A plant factory for moth pheromone production

Moths depend on pheromone communication for mate finding and synthetic pheromones are used for monitoring or disruption of pheromone communication in pest insects. Here we produce moth sex pheromone, using Nicotiana benthamiana as a plant factory, by transient expression of up to four genes coding for consecutive biosynthetic steps. We specifically produce multicomponent sex pheromones for two species. The fatty alcohol fractions from the genetically modified plants are acetylated to mimic the respective sex pheromones of the small ermine moths Yponomeuta evonymella and Y. padella. These mixtures are very efficient and specific for trapping of male moths, matching the activity of conventionally produced pheromones. Our long-term vision is to design tailor-made production of any moth pheromone component in genetically modified plants. Such semisynthetic preparation of sex pheromones is a novel and cost-effective way of producing moderate to large quantities of pheromones with high purity and a minimum of hazardous waste.

P heromones are environmentally friendly alternatives to traditional pesticides for the control of insect pests and indeed synthetic pheromones are produced in large amounts for this purpose 1 . Current standard approaches to pheromone synthesis either require the use of hazardous chemicals or may result in the production of hazardous waste byproducts 2,3 . We propose to overcome the problems inherent to synthetic pheromone production by designing and developing an innovative green chemistry alternative while minimizing hazards 4 . Our strategy involves the use of a cost-effective plant factory expressing a suite of biosynthetic enzymes for production of moth pheromones.
Female moths emit species-specific pheromone component blends that attract males of the same species over long distances 1 . A majority of the identified moth pheromone compounds consist of fatty acid derivatives produced de novo in the pheromone gland 5,6 . Their biosynthesis typically involves desaturation of fatty acids, reduction to primary alcohols and further modification to produce acetates or aldehydes. Great advances have been made during the last 15 years with respect to our understanding of the molecular basis of moth pheromone biosynthesis 7,8 . Heterologous expression systems have allowed the confirmation of the function of many desaturases [9][10][11][12][13] and fatty-acyl reductases (FARs) [14][15][16][17] . When expressed in yeast, the desaturases produce fatty acids of different chain lengths with double bonds in different positions. The functionally characterized FARs convert fatty-acyl moieties into fatty alcohols.
Transgenic plants have proven useful to express enzymes from insects. For example, transgenic plants produced a pheromone precursor upon introduction of a moth desaturase 18 and an aphid alarm pheromone has been produced from endogenous plant sesquiterpene by expression of an (E)-b-farnesene synthase cDNA 19 . We wanted to take this approach further and exploit the molecular toolbox currently available to produce specific multicomponent moth pheromones in Nicotiana benthamiana. This requires the coordinated expression of several genes to be successful. Producing stable lines of transformed plants that assemble these multistep recombinant pathways remains a major challenge. An alternative is to express the genes in leaves transiently via infiltration of the abaxial air spaces with Agrobacterium tumefaciens cultures harbouring each expression construct plus viral silencing suppressor protein P19 that inhibits the host cells' transgene-silencing apparatus 20 . This leaf-based expression format has been used in transient expression of the multicomponent metabolic pathway of polyunsaturated fatty acid metabolism 21 .
Here we first produce (Z)-11-hexadecenol (Z11-16:OH), (E)-11-tetradecenol (E11-14:OH), (Z)-11-tetradecenol (Z11-14:OH) and their corresponding acetates, by introducing all the necessary and sufficient genes into N. benthamiana plants. These acetates and alcohols have been identified as pheromone components in hundreds of moth species 22 . Specifically, we then produce plantderived mimics of the sex pheromones of two species of small ermine moths, Yponomeuta evonymella and Y. padella 23 and compare their behavioural activity to conventionally produced pheromones. Our series of proof-of-principle experiments demonstrate the feasibility of production of behaviourally active moth pheromones in plants.

Results
Assembly of moth pheromone biosynthetic pathways. In plant leaf cells, de novo fatty acid biosynthesis takes place in the chloroplasts. The sequential, two carbon atom elongation of the acyl chain esterified to acyl carrier protein (ACP) is catalysed by the fatty acid synthase complex. The majority of the acyl groups are cleaved from ACP at 16 and 18 carbon lengths by plastidial thioesterases, transported through the plastidial envelope and acylated to CoA at the outer envelope. The acyl-CoAs are then utilized for the synthesis of extraplastidic lipids by enzymes localized in the endoplasmic reticulum (ER) 24 . The enzymes of pheromone biosynthesis used in this study also reside in the ER in their natural environment 25 . Taking this into consideration, we thus took the following approach to pheromone production in N. benthamiana leaves (Fig. 1): By introducing a plastidial thioesterase, CpFATB2, derived from Cuphea palustris, that terminates the fatty acid biosynthesis by mainly hydrolyzing the thioester bond of myristoyl-ACP (14:ACP) 26 , we increased the amount of myristic acid exported from the plastid, thereby creating a cytosolic pool of 14:CoA available for pheromone biosynthesis. We also introduced D11 desaturases that specifically recognizes 14:CoA and 16:CoA to produce (E/Z)-11tetradecenoyl-CoA (E/Z11-14:CoA) and (Z)-11-hexadecenoyl-CoA (Z11-16:CoA). FARs from several moth species could then be introduced to reduce the E/Z11-14:CoA and Z11-16:CoA into E/Z11-14:OH and Z11-16:OH. Finally, a plant-derived acetyltransferase gene (EaDAcT) 27 was expressed to transfer an acetate group to the primary alcohols and form the (E/Z)-11tetradecenyl acetates (E/Z11-14:OAc) and (Z)-11-hexadecenyl acetate (Z11-16:OAc).
To evaluate the feasibility of our approach, we first expressed candidate genes individually and in combinations for each of the biosynthetic steps required for pheromone production to take place in the leaves. In this way, 11 genes were characterized ( Table 1). Fifty gene combinations were evaluated out of which 34 yielded products of interest. The amount of myristic acid (14:acyl) substrate available to insect pheromone component biosynthesis in N. benthamiana is very low. Upon expression of the thioesterase CpFATB2, the plants generally produced more than 200 mg of myristic acid per gram fresh leaf tissue, whereas the amount in control plants was less than 2 mg.
Next, we tested five desaturases exhibiting various specificity and functionality in our leaf chassis. It was shown previously that insect-derived desaturases involved in pheromone biosynthesis can be successfully expressed in tobacco 18 . The OnuD11 (ref. 12) showed no detectable activity in our leaf expression system, whereas the other four conferred on the plant the ability to produce amounts of unsaturated products ranging from 30 to 600 mg per gram fresh leaf tissue. Relative to the respective saturated precursor, the amount of product found was in the range of 9-50%. Among these four desaturases, we selected the AtrD11 and AveD11 (ref. 9) as candidates for subsequent larger scale production of pheromone. AtrD11 showed high substrate specificity towards 16:CoA and AveD11 specifically interacted with 14:CoA to produce a mixture of E11-14:CoA and Z11-14:CoA. The activity of CroD11 (ref. 10) was very similar to that of AveD11 but CroD11 produced a somewhat higher ratio of E11-to Z11-14:CoA, which was not suitable for the production of small ermine moth pheromones.
In the third step, we evaluated four FARs that have previously been shown to be active in moth pheromone biosynthesis [15][16][17] . The OnuFAR Z 15 and E 15 resulted in very low product yield when expressed in plants. HarFAR 17 showed higher conversion rates than YroFAR 16 . The reduction of fatty-acyl precursors to the corresponding fatty alcohols was much less efficient compared with the desaturation. Efforts were made to optimize the conversion rates by engineering variants of FARs. Of all the tested modifications (Methods), only the ER retention signal (KKYR), attached to the C-terminal end of the enzyme, improved the alcohol production over the original version (Table 1). Therefore, HarFAR_KKYR was selected as the most suitable candidate for the larger scale pheromone production experiment to be performed.
In the fourth and final step, a plant-derived diacylglycerol acetyltransferase (EaDAcT), cloned from the Burning Bush (Euonymus alatus) 27 was tested, as no acetyltransferase involved in moth pheromone biosynthesis has been characterized so far. EaDAcT is similar to the Jojoba wax synthase, which is a fattyacyl-CoA: fatty alcohol acyltransferase, and it may therefore possess the ability to acetylate long-chain alcohols. When expressed in our leaf transient expression system, it resulted in 5-10% conversion rate of 14C and 16C alcohol substrates ( Table 1, Fig. 2). We considered the acetate production to be on the low side, both in terms of absolute amounts and in relation to the amount of precursor alcohol and for extraction and implementation in the trapping of moths (Fig. 3). This particular step will eventually require further improvements and will most likely benefit from the future characterization of an insect enzyme.
Attractiveness of plant-derived moth pheromones. Using the most promising candidate genes, we aimed at specifically producing the Y. evonymella and Y. padella pheromones in amounts sufficient for attraction of male moths in the field. Y. evonymella and Y. padella are attracted to Z11-14:OAc/E11-14:OAc/14:OAc ¼ 100/20/60 and Z11-14:OAc/E11-14:OAc/Z11-16:OAc ¼ 100/34/400, respectively. Although they lack minor components necessary for maximal attraction, these three-component mixture blends of synthetic compounds have still been confirmed to be highly attractive and species specific when tested in field experiments 23,28 . To obtain enough pheromone for trapping of moths in the field, all the leaves (ca. 10) on a N. benthamiana plant were infiltrated. One plant was infiltrated with P19, CpFATB2, AveD11 and HarFAR to produce a mixture of 14C alcohols and another plant was infiltrated with P19, AtrD11 and HarFAR to produce 16C alcohols. The alcohol fractions from the leaves were isolated and acetylated chemically. The 14C alcohol-producing plant yielded more than 500 mg of 14C acetates plus a smaller amount of saturated 16:OAc; the 16C plant produced 1,280 mg Z11-16:OAc and a smaller amount of 16:OAc as determined by GC-MS analysis ( Table 2).
For trapping of Y. evonymella, the acetate mixture from the 14C alcohol-producing plant was used as it was, whereas a 1:4 mixture of the acetates from both 14C and 16C alcoholproducing plants were used for trapping of Y. padella. The behavioural activity of the plant-derived acetate mixtures was compared with the activity of conventionally produced synthetic three-component pheromone lures (positive controls) and blank traps (negative controls). Catches of Y. evonymella were very high. Traps baited with plant-derived pheromone attracted on average 130 males per trap, approximately half the catches obtained with conventional produced pheromone components. In the case of Y. padella, the population was obviously not as abundant in the trapping area but for this species the plantderived pheromone was not different from the conventionally produced pheromone in attractiveness. Blank traps trapped no males of either species (Fig. 3).

Discussion
Attraction of male moths to traps baited with plant-derived pheromones was high and specific, even if the 45/100 ratio of  23 . However, we predict that in the near future, it will be possible to modify the ratio between plant-produced isomers by engineering the desaturase and reductase genes. We have already demonstrated that substitution of one amino acid can cause remarkable changes in specificity of a moth FAR 8 . The genes and molecular techniques available to us thus constitute a powerful toolbox from which we can choose the elements to be used in the construction of genetically modified plants producing optimal pheromone component ratios.
Modifying and optimizing the acetate/alcohol ratio is more challenging. Most moth pheromones contain acetates as major components. Too high amounts of corresponding alcohols may significantly reduce or even completely abolish attraction. This means that either the conversion of alcohols into acetates has to be almost complete or we have to be able to control the relative release of alcohols and acetates from the plants. For the time being, we do not know anything about the release of moth sex pheromone components from intact plants. Even though transient expression has been shown to be an efficient system for generating various products 29 , stable transformation would be required to eliminate applications of expression constructs and to obtain a system where a larger part of the biomass produces the product of interest over a longer time. Optimization of expression with regards to promoter strength and temporal control or induction would be of importance in the optimization of product yield. Endogenous temporal control would require identification of a promoter that has an onset at a desired time point during plant development. In the case of inducible expression, several systems exist which involve application of low-cost compounds to induce gene expression, one example of such an inducer being methoxyfenozide 30 . Modifications of plant endogenous lipid metabolic processes to build an efficient flow from saturated substrates via desaturation, reduction and down-stream modifications such as acetylation, will involve coordinated expression of multiple constructs 31 in multiple copies. This is a research challenge, requiring the production of numerous transgenic lines and selection of elite events.
The current study paves the way for future production of moth pheromones in plant factories. Our experiments convincingly demonstrate that it is possible to produce highly attractive and species-specific moth pheromones making use of genetically modified plants (Fig. 2). This approach is just in its infancy, but the results are already promising. We now aim at developing tools to optimize ratios of the plant-produced pheromone compounds, increasing the yield and controlling what is released from the plants. Our long-term vision is to design tailor-made production of any moth pheromone component in stably transformed plants. Such sustainable preparation of sex pheromones from plants will be a novel and cost-effective way of producing moderate to large quantities of pheromones with high purity and a minimum of hazardous waste. Moving from transient expression to stable transformants will further improve the system. Genetically modified plants may eventually be used in intercropping as natural dispensers of insect pheromones and as part of a pushpull strategy 32,33 , provided that they can be shown to have no negative environmental side effects. Our proposed strategy is innovative and environmentally friendly. Although it contains fundamental research challenges, it has the potential to become  an economically sound part of many integrated pest management programs.

Methods
Plasmid constructs for transient expression. All the genes, including: CpFATB2, AtrD11,OnuD11, YroFAR, HarFAR, OnuFAR_E/Z and EaDAcT were amplified from cDNA with primers spanning from the start codon to the stop codon of the ORF, except AveD11, CroD11 and CpaE11, which were synthesized and codon optimized for the N. benthamiana codon bias (Invitrogen, Life Technologies). Truncation, gene fusion, ER targeting signal and retention signal attachments were achieved by PCR and fusion PCR 34 , as presented in the Methods. All genes and their modified versions were cloned into plant expression vector pXZP393 by Gateway recombination cloning technology (Invitrogen). After confirming the integrity of the constructs by sequencing, the expression clones were introduced into Agrobacterium tumefaciens strain GV3101 (MP90RK) by electroporation (1500 V mm À 1 , 5 ms, Eppendorf 2510).
In order to inhibit the host cells' transgene-silencing apparatus and extend transgene expression over a longer period of time with a higher degree of expression, the viral silencing suppressor protein P19 was co-introduced into the Agrobacterium mixtures 21 .
Outline of all the molecular manipulations performed on FARs. The used insect FARs display amino-acid sequence in the last B100 aa of the C-terminal region, which could form a transmembrane region or membrane anchor as predicted using TMHMM 35 . As a first modification (A), this region was truncated as it had been shown that a similar truncation of mouse FAR1, MmFAR1Dc, had rendered the heterologous expressed enzyme highly active (HarFAR was truncated at 352L-V353; YroFAR was truncated at 353K-F354) 36 . In a second modification (B), the C-terminal region was replaced by a corresponding region of either MmFAR1Dc or a Marinobacter aquaeolei VT8 FAR 37 (HarMmuFAR: SACNP-FHWGE; HarMaqFAR: ACNPI-SLGEF; YroMmuFAR: GTTNP-FHWGE; YroMaqFAR: GTTNP-ISLGE; aa sequence at junction are shown). In a third set of modifications (C), an ER retention signal was added to the C-terminal (HarFAR_KKSY: KKSYE-4KKSY; HarFAR_KKYR: KKSYE-4KKYR; YroFAR_KKYR: attaching KKYR to the C-terminal).
Although there is no direct evidence showing that the insect-derived fatty acyl desaturases involved in sex pheromone biosynthesis are present in the ER, their four transmembrane regions strongly suggest they are located in the ER 38 . With the aim of bringing the FARs nearer to the desaturases, different versions of gene fusions were produced: (Atr1-267)_HarFAR_KKYR: HRLWAH-MVVLTS (this construct fuses the first 89 aa of AtrD11 to HarFAR_KKYR, that contains the ER targeting signal and first transmembrane domain); AtrD11_(Atr1-267)_HarKKYR (this construct fuses AtrD11 with HarFAR_KKYR, with a long spacer, that is the first 89aa from AtrD11); AveYroFAR (this construct fuses AveD11 with YroFAR with no spacer); AtrHarFAR (this construct fuses AtrD11 with HarFAR with no spacer).
Agrobacterium infiltrations of Nicotiana benthamiana. Ten ml of Agrobacterium solution containing individual gene constructs were grown at 28°C with LB medium supplemented with appropriate antibiotics overnight in a 300 r.p.m. (revolutions per minute) incubator. Acetosyringone (final conc. 100 mM) was added to the overnight culture and grown for an additional 2-3 h to induce virulence genes encoded by the Agrobacterium genome. Bacteria were spun down at 3,000 g for 5 min at room temperature and resuspended in infiltration buffer (5 mM MgCl 2 , 5 mM 4-morpholineethanesulphonic acid, 100 mM acetosyringone, pH 5.7). Optical density (A 600 nm ) of each Agrobacterium culture was measured and the final concentration of each culture was adjusted equally to A 600 nm ¼ 0.2, in a total volume of 20 ml, by infiltration buffer as described before. The final mixture of Agrobacterium cells was drawn up into a 1 ml syringe (without needle) and infiltrate into the underside of a suitable leaf, with a gentle squeeze on the plunger and modest pressure on the leaf using a finger, the Agrobacterium solution was forced into the mesophyll spaces wetting the leaf. Three entire leaves of similar age, from 1-month-old N. benthamiana plants were infiltrated. For production of fatty alcohols for pheromone traps, all the leaves from one N. benthamiana plant were infiltrated. The plant was watered generously and maintained in growth chamber for 4 days.
Lipid analysis. Total lipids were extracted from ca. 1 g of leaf tissue using 3.75 ml of methanol/chloroform (2:1, v/v), in a glass tissue grinder. The crude extract was transferred to a glass tube and the grinder was rinsed with 1.25 ml of chloroform, which was then added to the extract. One ml of acetic acid (0.15 M) and 1.25 ml of water were added to the tube producing a biphasic mixture. Tubes were vortexed vigorously and centrifuged at 550 g for 2 min. The ca. 2.5 ml of chloroform phase containing the total lipids was transferred to a new glass tube.
Fatty acid methyl esters (FAMEs) were prepared from a portion of this total lipid extract. Thus, 100 ml of the extract was transferred to a new glass tube, and 10 ml (1 nmol ml À 1 ) internal standard 17:Me was added. The solvent was evaporated under a gentle nitrogen flow. Then 1 ml of sulphuric acid (2% in methanol, v/v) was added, and the reactant was vortexed vigorously followed by 1-h incubation at 90°C. After incubation, 1 ml of water was added, the mixture was vortexed and then 1 ml of hexane was added to extract the FAMEs.
Fatty alcohols (OH) and acetates (OAc) from the total lipid extract were separated by thin-layer chromatography (TLC). Ten ml of 17:OH (1 nmol ml À 1 ) and 3 mg of Z8-13:OAc were first added as internal standards to 300 ml of total lipid extract, which was then concentrated to ca. 50 ml under gentle nitrogen flow. The   concentrated sample was subsequently loaded onto a TLC plate (Silica gel 60, Merck, Germany) and separated with a mobile phase of hexane/diethyl ether/acetic acid (85:15:1, v/v/v). The bands were visualized by spraying water, and target gel areas were collected separately into 4-ml vials and extracted with 3.75 ml of methanol/chloroform (2:1, v/v) in sonication bath for 15 min. The extract was then centrifuged at 1,685 g for 3 min. The supernatant was transferred to a new tube and 1 ml acetic acid (0.15 M), 1.25 ml of chloroform and 1.25 ml of water were added to partition the lipids into chloroform. The chloroform phase containing alcohols or acetates, were transferred into new tubes and evaporated to dryness, followed by adding 40 ml of hexane to dissolve the compounds. FAMEs, alcohols and acetates were analysed and quantified by GC-MS on a Hewlett-Packard 6890GC coupled to a mass detector HP 5975, operated in electron impact mode (70 eV). All compounds were identified by comparison of their mass spectra and retention times to those of corresponding synthetic standards available from our laboratory collection. The GC was equipped with a polar HP-INNOWax capillary column (100% polyethylene glycol; 30 cm*0.25 mm*0.25 mm, Agilent Technologies), and helium was used as carrier gas (average velocity: 33 cm s À 1 ). The injection port was configured in splitless mode and the oven temperature was set to 80°C for 1 min, then increased at a rate of 10°C min À 1 up to 210°C, followed by a hold at 210°C for 15 min, and then increased at a rate of 10°C min À 1 up to 230°C followed by a hold at 230°C for 20 min; the transfer line temperature was 280°C. Data were analysed by the ChemStation software (Agilent, Technologies, USA).
Field trapping experiment. All the leaves (about 20 g of fresh weight) on one N. benthamiana plant, in two combinations named plant #72 and plant #76 were infiltrated to prepare sufficient amounts of plant-derived pheromone for field trapping. For plant #72, which produced 14:OH, Z11-14:OH, E11-14:OH and 16:OH, the gene constructs were P19, CpFATB2, AveD11 and HarFAR. For plant #76, which produced 16:OH and Z11-16:OH, the gene constructs were P19, AtrD11 and HarFAR. The fatty alcohols in the chloroform/methanol extracts of the leaves were isolated by TLC fractionation, and subsequently extracted by methanol/ chloroform from the recovered silica gel. The alcohol extract was evaporated to dryness under gentle nitrogen flow, and the residues were used for preparing corresponding acetates by adding 1 ml of acetyl chloride and incubating at room temperature for 20 min after a brief vortex mixing. After acetylation, the solvent was evaporated under gentle nitrogen stream and 1 ml of hexane was added to dissolve the acetylated products, which were subsequently washed by 5% NaHCO 3 and water, and finally dried over anhydrous Na 2 SO 4 .
The attractiveness of N. benthamiana-derived pheromones to male ermine moths was investigated in southern Sweden, in the vicinity of Lund on pastureland with shrubs of Crataegus sp., Euonymus europeus and Prunus spinosa (Y. padella 10 th -24 th , Aug, 2012), or along paths in a park with rich abundance of P. padus trees (Y. evonymella, 3 rd -19 th , Aug, 2012). Transparent plastic delta traps with sticky inserts (Csalomon, Budapest, Hungary) were used for the trapping experiments (five replicates for each treatment). Prior to loading the plant-derived pheromone compounds on the red rubber septa (Wheaton Science Products, Millville, NJ, USA), the absolute amounts and blend ratios were checked by GC-MS. A dose of 100 mg total acetates per septum was applied. Traps baited with synthetic acetates in same amounts and blend ratios were used as a positive control. The daily catch numbers were recorded and traps were redistributed within blocks after each check.
Statistical analysis. Data were processed by Excel 2011 and Minitab13. Before conducting the test, normality tests of the data were performed. Two sample t-tests were used, to compare the moth catches of plant-derived pheromones and synthetic pheromones, the fatty alcohol productions from wild-type HarFAR and modified HarFAR.