEs. Nevertheless, although crystals were obtained from the plastidial FAR Defective in Pollen Wall (DPW) from rice , no X-ray or NMR structure has yet been reported for any FAR, including non-plant FARs, hampering the understanding of substrate specificity at the amino acid level, and its modification for industrial applications.
Jojoba-type WSs belong to a superfamily of membranebound O-acyltransferases and are related to DGAT-type 1 enzymes. They can be found in various plants, microalgae, and protists, but only a few of them, such as OsWS1 from rice (Xia et al., 2015), EgWS from E. gracilis , and CzWS1 from Chromochloris zofingiensis (Xu et al., 2021), were analysed in detail.

FAR-WS fusion protein
In Tetrahymena species and related unicellular ciliate protozoa, FARs are often found fused to an acyltransferase domain. Dittrich- Domergue and coworkers (2014) showed that in Tetrahymena thermophila this bifunctional peroxisomal protein is involved in the early step of ether lipid biosynthesis: the N-terminal FAR domain produces a fatty alcohol, while the C-terminal domain generates sn-1-acyl-dihydroxyacetone phosphate. These two substrates are then used by a third enzyme, the alkyl-dihydroxyacetone phosphate synthase, to initiate ether lipid biosynthesis. The existence of a protein carrying both FAR and acyltransferase activity suggested that combining both FAR and WS domains in a single polypeptide may permit WE production by expressing a gene encoding a single bifunctional protein. Such a strategy indeed appeared successful for the engineering of WE production in plant seeds and leaves (see below).

Oilseed platforms for wax ester production
The first successful production of WEs in a non-WE-storing oilseed plant was achieved using Arabidopsis (Lardizabal et al., 2000). This proof-of-concept report initiated further studies on the possibility of producing WEs using oilseed crops such as Camelina sativa, Crambe abyssinica, Brassica carinata, and Lepidium campestre, most of which were carried out within the EU 7FP international project ICON (Industrial crops producing added value oils for novel chemicals; https://cordis.europa.eu/project/id/211400). These species were selected as oil crop platforms for WE production for several reasons (reviewed in Carlsson, 2009;Bansal and Durrett, 2016;Samarappuli et al., 2020). Most importantly, they are cultivated only for industrial purposes, which minimizes the risk of admixing into the food oil crops (Carlsson, 2009). In addition, these crops have favorable agronomic properties, interesting seed fatty acid profiles, and can be grown on marginal land (Zhu et al., 2016;Ivarson et al., 2017).

FAR and WS combinations and wax ester yields
To date, different attempts to engineer plants for WE production have been reported. The yields and composition of WEs accumulated in the host species harboring various combinations of WE-synthesizing genes are summarized in Table 1 (see Box 2 for FAR and WS nomenclature).
First studies focused on maximizing WE yields and tailoring WE composition by employing various combinations of FARs and WSs of different origin. The mouse FAR (MmFAR1) and WS (MmWS), which showed preference for 16-18C saturated and unsaturated substrates (Cheng and Russell, 2004a, b), were used for the synthesis of WEs composed of long carbon chains with a maximum of one double bond per alcohol and acid moiety, which are highly suitable for industrial applications (Heilmann et al., 2012). Since MmFAR1 and MmWS localize to different organelles (peroxisomes and the endoplasmic reticulum, respectively), a C-terminal peroxisomal targeting signal was removed from MmFAR1, and both enzymes were fused with Arabidopsis oleosin 3, a lipid droplet protein. Both genes were expressed under the seed-specific napin promoter. Co-targeting MmFAR and MmWS to lipid droplets resulted in 2-fold increase in WE accumulation in Arabidopsis seeds compared with the unmodified enzymes (from 22 to 45 mg g −1 seed), without affecting the WE composition. In both cases, the most abundant WE species were gondoyl linoleate (20:1-18:2), oleyl linoleate (18:1-18:2) and arachidyl linoleate (20:0-18:2) (Heilmann et al., 2012). A further increase in WE yield was obtained (70 mg g −1 seed) when the jojoba WS (ScWS) was used in combination with the major FAR providing fatty alcohols for WE production in M. aquaeolei (MaFAR, Maqu_2220;Hofvander et al., 2011). Accumulated WEs were mainly composed of 18:1 and 20:1 alcohol moieties and 20:1 and 18:1 acyl moieties, with a predominance of oleyl gondoate (18:1-20:1) (Iven et al., 2013).
The efficiency of different combinations of MmFAR, MaFAR, MmWS, and ScWS in producing WEs was then tested in Arabidopsis and C. sativa seeds (Iven et al., 2016). The expression of MaFAR and ScWS genes resulted in the highest levels of WEs, reaching 108 mg g −1 seed in Arabidopsis and 47 mg g −1 seed in C. sativa. In contrast, transgenic lines with oleosin-fused MmFAR combined with MmWS or ScWS produced considerably less WEs (up to 33 mg g −1 in Arabidopsis and up to 21 mg g −1 seed in C. sativa), which suggested that a low efficiency in fatty alcohol formation limited WE synthesis. The overall composition of accumulated WEs in Arabidopsis and C. sativa was similar, but the WE profiles were partially influenced by the substrate specificity of the utilized enzymes. For example, combinations with ScWS incorporated mainly 20:1, whereas WEs produced by MmWS contained mainly unsaturated C18 acyl moieties (Iven et al., 2016).

Tailoring wax ester composition in seeds
Several strategies were developed in parallel not to increase the WE yields, but for producing tailored WEs by modifying the acyl-chain pools used as substrates by FARs and WEs. Approaches targeting either fatty acid synthesis in plastids or different pathways involving endoplasmic reticulum-localized enzymes were tested (Fig. 3).
Accumulation of WEs enriched in very long acyl-chains was first attempted in Arabidopsis using the FAR and WS from jojoba together with a 3-ketoacyl-CoA synthase (KCS or FAE1, a component of the fatty acid elongase complex) from Lunaria annua to increase very long-chain fatty acid content in the acyl-CoA pool (Lardizabal et al., 2000). A similar approach was later used to establish jojoba-like WE production in C. abyssinica, C. sativa, high erucic acid B. carinata (with FAD2-RNAi suppression and C. abyssinica FAE1 overexpression) and L. campestre (Zhu et al., 2016;Ivarson et al., 2017;Li et al., 2019). WEs accumulated in C. abyssinica, C. sativa, and B. carinata, co-expressing ScFAR and ScWS, accounted for approximately 25% of the total oil content. Individual C. abyssinica seeds containing more than 50% of WEs were also identified. WEs accumulated in C. abyssinica and B. carinata were mainly composed of 20:1-FA, 22:1-FA and 22:1-OH, whereas most abundant WEs accumulated in C. sativa consisted of 20:1-FA, 24:0-FA and C22-C24 saturated and unsaturated fatty alcohols. Further modifications of the WE composition were achieved by addition of FAE1 gene from jojoba (ScFAE1) or L. annua (LaFAE1) and FAD2-RNAi suppression, which resulted in increased content of WEs with C24 acyl and alcohol moieties, and decrease in saturated and polyunsaturated components (Zhu et al., 2016). Through direct selection based on WE content, C. abyssinica lines expressing ScFAR/ScWS, ScFAR/ScWS/ScFAE1 and ScFAR/ScWS/CaFAD2-RNAi combinations with highly stabilized WE levels were further developed (Li et al., 2019). Detailed analysis of the seed oil of T 7 generations revealed that the average WE content in these lines was 25%, 18%, and 29%, respectively (Li et al., 2019).
Studies on L. campestre demonstrated that wild species have a high potential as new industrial crops for engineered WE production. Lines harboring ScFAR and ScWS genes accumulated mainly C42 and C44 WEs with levels reaching 44.7 mg g −1 seed. Addition of ScFAE1 led to higher WE amounts (up to 86 mg g −1 seed) with increased content of C46 and C48 species. Interestingly, these C46 and C48 WEs were accumulated in the seed coat rather than in the embryo, whereas other species, such as 42:2 or 44:2 were detected both in the embryo and the seed coat (Ivarson et al., 2017).
Altogether, these results suggest that modifying the acyl-CoA pool composition can tremendously affect the final WE composition, but at the same time, its impact on the WE yields can vary from negative to positive (Table 1). In addition, it seems that reconstituting WE synthesis in mutant backgrounds or by down-regulating competing activities had a stronger effect on the WE composition than overexpressing new activities like thioesterases.

Modifying FAR and WS specificities
Based on to the WE profiles obtained in different plant species and backgrounds with the various gene combinations, it can be concluded that WE composition is mainly influenced by the availability of acyl-CoAs, and to a lesser extent by the substrate specificity of FARs and WSs. However, matching the enzymes' activity to the available substrates increases the likelihood of their efficient incorporation into WEs. The rapidly growing number Box 2. Nomenclature of enzymes used for wax ester synthesis in plants Since various combinations of enzymes derived from different organisms have been used for WE production in plants, a unified nomenclature is important for data comparison. Most of the abbreviations used in the papers discussed in this review follow a two-letter prefix convention with the first letter uppercase and the second letter lowercase (for example, MaFAR-fatty acyl reductase from Marinobacter aquaeoli; ScWSwax synthase from Simmondsia chinensis). Therefore, we adopted this convention here. However, it should be noted that some studies used a different nomenclature to facilitate the analysis of the results (for example in Ruiz-Lopez et al., 2017). Some discrepancies in the literature may also occur in the case of mammalian and bacterial WE-synthesizing enzyme. Although in 2005 it was proposed to use the name acyl-CoA wax alcohol acyltransferase (AWAT) for mammalian wax synthases (Turkish et al., 2005), both abbreviations MmWS and AWAT2 frequently appear for the mouse WS. The variations in abbreviations used for bacterial wax synthases (WS, WS/DGAT, WSD) result from bifunctionality of these enzymes. In addition, for some enzymes, other names are commonly used (for example, AtfA for WS/DGAT from Acinetobacter baylyi or Maqu_2220 for one of the FARs from Marinobacter aquaeolei).
of sequenced genomes has enabled the identification and characterization of putative FARs and WSs from different organisms. The extensive available data on their properties, substrate specificity, and activity in different expression systems provide a useful toolbox for designing the production of tailored WEs. Additionally, several studies showed that the substrate specificity of FARs and WSs can be engineered. Arabidopsis FAR5 and FAR8 have high (85%) amino acid sequence similarity but possess distinct substrate specificities towards 18:0-CoA and 16:0-CoA, respectively. Domain-swaps and site specific mutations revealed that two amino acid substitution (A355L and V377M) in FAR5 sequence changed enzyme specificity from 18:0-CoA to 16:0-CoA, whereas reciprocal substitution in FAR8 had the opposite effect (Chacón et al., 2013). Different amino acid substitutions also enabled alteration of substrate specificity of bacterial WS/DGATs. Selection of residue potentially affecting the substrate specificity of M. aquaeolei WS/ DGAT (Ma1) was based on sequence comparison with PapA5 acyltransferase from Mycobacterium tuberculosis. It was shown that specific changes in residues at positions 360, 356, and 405 can shift Ma1 enzyme selectivity for short-chain, mediumchain, branched, and aromatic fatty alcohols (Barney et al., 2013(Barney et al., , 2015. Further residue substitutions, identified on the basis of the crystal structure of M. aquaeolei WS/DGAT, were introduced within the acyl-CoA-binding pocket and led to increased preference towards shorter acyl-CoAs (Petronikolou and Nair, 2018). Enzyme selectivity can also be modified by fusion with the domains determining substrate specificity, which was shown for the mouse AWAT2 carrying segments of mouse DGAT2 (Kawelke and Feussner, 2015), and for A. baylyi WS/ DGAT fused with two transmembrane domains from mouse AWAT2 (Yu et al., 2018).
To conclude, further development in metabolic engineering of tailored WE production in plants will most likely focus on combining efficient strategies to modify the acyl-CoA and/or acyl-ACP pools with suitable FARs and WSs.

Limitations for wax ester accumulation in seeds
Studies on WE accumulation in different oilseed crops demonstrated that high WE contents negatively impacted seed germination. C. sativa MaFAR/ScWS-expressing lines with high WE amounts had white cotyledons and their germination was delayed (Iven et al., 2016;Yu et al., 2018). In the case of C. abyssinica, a decreased frequency of seed germination, reduced growth, and brown spots on the cotyledons were observed for seeds containing over 35% WEs in their oil, especially for the lines harboring the ScFAR/ScWS/ScFAE1 combination ( Li et al., 2019). In field and greenhouse trials, C. abyssinica lines expressing ScFAR, ScWS, and ScFAE1 had reduced seed yields and oil contents, and showed lower germination rates as well as delayed flowering and seed maturation. Similar effects were observed for C. sativa ScFAR/ScWS/LaFAE1expressing lines grown in a greenhouse (Zhu et al., 2016). Hampered germination was also reported for L. campestre seeds with high WE content. Interestingly, the embryos of the transgenic lines displayed disrupted neutral lipid packaging (Ivarson et al., 2017).
The results above strongly suggest that the impaired seed germination and seedling establishment represent a bottleneck in accumulating very high WE levels in the seeds of oilseed crops. The lowest seed germination rates and seed yields were observed for transgenic C. abyssinica and C. sativa lines harboring additionally ScFAE1 or LaFAE1 gene. Since this additional expression resulted in higher amounts of WEs composed of C24 acyl and alcohol moieties, it was postulated that the production of very long-chain fatty acids and alcohols may have a negative impact on lipid metabolism in developing and germinating seeds (Zhu et al., 2016;Li et al., 2019). The accumulation of potentially toxic fatty alcohols and/or the inability of the host seed enzymatic machinery to metabolize WEs may be the origin of the observed impaired germination and seedling growth. In jojoba, the WE mobilization pathway relies on the activity of three enzymes. First, a lipase (wax ester hydrolase) hydrolyses WEs to fatty acids and fatty alcohols. Then, before entering β-oxidation, the free fatty alcohols need to be converted to the corresponding fatty acids, which is catalysed by a fatty alcohol oxidase and a fatty aldehyde dehydrogenase (Rajangam et al., 2013). Since jojoba lipases exhibited high activity towards both WEs and TAGs (Kawiński et al., 2021), it cannot be ruled out that lipases from oilseed crops have the ability to hydrolyse WEs. Therefore, introduction of the jojoba fatty alcohol oxidation pathway to WE-storing crops could promote WE degradation and thus improve seed germination. Finally, disturbances in seed yield, germination, and seedling establishment may also be a result of improper packaging of WEs into lipid droplets, which was observed for L. campestre (Ivarson et al., 2017). The recently released complete jojoba genome (Sturtevant et al., 2020) represents an excellent resource to better understand how this unique plant optimally packs and stores WE in its seeds, and efficiently degrades them for germination.

Extraction and properties of wax esters accumulated in oilseeds
To compete with synthetic WEs, plant-based WEs need to have favorable properties for industrial applications, and methods for their extraction from seeds should be simple and cheap. The oil from WE-accumulating seeds can be recovered using standard solvent extraction preceded by seed crushing or pressing (Ruiz-Lopez et al., 2017;Shirani et al., 2020). The crucial step is the separation of WEs from TAGs. For WEs with melting points higher than TAGs, it is possible to use a process called winterization, which consists of gradual cooling of the oil resulting in crystallization of WEs (Zhu et al., 2016). However, this method is not efficient enough for shorter WEs, which are characterized by lower melting points. Ruiz-Lopez et al. (2017) tested the efficiency of molecular distillation on a pilot-scale using 1200 C. sativa plants accumulating mixtures of C30-C40 WEs in the seeds (up to 30 mg g −1 seed). The procedure allowed recovery of over 80% of WEs from refined oil, whereas only 20% of WEs were extracted by winterization (Ruiz-Lopez et al., 2017). Transgenic C. abyssinica plants containing 20% of WEs in their oil were used for development of another method of WE separation, which included a mild methylation step followed by short-path distillation. Addition of the purified wax esters to conventional crambe oil to the concentration of 15 wt% improved its temperature stability, oxidative resistance, wear resistance and lubrication properties at elevated temperatures (Shirani et al., 2020). These findings demonstrate that WEs accumulated in seeds of transgenic oil crops can be extracted using rather simple procedures, and used for the production of added-value lubricants.

Wax ester production in leaves
Whereas the worldwide demand for vegetable oil is increasing yearly, the arable land available for growing oilseed crops remains limited. In that context, producing lipids of interest in the vegetative tissues of high biomass crops has been proposed as a promising alternative (Carlsson et al., 2011;Mitchell et al., 2020). WE production in green tissues was first assayed by transiently co-expressing FAR and WS genes in Nicotiana benthamiana leaves. Aslan et al. (2014) tested several combination of two different FARs together with two WSs: AtFAR6 from Arabidopsis (Doan et al., 2012) or MaFAR from M. aquaeolei (Maqu_2220) together with AtPES2 from Arabidopsis (Lippold et al., 2012) or MhWS2 from M. hydrocarbonoclasticus. In order to address the enzymes of prokaryotic origin to the chloroplast, the authors fused the transit peptide sequence of AtFAR6 at their N-terminus (yielding tpMaFAR and tpMhWS).
The highest WE amount (1.62 nmol mg −1 FW, corresponding to approximately 0.9% of leaf DW, 5 days postinfiltration) was achieved using tpMaFAR in combination with AtPES2. Expressing AtFAR6 resulted in WEs mainly composed of 16:0-OH, while combinations using tpMaFAR produced WEs containing both 16:0-OH and 18:0-OH. At the fatty acid level, WEs produced upon tpMhWS expression mainly contained 16:0 and 18:0, while expressing AtPES2 resulted in WEs mainly comprising medium chain (12:0 and 14:0) fatty acids. Transition electron microscopy analyses indicated that WEs accumulated in the chloroplasts as aggregates of various shapes. The authors also tested the activity of a fusion polypeptide harboring the catalytic parts of both MaFAR and MhWS. Expression of tpMaFAR::MhWS resulted in WE yields in the same range as when expressing tpMaFAR and MhWS as separate polypeptides. Co-expression of Arabidopsis WRINKLED1 gene (AtWRI1), a master positive regulator of fatty acid biosynthesis (Focks and Benning, 1998), generally did not increase WE yields, even though higher accumulation of TAGs, suggesting increased fatty acid production, was observed. Altogether, this first study showed that it is possible to divert de novo fatty acid biosynthesis in the chloroplast to WE synthesis, and that using enzymes differing in their substrate preferences may permit the production of specific mixtures of WEs.
In a following study, Aslan et al. (2015a) attempted to additionally inhibit KASII expression using RNAi technology in leaves of N. benthamiana. KASII is involved in C16 to C18 fatty acid conversion within the plastid. Although transient RNAi approaches led to an almost complete inhibition of KASII expression, only a moderate increase in WE production was observed for enzyme combinations containing AtFAR6 (from 0.94 to 1.63 nmol mg −1 FW). Similarly, the 16 to 18 ratio was increased upon expression of AtFAR6, but not when tpMaFAR was used. In agreement with a previous study showing that strong seed-specific hairpin-RNAi reduction of KASII expression led to lethality (Pidkowich et al., 2007), leaves agro-infiltrated with the KASII-RNAi construct displayed bleaching symptoms.
Finally, Aslan et al. (2015b) tackled WE production in green tissues, not using transient expression, but stably transforming tobacco plants. In this study, they expressed a gene encoding a fusion protein between two bacterial enzymes (tpMaFAR::MhWS) under the control of the cauliflower mosaic virus 35S promoter in N. benthamiana. The best transgenic plants obtained showed an 8-fold increase in WEs at the whole plant level, reaching 0.28 µmol g −1 FW in leaves. WEs were also detected in stems, predominantly in the middle part. As expected from previous works, the produced WEs were mainly composed of C16 and C18 fatty acids and fatty alcohols. Nevertheless, the detection of very long-chain fatty acids (20:0 and 22:0) in the WE fatty acid fraction suggests that although the fusion polypeptide was targeted to the plastids, some of the WE production might have occurred outside. The authors also showed that the chlorotic leaf phenotype observed in some lines producing WEs was most likely caused by accumulation of free fatty alcohols. This suggests that future strategies should use highly active WSs to efficiently convert all the fatty alcohols produced into WEs, thus avoiding negative effects on plant health and increasing WE yields.

Conclusions and future prospects
The past two decades have witnessed impressive progress in the metabolic engineering of WE production in plants. Pioneering studies successfully demonstrated the feasibility of WE accumulation in seeds of TAG-storing crops, such as C. sativa, C. abyssinica, B. carinata, and L. campestre. Additionally, using a wide range of WE-synthesizing enzymes with desired specificities together with different approaches to modulate the acyl-CoA pool enabled the specific accumulation of certain WE molecular species. However, the strategies developed so far have only yielded viable transgenic crops with limited levels of WEs. Therefore, further research should focus on increasing WE accumulation by identifying and overcoming possible bottlenecks (see Box 3). Lipidomics, metabolomics, and visualization of the WE spatial distribution in situ using MALDI-MSI will certainly advance our understanding of implementing WE metabolism in transgenic seeds, and unravel underlying limitations. Localizing both FAR and WS enzymes to the same subcellular compartment (or even subdomains of compartments) to increase their physical proximity, using single enzymes or fusion proteins, might improve substrate channeling and thus WE yields. The competitiveness of plant-based WE production could also be increased by establishing the synthesis of WEs with even higher lubrication properties, such as WEs with hydroxy groups or branched acyl-chains. Such WEs have already been described in nature, and some FAR and WS enzymes with suitable specificities have been identified in birds, plants, and protists. Nevertheless, their potential, as well as that of many so far uncharacterized enzymes, for producing tailored WE in plants still needs to be evaluated. In addition, molecular modeling and directed evolution approaches may allow optimization of the substrate specificity of certain FAR and WS enzymes. Most importantly, a better understanding of why seeds with high WE contents are impaired in germination is essential for generating valuable transgenic plant lines stable over multiple generations. Genes allowing normal germination despite high levels of seed WEs must be present within the recently published jojoba genome. Another strategy to be tested is to produce WEs in the vegetative tissues of high biomass crops such as sorghum or sugarcane, optimally Box 3. Current and future challenges in further development of wax ester production in plants • Identifying and overcoming the reasons for impaired germination and seedling growth of transgenic plant lines with high WE content. • Improving WE packaging in seeds. • Increasing WE levels in seed oil. • Engineering synthesis of WE blends with defined composition suited for industrial applications. • Establishing production of unusual WEs with outstanding lubrication properties. • Further optimization of WE production in vegetative tissues.
during senescence when thylakoid membranes and lipids are remobilized. To conclude, even if the first attempts to modify plant lipid metabolism towards the production of WEs have been promising, a lot more still needs to be achieved in order to establish sustainable sources of bio-lubricants in plants.