Characterization of the rice carotenoid cleavage dioxygenase 1 reveals a novel route for geranial biosynthesis

Authors


S. Al-Babili, Institute for Biology II, Cell Biology, Albert-Ludwigs University of Freiburg, Schaenzlestr. 1, D-79104 Freiburg, Germany
Fax: +49 761 203 2675
Tel: +49 761 203 8454
E-mail: salim.albabili@biologie.uni-freiburg.de

Abstract

Carotenoid cleavage products – apocarotenoids – include biologically active compounds, such as hormones, pigments and volatiles. Their biosynthesis is initiated by the oxidative cleavage of C–C double bonds in carotenoid backbones, leading to aldehydes and/or ketones. This step is catalyzed by carotenoid oxygenases, which constitute an ubiquitous enzyme family, including the group of plant carotenoid cleavage dioxygenases 1 (CCD1s), which mediates the formation of volatile C13 ketones, such as β-ionone, by cleaving the C9–C10 and C9′–C10′ double bonds of cyclic and acyclic carotenoids. Recently, it was reported that plant CCD1s also act on the C5–C6/C5′–C6′ double bonds of acyclic carotenes, leading to the volatile C8 ketone 6-methyl-5-hepten-2-one. Using in vitro and in vivo assays, we show here that rice CCD1 converts lycopene into the three different volatiles, pseudoionone, 6-methyl-5-hepten-2-one, and geranial (C10), suggesting that the C7–C8/C7′–C8′ double bonds of acyclic carotenoid ends constitute a novel cleavage site for the CCD1 plant subfamily. The results were confirmed by HPLC, LC-MS and GC-MS analyses, and further substantiated by in vitro incubations with the monocyclic carotenoid 3-OH-γ-carotene and with linear synthetic substrates. Bicyclic carotenoids were cleaved, as reported for other plant CCD1s, at the C9–C10 and C9′–C10′ double bonds. Our study reveals a novel source for the widely occurring plant volatile geranial, which is the cleavage of noncyclic ends of carotenoids.

Abbreviations
CCD

carotenoid cleavage dioxygenase

GST

glutathione S-transferase

NIST

National Institute of Standards and Technology

OsCCD1

Oryza sativa carotenoid cleavage dioxygenase 1

SPME

solid-phase microextraction

Carotenoids are isoprenoid pigments synthesized by all photosynthetic organisms and some nonphotosynthetic bacteria and fungi. In plants, carotenoids are essential in protecting the photosynthetic apparatus from photo-oxidation, and represent essential constituents of the light-harvesting and of the reaction center complexes [1–4]. Carotenoids are also the source of apocarotenoids [5–7], which are physiologically active compounds, including the ubiquitous chromophore retinal, the chordate morphogen retinoic acid and the phytohormone abscisic acid as the best-known examples. Further carotenoid-derived signaling molecules are represented by strigolactones, a group of C15 apocarotenoids attracting both symbiotic arbuscular mycorrhizal fungi and parasitic plants [8,9] and, as recently shown, exerting functions as novel plant hormones regulating shoot branching [10,11]. In addition, the development of arbuscular mycorrhiza is also accompanied by accumulation of cyclohexenone (C13) and mycorradicin (C14) derivatives [12], all of which are apocarotenoids conferring a yellow pigmentation to infected roots [13]. Apocarotenoids, such as bixin in Bixa orellana [14] and saffron in Crocus sativus [15], are plant pigments of economic value.

The synthesis of apocarotenoids is initiated by the oxidative cleavage of double bonds in carotenoid backbones, generally catalyzed by carotenoid oxygenases, nonheme iron enzymes that are common in all taxa [5–7,16]. VP14 (viviparous14) from maize, which catalyzes the formation of the abscisic acid precursor xanthoxin by cleaving 9-cis-epoxy carotenoids [17], is the first identified member of this enzyme class. On the basis of their substrate specificity, VP14 and its orthologs have been termed 9-cis-epoxy-carotenoid dioxygenases. Plants possess a second group of carotenoid oxygenases, carotenoid cleavage dioxygenases (CCDs), which act on different carotenoid substrates. The CCDs of higher plants contribute to diverse physiological processes, including the regulation of the outgrowth of lateral shoot buds [18–20] and plastid development [21].

Plants release volatile apocarotenoids, including C13 ketones such as β-ionone and damascone [22], which constitute an essential aroma note in tea, grapes, roses, tobacco and wine [23]. Such compounds may arise by unspecific oxidative degradation or lipid co-oxidation processes, involving lipoxygenases [24]. Alternatively, they are produced by double bond-specific cleavage reactions mediated by peroxidases [25] or by CCDs such as members of the plant CCD1 subfamily. Plant CCD1s cleave numerous cyclic and linear all-trans-carotenoids at the C9–C10 and C9′–C10′ double bonds into C14 dialdehydes, which are common to all carotenoid substrates, and two variable end-group-derived C13 ketones [6,7,16]. The wide substrate specificity of plant CCD1s allows the production of divergent volatile C13 compounds, including β-ionophores, α-ionones, pseudoionone and geranylacetone. The first member of the CCD1 subfamily was identified from Arabidopsis thaliana [26], and was later shown to act as a dioxygenase [27]. Sequence homology then allowed the identification and charcterization of orthologs from several plant species, such as crocus, tomato, grape, melon, petunia and maize [15,28–32].

The biological function of CCD1s was confirmed by loss-of-function experiments in tomato fruits and petunia flowers, leading to decreased emission of β-ionone [27,31]. Moreover, recent studies on the CCD1 from maize indicated its involvement in the formation of cyclohexenone and mycorradicin in mycorrhizal roots [33]. Underscoring a role of CCD1 in carotenoid catabolism, seeds of Arabidopsis ccd1 mutants contained elevated carotenoid levels [34]. This suggested that the modification of CCD1 expression is instrumental for altering volatile production contributing to taste, or in increasing the carotenoid content in crop plant tissues where elevated provitamin A carotenoid levels are aimed for, such as high-β-carotene tomato [35,36], canola [37], golden rice [38] or golden potato [39]. Hence, the identification of substrates and cleavage sites of CCD1s from crop plants is considered to be a worthwhile approach.

It has recently been discovered that plant CCD1s exert additional activity at the C5–C6 and/or the C5′–C6′ double bonds of acyclic carotenoids, leading to the formation of the C8 ketone 6-methyl-5-hepten-2-one [32]. In this study, we investigated the enzymatic activities of the sole CCD1 [Oryza sativa CCD1 (OsCCD1)] occurring in rice. Our study revealed the C7′–C8′ double bond of linear and monocyclic carotenoids to be an additional novel cleavage site of OsCCD1, leading to geranial and indicating a novel plant route for the formation of this widespread volatile compound.

Results

Purified OsCCD1 cleaves acyclic apolycopenals at the C7–C8 double bond

To investigate the activity of OsCCD1, the corresponding cDNA was cloned and expressed as a glutathione S-transferase (GST)-fusion protein in Escherichia coli cells. However, the insolubility of the fusion protein, which could not be improved by modulating the expression conditions, hampered its purification. Therefore, the GST–OsCCD1 fusion was expressed in BL21(DE3) E. coli cells harboring the vector pGro7, which encodes the chaperones groES–groEL, enhancing correct folding. This resulted in a striking improvement of the GST–OsCCD1 solubility, allowing protein purification (Fig. S1).

It has been shown that CCD1s from Arabidopsis and maize maintain their regional specificity in cleaving the C9–C10 double bond with the synthetic apocarotenoid β-apo-8′-carotenal (C30), forming β-ionone (C13) and the C17 dialdehyde apo-8,10′-carotendial [27,32]. To test the cleavage activity of OsCCD1, in vitro assays were performed with this substrate, using purified enzyme. Subsequent HPLC analyses (data not shown) revealed a cleavage pattern identical to that of the plant CCD1s mentioned above. To determine the impact of the chain length and of the ionone ring on the cleavage pattern, purified OsCCD1 was incubated with β-apo-10′-carotenal (C27), which is shorter than β-apo-8′-carotenal (C30), and the acyclic substrates apo-10′-lycopenal (C27) and apo-8′-lycopenal (C30). HPLC analyses of the incubation with the cyclic β-apo-10′-carotenal (Fig. 1; substrate I) revealed an almost complete conversion of this substrate (Table S1) and the formation of the C14 dialdehyde apo-10,10′-carotendial (rosafluene dialdehyde; product 1) and β-ionone (C13; product 2), as confirmed by LC-MS and GC-MS analyses, respectively (data not shown). The relatively low amount of the C14 dialdehyde is probably a result of instability. The formation of β-ionone from β-apo-8′-carotenal (C30) and β-apo-10′-carotenal (C27) suggested that the cleavage of the C9–C10 double bond occurs independently of the chain length of monocyclic apocarotenals, pointing to the ionone ring as a determinant governing the reaction site.

Figure 1.

 (A) HPLC analyses of OsCCD1 in vitro assays with synthetic apocarotenoids. The cyclic β-apo-10′-carotenal (C27, I) was cleaved into apo-10,10′-carotendial (C14, 1) and β-ionone (C13, 2); the lower amount of the former is probably a result of its instability. The acyclic substrates apo-10′-lycopenal (C27, II) and apo-8′-lycopenal (C30, III) were converted into pseudoionone (3) and three dialdehydes of different chain lengths (1, 4 and 5 from II; 4, 6 and 7 from III), as indicated by their UV/visible-spectra (shown in the insets) and retention times. (B) Structures of the synthetic substrates showing the cleavage. I, β-apo-10′-carotenal; II, apo-10′-lycopenal; III, apo-8′-lycopenal. The substrates were cleaved at three different double bonds, including the novel site (b, shaded). Cleavage of the C9–C10 double bond (a) results in products 1 and 2 from I, products 1 and 3 from II, and products 3 and 4 from III. The cleavage of (b) leads to product 4 from II, and product 6 from III. The cleavage of the C5–C6 double bond (c) leads to product 5 from II, and product 7 from III. (C) Structures of the cleavage products detected: 1, apo-10,10′-carotendial (C14 dialdehyde); 2, β-ionone (C13); 3, pseudoionone (C13); 4, apo-8,10′-carotendial (C17 dialdehyde); 5, apo-6,10′-carotendial (C19 dialdehyde); 6, apo-8,8′-carotendial (C20 dialdehyde); 7, apo-6,8′-carotendial (C22 dialdehyde).

The two acyclic substrates were cleaved almost completely within 30 min (Table S1), proving them to be as suitable as the monocyclic apocarotenals. However, as shown in the HPLC analyses (Fig. 1), the cleavage of apo-10′-lycopenal (C27; substrate II) led to a more complex mixture of products, including three compounds (products 1, 4 and 5) identified as dialdehydes on the basis of the fine structure of the corresponding UV–visible spectra. The three dialdehydes differed in their chain lengths, as indicated by their retention times and absorption maxima. Products 1 and 5 were supposed to represent a C14 and a C19 dialdehyde, respectively. These are expected to arise upon cleavage at the known plant CCD1 sites (C9–C10 and C5–C6). The retention time of product 4 indicated a chain length between C14 and C19. This pointed to the recognition of a novel cleavage site, at the C7–C8 double bond, between the above mentioned C9–C10 and C5–C6 positions, yielding a C17 dialdehyde. To confirm their nature, the dialdehyde products, 1, 4 and 5, were purified and analyzed by LC-MS. In order to stabilize the C14 dialdehyde (product 1), it was derivatized with O-methyl-hydroxylamine-hydrochloride prior to LC-MS analyses. As shown in Fig. 2, derivatized product 1 exhibited an [M + (NCH3)2 + H]+ molecular ion of mass 275.17 (substrate I), consistent with the C14 dialdehyde structure, and the molecular ions for products 4 and 5 (Fig. 2; substrates II and III) proved their identities as C17 and C19 dialdehydes, respectively.

Figure 2.

 LC-MS identification of the dialdehydes produced in vitro. The three dialdehydes formed from apo-10′-lycopenal in vitro (Fig. 1; products 1, 4 and 5) represent a C14, a C17 and a C19 dialdehyde, respectively, as suggested by their molecular ions of mass 275.17 (I, [M + (NCH3)2 + H]+; product 1), 257.13 (II, [M + H]+; product 4) and 283.14 (III, [M + H]+; product 5).

HPLC analyses of the incubation with apo-8′-lycopenal (C30; substrate III in Fig. 1) confirmed novel cleavage of the C7–C8 double bond. As shown in Fig. 1, the reaction led to an equivalent series of C17, C20 and C22 dialdehydes (products 4, 6 and 7), corresponding to the cleavage of the C9–C10, C7–C8 and C5–C6 double bonds, respectively. The nature of these dialdehyde products was confirmed by LC-MS analyses (data not shown).

Cleavage of the three double bonds described above must also lead to the three different mono-oxygenated products pseudoionone (C13), geranial (C10) and 6-methyl-5-hepten-2-one (C8) (for structures, see Fig. 4). Pseudoionone was found by HPLC analysis (Fig. 1; product 3) and its presence was further demonstrated by GC-MS analyses, which also showed the formation of geranial and and 6-methyl-5-hepten-2-one (data not shown).

Figure 4.

 GC-MS analyses of volatile OsCCD1 products. Volatile compounds produced in lycopene-accumulating and OsCCD1-expressing cells were collected using SPME and subjected to GC-MS (IIII). As suggested by the [M]+ molecular ions (bold) and comparison of mass spectra with the NIST library, the in vivo activity of OsCCD1 led to the volatiles 6-methyl-5-hepten-2-one (C8, I), geraniol (C10, II) and pseudoionone (C13, III). IV represents the detection of geranial (C10) produced from 3-OH-γ-carotene in vitro.

OsCCD1 mediates double cleavage of different site combinations in 3-OH-γ-carotene and lycopene

To determine the cleavage sites in natural substrates, purified OsCCD1 was incubated with the bicyclic zeaxanthin, the monocyclic 3-OH-γ-carotene and the acyclic lycopene. As shown in Fig. 3, OsCCD1 converted zeaxanthin (substrate I) into the two products 3-OH-β-ionone (C13; product 1) and rosafluene-dialdehyde (C14; product 2), as confirmed by LC-MS and GC-MS analyses, respectively (data not shown). This suggested that OsCCD1 cleaves the C9–C10 and C9′–C10′ double bonds of bicyclic carotenoids, like other plant orthologs.

Figure 3.

 (A) HPLC analyses of OsCCD1 in vitro incubations with three natural carotenoids. UV–visible spectra of the products are shown in the insets. The bicyclic zeaxanthin (I) was converted into 3-OH-β-ionone (1) and apo-10,10′-carotendial (C14, 2). The products 3-OH-β-ionone (C13, 1), apo-10,10′-carotendial (C14 dialdehyde, 2), pseudoionone (C13, 3), apo-8,10′-carotendial (C17 dialdehyde, 4) and apo-6,10′-carotendial (C19 dialdehyde, 5) were obtained from 3-OH-γ-carotene (II). The cleavage of lycopene (III) led to pseudoionone (C13, 3), apo-8,10′-carotendial (C17 dialdehyde, 4) and apo-6,10′-carotendial (C19 dialdehyde, 5). Longer incubations with lycopene also resulted in the accumulation of the C14 dialdehyde apo-10,10′-carotendial corresponding to compound 2 (not shown). (B) Structures of the substrates showing the cleavage sites, including the novel C7–C8/C7′–C8′ double bonds (b, b′, shaded). I, zeaxanthin; II, 3-OH-γ-carotene; III, lycopene. OsCCD1 cleaves zeaxanthin at the C9–C10/C9′–C10′ double bonds (a/a′). 3-OH-γ-Carotene (II) is cleaved at the C9–C10 double bond (a) in combination with the C9′–C10′ double bond (a′), the C7′–C8′ double bond or the C5′–C6′ (c′) double bond. Lycopene (III) is cleaved either at the C9–C10 double bond (a) in combination with one of the three double bonds C9′–C10′ (a′), C7′–C8′ (b′) or C5′–C6′ (c′), or at the C9′–C10′ double bond (a′) in combination with one of the a, b or c sites. (C) Structures of the cleavage products detected.

Although it occurred at lower conversion rates than with the synthetic substrates (Table S1), we clearly observed the accumulation of the C17 and C19 dialdehydes (Fig. 3; products 4 and 5, respectively) from 3-OH-γ-carotene (Fig. 3; substrate II). Owing to its instability, the third dialdehyde (C14), expected from the cleavage of the C9–C10 and C9′–C10′ double bonds, occurred at low levels only (Fig. 3; product 2). The formation of the C14, C17 and C19 dialdehydes (for structures, see Fig. 3C) suggested a single cut at the C9–C10 double bond of the ring-bearing side of 3-OH-γ-carotene in combination with several cleavage options in the linear half of the molecule, namely at the C9′–C10′, C7′–C8′ and C5′–C6′ double bond. The occurrence of the C17 dialdehyde confirmed the novel site at the C7′–C8′ double bond observed with apolycopenals and implied the formation of geranial. Accordingly, the GC-MS analyses of the in vitro assays pointed to the conversion of 3-OH-γ-carotene into geranial (Fig. 4; substrate IV), as indicated by the detection of the expected [M]+ molecular ion of mass 152.3 and the fragmentation pattern (Fig. 4; substrate IV), which was correctly recognized by the National Institute of Standards and Technology (NIST) library (mass spectral search program Version 2.0). In addition, these GC-MS analyses (data not shown) revealed the formation of the known plant CCD1 mono-oxo products 6-methyl-5-hepten-2-one, 3-OH-β-ionone and pseudoionone, the latter two of which had already been detected in the HPLC analyses (Fig. 3; products 1 and 3).

The high lipophilicity of lycopene did not allow the use of octyl-β-glucoside as detergent in the corresponding in vitro assays. Therefore, lycopene micelles were produced with a Triton X mixture. The products formed from this substrate (Fig. 3; substrate III, products 3, 4 and 5) suggested the cleavage of the double bond combinations C9–C10/C7′–C8′, C9–C10/C5′–C6′ and their symmetrical counterparts. The C14 dialdehyde formed by cleavage of the C9–C10/C9′–C10′ double bond combination was only detectable in longer incubations (data not shown). GC-MS analyses of the lycopene incubations demonstrated the formation of pseudoionone and 6-methyl-5-hepten-2-one. However, geranial could not be detected, although the formation of the C17 dialdehyde confirmed the cleavage of lycopene at the novel C7–C8 or C7′–C8′ site.

OsCCD1 catalyzes the formation of three different volatiles from lycopene in vivo

To confirm cleavage at the C7–C8 or C7′–C8′ double bond in vivo, OsCCD1 was expressed in lycopene-accumulating E. coli cells, and volatile compounds were collected from the medium and analyzed by GC-MS. As shown in Fig. 4, the activity of the enzyme resulted in the accumulation of 6-methyl-5-hepten-2-one (I), the reduced form of geranial, geraniol (II), and pseudoionone (III). The three compounds, undetectable in the controls, were identified by their correct [M]+ molecular ions and by comparing the mass spectra with the NIST library. HPLC analyses of the corresponding cell pellets revealed the accumulation of a complex mixture of compounds, tentatively identified as dialcohols corresponding to three dialdehydes described above (data not shown).

OsCCD1 exhibits different preferences for the three cleavage sites

To gain insights into the preference of OsCCD1 for the three cleavage sites, we determined the relative amounts of the C17, C20 and C22 dialdehyde products formed upon incubation with apo-8′-lycopenal (C30), corresponding to C9–C10, C7–C8 and C5–C6 double bond recognition, respectively (Fig. 1B; substrate III). The enzyme exhibited by far the highest preference for the C9–C10 double bond in vitro, as suggested by the predominance of the C17 dialdehyde, which accounted for about 90% of the total dialdehyde products (Fig. 5A). The relative amount (about 7%) of the C20 dialdehyde was much higher than of the C22 dialdehyde (about 2%), indicating a higher affinity for the novel C7–C8 double bond than for the C5–C6 double bond. The instability of the C14 dialdehyde arising from the double cleavage at the C9–C10/C9′–C10′ double bonds in lycopene and 3-OH-γ-carotene hampered determination of the preference for these sites, and allowed only a comparison of the C9–C10/C7′–C8′ and C9–C10/C5′–C6′ cleavage products corresponding to the C17 and C19 dialdehydes, respectively. The higher relative amount of the C17 dialdehyde indicated a higher preference for the C7–C8 than for the C5–C6 double bond in 3-OH-γ-carotene, whereas the opposite tendency was observed with lycopene and apo-10′-lycopenal (Fig. 5B).

Figure 5.

 Determination of the relative amounts of dialdehyde products. (A) Incubations with apo-8′-lycopenal (C30). The peak areas of the three dialdehydes (C17, C20 and C22) formed from apo-8′-lycopenal were calculated at a max. plot of 350–550 nm. The values represent the proportion of each dialdehyde in the sum of the three peak areas. (B) Incubations with apo-10′-lycopenal (C27), 3-OH-γ-carotene and lycopene. The values represent ratios of the C17 and C19 dialdehydes in the sum of their peak areas calculated by integrating each peak at its individual λmax. Data represent the average of six independent incubations.

Discussion

Plant CCD1s are known to catalyze the cleavage of the C9–C10 and the C9′–C10′ double bonds of several carotenoids. Recently, it was shown that these enzymes can also cleave at the C5–C6 and/or the C5′–C6′ double bonds in lycopene [32]. This was deduced from GC-MS analyses of lycopene-accumulating E. coli cells expressing different plant CCD1s and from in vitro incubations with a CCD1 from maize (ZmCCD1), showing in both cases the formation of 6-methyl-5-hepten-2-one (C8), besides pseudoionone. Geranial (C10) was not detected, and cleavage of the C7–C8/C7′–C8′ double bonds was therefore excluded [32]. On the basis of our recent work on a Nostoc carotenoid oxygenase producing geranial and derivatives thereof from monocyclic carotenoids [40], we assumed that plant CCD1s may also be able to cleave the C7–C8/C7′–C8′ double bonds. Here, we demonstrate that the rice enzyme OsCCD1 cleaves linear ends of carotenoids at three different double bond positions, including the C7–C8/C7′–C8′ double bonds, leading to geranial.

To avoid further metabolization of products that can occur in vivo, we relied first on in vitro incubations using purified enzyme, which allowed clear identification of the products formed. In a first approach, we checked the site specificities using synthetic apocarotenals packed in octyl-β-glucoside micelles. This enabled us to observe the cleavage of the C7–C8 double bond. However, the confirmation of this novel activity with the highly lipophilic lycopene and γ-carotene required the use of different detergents. The best activities were obtained with micelles produced with a Triton X mixture, following the protocol of Prado-Cabrero et al. [41]. The accumulation of the C17 dialdehyde in the lycopene assays confirmed the cleavage at the C9–C10/C7′–C8′ double bonds. However, the activities determined were still low in comparison to the incubations with zeaxanthin, and did not allow clear identification of geranial. Furthermore, we did not detect significant conversion of lycopene in the 30 min incubations used to determine the substrate preferences of the enzyme (Table S1). This weak activity is probably due to the use of the Triton X mixture, which was necessary to soulibilize lycopene, but led to an overall reduction of enzyme activity (Table S1). Therefore, we used the more polar 3-OH-γ-carotene, which could be incorporated into octyl-β-glucoside micelles and which was readily cleaved at the double bond combinations C9–C10/C9′–C10′, C9–C10/C7′–C8′ and C9–C10/C5′–C6′.

The formation of multiple dialdehyde products allows some conclusions to be drawn on the site preferences of OsCCD1. For instance, the C14, C17 and C19 dialdehydes formed from lycopene and 3-OH-γ-carotene arise from cleavage of the C9–C10 double bond, which is combined with the C9′–C10′, C7′–C8′ or C5′–C6′ double bonds. The C9–C10/C9′–C10′ double bonds constitute the main site, as suggested by the determination of the relative amounts of the dialdehydes produced from apo-8′-lycopenal; the C9–C10/C9′–C10′ double bonds also constitute the sole cleavage site in ring-bearing moieties of substrates. This preference may explain the absence of the C20, C24 and C22 dialdehydes in the lycopene incubation, which would be expected if the cleavage reactions occurred only at the C7–C8/C7′–C8′ and/or C5/C6/C5′–C6′ double bonds. A further conclusion is that the first cleavage site plays a role in the determination of the second one in acyclic substrates. This is shown by the different preferences for the C7–C8 and C5–C6 double bonds (Fig. 5) in apo-10′-lycopenal (C27) and apo-8′-lycopenal (C30), which mimic a lycopene molecule cleaved at the C9′–C10′ and C7′–C8′ double bonds, respectively. Moreover, comparison of the relative amounts of the C19 and C17 dialdehydes accumulated in the incubations with the natural substrates lycopene and 3-OH-γ-carotene indicates that the preference of OsCCD1 for the C5′–C6′ and C7′–C8′ double bonds depends on the nature of the substrate.

The cleavage of a sole double bond in the cyclic moiety of 3-OH-γ-carotene and of monocyclic β-apocarotenals with different chain lengths indicates that the β-ionone end-group may determine the reaction site in the polyene chain. This may provide an explanation for the ‘wobbling’ of the enzyme on linear substrate moieties, where three different double bonds are being recognized. Similar results were obtained with the cyanobacterial enzyme SynACO, representing up to now the only carotenoid oxygenase with a known crystal structure [42]. SynACO cleaved β-apocarotenoids with different chain lengths at a sole site, the C15–C15′ double bond, leading to retinal and derivatives thereof [43]. In contrast, apolycopenals were cleaved at multiple positions, including the C15–C15′ double bond, and indicating ‘wobbling’ of the enzyme (S. Ruch, S. Al-Babili and P. Beyer; unpublished data).

Owing to the high sequence similarity of plant CCD1s, it appears likely that the cleavage of the C7–C8/C7′–C8′ double bonds of linear substrates is not unique to OsCCD1. Apart from the formation of geranial, the OsCCD1 cleavage reactions were identical to those of other plant CCD1s, as supported by the formation of pseudoionone, 6-methyl-5-hepten-2-one and β-ionones. Geranial is biologically active and known to exert antifungal and antimicrobial activities [44,45]; it represents a major volatile of tomato fruits [46] and roses [47]. Moreover, citral, the mixture of geranial and its cis-isomer neral, is a major component of the aroma of lemon grass and other lemon-scented plants. Geranial is synthesized from geranyl diphosphate by the enzymes geraniol synthase [48] and geraniol dehydrogenase [49]. The enzymatic cleavage of monocyclic and acyclic carotenoids into geranial represents a novel biosynthetic route, and may provide an explanation for the impact of lycopene accumulation on the emission of geranial, as observed in the fruits of several tomato and watermelon varieties [46], as well as in transgenic tomato fruits, where elevated carotenoid levels were accompanied by an increased emission of citral [50]. The possible synthesis of geranial by tomato CCD1s is now under investigation.

Experimental procedures

Cloning procedures

Five micrograms of total RNA, isolated from 14-day-old seedlings (O. sativa var. japonica cv. TP309), was used for cDNA synthesis using SuperScript III RnaseH (Invitrogen, Paisley, UK), according to the instructions of the manufacturer. Two microliters of cDNA was then applied for the amplification of OsCCD1 (accession no. AK066766, encoded by Os12g0640600), using the primers CCD-1 (5′-ATGGGAGGCGGCGATGGCGATGAG-3′) and CCD-2 (5′-TCACGCTGATTGTTTTGCCAGTTG-3′). The PCR reaction was performed with 100 ng of each primer, 200 μm dNTPs and 1 μL of Advantage cDNA Polymerase Mix (BD Biosciences, San Jose, CA, USA) in the buffer provided, as follows: 2 min of initial denaturation at 94 °C, followed by 32 cycles of 30 s at 94 °C, 30 s at 58 °C, and 2 min at 68 °C, and 10 min of final polymerization at 68 °C. The obtained PCR product was purified using GFX PCR DNA and a Gel Band Purification Kit (Amersham Biosciences, Piscataway, NJ, USA), and cloned into the pCR2.1–TOPO vector and pBAD/TOPO (Invitrogen) to yield pCR–OsCCD1 and pBAD–OsCCD1, respectively. The nature of the cDNA was verified by sequencing. To express OsCCD1 as a GST-fusion protein, the corresponding cDNA was excised as an EcoRI fragment from pCR–OsCCD1, and then ligated into accordingly digested and alkaline phosphatase-treated pGEX–5X-1 (Amersham Biosciences) to yield pGEX–OsCCD1.

Protein expression and purification

pGEX–OsCCD1 was transformed into BL21(DE3) E. coli cells harboring the plasmid pGro7 (Takara Bio Inc.; Mobitec, Göttingen, Germany), which encodes the groES–groEL chaperone system under the control of an arabinose-inducible promoter. Overnight cultures (2.5 mL) were inoculated into 50 mL of 2 × YT medium containing 0.2% (w/v) arabinose, grown at 28 °C to a D600 nm of 0.7, and induced with 0.2 mm isopropyl thio-β-d-galactoside for 4 h. Cells were harvested by centrifugation (10 min, 6000 g), resuspended in 4 mL of NaCl/Pi (pH 7.3), and lysed using a French press. Six milliliters of NaCl/Pi (pH 7.3) containing 1% Triton X-100 was then added, and the suspension was incubated at room temperature for 30 min. After centrifugation for 10 min at 12 000 g, the fusion protein was purified using glutathione–Sepharose 4B (Amersham Biosciences), according to the manufacturer’s instructions. OsCCD1 was then released by overnight treatment with the protease factor Xa in NaCl/Pi (pH 7.3) at room temperature, according to the manufacturer’s instructions (Amersham Biosciences). The protein eluate contained approximately 50% purified OsCCD1. Purification steps and protein expression were monitored by SDS/PAGE. The control strain expressed only GST.

Enzyme assays

Synthetic substrates were kindly provided by BASF (Ludwigshafen, Germany). Lycopene was obtained from Roth (Karlsruhe, Germany). Zeaxanthin and 3-OH-γ-carotene were isolated from E. coli cells transformed with carotenoid biosynthetic genes (unpublished data). The substrates were purified using TLC, and quantified spectrophotometrically at their individual λmax values, using extinction coefficients calculated from E1% [51]. Protein concentrations were determined using the BioRad protein assay kit (BioRad, Hercules, CA, USA).

In vitro assays contained 40 μg of purified enzyme eluate at substrate concentrations of 80 μm (lycopene and 3-OH-γ-carotene) or 40 μm (zeaxanthin and synthetic substrates). For the production of lycopene micelles, dried substrate was resuspended in 200 μL of benzene and mixed with 150 μL of an ethanolic detergent mixture consisting of 0.7% (v/v) Triton X-100 and 1.6% (v/v) Triton X-405. The mixture was then dried using a vacuum centrifuge to produce a carotenoid-containing gel. The gel was resuspended in 110 μL of 2× incubation buffer containing 2 mm tris(2-carboxyethyl)phosphine, 0.6 mm FeSO4 and 2 mg·mL−1 catalase (Sigma, Deisenhofen, Germany) in 200 mm Hepes/NaOH (pH 7.8). One hundred microliters of this lycopene suspension was then used in the in vitro assay, which was started by adding water and purified OsCCD1 to obtain a final volume of 200 μL. 3-OH-γ-Carotene, zeaxanthin and apocarotenoids were solubilized using octyl-β-glucoside at a final concentration of 1% (v/v). For this purpose, substrates were mixed with 50 μL of a 4% octyl-β-glucoside ethanolic solution, dried using a vacuum centrifuge, and resuspended in 100 μL of the 2× incubation buffer mentioned above. Water and purified OsCCD1 were then added to obtain the final volume of 200 μL. Depending on the substrates, the incubations were performed at 28 °C for 4 h (lycopene and 3-OH-γ-carotene), 2 h (zeaxanthin) or 30 min (synthetic substrates). Reactions were stopped by adding two volumes of acetone. Lipophilic compounds were partitioned against petroleum ether/diethyl ether 1 : 4 (v/v), vacuum-dried, and dissolved in 40 μL of chloroform. HPLC analyses were then performed using 20 μL of the extracts. For GC-MS analyses, volatile compounds were collected with solid-phase microextraction (SPME) fibers (100 μm polydimethylsiloxane; Sigma-Aldrich) for 30 min.

Conversion rates were determined in 30 min incubation assays using 30 μg of purified enzyme eluate. For quantification, 200 μL of an acetonic solution of α-tocopherole acetate (1 mg·mL−1) was added as internal standard to each assay prior to extraction. The conversion rates were determined by calculating the decrease of substrate peak areas measured at their individual λmax values using the max plot function of the software empower pro (Waters, Eschborn, Germany). Peak areas were normalized relative to the peak area of the internal standard, which was quantified at its absorption maximum of 285 nm.

In vivo test using lycopene-accumulating E. coli cells

Lycopene-accumulating XL1-Blue E. coli cells (unpublished data), harboring the corresponding biosynthetic genes from Erwinia herbicola, were transformed with pBAD–OsCCD1 or with pBAD–TOPO as a negative control. Overnight cultures were used to inoculate 50 mL of LB medium. Bacteria were grown at 28 °C to a D600 nm of 0.5, and induced with 0.08% (w/v) arabinose. Cells were harvested after 6 h, and volatile compounds were collected by introducing the SPME fiber into the cell-free medium for 30 min. For HPLC analyses, cells were harvested after 4 h, and carotenoids were extracted and processed as described above.

Analytical methods

For HPLC analyses, a Waters system equipped with a photodiode array detector (model 996) was used. A C30-reversed phase column (YMC Europe, Schermbeck, Germany) was developed with solvent system B [MeOH/t-butylmethyl ether/water (60 : 2 : 20, v/v/v)] and solvent system A [MeOH/t-butylmethyl ether (1 : 1, v/v)] at a flow rate of 1 mL·min−1, using a linear gradient from 100% solvent B to 43% solvent B within 45 min, and then to 0% solvent B within 1 min. The final conditions were maintained for 26 min at a flow rate of 2 mL·min−1, and this was followed by re-equilibration.

LC-MS analyses of compounds collected from HPLC were performed using a Thermo Finnigan LTQ mass spectrometer coupled to a Surveyor HPLC system consisting of a Surveyor Pump Plus, Surveyor PDA Plus and Surveyor Autosampler Plus (Thermo Electron, Waltham, MA, USA). Separations were carried out using a YMC-Pack C30-reversed phase column (150 × 3 mm internal diameter, 3 μm). Separation and identification of the C17 and C19 dialdehydes was performed as described in [40]. The oxime of the C14 dialdehyde was produced by adding 50 μL of O-methyl-hydroxylamine-hydrochloride (15 mg·mL−1) to an MeOH solution of the HPLC-purified compound, and then incubating for 20 min at 50 °C. The product was then partitioned against petroleum ether/diethyl ether 1 : 4 (v/v). The identification of the oxime was carried out according to [40].

GC-MS analyses were carried out with a Finnigan Trace DSQ mass spectrometer coupled to a Trace GC gas chromatograph equipped with a 30 m Zebron ZB 5 column (5% phenylpolysilanoxane/95% dimethylpolysilanoxane, 0.25 mm internal diameter, and 0.25 μm film thickness; Phenomenex, Aschaffenburg, Germany). The temperature program used was as follows: 50 °C held isocratically for 5 min, followed by a ramp of 25 °C·min−1 to a final temperature of 340 °C, which was maintained for an additional 5 min. The He carrier gas flow was maintained at 1 mL·min−1 using a split flow of 1 : 20. The splitless time was 3 min, and the injector oven temperature was set at 220 °C. Standard electrospray ionization was used at an ion source potential of 70 eV and with an ion source temperature of 200 °C. Identification of compounds was done by comparing the mass spectra with the NIST database.

Acknowledgements

This work was supported by the HarvestPlus programme (http://www.harvestplus.org) and by the Deutsche Forschungsgemeinschaft (DFG), Grant 892/1-3. We are indebted to J. Mayer for valuable discussions. We thank H. Ernst for providing the synthetic substrates and E. Scheffer for skilful technical assistance.

Ancillary