Characterization of a Novel Carotenoid Cleavage Dioxygenase from Plants*

The plant hormone abscisic acid is derived from the oxidative cleavage of a carotenoid precursor. Enzymes that catalyze this carotenoid cleavage reaction,nine- c is epoxy-carotenoid dioxygenases, have been identified in several plant species. Similar proteins, whose functions are not yet known, are present in diverse organisms. A putative cleavage enzyme from Arabidopsis thaliana contains several highly conserved motifs found in other carotenoid cleavage enzymes. However, the overall homology with known abscisic acid biosynthetic enzymes is low. To determine the biochemical function of this protein, it was expressed in Escherichia coli and used for in vitro assays. The recombinant protein was able to cleave a variety of carotenoids at the 9–10 and 9′-10′ positions. In most instances, the enzyme cleaves the substrate symmetrically to produce a C14 dialdehyde and two C13 products, which vary depending on the carotenoid substrate. Based upon sequence similarity, orthologs of this gene are present throughout the plant kingdom. A similar protein in beans catalyzes the same reaction in vitro. The characterization of these activities offers the potential to synthesize a variety of interesting, natural products and is the first step in determining the function of this gene family in plants.

The plant hormone abscisic acid is derived from the oxidative cleavage of a carotenoid precursor. Enzymes that catalyze this carotenoid cleavage reaction, nine-cis epoxy-carotenoid dioxygenases, have been identified in several plant species. Similar proteins, whose functions are not yet known, are present in diverse organisms. A putative cleavage enzyme from Arabidopsis thaliana contains several highly conserved motifs found in other carotenoid cleavage enzymes. However, the overall homology with known abscisic acid biosynthetic enzymes is low. To determine the biochemical function of this protein, it was expressed in Escherichia coli and used for in vitro assays. The recombinant protein was able to cleave a variety of carotenoids at the 9 -10 and 9-10 positions. In most instances, the enzyme cleaves the substrate symmetrically to produce a C 14 dialdehyde and two C 13 products, which vary depending on the carotenoid substrate. Based upon sequence similarity, orthologs of this gene are present throughout the plant kingdom. A similar protein in beans catalyzes the same reaction in vitro. The characterization of these activities offers the potential to synthesize a variety of interesting, natural products and is the first step in determining the function of this gene family in plants.
Apocarotenoids are a class of compounds derived from the oxidative cleavage of carotenoids (1). This is a structurally diverse class of compounds that are widely distributed in nature. The assortment of apocarotenoids results from the large number of carotenoid precursors (more than 600 carotenoids have been identified), variations in the site of cleavage, and modifications subsequent to cleavage. Abscisic acid (ABA) 1 , which is derived from the oxidative cleavage of a 9-cis-epoxycarotenoid, is necessary for seed development and adaptation to environmental stresses in plants (2). Although historically referred to as retinoids, vitamin A and related compounds are derived from the cleavage of ␤-carotene (3).
Plants produce a number of volatile apocarotenoids. These compounds probably serve a function as insect attractants and are valued as flavor/aroma compounds. Apocarotenoid pig-ments also have some economic value. For example, bixin (annatto) is commonly used as a natural food coloring, and crocin is the major pigment in saffron (4).
The formation of apocarotenoids may result from nonspecific mechanisms, such as lipoxygenase cooxidation or photooxidation. Enzymes capable of cleaving carotenoids at specific sites are believed to be involved in the synthesis of a number of apocarotenoids. The Vp14 gene, which encodes an ABA biosynthetic enzyme in maize (5,6), was the first carotenoid cleavage enzyme to be cloned from any organism. Since the characterization of VP14, ABA biosynthetic enzymes have been identified in several plants (7)(8)(9)(10). Based upon sequence similarity, an enzyme necessary for vitamin A biosynthesis has been identified in Drosophila (11) and vertebrates (12). Another carotenoid cleavage enzyme in animals that reacts with the 9 -10 double bond has also been identified (13). The definitive demonstration of both symmetric and asymmetric cleavage pathways appears to resolve a long standing debate concerning vitamin A biosynthesis (14).
A number of hypothetical proteins that are similar to these carotenoid cleavage enzymes have also been identified in the sequence data bases, but their biochemical functions have not yet been demonstrated. In the Arabidopsis genome sequence, nine potential carotenoid cleavage enzymes have been identified. Most of these genes have been annotated as neoxanthin cleavage enzymes or nine-cis-epoxy-carotenoid dioxygenases (NCEDs). Both terms imply an involvement in ABA biosynthesis. Although it is likely that two or more genes encode ninecis-epoxy-carotenoid dioxygenases, it is doubtful that all nine genes are involved in ABA biosynthesis. Some of these proteins may catalyze carotenoid cleavage reactions not involved in ABA biosynthesis. Others might catalyze a double bond cleavage reaction of substrates other than carotenoids, such as the lignostilbene dioxygenases from prokaryotes (15,16). In this paper, a putative carotenoid cleavage enzyme from Arabidopsis was expressed in Escherichia coli. It is shown that recombinant protein catalyzes a carotenoid cleavage reaction different from the one in ABA biosynthesis. The gene symbol CCD, Carotenoid Cleavage Dioxygenase, was adopted in this paper to distinguish this gene family from genes that encode ABA biosynthetic enzymes.

EXPERIMENTAL PROCEDURES
Constructs-A truncated cDNA for the AtCCD1 gene was obtained from the Arabidopsis expressed sequence tag collection (N95924). The full-length cDNA sequence (19) (AJ005813) and the genomic sequence (AL163818; gene identification MAA211 50) have also become available. A full-length clone was obtained by reverse transcription-polymerase chain reaction using RNA from light-grown seedlings. The polymerase chain reaction product was amplified with Pfu polymerase (Stratagene) and the following primer sequences: 5Ј-CATGGCGGAGAAACTCAGTG-3Ј and 5Ј-TTATATAAGAGTTTGTTCCTGG-3Ј. The amplified fragment was cloned into the SmaI site of pBluescript SK (Stratagene) and pGEX-2T (Amersham Pharmacia Biotech) to produce pCZY1 and pCZY2, respectively. A fragment of the PvCCD1 gene in beans, previously called PvNCED2, was amplified with degenerate primers as previously described (7). The flanking sequences were obtained by polymerase chain reaction from a bean cDNA library with one primer annealing to the initial polymerase chain reaction product and either a T7 or T3 primer annealing to the vector. Once the 5Ј and 3Ј sequences were determined, a full-length clone (AY029525) was amplified from the bean library with the following primer sequences: 5Ј TGGATCCATGGGGGATGATGG 3Јand 5Ј-TG-GATCCTCACAGTTTTGCTTG-3Ј. The BamHI restriction sites added on to the primer were used to clone into the BamHI site of pGEX-2T (Amersham Pharmacia Biotech) to create pXQ1.
Protein Expression and Enzyme Assays-A 5-ml culture of pCZY2 or pXQ1 was used to inoculate a 100-ml culture in 2ϫ YT medium (per liter: 16 g of tryptone, 10 g of yeast extract, and 5 g of NaCl). Cultures were grown at 37°C until an A 600 of 0.7 was reached. Expression of the protein was induced by the addition of 0.2 mM isopropyl ␤-D-thiogalactopyranoside, and the cultures were grown at 28°C for an additional 3-5 h. The E. coli cells were harvested by centrifugation, resuspended in Tris-buffered saline, and lysed in a French press. The recombinant protein was purified with glutathione S-transferase-agarose (Sigma) and then released by cleavage with thrombin for 7 h at 4°C. The carotenoid substrates were extracted and purified as previously described (17). Assays contained 0.1% Triton X-100, 0.5 mM FeSO4, 5 mM ascorbate, and the appropriate carotenoid substrate in 100 mM Bis Tris, pH 7.0. The assay products were partitioned into ethyl acetate and analyzed by HPLC or thin-layer chromatography as previously described (5). The mass spectra for C 27 compounds were obtained by direct inlet as previously described (5). For gas chromatography-mass spectrometry of the C 13 (18) were cotransformed into E. coli strain JM109 with the pCZY1 or pCZY2 constructs described above. An overnight culture of 2 ml was used to inoculate 50 ml of LB medium containing the appropriate antibiotics and 0.1 mM isopropyl ␤-D-thiogalactopyranoside. The cultures were grown at 30°C for 24 h. The E. coli culture was centrifuged, and the cell pellet was resuspended in an equal volume of formaldehyde. An equal volume of methanol was then added, followed by two volumes of ethyl acetate. The phases were separated by the addition of water, and the ethyl acetate phase was retained for HPLC analysis.
Following centrifugation, the medium from the cultures was partitioned twice with an equal volume of ethyl acetate. The ethyl acetate was evaporated under vacuum, and the extracts were saponified prior to HPLC analysis. The HPLC fractions were dried, resuspended in the appropriate solvent, and used for UV-visible or fluorescence spectroscopy. A rosafluene standard was prepared by the reduction of the C 14 dialdehyde with NaBH 4 .

RESULTS
In Vitro Characterization of AtCCD1-The AtCCD1 gene, originally called AtNCED1 (19) in Arabidopsis, encodes a protein with limited sequence similarity to the carotenoid cleavage enzymes in ABA biosynthesis (31% identical and 42% similar to VP14). Although it was suggested that the gene is involved in ABA biosynthesis (19), the biochemical function of the protein has not yet been reported. Hypothetical proteins that share much higher similarity to known ABA biosynthetic enzymes have since been identified in the Arabidopsis genome sequence.
A comparison of the AtCCD1 protein with hypothetical proteins from other plant species indicates that it is a member of a highly conserved subfamily of carotenoid cleavage-related proteins (Fig. 1). The AtCCD1 protein is 67% identical and 74% similar to a protein from avocados, PaNCED2 (9). A protein from beans, PvCCD1 (this paper), is 78% identical and 86% similar to AtCCD1. In addition, the expressed sequence tags FIG. 1. Alignment of the AtCCD1 protein with related sequences. PaNCED2 was previously identified in avocados (AF224670) (9); PvCCD1 (AYO29525) is an apparent ortholog of AtCCD1 from beans; AtNCED5 is a potential ortholog of VP14 that was identified in the Arabidopsis genome sequence (AC074176); AnaCCD1 is an open reading frame from the cyanobacterium Anabaena PCC 7120 (c307: sequence 1467-2894).

FIG. 2. Thin-layer chromatography analysis of assays with the recombinant AtCCD1 protein and various carotenoid substrates.
The substrates used were ␤,␤-carotene (␤-Car), lutein (Lut), zeaxanthin (Zeax), all-trans-violaxanthin (tViol), 9-cis-violaxanthin (9cViol), and 9Ј-cis-neoxanthin (9cNeox). Enzyme assay products were separated on a thin-layer silica plate that was developed in hexane, ethyl acetate, and 2-propanol (70:20:10). Following chromatography, the plate was sprayed with 2,4-dinitrophenylhydrazine to detect aldehydes and ketones. The products were also characterized by a combination of HPLC, UV-visible spectroscopy, and mass spectrometry. Several of the characterized products are labeled on the chromatogram (see Fig. 3). Residual substrates are indicated by asterisks. from a variety of plants share a high degree of similarity to AtCCD1 at the nucleotide and the deduced amino acid level. Based upon limited sequence data, genes with greater than 80% nucleotide identity are present in the expressed sequence tags of numerous plant species. A hypothetical protein from Anabaena PCC 7120, AnaCCD1, also shares some sequence similarity with AtCCD1 (38% identity and 48% similarity).
To determine the potential function of AtCCD1, the gene was cloned into a glutathione S-transferase fusion vector for expression in E. coli, and the recombinant protein was assayed for cleavage activity with a variety of carotenoid substrates. The assay products were separated by thin-layer chromatography and sprayed with 2,4-dinitrophenylhydrazine to visualize aldehyde and ketone products (Fig. 2). To generate the reaction scheme (Fig. 3), the assay products were also characterized by a combination of HPLC, UV-visible spectroscopy, and mass spectrometry. A variety of C 13 products, resulting from cleavage at the 9 -10 and the 9Ј-10Ј positions, were identified (e.g. II, V, and VI). In assays containing lutein, zeaxanthin, and alltrans-violaxanthin (Fig. 2, lanes 2, 3, and 4, respectively), the products did not separate from the carotenoid substrates by TLC. It was, however, possible to isolate these products by HPLC for further characterization. The red spot (I) in Fig. 2, lanes 1-6 was the C 14 dialdehyde resulting from symmetrical cleavage at the 9 -10 and 9Ј-10Ј positions (Fig. 3). This C 14 dialdehyde was a major product when ␤-carotene, lutein, zeaxanthin, or all-trans-violaxanthin was used as a substrate. The enzyme did not cleave as well adjacent to a 9-cis double bond or an allenic bond found in some carotenoids. In assays containing the 9-cis isomer of violaxanthin (Fig. 2, lane 5), the major product was the C 27 epoxy-apocarotenal (III) that results from a single cleavage distal to the cis double bond. When 9Ј-cisneoxanthin was used as a substrate (Fig. 2, lane 6), both a C 27 epoxy-apocarotenal (III) and a C 27 allenic-apocarotenal (IV) were produced.
FIG. 4. E. coli strains engineered to accumulate zeaxanthin and coexpressing CCD1 proteins. "AtCCD1" is a strain coexpressing pCZY1, and "PvCCD1" is a strain coexpressing pXQ1. The control strain contains the pGEX-2T vector with no insert. natural product of roses, which would be produced by the reduction of the C 14 dialdehyde to the corresponding dialcohol (20). Reduction of the C 14 dialdehyde with NaBH 4 did produce a compound with the same retention time and absorption spectrum (data not shown). Peak 2, which has a spectrum shifted to a slightly lower wavelength, is probably a cis isomer.
A gene sharing high sequence similarity to AtCCD1 was cloned from Phaseolus vulgaris and named PvCCD1, previously PvNCED2 (7). The PvCCD1 protein was also expressed in E. coli (Fig. 4) and shown to catalyze the symmetric cleavage of zeaxanthin to form the C 14 dialdehyde. DISCUSSION The recombinant AtCCD1 protein cleaves a variety of carotenoids at the 9 -10 and 9Ј-10Ј double bonds. With most substrates tested, the carotenoid is symmetrically cleaved to produce a C 14 dialdehyde. The prevalence of the C 14 product (symmetric cleavage) relative to the C 27 product (single cleavage) may indicate that the enzyme functions as a dimer, simultaneously cleaving at both ends. The lignostilbene dioxygenases from prokaryotes, which share sequence similarity to the carotenoid cleavage enzymes, are known to function as dimers (16). In the case of 9-cis-violaxanthin and 9Ј-cis-neoxanthin, the C 27 apocarotenals are major products. Cleavage adjacent to a cis double bond or the allenic bond of neoxanthin appears to be impaired.
The structural variations in carotenoids exist mainly in the rings. Because AtCCD1 is able to cleave a variety of carotenoids, a number of interesting C 13 ring products are generated in the in vitro assays. The grasshopper ketone (V), derived from the cleavage of neoxanthin, is regurgitated by some species of grasshoppers (21,22). The unpalatable taste of this compound may deter ants and other predators. Other C 13 compounds are valued as flavor and aroma constituents of a variety of foods (23). Some C 13 cleavage products are effective insect lures (24,25) and may serve as insect attractants when produced by plants.
Probable orthologs of the AtCCD1 gene are present in a wide range of plants. An ortholog in beans is able to catalyze the same reaction. The substrate of these enzymes in planta and the function of the apocarotenoid products are not yet known. However, a number of apocarotenoids have been identified in plants that are derived by an AtCCD1-like activity. The C 14 dialdehyde is the probable precursor of rosafluene, a highly fluorescent compound found in some varieties of roses (26). The C 14 product is also the expected precursor of mycorradicin, a yellow pigment that accumulates in the roots of plants inoculated with arbuscular mycorrhizal fungi (27). Several C 13 cleavage products are also abundant in the roots of infected plants.
Although Arabidopsis does not appear to form mycorrhizal associations, an AtCCD1-like protein probably catalyzes the synthesis of these apocarotenoids in other plant species.
A C 13 product of zeaxanthin or lutein cleavage, 3-hydroxy-␤ionone, accumulates in etiolated bean seedlings on exposure to light, and it has been suggested that this compound may have a function in the light-induced inhibition of hypocotyl elongation (28,29). The AtCCD1 gene is also induced by light in etiolated seedlings. 2 With the characterization of the AtCCD1 activity in vitro, it is now possible to speculate on the possible function of this gene in plants. By altering the expression of this gene in transgenic plants, it should be possible to characterize the apocarotenoid products and determine their biological role(s).