(−)-α-Bisabolol, a sesquiterpene alcohol, is a major ingredient in the essential oil of chamomile (Matricaria recutita) and is used in many health products. The current supply of (−)-α-bisabolol is mainly dependent on the Brazilian candeia tree (Eremanthus erythropappus) by distillation or by chemical synthesis. However, the distillation method using the candeia tree is not sustainable, and chemical synthesis suffers from impurities arising from undesirable α-bisabolol isomers. Therefore enzymatic synthesis of (−)-α-bisabolol is a viable alternative. In the present study, a cDNA encoding (−)-α-bisabolol synthase (MrBBS) was identified from chamomile and used for enantioselective (−)-α-bisabolol synthesis in yeast. Chamomile MrBBS was identified by Illumina and 454 sequencing, followed by activity screening in yeast. When MrBBS was expressed in yeast, 8 mg of α-bisabolol was synthesized de novo per litre of culture. The structure of purified α-bisabolol was elucidated as (S,S)-α-bisabolol [or (−)-α-bisabolol]. Although MrBBS possesses a putative chloroplast-targeting peptide, it was localized in the cytosol, and a deletion of its N-terminal 23 amino acids significantly reduced its stability and activity. Recombinant MrBBS showed kinetic properties comparable with those of other sesquiterpene synthases. These data provide compelling evidence that chamomile MrBBS synthesizes enantiopure (−)-α-bisabolol as a single sesquiterpene product, opening a biotechnological opportunity to produce (−)-α-bisabolol.
Terpenoids (or isoprenoids) are the largest and structurally most diverse group of secondary metabolites, with more than 55000 listed compounds . In Nature, terpenoids have essential roles, such as those involved in protein prenylation (e.g. RAS prenylation), plant hormonal regulation (e.g. gibberellin, brassinosteroids, cytokinin and abscisic acid), photosynthetic light harvesting (e.g. carotenoid and chlorophyll), and respiratory electron transport (e.g. ubiquinone) [2–4]. Terpenoids often function as toxins, growth inhibitors or deterrents to protect the producers from competing plants or herbivores . Terpenoids also have been extensively exploited by humans for their beneficial functions as pharmaceuticals (e.g. paclitaxel and artemisinin), flavours and fragrance compounds (e.g. menthol, patchoulol and nootkatone) .
In Nature, terpenoids are synthesized from successive condensations of five-carbon building units. These are IPP (isopentenyl diphosphate) and its isomer, DMAPP (dimethyl allyl diphosphate), both of which are synthesized in two independent pathways: the cytosolic MVA (mevalonate) pathway and the plastidic MEP (2-C-methyl-D-erythritol 4-phosphate) pathway [3,7]. Starting with DMAPP, the head to tail condensation reaction is catalysed by prenyltransferase, and one to three IPP units are added to yield three prenyl diphosphate molecules: C10 GPP (geranyl diphosphate), C15 FPP (farnesyl diphosphate) and C20 GGPP (geranylgeranyl diphosphate). These prenyl diphosphates are converted by TPS (terpene synthase) into various terpene hydrocarbons [8,9]. Isoprenoids are classified according to the number of carbons as monoterpenes (C10), sesquiterpenes (C15) and diterpenes (C20). As terpenes have multiple chiral centres and complex hydrocarbon backbones, chemical synthesis of terpenes is not usually practical, and hence the identification and utilization of unique TPSs are central in contemporary pharmaceutical and chemical industries .
Chamomile (Matricaria recutita) is one of the most commonly consumed herbal medicines . The therapeutic values of chamomile come from its essential oil component, (−)-α-bisabolol . (−)-α-Bisabolol, a sesquiterpene alcohol, is naturally occurring in the Brazilian candeia tree (Eremanthus erythropappus) and also in German chamomile (Matricaria recutita) [13,14]. This compound has been shown to possess various pharmaceutical activities (e.g. antibacterial, antiseptic and anti-inflammatory) and skin-soothing and -moisturizing properties [15–18]. At present, the majority of (−)-α-bisabolol, sold as ‘natural α-bisabolol’, is produced by distillation of candeia tree leaves . However, sustainability of this Brazilian indigenous plant has raised an environmental issue in recent years . (−)-α-Bisabolol can be chemically synthesized, but the synthetic (−)-α-bisabolol contains other diastereomers [(+)-α-bisabolol and (+/−)-epi-α-bisabolol] and undesirable by-products from chemical synthesis procedures. Therefore synthetic production of ‘natural identical’ (−)-α-bisabolol requires an economically unfavourable purification procedure to achieve a high optical purity. Nonetheless, the synthetic supply of (−)-α-bisabolol and its stereoisomers have recently been replacing the natural (−)-α-bisabolol production from the Brazilian candeia tree.
α-Bisabolol possesses two chiral centres at C1 and C7 (Figure 1). Depending on the configuration of these two chiral carbons, four possible stereoisomers of α-bisabolols, (+/−)-α-bisabolol and (+/−)-epi-α-bisabolol, can be formed . Previously, two α-bisabolol synthase genes were identified from Aztec herb (Lippia dulcis) and bacterium Streptomyces citricolor, and the products from their recombinant enzymes were elucidated as (+)-epi-α-bisabolol and (−)-epi-α-bisabolol respectively [21,22]. Although a specific configuration was not elucidated, one other α-bisabolol synthase was also found from Artemisia annua , and dual-functional bisabolene/α-bisabolol synthases which synthesize only a trace amount of α-bisabolol were isolated from Arabidopsis and Santalolum spp. [24,25]. To date, a sesqui-TPS (sesquiterpene synthase) synthesizing pure (−)-α-bisabolol as a single terpene product has not been identified.
Structures of four α-bisabolol isomers
From chamomile, five TPSs have recently been characterized . However, (−)-α-bisabolol synthase was not included among the TPSs characterized from chamomile. In the present study, we identified a gene encoding chamomile α-bisabolol synthase (MrBBS) using next-generation (Illumina and 454) sequencing. This enzyme can synthesize enantiopure (−)-α-bisabolol as a single terpene product. Furthermore, as proof-of-concept, microbial and in planta production of (−)-α-bisabolol was demonstrated.
Chamomile (M. recutita) was grown in a phyto-chamber maintained with a 16 h light/8 h dark photoperiod at 23°C. Stem, leaf, root and flower were harvested from 6-week-old M. recutita. Nicotiana benthamiana for transient expression experiments was grown for 5–6 weeks in a phyto-chamber with a 16 h light/8 h dark photoperiod at 23°C.
DNA construct for yeast expression
Six TPS clones were amplified from 20 ng of cDNA using the primers listed in Supplementary Table S1 (http://www.biochemj.org/bj/463/bj4630239add.htm) (1–12). Amplified PCR products were cloned into pMD-20 (TaKaRa Bio), and cDNA sequences were confirmed by sequencing. The TPS cDNAs were digested using restriction enzymes and ligated into the respective restriction digestion sites of pESC-Leu2d vector. For fused MrBBS, MBP (maltose-binding protein), GST and TRX (thioredoxin) were amplified using primers 28–33 using templates of pMAL-c2X (NEB) pGEX-2T (GE Healthcare) and pET32a (Novagen) respectively. Amplified PCR products were digested appropriately and ligated to the N-terminus of MrBBS in pESC-Leu2d vector. To characterize each sesqui-TPS, the constructs and empty vector were separately transformed into the EPY300 yeast strain by the lithium acetate/PEG transformation method [27–29].
In vivo yeast characterization and bisabolol quantification
Transgenic yeast were inoculated in 2 ml of SC (synthetic complete) medium (supplemented with 2% glucose and lacking methionine, histidine and leucine) and grown overnight at 30°C. The transgenic yeast culture was diluted 50-fold into 30 ml of SC medium (with 1.8% galactose, 0.2% glucose, 2 mM methionine and lacking histidine and leucine). To sequester the volatile sesquiterpene product, 3 ml of dodecane was overlaid on to the culture, and incubated at 30°C and 200 rev./min for 3 days. Cultures were transferred to a 50 ml Falcon tube and centrifuged at 3000 g for 5 min. The dodecane layer was separated and diluted 100-fold in hexane for GC–MS analysis. For quantification, the cultured medium was extracted with ethyl acetate and analysed by GC–FID (flame ionizer detector) (Shimadzu) against the calibration curve constructed from commercial (−)-α-bisabolol (Sigma) using the following temperature programme: initial temperature 60°C (2 min hold), increase to 300°C at 5°C·min−1, and 300°C held for 10 min. The analysis was conducted using a ZB-5 GC column (0.25 μm film thickness, 0.25 mm×30 m) and nitrogen as a carrier gas at a flow rate of 1 ml·min−1.
Protein expression in Escherichia coli and purification
To obtain recombinant MrBBS–His6, the ORFs of full-length MrBBS, 23- or 52-residue-deleted MrBBS were cloned into pET21 vector (Novagen), using the Gibson assembly cloning kit (NEB) and the primers listed in Supplementary Table S1 (13–16). Recombinant pET21-MrBBS-His6 construct was transformed into E. coli BL21 codon plus-RIL cells (Stratagene) and selected with ampicillin (100 μg·ml−1) and chloramphenicol (36 μg·ml−1). A single transformed colony was cultured in LB medium to a D600 of 0.5. MrTPS (chamomile TPS) expression was then induced with 0.4 mM IPTG by incubation at 17°C for 16 h. The cells were harvested by centrifugation and resuspended in lysis buffer [20 mM Hepes (pH 7.6), 10 mM MgCl2, 500 mM NaCl, 5 mM 2-mercaptoethanol, 20 mM imidazole, 10% (w/v) glycerol and 1 mM PMSF]. After cell lysis by sonication and centrifugation at 4000 g for 10 min at 4°C, the clear supernatant was loaded on to a 1 ml HisTrap HP column (GE Healthcare), which was pre-equilibrated with wash buffer [20 mM sodium phosphate (pH 7.4), 10 mM MgCl2, 500 mM NaCl, 5 mM 2-mercaptoethanol, 20 mM imidazole and 15% (w/v) glycerol]. After the column was washed with a 10-fold column volume of the wash buffer, the bound protein was eluted by a linear gradient of imidazole up to 500 mM. Protein in each fraction was confirmed using SDS/PAGE. Fractions containing the sesqui-TPS were pooled and dialysed at 4°C for 24 h to remove imidazole.
In vitro enzyme assay and kinetics
For the in vitro enzyme assay, 500 μl of reaction mixture was prepared with 20 μg of purified recombinant protein, 100 μM FPP (Echelon Biosciences), 50 mM Tris/HCl (pH 7.5) and 10 mM MgCl2. The mixture was incubated at 30°C for 2 h, followed by extraction with 500 μl of pentane (vortex-mixed and centrifuged at 4000 g for 2 min). The aqueous layer was extracted twice more with 500 μl of pentane each. The collected pentane layer was concentrated and directly injected into GC–MS. To determine kinetic parameters, 1 μg of protein was used in a 100 μl reaction mixture containing 25 mM Hepes (pH 7.4), 10 mM MgCl2 and 0.5–25 μM FPP. The FPP was prepared by diluting 23 Ci·mmol−1 [1-3H]FPP (PerkinElmer) 100-fold with unlabelled FPP. The mixture was incubated at 30°C for 10 min, after which the mixture was mixed with 100 μl of stop solution (4 M NaOH and 1 M EDTA). After quenching for 10 min, 900 μl of pentane was added, vortex-mixed for 10 s and centrifuged at 15000 g for 1 min. Then, 400 μl of the pentane layer was taken and mixed with 3.5 ml of scintillation cocktail. The 3H radioactivity was monitored using a Beckman LS6500 liquid scintillation counter. The Michaelis–Menten plot was created and the kinetic parameters were calculated by using the Enzyme Kinetics module in SigmaPlot 12 (Systat).
For α-bisabolol purification, MrBBS-expressing transgenic yeast was cultured in a total of 5 litres of culture without dodecane overlay. After 3 days of culture at 30°C, the culture medium was decanted to a separatory funnel and extracted twice with 1 litre of ethyl acetate. The combined ethyl acetate fraction was concentrated on a rotary evaporator to 2 ml, which was loaded on to a silica column (24 mm×225 mm, 15 g of silica gel 60, 40–63 μm) pre-washed with five column volumes of hexane. Then the column was eluted stepwise with 60 ml each of 0%, 5% and 10% ethyl acetate in hexane. Each fraction was analysed by TLC with hexane and ethyl acetate (9:1, v/v) as elution solvent and was visualized under iodine vapour. The fractions containing α-bisabolol were combined and loaded on to a pre-coated silica gel 60 F254 plate (175–225 μm, Merck), which was eluted by hexane and ethyl acetate (9:1, v/v). The α-bisabolol-containing fraction was extracted with ethyl acetate. Evaporation of the solvent gave 34 mg of purified bisabolol suitable for spectrometric analyses.
1D and 2D NMR [1H, 13C, 1H-1H COSY, DEPT (distortionless enhancement by polarization transfer), HSQC and HMBC (heteronuclear multiple bond correlation)] experiments were conducted on a JNM-LA400 spectrometer (Jeol) with [2H]chloroform CDCl3 as solvent. The optical rotation was determined by using a Jasco P-1020 polarimeter (Jasco). Authentic (−)-α-bisabolol was used as standard (Sigma). To analyse sesquiterpene products from MrBBS-expressing yeast, the diluted dodecane layer was directly injected into GC–MS (PerkinElmer Clarus 680 GC system coupled to a PerkinElmer 600T mass spectrometer) with the following temperature programme: initial temperature of 50°C (5-min hold), increased to 200°C at 2°C·min−1, and ramped to 300°C at 50°C·min−1 (15 min hold). The GC column was TG-5MS (0.25 μm film thickness, 0.25 mm×30 m, Thermo Scientific). Retention times of the products were compared with those of authentic (−)-α-bisabolol, and the fragmentation pattern was searched against the NIST11 database. To identify the in vitro enzymatic product of MrBBS, 2 μl of pentane extract was directly analysed by GC–MS (Agilent 6890N gas chromatography system and Agilent 5975B mass spectrometer) with an authentic standard in the following temperature program: injection at 250°C, and ramped from 40°C to 250°C at a rate of 10°C·min−1. The column used was DB1-MS (0.25 μm film thickness, 250 μm×30 m) and helium was used as a carrier gas with a flow rate of 1 ml·min−1. For chiral analysis, a previously published GC–MS method and chiral column were used .
RNA isolation and expression analysis
Harvested plant material was ground in liquid nitrogen and extracted using TRIzol® reagent (Invitrogen). A 1 μg sample of total RNA was used to synthesize first-strand cDNA with the cDNA synthesis kit (J&M). Bisabolol synthase transcript level was determined using qRT-PCR (quantitative reverse transcription–PCR) on a Rotor-Gene 2000 instrument (Corbett Research) using 50 ng of cDNA with 2× QuantiMix SYBR kit (J&M) and primers 17/18. All qRT-PCRs were performed in triplicate. The endogenous actin gene was used as internal control with primers 19/20. qRT-PCR primers are listed in Supplementary Table S1. Relative gene expression was calculated using the ΔΔCT method.
Transient expression in N. benthamiana leaves
MrBBS and its N-terminally deleted versions were PCR-amplified with specific primers (21–24) and cloned into pDONR221, and then transferred to pK7WG2D using the Gateway system (Invitrogen). The recombinant vectors were transformed into Agrobacterium tumefaciens LBA4404, and transformed cells were cultured overnight at 28°C in 50 ml of LB medium containing rifampicin (20 μg·ml−1) and spectinomycin (100 μg·ml−1). The cultures were centrifuged at 3500 g for 10 min, and resuspended in buffer (10 mM Mes, 10 mM MgCl2 and 150 μM acetosyringone, pH 5.6) to adjust the D600 to 0.5. The suspension was incubated further at room temperature for 8 h. The culture was infiltrated into the abaxial side of 6-week-old N. benthamiana leaves with a 1 ml syringe. After 4 days, leaves were harvested and ground in liquid nitrogen with a pestle and mortar, followed by vortex-mixing in ethyl acetate and sonication for 15 min. The extracts were centrifuged at 3500 g for 10 min and dehydrated with Na2SO4. GC–MS analysis conditions were the same as above for in vivo yeast characterization except for the temperature programme (1 min hold at 130°C, linear increase to 230°C at 10°C·min−1 and subsequent linear increase to 310°C at 20°C·min−1 with a 10 min hold).
Subcellular localization of bisabolol synthase
Full-length MrBBS and a DNA fragment encoding its N-terminal 52 residues were amplified by PCR (primers 21–27). The PCR products were cloned into pEAQ-GFP vector , and infiltration experiments were performed as described above. Confocal microscopy was performed with a multi-photon confocal laser-scanning microscope (DE/LSM 510 NLO, Carl Zeiss). Infiltrated N. benthamiana leaves were excited with a 488 nm laser at 10%. Emission was detected between 500 and 530 nm. The chlorophyll autofluorescence was detected at 560 nm.
Bioinformatic analysis to identify candidate sesqui-TPS genes
Transcriptome data of M. recutita were obtained from the publicly available PhytoMetaSyn project (http://www.phytometasyn.ca). Putative sesquiterpene synthase (sesqui-TPS) genes were screened from the assembled dataset using tBLASTn. Subcellular localization of MrBBS was predicted by ChloroP (http://cbs.dtu.dk/services/ChloroP), TargetP (http://cbs.dtu.dk/services/TargetP) and PSORT (http://psort.hgc.jp/form2.html). For quantitative mapping, the ‘Map reads to reference’ tool in the CLC Genomics Workbench (5.5.1) was employed using a single read Illumina data. The parameters used were: mismatch cost, 2; insertion cost, 1; deletion cost, 1; length fraction, 0.8; similarity fraction, 0.98.
Sesquiterpene synthase isolation from chamomile cDNA
The focus of the present study was to identify (−)-α-bisabolol synthase from chamomile (Matricaria recutita), an Asteraceae family plant species known to synthesize (−)-α-bisabolol as one of the main floral terpene products. Using chamomile floret mRNAs, a half plate of 454 and one lane of Illumina sequencing were carried out to determine deep transcript sequences. From 454 titanium sequencing, 735449 reads with an average length of 480 bp were generated, and 602979 cleaned reads were assembled using the MIRA algorithm , yielding 44324 contigs. From Illumina GAII sequencing, 64608668 reads at an average length of 108 bp were sequenced, and 54992498 cleaned reads with an average length of 81 bp were assembled using the Velvet algorithm , yielding 59718 unique transcripts. The BLAST-searchable assembled transcript data for both 454 and Illumina data are accessible through the PhytoMetaSyn website (http://www.phytometasyn.ca) [33,34].
From these assembled transcripts, eight sesqui-TPS cDNAs encoding full-length enzymes were identified, and subsequently all eight cDNAs were PCR-cloned from chamomile floral tissues. These cDNAs were named MrTPS1–MrTPS8. Of these, two sesqui-TPS clones, MrTPS3 and MrTPS5, displayed >99% sequence identity with the previously reported chamomile sesqui-TPSs, and these were shown to encode (−)-germacrene D synthase and germacrene A synthase respectively . Therefore these two clones were excluded from subsequent functional analysis, and the remaining six clones (MrTPS1, MrTPS 2, MrTPS 4 and MrTPS 6–MrTPS8) were subjected to functional expression in yeast.
Activity screening of sesquiterpene synthases in yeast
In order to assess whether (−)-α-bisabolol synthase is present among the isolated cDNAs, the ORFs of the six clones were expressed under the Gal1 promoter of the pESC-Leu2d plasmid in yeast . The plasmid harbouring each sesqui-TPS cDNA was used to transform the EPY300 yeast strain engineered to endogenously produce an elevated level of FPP, a sesqui-TPS substrate [28,29]. Volatile terpenes produced from the transgenic yeast cultures were sequestered by dodecane overlaid on the culture, and terpenes trapped in the dodecane were analysed by GC–MS. GC–MS analyses showed that yeast strains expressing MrTPS1, MrTPS4 and MrTPS6 synthesize terpene products different from those of the vector control, demonstrating the synthesis of novel terpenes by these chamomile sesqui-TPS enzymes (Figure 2A). However, volatile product profiles of the yeast expressing MrTPS2, MrTPS7 and MrTPS8 were the same as the control (results not shown). The EI (electron impact) spectral matches of the new volatiles against the MS database suggested that the MrTPS1, MrTPS4 and MrTPS6-expressing yeast strains synthesize α-bisabolol, bicyclogermacrene and β-farnesene respectively, as major terpene products (Figures 2C–2E). Of particular interest is the MrTPS1-expressing yeast strain showing a dominant peak with an EI–mass spectrum identical with that of α-bisabolol in the database. To confirm this, an authentic (−)-α-bisabolol standard was analysed by GC–MS together with the MrTPS1 product, and the MrTPS1 product showed an identical mass-fragmenting pattern and comparable retention time with those of the (−)-α-bisabolol standard (Figures 2B and 2C). Aside from the farnesol (Figure 2, peak 4) derived from the FPP substrate in yeast, α-bisabolol was a single terpene produced by MrTPS1. On the basis of this result, MrTPS1 was named chamomile α-bisabolol synthase (MrBBS).
In vivo screening of sesqui-TPSs isolated from chamomile in engineered yeast
Relative expression of MrBBS in chamomile
A recent chemical analysis of chamomile has shown that α-bisabolol and its oxidized products, bisabolol oxide A and B, are the most abundant terpenoids in ray and disk florets among different plant tissues (63–81% of total terpenes) . It is therefore predicted that the expression of MrBBS is highest in chamomile florets. To determine whether MrBBS transcripts are most abundant in floral tissues, qRT-PCR analysis was conducted in various tissues, including three different developmental stages of flowers. qRT-PCR results showed that MrBBS transcripts were 24–58-fold higher in florets than in leaf (Figure 3). In different floral stages, open flowers had an approximately 2.5-fold higher level of MrBBS expression than in immature flowers.
Relative MrBBS transcript levels by floral developmental stages and organs
In addition to qRT-PCR data, massive Illumina sequencing enabled us to compare the relative expression levels of eight MrTPS genes in chamomile floral tissue. To do so, approximately 32.3 million Illumina single reads were quantitatively mapped on reference sequences of the eight sesqui-TPSs. The abundance of their transcripts varied between 320 and 5500 reads per million total reads, with MrBBS being the most abundant transcript in chamomile floral tissue (Table 1). These expression studies showed that MrBBS is predominantly expressed in the open floral tissue, in which MrBBS is the most abundant transcript among the eight sesqui-TPSs identified.
|Name||ORF length||Unique read number||Read number per million total reads|
|Name||ORF length||Unique read number||Read number per million total reads|
De novo α-bisabolol production in yeast
To elucidate the stereo-configuration of α-bisabolol, it was essential to acquire a sufficient amount of MrBBS-produced α-bisabolol in high purity. Before purification, the level of α-bisabolol production from the transgenic EPY300 yeast strain was examined. In a simple shaking flask condition, when the whole culture was extracted using organic solvent, 8.1±0.4 mg (n=5) of α-bisabolol was produced from 1 litre of culture after 4 days of cultivation (Supplementary Figure S1A at http://www.biochemj.org/bj/463/bj4630239add.htm). Although this productivity was satisfactory, we further sought yield improvement. It has been reported that the translational fusion of soluble protein to the N-terminus of sesqui-TPS can markedly increase in vivo production of sesquiterpene in yeast, probably because of increased protein solubility [35,36]. Therefore commonly used soluble protein (i.e. MBP, GST or TRX) was fused to the N-terminus of MrBBS, and the relative α-bisabolol production from the yeast lines expressing these three fusion genes were compared (Supplementary Figure S1B). MBP-fused MrBBS showed that α-bisabolol production was comparable with that of non-fused MrBBS, but GST- and TRX-fused MrBBS showed a lower level of α-bisabolol production (25% and 32% of non-fused MrBBS respectively) (Supplementary Figure S1C). We concluded that yeast expressing non-fused native MrBBS was optimal for α-bisabolol production, and hence it was used for subsequent α-bisabolol purification.
Purification and structural elucidation of (−)-α-bisabolol
GC–MS data strongly suggested that the MrBBS product is α-bisabolol. To unambiguously distinguish between the four stereoisomers of α-bisabolol (Figure 1), 34 mg of α-bisabolol was purified from the whole yeast culture using a silica column, and the purified product was subjected to 1D and 2D NMR and optical rotation analysis in comparison with the authentic (−)-α-bisabolol standard. It is known that α-bisabolol and its diastereomer, epi-α-bisabolol, have slight but consistent differences in their 13C- and 1H-chemical shifts. In particular, signals of H3 and H15, and C2, C4, C6, C8 and C14 are sufficiently different between these two isomers [20,21,37]. In 13C- and 1H-NMR analyses, the chemical shift values of the MrBBS-product were shown to coincide completely with those of (−)-α-bisabolol standard [full 13C- and 1H-NMR signal assignments of MrBBS-product and (−)-α-bisabolol are given in Table 2]. Thus the MrBBS-product was clearly determined to be α-bisabolol and not epi-α-bisabolol. Since α-bisabolol has two enantiomers, (+)- and (−)-α-bisabolol, with identical chemical shifts, determination of the absolute configuration of MrBBS product was addressed by optical rotation measurement. In this analysis, the observed optical rotation value, [α]25−65.8 (ethanol), was sufficiently close to [α]25−67.6 from the authentic (−)-α-bisabolol standard. This observation conclusively established the stereochemistry of the MrBBS product as (−)-α-bisabolol or (−)-(1S,7S)-bisabolol, which is currently commercialized as ‘natural identical’ α-bisabolol.
|2a/b||1.76 (m)/1.95 (m)||27.15|
|6a/b||1.25 (m)/1.89 (m)||23.51|
|12||1.59 (d, J=1.28Hz)||17.91|
|2a/b||1.76 (m)/1.95 (m)||27.15|
|6a/b||1.25 (m)/1.89 (m)||23.51|
|12||1.59 (d, J=1.28Hz)||17.91|
The N-terminal motif in MrBBS influences its catalytic activity
It is well established that sesqui-TPS localizes to the cytosol in agreement with its utilization of the cytosolic MVA-pathway-derived FPP. Intriguingly, however, computational prediction of subcellular location of MrBBS indicated the presence of chloroplast-targeting sequences. The TargetP and ChloroP algorithms predicted an N-terminal 52-residue plastid-targeting peptide, whereas the PSORT algorithm predicted a N-terminal 14-residue plastid-targeting peptide. These in silico predictions suggested possible plastidic localization of MrBBS. Therefore we investigated the function of these unusual putative plastid-targeting peptides.
First, to examine whether the N-terminal motif affects MrBBS activity in planta, two truncated versions of MrBBS were constructed and transiently expressed in N. benthamiana leaves. The Δ23MrBBS version lacked 23 residues in the characteristic signature for plastidic localization, including 11 serine and threonine amino acids. The second Δ52MrBBS version lacked the 52 residues containing the plastid-targeting transit peptide predicted by TargetP and ChloroP. When (−)-α-bisabolol was measured from tobacco leaves 4 days after infiltration, full-length MrBBS-expressing leaves accumulated ~10 μg of (−)-α-bisabolol per g of fresh weight (Figure 4A, and Supplementary Figure S2 at http://www.biochemj.org/bj/463/bj4630239add.htm). However, the leaves expressing Δ23MrBBS accumulated a 41-fold lower level of (−)-α-bisabolol than those expressing full MrBBS. From the leaves expressing Δ52MrBBS, no (−)-α-bisabolol was detected. These results suggested that a short N-terminal 23-residue peptide of MrBBS influenced in planta (−)-α-bisabolol-synthesizing activity either by targeting MrBBS to plastid or by directly affecting MrBBS activity.
Characterization of truncated forms of the MrBBS enzyme
Next, the plastidic localization of MrBBS was investigated by expressing the protein as a GFP fusion. Full-length MrBBS or its N-terminal 52 residues was fused to GFP by C-terminal tagging, followed by transient expression in N. benthamiana leaves. Confocal microscopy, however, showed that the GFP signal did not overlap with chloroplastic autofluorescence in both the full and 52-residue-fused constructs (Figure 5). Therefore MrBBS is not localized to the plastid, but rather is localized to cytosol similarly to other sesqui-TPS enzymes despite the presence of the putative plastid-targeting motif on its N-terminus. Furthermore, it can be inferred that a short N-terminal peptide directly influences MrBBS catalytic activity.
Subcellular localization of MrBBS in N. benthamiana mesophyll cells
Finally, in order to probe the role of the short N-terminal peptide on MrBBS activity, full-length MrBBS, Δ23MrBBS or Δ52MrBBS was expressed and purified from E. coli by its His6 tag appended to the C-terminus of MrBBS. During the purification procedure, it was apparent that the truncated forms of MrBBS were either unstable or prone to aggregation in E. coli since significantly less recombinant enzyme could be purified from Δ23MrBBS-expressing bacterial cells (Figure 4B). Δ52MrBBS recombinant enzyme could not be purified from the soluble fraction, although it was present in the insoluble fraction (results not shown). When the specific activities of the full-length MrBBS and Δ23MrBBS were measured using the same amount of enzyme, Δ23MrBBS specific activity was two orders of magnitude lower than that of full-length MrBBS, indicating that the catalytic activity of Δ23MrBBS was severely compromised (Figure 4C).
Taken together, these data suggest that the N-terminal transit-peptide-like motif does not have a role in localizing MrBBS to plastids, but directly influences MrBBS stability and/or catalytic activity. A small deletion as short as 23 residues was sufficiently detrimental to MrBBS activity in vitro and in planta.
MrBBS enzyme characterization
Using purified full-length MrBBS-His6 recombinant enzyme, kinetic properties of MrBBS were examined. MrBBS kinetic parameters were determined to have a Km of 3.6 μM for FPP and a kcat of 4.6×10−3. This is comparable with other reported sesqui-TPSs (Supplementary Figure S3 at http://www.biochemj.org/bj/463/bj4630239add.htm). Kinetic parameters of Δ23MrBBS, however, could not be determined because of its negligible activity. The in vitro MrBBS-synthesized α-bisabolol was more carefully examined using a chiral column and the GC programme previously tuned to clearly separate different α-bisabolol isomers . Besides (−)-α-bisabolol, only one other terpene, putatively assigned as β-farnesene by database search (91% spectral match), was identified in the GC–MS analysis, but this minor terpene constituted only 2% of the total terpenoids (Supplementary Figure S4 at http://www.biochemj.org/bj/463/bj4630239add.htm). No evidence of another isomer formation, other than (−)-α-bisabolol, was obtained in this chiral GC–MS analysis (Supplementary Figure S4). Therefore the MrBBS enzyme in vitro synthesizes enantiopure (−)-α-bisabolol with 98% terpene purity, consistent with in vivo data in yeast and tobacco.
(−)-α-Bisabolol, a widely added component of cosmetic products owing to its tissue-healing and -moisturizing properties, is isolated from the Brazilian candeia tree. Recent concerns over the sustainability of the candeia tree populations in Brazil have resulted in calls for the chemical synthesis of the bisabolol to meet commercial demands. However, the chemically synthesized α-bisabolol requires a costly purification process to obtain natural identical (−)-α-isomer. Therefore sustainable and economic production of (−)-α-bisabolol has become an important issue in the speciality chemical industry. In the present study, we identified a novel stereoselective (−)-α-bisabolol synthase (MrBBS) from chamomile and demonstrated its use for microbial and in planta (−)-α-bisabolol synthesis.
MrBBS is a unique enzyme of enantioselective (−)-α-bisabolol synthesis
One key characteristic of chamomile MrBBS, relevant to biotechnological uses, is its exclusive synthesis of (−)-α-bisabolol as a single terpene product. Only a negligible amount of farnesene was detected as a co-product. It has been reported that some TPSs can synthesize dozens of products. In some extreme cases, Medicago truncatula TPS generated 27 products and Abies grandis TPS generated 52 products [38,39]. With regard to TPS-mediated bisabolol synthesis, α-bisabolene synthases from Sandalwood and Arabidopsis synthesize α-bisabolol as a minor co-product [24,25]. Therefore, before the present study, it was reasonably postulated that (−)-α-bisabolol in chamomile was synthesized by one or more multifunctional TPSs, which could make it difficult to utilize (−)-α-bisabolol TPS for biotechnological purposes. Our MrBBS characterization data, however, showed that a single enzyme, MrBBS, is sufficient to produce a single enantioselective (−)-α-bisabolol as a major catalytic product (>98%). Although sesqui-TPSs synthesizing α-bisabolol as a single major product have been cloned and characterized previously, these enzymes synthesized structurally different undesirable isomers (Lippia dulcis and Streptomyces citricolor) or α-bisabolol of unknown stereochemistry (Artemisia annua) [21–23]. At present, the catalytic feature of MrBBS [i.e. formation of a single enantiopure (−)-α-bisabolol] is unique and suitable for biotechnological synthesis of natural identical (−)-α-bisabolol. As a proof-of-concept, MrBBS cDNA was expressed in yeast and tobacco, and we confirmed that (−)-α-bisabolol titres of 8 mg/l of yeast culture and 10 μg/g of tobacco fresh weight can be reached. These are the first demonstrations of enantiopure (−)-α-bisabolol synthesis beyond that observed by the candeia tree and chamomile.
Proposed mechanism of (−)-α-bisabolol synthesis by MrBBS
The (1S)-configuration of (−)-α-bisabolol is reminiscent of the cyclization mechanism of amorphadiene synthase, in which the initial formation of (1S)-bisabolyl carbocation is followed by hydride shifts to form amorphadiene skeleton . On the basis of this relationship, the mechanism of action of MrBBS is proposed in Figure 6. FPP is isomerized to NPP (nerolidyl diphosphate) to assume the correct conformation for the initial formation of a six-membered ring so that the overlap of the p-orbital at the C6 Re face with the C1 p-orbital would result in the formation of carbocation with (1S)-configuration. Quenching of this carbocation by a water molecule approached from the Si face at C7 completes the generation of (−)-α-bisabolol. We have previously cloned LdBBS [(+)-epi-α-bisabolol synthase from Lippia dulcis] . For the synthesis of (+)-epi-α-bisabolol by LdBBS, NPP would fold in such a way that the C1 p-orbital overlaps with the p-orbital of the C6 Si face to result in an intermediate carbocation with (1R)-configuration (Figure 6). Addition of water to the cationic C7 Si face would lead to (1R,7S)-configuration. Therefore a comparison of 3D structures of active sites of MrBBS and LdBBS would provide a unique opportunity to identify structural elements that guide the correct folding of FPP (or NPP) required to arrive at different configurations at C1 of bisabolol.
Proposed mechanism of MrBBS
The N-terminal motif of MrBBS influences its catalytic activity
Computational analyses of MrBBS predicted the N-terminal 52 amino acid residues to be a chloroplast-targeting transit peptide. There is precedence of chloroplast-targeted sesqui-TPS in wild tomato (Solanum habrochaites) . In this case, the chloroplast-localized FPPS (FPP synthase) that contained the predicted plastid-localizing motif synthesizes Z,Z-FPP (or cis,cis-FPP), which is an isomer of common E,E-FPP (or trans,trans-FPP). Subsequently, chloroplast-localized sesqui-TPS catalyses the formation of santalene and bergamotene from Z,Z-FPP. Since in silico analyses of MrBBS sequences consistently predicted it to be a plastid-localizing enzyme, we suspected that MrBBS could also localize to the chloroplast; however, no evidence of chloroplast localization was obtained by expression of a GFP fusion to MrBBS, and MrBBS appears to localize in the cytosol. Moreover, in contrast with the S. habrochaites FPPS, the chamomile FPPS, identified from 454 and Illumina transcript datasets, does not possess a plastid-localizing transit peptide (results not shown). On the basis of these lines of evidence, we conclude that the computational analysis incorrectly predicted the MrBBS localization in this case.
Interestingly, the deletion of short 23 N-terminal residues in MrBBS effectively abolished its catalytic activity. On one hand, this result reinforced that the serine/threonine-rich motif is not a transit peptide since it is known that the transit peptide does not affect catalytic activity of TPS (see below). On the other hand, such a dramatic loss of sesqui-TPS activity is unexpected because the serine/threonine-rich 23-residue peptide is distal to the highly conserved (RX)X8W motif in sesqui-TPS. In mono-TPS (monoterpene synthase), this (RX)X8W motif is more refined to the RRX8W motif, and changing the RR motif to RP or RA, or removing the RR residues in limonene synthase (mono-TPS) results in a complete loss of catalytic activity . Although this RRX8W motif has functional significance in mono-TPS, a long peptide (50–90 residues, part of which is a transit peptide) proceeding the RRX8W motif is not required for mono-TPS catalytic activity. Hence entire residues can be removed without altering mono-TPS enzyme activity. In contrast, typical sesqui-TPS has a much shorter peptide in front of the (RX)X8W motif. For example, only five amino acids are present in front of the (RX)X8W motif in LdBBS  (Supplementary Figure S5 at http://www.biochemj.org/bj/463/bj4630239add.htm). Because of such an ambiguous status, the functional implication of the N-terminal sequence of sesqui-TPS has so far been overlooked.
Our data indicated that a short 23-residue peptide of MrBBS has a direct impact on its activity and may also influence the stability and/or solubility of the enzyme. It is not clear how the short peptide confers increased catalytic activity to MrBBS, although the N-terminal peptide could impose proper protein-folding and substrate-binding capability of the enzyme. Curiously, we have found that some sesqui-TPSs from the Asteraceae family also display this signature in their N-terminus [e.g. AaFAS (A. annua farnesene synthase) and MrTPS6/MrFAS (chamomile farnese synthase), see Supplementary Figure S5], indicating that this N-terminal signature may be retained in a small subset of sesqui-TPSs. We attempted to model the N-terminal peptide of MrBBS using the known crystal structure of 5-epi-aristolochene synthase in hopes to understand its role. However, X-ray crystallographic analyses of 5-epi-aristolochene synthases gave an indication of structural flexibility of the N-terminal region , which makes it difficult to implicate the role of the N-terminal region of sesqui-TPS at this moment. On the basis of our results, we propose that the serine/threonine-rich motif positioned upstream of the (RX)X8W motif does not encode a plastid-targeting transit peptide, but rather is critical for proper MrBBS function.
We thank Mohamed Attia (University of Calgary) for in silico identification of MrTPS cDNAs and Dr Douglas Muench (University of Calgary) for help with revision of the paper.
dimethyl allyl diphosphate
(+)-epi-α-bisabolol synthase from Lippia dulcis
chamomile (Matricaria recutita) α-bisabolol synthase
chamomile (Matricaria recutita) TPS
quantitative reverse transcription–PCR
Young-Jin Son characterized MrBBS in yeast, purified (−)-α-bisabolol, and examined MrBBS localization. Moonhyuk Kwon characterized recombinant MrBBS from E. coli and plants. Dae-Kyun Ro performed in silico analysis, and Soo-Un Kim elucidated (−)-α-bisabolol structure. Dae-Kyun Ro and Soo-Un Kim designed the study. All authors contributed to the preparation of the paper.
This work was supported by the Next-Generation BioGreen 21 Program [Systems and Synthetic Agrobiotech Center grant number PJ009549032014], Rural Development Administration, Republic of Korea to S.-U.K. and by Genome Canada [grant number PP780422], Genome Alberta, Government of Alberta [grant number PP789067], Genome Quebec, McGill Innovation Centre, MDEIE (Ministére du Développement Économique, Innovation et Exportation) and Canada Research Chair programme to D.-K.R.
Nucleotide sequence of identified Matricaria recutita sesquiterpene synthases have been submitted to the DDBJ, EMBL, GenBank® and GSDB Nucleotide Sequence Databases with accession numbers: KJ020282 [MrTPS1, (−)-α-bisabolol synthase], KJ020284 (MrTPS4, bicyclogermacrene synthase) and KJ020283 (MrTPS6, β-farnesene synthase).