Carotenoids are the precursors of important fragrance compounds in flowers of Osmanthus fragrans Lour. var. aurantiacus, which exhibit the highest diversity of carotenoid-derived volatiles among the flowering plants investigated. A cDNA encoding a carotenoid cleavage enzyme, OfCCD1, was identified from transcripts isolated from flowers of O. fragrans Lour. It is shown that the recombinant enzymes cleave carotenes to produce α-ionone and β-ionone in in vitro assays. It was also found that carotenoid content, volatile emissions, and OfCCD1 transcript levels are subjected to photorhythmic changes and principally increased during daylight hours. At the times when OfCCD1 transcript levels reached their maxima, the carotenoid content remained low or slightly decreased. The emission of ionones was also higher during the day; however, emissions decreased at a lower rate than the transcript levels. Moreover, carotenoid content increased from the first to the second day, whereas the volatile release decreased, and the OfCCD1 transcript levels displayed steady-state oscillations, suggesting that the substrate availability in the cellular compartments is changing or other regulatory factors are involved in volatile norisoprenoid formation. Furthermore, the sensory evaluation of the aroma of the model mixtures suggests that the proportionally higher contribution of α-ionone and β-ionone to total volatile emissions in the evening is probably the reason for the increased perception by humans of the scent emission of Osmanthus flowers.
Osmanthus fragrans Lour. is a shrub native to East Asia, and horticultural varieties can be found from Japan through China, Indo-China, Thailand, and India, to the Caucasus region. The petals of the evergreen Oleaceae flowers show by far the highest diversity of carotenoid-derived aroma compounds among the flowering plants (Kaiser, 2002). Because of its unique scent, commercial extracts are in high demand for use in the production of expensive perfumes and cosmetics. In China, the essential oils are used for flavouring tea, wine, and foods. The dominant compound in the essential oils of O. fragrans is β-ionone (Wang et al., 2009), however, how it is synthesized in these flowers is not known.
The contribution of CCD1 enzymes in norisoprenoid formation in flowers and fruits of other species has been demonstrated (Schwartz et al., 2001; Simkin et al., 2004a, b, 2008; Mathieu et al., 2005; Ibdah et al., 2006; Kato et al., 2006; Vogel et al., 2008; García-Limones et al., 2009; Huang et al., 2009,b). Carotenoid cleavage dioxygenases (CCDs) typically exhibit a high degree of regio-specificity for double bond positions and can cleave multiple substrates. There are examples of CCDs that can cleave multiple double bonds, i.e. enzymes of the CCD1 family are involved in the cleavage of the 5,6 (5′,6′); 7,8 (7′,8′); and 9,10 (9′10′) double bonds to produce divergent volatiles. LCD from Bixa orellana, ZmCCD1 from Zea mays, AtCCD1 from Arabidopsis thaliana, and LeCCD1 from Lycopersicon esculentum cleave lycopene at the 5,6 (5′,6′) double bonds (Bouvier et al., 2003, Vogel et al., 2008). OsCCD1 enzymes from rice can cleave the 7,8 (7′,8′) double bonds of the non-cyclic carotenoid lycopene (Ilg et al., 2009). A substantial number of enzymes involved in the cleavage of the 9,10 (9′10′) double bonds of carotenoids have been identified, such as AtCCD1 from Arabidopsis thaliana (Schwartz et al., 2001); PhCCD1 from Petunia hybrida (Simkin et al., 2004a); LeCCD1 from Lycopersicon esculentum (Simkin et al., 2004b); VvCCD1 from Vitis vinifera (Mathieu et al., 2005); CmCCD1 from Cucumis melo (Ibdah et al., 2006); CitCCD1 from Citrus limon, Citrus sinensis, and Citrus unshiu (Kato et al., 2006); CcCCD1 from Coffea canephora and CaCCD1 from Coffea arabica (Simkin et al., 2008); ZmCCD1 from Zea mays (Vogel et al., 2008); FaCCD1 from Fragaria ananassa (García-Limones et al., 2009); or RdCCD1 from Rosa damascena (Huang et al., 2009,b). Moreover, CCD4 enzymes from Crocus sativus, Rosa damascena, Osmanthus fragrans, Malus domestica, Chrysanthemum morifolium (Rubio et al., 2008; Huang et al., 2009,a) and CCD7 and CCD8 from Arabidopsis thaliana (Schwartz et al., 2004) can cleave their carotenoid or apocarotenoid substrates at the 9,10 (9′10′) double bonds. The role of CCD7 and CCD8 in the production of downstream metabolites involved in branching was known (Schwartz et al., 2004) before the carotenoid-derived strigolactones were identified to be involved in the stimulation of the colonization of arbuscular mycorrhizal fungi (Akiyama et al., 2005), the germination of parasitic plant seeds (Bouwmeester et al., 2007), and bud outgrowth (Umehara et al., 2008).
Although both CCD1 and CCD4 cleave their substrates at the same 9,10 (9′10′) double bonds, CCD4 enzymes only cleave cyclic non-polar carotenoids and apocarotenoids such as β-carotene and do not cleave xanthophylls and non-cyclic carotenoids such as zeaxanthin and lycopene (Rubio et al., 2008; Huang et al., 2009,a). Moreover CCD1 enzymes are cytoplasmic enzymes, whereas CCD4 enzymes carry a targeting sequence and are located in the plastids (Auldridge et al., 2006; Rubio et al., 2008). Hence, CCD4 enzymes have access to carotenoids located in the plastids. However, recombinant CCD4 isoforms oxidize different substrates; for example, AtCCD4 from A. thaliana and RdCCD4 from R. damascena prefer apocarotenoids and CmCCD4a from Chrysanthemum morifolium and MdCCD4 from M. domestica prefer carotenoids and are suggested to exhibit different biochemical functions (Huang et al., 2009,a). Recombinant CCD1 enzymes can utilize either carotenoids or apocarotenoids in vitro (Huang et al., 2009,b). However, it was recently suggested that the in vivo substrates of CCD1 are C27-apocarotenoids. RNAi-mediated MtCCD1 repression in mycorrhizal roots of Medicago truncatula caused an accumulation of C27-apocarotenoids and, therefore, have been suggested to be the major substrates for CCD1 enzymes in planta (Floss et al., 2009).
Previously α-carotene and β-carotene were identified as the two dominant carotenoids in petals of O. fragrans flowers (Baldermann, 2008). These two carotenes contribute to more than 90% of the total amount of carotenoids in flowers of O. fragrans. β-iοnone and α-ionone, two major ionones emitted from flowers of O. fragrans (Wang et al., 2009), are the proposed reaction products of the cleavage of the 9,10 (9'10') double bond of α-carotene and β-carotene (Fig. 1).
In petunia flowers, β-ionone emission was correlated with the transcript levels of PhCCD1 and in chrysanthemum flowers the white colour was associated with the transcript levels of CmCCD4a (Simkin et al., 2004,a; Ohmiya et al., 2006). None of these studies investigating the enzymatic carotenoid cleavage in flowers included the determination of the relative levels of the substrates (carotenoids) or reaction products (ionones), in addition to the analysis of the transcript levels.
The OfCCD4 from O. fragrans showed only very low activity with carotenoids and apocarotenoids and it is suggested that isoforms of CCD4 enzymes probably possess different biological functions (Huang et al., 2009a). It is therefore hypothesized that a member of the CCD1 family might be involved in the C13-norisoprenoid formation in flowers of O. fragrans. Its gene was identified and the enzyme it encodes was functionally characterized. The determination of the relative levels of the substrates and reaction products in addition to the analysis of the transcript levels of OfCCD1 by quantitative real-time PCR over the flowering period provided detailed information regarding the role of OfDDC1 in fragrance formation in flowers of O. fragrans.
Materials and methods
The flowers of Osmanthus fragrans Lour. var. aurantiacus were collected from the grounds of Shizuoka University, Japan during the flowering period in autumn 2006 and 2008. Flowers releasing the strongest odour during the unfurling process (stages 4 and 5), after changing colour from yellow to orange, were used for the detailed studies (Fig. 2).
Freshly cut flowering branches without leaves (stage 4) were exposed to constant temperature (22 °C) and relative humidity (70%). Samples were either subjected to a 12/12 h light/dark regime for 48 h or to continuous light or dark periods for 24 h. The light intensity inside the incubator was set to 80 μmol m−2 s−1. At least 4 g of O. fragrans flowers (8.4±3.2 mg and 6.8±0.8 mm per flower) were collected in intervals of 4 h and directly frozen with liquid nitrogen. All samples were stored at –80 °C prior to analysis.
Isolation and sequence analysis of OfCCD
The first strand cDNA was synthesized from 1 μg total RNA using the SMART RACE cDNA Amplification Kit (Clontech, Laboratories, Palo Alto, CA) according to the manufacturer's instructions. The cDNA fragments of OfCCD1 genes were amplified by PCR with the cDNA template and the primers that have been reported previously (see Supplementary Table S1 at JXB online) (Kato et al., 2006). The PCR product was purified by Microspin™ columns (Amersham Bioscience, Piscataway, NJ) and the amplified cDNAs of the 3′ and 5′ were cloned with the TOPO TA-Cloning Kit (Invitrogen, San Diego, CA) and sequenced. End-to-end PCR was performed with primers designed from the cDNA sequences obtained by RACE-PCR.
Expression and purification of the recombinant protein
The cDNA of OfCCD1 for the expression of recombinant proteins was amplified by PCR with the primers shown in Supplementary Table S1 at JXB online. The cDNA fragments were cloned into EcoRI and XhoI/BamHI sites of the pGEX-6P-1 plasmid (Amersham Bioscience). The plasmids were transformed into E. coli strain XL1-Blue cells. For protein expression, 2 ml of an overnight culture was used to inoculate 200 ml of YT medium (8 g l−1 tryptone, 10 g l−1 yeast extract, 5 g l−1 NaCl) containing the appropriate antibiotics. The cultures were grown at 27 °C until an OD600 of 0.6 was reached. The expression of the proteins was induced by addition of 200 μl of 100 mM isopropyl-β-D-thiogalactoside (IPTG).
To simplify the enzyme assay, cultures were alternatively grown with the addition of 1000 μl 100 mM FeSO4, and 100 μl 100 mM IPTG at 16 °C for 18 h. The E. coli cells were harvested by centrifugation and immediately frozen in liquid nitrogen. The cells were suspended in 20 ml phosphate buffered saline (140 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4) and 5 μl (7.5 U ml−1 suspension) lysozyme (rLysozyme, Novagen, Darmstadt, Germany) and 25 μl (20 U ml−1 suspension) cold active nuclease (Cryonase, Takara Bio Inc, Shiga, Japan) were added. After incubation for 30 min at room temperature the lysate was sonicated (6×30 s) (Ultrasonic Homogenizer, SMT Co., Ltd, Tokyo, Japan). Subsequently 1 ml 20% Triton X-100 (v/v) was added and the lysate moderatly shaken on ice for 30 min. The cell debris was removed by centrifugation at 20 000 g for 60 min and the recombinant protein bound to Sepharose 4B (Amersham Bioscience). The column was washed with 10 ml of phosphate buffered saline and 10 ml cleavage buffer (50 mM TRIS-HCl, 150 mM NaCl, 1 mM dithiothreitol, 0.05% Triton X-100 (v/v), and 1 mM ETDA in the case of cultivation of XL1-Blue in the absence of ferrous iron). The recombinant protein was obtained after digestion with PreScission Protease (Amersham Bioscience) at 5 °C overnight. The purity of the recombinant protein was analysed by SDS-PAGE on 12.5% polyacrylamide gel (e-PAGE 12.5%, Tokyo, Japan) using the Precission Plus Protein Dual Colour Standard (Bio-Rad, Tokyo, Japan) as marker. The proteins were stained with Bio-Safe Coomassie Blue G-250 Stain (BioRad) following the manufacturer's instructions.
Analysis of carotenoids
The method used to analyse the carotenoids was previously published in detail (Taylor et al., 2006). Briefly, at least 4 g flowers petals were ground in liquid nitrogen and 20 mg were transferred to a micro-centrifuge tube containing 350 ng of the internal standard β-apo-8'-carotenal. Firstly, 100 μl methanol and then 100 μl 50 mM TRIS-HCl (pH 8.0) containing 1 M NaCl were added. The carotenoids were extracted three times with 400 μl chloroform. The samples were stored under argon atmosphere at –80 °C prior to analysis. For HPLC analysis the samples were dissolved in 50 μl chloroform-methanol (1:4 v/v).
The carotenoids were analysed on a Jasco HPLC-PDA system (Tokyo, Japan) and separated on a C30-column (YMC Co. Ltd Japan, 4.6×250 mm, 5 μm). Mixtures of methanol methyl-tert-butyl-ether and water in different volume ratios (solvent A: 81/15/4 and solvent B: 6/90/4) were used as the mobile phases at a flow rate of 0.8 ml min−1. The carotenoids were separated in gradient mode from 30% to 100% solvent B within 20 min. Quantification was achieved from dose–response curves and identification by co-chromatography with references substances.
Total RNA extraction, reverse transcription, and real-time quantitative PCR
The total RNA was extracted from at least 4 g flower petals according to the method described by Ikoma et al. (1996). The genomic DNA was removed by on-column DNA digestion during the purification of the RNA using the RNeasy Mini Kit (Qiagen, Tokyo, Japan) according to the specifications given by the manufacturer.
The first strand cDNA was synthesized from 200 ng purified RNA using radom hexamers at 37 °C for 60 min and TaqMan reverse transcription reagents (Applied Biosystems, Tokyo, Japan). TaqMan MGB probes and primers were designed on based on common sequences using the Primer express software (Applied Biosystems; see Supplementary Table S1 at JXB online). For endogenous control, the TaqMan ribosomal RNA control reagent VIC probe (Applied Biosystems) was used. TaqMan real-time PCR was carried out with the TaqMan Universal PCR Master Mix (Applied Biosystems) using the ABI PRISM 7000 instrument (Applied Biosystems). The PCR program included an initial step of 50 °C for 2 min, a 10 min denaturation step at 95 °C and then 40 cycles of 15 s of denaturation at 95 °C and 1 min of hybridization/polymerization at 60 °C. The relative expression ratios were calculated using the ABI PRISM 7000 sequence detection software (Applied Biosystems) and normalized using the 18S ribosomal RNA results. Real-time quantitative PCR was performed in three replicates for each sample.
Headspace sampling and analysis of volatiles
The volatiles emitted by O. fragrans flowers were collected by dynamic headspace sampling. Freshly cut flowering branches, after removal of the leaves, were placed into headspace sampling units and filtered air (Charcoal filter) was pumped at a flow rate of 100 ml min−1 through the sampling unit. The volatiles were trapped on TenaxTA (180 mg) and analysed by GC-MS equipped with a thermal desorption system (TDS, Gerstel GmbH and Co. KG) under the following operating conditions: desorption temperature 260 °C, desorption time for 1 min, and split ratio of 15:1. The GC was equipped with a capillary TC-WAX column (GL Sciences Inc., Japan), 60 m×0.25 mm ID, and 0.5 μm film thickness. Helium was used as a carrier gas at a flow rate of 1.7 ml min−1. The temperature program of the oven was set to 40 °C for 5 min, then 3 °C min-1 up to 230 °C, and kept at this temperature for 60 min. The mass scan range was m/z 29–500 and the electric potential was set to EI 70 eV. Under these conditions, α- and β-ionones were detected at 59.6 min and 62.7 min, respectively.
Enzyme assays of recombinant proteins
The enzymatic activity of the recombinant OfCCD1 enzyme was assayed according to the method by Kato et al. (2006). For the enzymes obtained after cultivation of the E. coli cells in the presence of ferrous iron the activities were screened following the method described by Fleischmann et al. (2002).
Identification of the volatile reaction products of recombinant proteins
The volatile reaction products of the assay mixtures were analysed after solid phase micro-extraction (SPME) by gas chromatography mass spectrometry (GC-MS). Therefore a SPME fibre coated with 100 μm polymethylsiloxane (Supelco, Bellefonte, PA) was introduced into a headspace vial containing 2 ml enzymatic reaction mixture and 1 ml saturated sodium chloride solution and stirred for 1 h at 35 °C. The volatiles absorbed onto the fibre were analysed by GS-MS using a capillary Suplecowax column (GL Sciences Inc., Japan, 30 m×0.25 mm ID, 0.25 μm film thickness). The temperature program of the oven was set as follows: 50 °C maintained for 3 min, 5 °C min−1 up to 190 °C, 40 °C min−1 up to 240 °C, and held for 3 min. The mass scan range was set to m/z 50–300 and the electric potential to 1.00 kV. α-Ionone and β-ionone were detected at 23.1 min and 25.1 min, respectively.
For sensory evaluation, three model mixtures simulating O. fragrans odour were evaluated by 23 panelists (16 male and 7 female). The three model samples contained different amounts of β-ionone, α-ionone, linalool, linalool oxides (furanoids), and γ-decalactone in ratios and concentrations comparable to the emitted volatiles at 02.00 h, 10.00 h, and 18.00 h (indicated by 1, 2, and 3 in Fig 7A). The exact compositions of the model mixtures are listed in Supplementary Table S2 at JXB online. To consider the different amounts of emitted volatiles, 0.06 g, 2.0 g, and 0.6 g of mixtures 1, 2, and 3 were diluted in ethanol (w/w) to give 10 g of stock solutions 1, 2, and 3, respectively. The three concentrates were diluted 1:10 with ethanol (w/w) and subsequently with MilliQ water until their odour intensities were felt to be the same as living flowers (100 ppm). For sensory evaluation, 10 g of samples in concentrations of 0.1 ppm, 1 ppm, and 10 ppm in ascending order were presented to the panelists in closed sensory vials (total volume 50 ml), coded by a random three-digit number. The panelists were asked to evaluate the intensity of the samples from 1 (none) to 5 (very strong). 10 g of Milli-Q water (intensity 1) and model mixture 2 in a concentration of 100 ppm (intensity 5) were provided as reference samples. Model mixture 2 (100 ppm) was used because it simulates the aroma of O. fragrans flowers at the time of highest volatiles emission and the odour of this concentration was evaluated to be similar to living flowers. The results were averaged and analysed by ANOVA (analysis of variance) and Tukey's multiple comparison test. A probability level of 5% (P <0.05) was considered as significant.
Isolation and functional characterization of OfCCD1
To identify CCD homologues in O. fragrans flowers, degenerate oligonucletides based on conserved CCD sequences and amplified cDNA fragments of RNA isolated from O. fragrans flowers were designed. A full-length cDNA was subsequently obtained by RACE-PCR using gene-specific primers. The nucleotide sequence of this cDNA encodes a predicted protein of 563 residues. Phylogenetic analyses showed that the protein encoded by this cDNA clusters with other plant CCD1 enzymes (Fig. 3). The cDNA was therefore designated as OfCCD1.
To determine whether OfCCD1 encodes a functional CCD, the cDNA was transferred into a glutathione pGEX-6P-1 fusion vector for expression in E. coli. The recombinant protein was then purified using affinity chromatography. SDS-PAGE analysis on a 12.5% acrylamide gel identified a single band with a calculated molecular size of 65 kDa (see Supplementary Fig. S2 at JXB online). This was in accordance with a predicted molecular mass of 64 kDa.
Two in vitro assays were used to determine the cleavage activity of the recombinant protein. The first assay utilized ferrous iron, catalase, and ascorbic acid, and OfCCD1* purified from E. coli cells grown and induced under standard conditions (20 μM isopropyl β-D-thiogalactoside (IPTG), and 27 °C for 6 h). In the second assay, OfCCD1 was purified from E. coli cells induced by the addition of reduced amounts of IPTG (10 μM) and the bacteria were grown at 16 °C for an additional 18 h in the presence of ferrous iron (100 μM). The second enzymatic reaction buffer did not contain additional compounds (ferrous iron, catalase, and ascorbic acid) and the enzyme assay was carried out according to Fleischmann et al. (2002). After cultivation of the E. coli cells in the absence of ferrous iron, the isolated enzymes (OfCCD1*) showed no activity due to the lack of ferrous iron (Fig. 4). The rate of β-carotene degradation was similar to the chemical degradation of β-carotene under our experimental conditions (blank). After the addition of ferrous iron, catalase, and ascorbic acid to the buffer-substrate mixture, OfCCD1* activity could be detected, however, to obtain a comparable decrease of the initial amount of β-carotene for OfCCD1 and OfCCD1* longer reaction times were necessary (Fig. 4). OfCCD1 isolated from liquid cultures containing ferrous iron yielded active recombinant OfCCD1 enzymes that degraded β-carotene faster and without the supplement of additional ferrous iron (Fig. 4). However, high activities were only obtained directly after isolation and a stabilization of the enzymes with glycerol and ascorbate was necessary for storage. Because ascorbate also protects carotenoids against oxidation Km and Vmax values were not determined.
The volatile enzymatic reaction products of the cleavage of β-carotene and α-carotene were analysed by SPME-GS-MS. β-Ionone was detected in the headspace of the reaction mixtures after the addition of β-carotene as substrate, and both α- and β-ionone were detected as volatiles in the headspace after applying α-carotene as the substrate (Fig. 5). Other putative volatile reaction products derived from the carotenoid cleavage, such as β-cyclocitral resulting from the cleavage of the 7,8 (7′,8′) double bond were not detected. These results indicate that the activity of OfCCD1 is similar to that of other CCD1 enzymes involved in the cleavage of the 9,10 (9′,10′) double bonds of cyclic carotenoids (Fig. 5A, B).
Cleavage activity against α-carotene and β-carotene
Other CCD1 enzymes from Arabidopsis thaliana (Schwartz et al., 2001), tomato (Simkin et al., 2004,b), melon (Ibdah et al., 2006), maize (Vogel et al., 2008), and roses (Huang et al., 2009,a) showed a broad substrate specificity against carotenoids and apocarotenoids. All CCD1 enzymes cleave at the 9,10 (9′,10′) double bonds. The widest range of substrates was tested with the rose CCD1, which cleaved symmetric carotenoids at both ends simultaneously (Huang et al., 2009b). The RdCCD1 exhibited different affinities against the end group moieties of pseudo-symmetric molecules, except for the pseudo-symmetric xanthophyl lutein where similar levels of the reaction products 3-hydroxy-α-ionone and 3-hydroxy-β-ionone were observed. So far, in no study were the symmetric and pseudo-symmetric carotenes β- and α-carotene used as substrates. The ratio of α- and β-ionone of the reaction by OfCCD1 with α-carotene was approximately 1.7:1, indicating that the preferred cleavage site was the α-ionone ring moiety (Fig. 5).
Changes in OfCCD1 transcript levels
To determine the change in the OfCCD1 transcript levels over time, RNA was isolated from petals harvested in intervals of 4 h over 48 h. In addition, RNA was isolated from petals of cut flowering branches subjected either to 24 h light or to 24 h continuous dark periods. OfCCD1 transcript levels were determined by qRT-PCR. OfCCD1 steady-state transcript levels increased during the light periods and reached their maximal levels either at 12.00 h (noon) or 16.00 h (Fig. 6A). When the branches were subjected to constant darkness for 24 h, the transcript levels increased over time (Fig. 6B), even though at a reduced level compared with the OfCCD1 transcript level changes detected during the 12/12 h (dark/light) photoperiods (Fig. 6A). The maximum transcript levels were observed after 20 h incubation in continuous darkness, which was somewhat delayed compared with the flowers subjected in parallel to 12/12 h (dark/light) photorhythmic conditions (Fig. 6A, B). When the flowers were placed in constant light, lower steady-state transcript levels and changes at lower amplitude were observed (Fig. 6B). The maximal OfCCD1 transcript levels in flower petals of branches placed into 24 h continuous light were detected after an 8 h incubation period (Fig. 6B). At this time the flowers which were subjected to 12/12 h (dark/light) photoperiods exhibited the lowest transcript levels during the dark period (Fig. 6A).
To confirm that the OfCCD1 peak equals the transcript levels during day, the OfCCD1 transcripts of flowers grown outside and picked at 14.00 h were analysed. The transcript levels in O. fragrans flowers picked outside at the same times at various flowering stages (2, 4, 7, and 10; Fig. 2) changed 0.2-fold (2 arbitrary units; see Supplementary Fig. S1 at JXB online), whereas the transcripts varied up to 3.5-fold (21 arbitrary units) between the light and dark periods (Fig. 7A).
Changes in the carotenoid content in Osmanthus fragrans flowers
To determine the changes in the concentrations of α- and β-carotene, previously identified as the two major carotenoids (Baldermann, 2008), cut flowering branches were subjected to controlled environmental conditions and the concentrations of α- and β-carotene were analysed at intervals of 4 h. The concentrations of both α- and β-carotene increased in the presence of light (Fig. 6C, D), indicating that carotenoid biosynthesis in the flowers of O. fragrans is influenced by light. During the dark period, little change in carotenoid concentrations were observed and the levels remained nearly at the values reached during the previous light period.
Although OfCCD1 transcript levels and carotenoid concentrations peak with an offset of 4 h, the carotenoid content decreased or remained at a relatively low level, or example, at 12.00 h and 16.00 h of the first and second days, respectively (12/12 h (dark/light) photoperiods, Fig. 6A, C).
To test the effect of light on the carotenoid content, flowers were incubated under continuous 24 h dark or light (Fig. 6D). A nearly steady increase in carotenoids was obtained inside the flowers in the absence of light and only small changes were observed under continuous photoemission (Fig. 6D). Lower carotenoid concentrations in flowers subjected to 24 h continuous light or dark were detected compared with the flowers subjected to 12/12 h light/dark regime. The carotenoid content decreased at the peak of the OfCCD1 transcript levels at 04.00 h and 12.00 h under continuous illumination (Fig. 6B, D).
Volatile emission and α-ionone and β-ionone release in flowers of O. fragrans
The cut flowering branches subjected to 12/12 h (dark/light) photoperiods released maximum amounts of volatiles shortly after the beginning of the light period, following by a decrease until the lowest release during the dark period at 06.00 h (Fig. 6E). To test if the release of volatiles was regulated by circadian mechanisms, the cut flowering branches were subjected to a regime of 24 h constant light or constant dark. Flowers subjected to constant dark (Fig. 6F) showed a similar emission pattern to those flowers subjected to 12/12 h (dark/light) photoperiods (Fig. 6E). In both cases, the maximum levels of released volatiles were detected after 12 h. The results indicate that the release of volatiles is regulated by both light and circadian mechanisms. Flowers subjected to continuous light reached their maximum emission after 12 h, followed by a decrease in the emission over the rest of the experimental period. The scent emission decreased strongly between the first and second days when the samples were subjected to 12/12 h (dark/light) photoperiods.
The emission of the two primary cleavage products of the major carotenes of Osmanthus flowers, α- and β-ionones (Fig. 1) were next examined under the different photorhythmic conditions (Fig. 6G, H). As with the total emission of volatiles, the release of ionones was higher during the light periods. Flowers subjected to a 24 h continuous light regime emitted more ionones compared with those flowers treated in parallel under 12/12 h (dark/light) photoperiods (Fig. 6G, H).
Compared with the total volatile emission, the emission of ionones was higher in the early evening, which means that the contribution of the ionones to the total volatiles increases during the day (Fig. 7A). Volatile norisoprenoids are characterized by extremely low odour detection thresholds in humans. To evaluate changes in the scent of flowers of Osmanthus fragrans at different day times, the odour intensities of three model mixtures reflecting the floral scent at 02.00, 10.00, and 18.00 h (1, 2, and 3 in Fig. 7A; see Supplementary Table S2 at JXB online) were subjected to sensory evaluation. Although the amount of emitted volatiles were much higher at 10.00 h, the model mixtures 2 (10.00 h) and 3 (18.00 h) were evaluated as similar, but significantly different from model mixture 1 (02.00 h) (Fig. 7B).
Isolation and functional analysis of OfCCD1
It is well known that the colour of yellow flowers is often caused by the presence of large amounts of carotenoids. Some flowers also have a broad variety of carotenoid-derived scent compounds, as in the case of O. fragrans which has the highest diversity of carotenoid-derived scent compounds among 1250 flowering plants investigated (Kaiser, 2002). Hence, it was of special interest to elucidate the biosynthesis of these compounds in O. fragrans. Since a previous report indicated that a 75% decrease in β-ionone formation was observed in transgenic petunia plants in which PhCCD1 gene expression was inhibited (Simkin et al., 2004a), the possible role of OfCCD1 on ionone biosynthesis was examined in O. fragrans.
Apart from the formation of volatile C13-norisoprenoids through the action of CCD1, enzymatic cleavage of the 9,10 (9′,10′) double bond has also been demonstrated for CCD4 enzymes from C. sativus, R. damascena, C. morifolium, M. domestica, and A. thaliana (Rubio et al., 2008; Huang et al., 2009a).
Because the O. fragrans CCD4 showed very low activity against carotenoids and apocarotenoids (Huang et al., 2009b) this study focused on the functional characterization of the OfCCC1 enzyme. The putative amino acid sequence of OfCCD1 exhibited the conserved histidine residues of the active centre of CCDs and in the presence of ferrous iron, the recombinant enzymes showed cleavage activity towards the two dominant carotenoids (β-carotene and α-carotene) found in flowers of O. fragrans.
CCD1 enzymes cleave symmetric and pseudo-symmetric carotenoids differently. Based on the observations of Kloer and Schulz (2006), it was suggested that pseudo-symmetric molecules undergo a two-step cleavage. First, the enzyme cleaves the C40 substrate once, releasing the products, and then it binds to the primary non-volatile apocarotenoid and cleaves it for a second time. It should be noted that recent studies suggest that CCD1 enzymes cleave the primary derived cleavage products (C27-apocarotenoids) in the cytosol in vivo (Floss et al., 2009). In vitro, CCD1 can cleave either carotenoids or apocarotenoids (Huang et al., 2009,b). In roses, the non-volatile reaction products (C27-apaocarotenoids) of the first cleavage were only detected when the substrates contained different moieties at their ends. In this study, the symmetric β-carotene and the pseudo-symmetric α-carotene were used as substrates and the cleavage of α-carotene resulted in higher amounts of α-ionone, suggesting that the first site of cleavage is the one with the α-ionone moiety. However, in the case of the rose enzyme (RdCCD1), the same amounts of the reaction products 3-hydroxy-α-ionone and 3-hydroxy-β-ionone were obtained from pseudo-symmetric xanthophyll lutein (Huang et al., 2009b).
Photorhythmic changes of OfCCD1 transcript levels, carotenoid concentrations, and volatile emission
Photorhythmic volatile emission in plants has been demonstrated in several flowering plants (Matile and Altenburger, 1988; Loughrin et al., 1990; Helsper, 1998;, Picone et al., 2004; Dudavera and Pichersky, 2006). In general, nocturnally pollinated flowers tend to have maximum scent emission during the dark period, whereas the diurnally pollinated flowers release higher amounts of volatiles during the day. Volatile emission can be regulated either by light or by endogenous circadian mechanisms, mostly controlled at the gene expression transcription level (Hendel-Rahmanim et al., 2007). One group of plant enzymes is characterized by an increase in activity in young flowers and a decline during ageing, while a second group of enzymes show little or no decline at the end of the flower's life (Dudavera and Pichersky, 2006). During the floral development of Ipomoea sp., I. obscura, and I. nil, the CCD1 and CCD4 transcript levels decreased (Yamamizo et al., 2009). In the case of OfCCD1, the steady-state transcript levels are subjected to circadian mechanisms and have a peak during the day.
The concentrations of α-carotene and β-carotene also underwent photorhythmic changes. It is interesting to note that there is a negative correlation between the abundance of OfCCD1 mRNA and the concentrations of the substrates (α-carotene and β-carotene). In O. fragrans flowers, the carotenoid levels remained low or decreased if the transcript levels of OfCCD1 were high. The carotenoid content increased over the experimental interval and reached the maximal concentration under light conditions. The light/dark regulation of carotenoid biosynthesis was investigated in red pepper, where all transcript levels of genes involved in the carotenoid biosynthesis decreased under dark conditions (Simkin et al., 2003). In citrus fruits, the transcript levels of genes encoding enzymes involved in carotenoid biosynthesis as well as CCD transcript levels increased during e fruit maturation (Kato et al., 2007). In chrysanthemums, a negative correlation between CmCCD4a mRNA abundance and carotenoid content was observed. However, recently obtained results during the flower development of Ipomoea sp., I. obscura, and I. nil, suggest that the flower colour cannot be correlated to carotenoid degradation activity in Ipomoea plants (Yamamizo et al., 2009). In O. fragrans, the OfCCD4 showed very low activity against carotenoids and apocarotenoids (Huang et al., 2009a) and hence, the contribution to the biodegradation of carotenoids is unclear. However, the transcript levels were quite similar to those of OfCCD1 (M Kato, unpublished results). Hence, based on the work presented here, it might be suggested that, in Osmanthus flowers, the slight decrease in α-carotene and β-carotene levels observed in the light periods is at least partly due to the high activity of OfCCD1.
In petunia corollas, a correlation between mRNA abundance and β-ionone emission was observed (Simkin et al., 2004,a). However, emission was still increasing when transcript levels began decreasing during the afternoon. This study provides a similar observation in Osmanthus flowers, where the β-ionone emission remained at high levels after the transcript levels of OfCCD1 had already decreased during the day. It was suggested that there might be some limitation due to the substrate availability. Carotenoids are synthesized in the plastids whereas the CCD1 enzymes are located in the cytosol and therefore the cytosolic CCD1 enzymes have access only to those carotenoids distributed on the outer envelope of plastids, where, for example, significant amounts of β-carotene have been detected in pea chloroplasts (Markwell et al., 1992). In O. fragrans flowers, the carotenoid concentrations increased over the flowering period and, hence, a limiting factor for the reaction of OfCCD1 with the substrates could be the access inside the cell compartments.
Another regulatory factor could be the catalytic efficiency of enzymes with their substrates. Carotenoid cleavage enzymes purified from plant tissues exhibit different affinities towards β-carotene. For example the Km values for β-carotene obtained for carotenoid cleavage enzymes isolated from different fruits varied from 11.0 μM l−1 for quince fruit, 6.6 μM l−1 for nectarine, and 3.6 μM l−1 for star fruit, respectively (Fleischmann et al., 2002, 2003; Baldermann et al., 2005). In Osmanthus flowers, the carotenoid content increased and steady-state maximal transcript levels were observed under light conditions, whereas the emission of ionones, as enzymatic reaction products, decreased over the flowering period. It might be suggested that the catalytic efficiency of the OfCCD1 enzymes with their substrates is another regulatory factor.
Our results demonstrate that OfCCD1 in flowers of Osmanthus fragrans Lour. is probably involved in the oxidative cleavage of carotenoids to produce the volatile scent compounds α- and β-ionone. However, detailed analysis of carotenoids as putative precursors, transcript levels of OfCCD1, and volatile emissions indicate that the activity of this enzyme is not sufficient to account for the total emission of these volatiles. Additional work is needed to clarify the contribution of other carotenoid cleavage enzymes to ionone emission and identify the in vivo substrates.
Changes in β-ionone and α-ionone emission in relation to sent perception
Osmanthus flowers release their volatiles under light conditions. The analysis showed that the highest total volatile emission occurs in the morning, and total emission is lower in the afternoon. The release of β-ionone and α-ionone also strongly increased in the presence of light in the morning, and remained at a high level when the total volatile emission began decreasing during the afternoon. Because β-ionone (0.007 μg l−1; Buttery et al., 1990) and α-ionone (0.4 μg l−1, Teranishi and Buttery, 1987) have very low odour perception thresholds for humans, in water those compounds exhibit a strong impact on floral scents. The sensory evaluation of model mixtures reflecting the floral scent of O. fragrans flowers at 02.00, 10.00, and 18.00 h demonstrated that the scent in the morning and early evening is considered as similar, although the total volatile emission had decreased by approximately 3-fold. A similar example is the low amount of C13 norisoprenoids in rose, which nonetheless make a strong contribution to the scent; while constituting less than 1% of the total volatiles, they contribute to more than 90% to the scent impression by humans (Ohloff and Demole, 1987). Hence, the increasing amounts of α-ionone and β-ionone in relation to the total volatiles in the early evening are likely to be responsible for the stronger smell in the afternoon or early evening.
Supplementary data are available at JXB online.
Supplementary Table S1. Primer sequences used for OfCCD1 gene cloning and analysis by TaqMan® real time quantitative PCR assay.
Supplementary Table S2. Composition of model mixtures for sensory evaluation of O. fragrans aroma.
Supplementary Fig. S1.OfCCD1 transcript levels of flowers at stages 2, 4, 7, and 10 at 14.00 h.
Supplementary Fig. S1. SDS-PAGE of purified OfCCD1.
This work was supported by the Japan Society for the Promotion of Science (JSPS 07434). We thank Dr Eran Pichersky for useful comments on the manuscript and Dr Toshiyuki Ohnishi for valuable discussions.