Evidence for Abscisic Acid Biosynthesis in Cuscuta reﬂexa , a Parasitic Plant Lacking Neoxanthin 1[W][OA]

Abscisic acid (ABA) is a plant hormone found in all higher plants; it plays an important role in seed dormancy, embryo development, and adaptation to environmental stresses, most notably drought. The regulatory step in ABA synthesis is the cleavage reaction of a 9-cis-epoxy-carotenoid catalyzed by the 9-cis-epoxy-carotenoid dioxygenases (NCEDs). The parasitic angiosperm Cuscuta reﬂexa lacks neoxanthin, one of the common precursors of ABA in all higher plants. Thus, is C. reﬂexa capable of synthesizing ABA, or does it acquire ABA from its host plants? Stem tips of C. reﬂexa were cultured in vitro and found to accumulate ABA in the absence of host plants. This demonstrates that this parasitic plant is capable of synthesizing ABA. Dehydration of detached stem tips caused a big rise in ABA content. During dehydration, 18 O was incorporated into ABA from 18 O 2 , indicating that ABA was synthesized de novo in C. reﬂexa . Two NCED genes, CrNCED1 and CrNCED2 , were cloned from C. reﬂexa . Expression of CrNCED s was up-regulated signiﬁcantly by dehydration. In vitro enzyme assays with recombinant CrNCED1 protein showed that the protein is able to cleave both 9-cis-violaxanthin and 9 # -cis-neoxanthin to give xanthoxin. Thus, despite the absence of neoxanthin in C. reﬂexa , the biochemical activity of CrNCED1 is similar to that of NCEDs from other higher plants. These results provide evidence for conservation of the ABA biosynthesis pathway among members of the plant kingdom.

Abscisic acid (ABA) is found in all higher plants and algae and is also produced by some fungi (Oritani and Kiyota, 2003;Schwartz and Zeevaart, 2004;Nambara and Marion-Poll, 2005). In higher plants, ABA is involved in seed dormancy, embryo development, and adaptation to various abiotic stresses. ABA is a sesquiterpenoid (C 15 ). In some fungi, there is a direct pathway from isopentenyl pyrophosphate (C 5 ) via farnesyl pyrophosphate (C 15 ; Oritani and Kiyota, 2003). Higher plants synthesize ABA via the C 40 indirect pathway. A C 40 carotenoid is oxidatively cleaved to form a C 25 byproduct and the C 15 precursor of ABA, xanthoxin. Biochemical and molecular evidence has shown that the cleavage reaction is the rate-limiting step in the ABA biosynthetic pathway.
Although the pathway of ABA biosynthesis in higher plants has been well established, there are still a few unresolved questions. One is the endogenous substrate of the cleavage reaction. The biochemical evidence has indicated that the C 40 substrate for production of biologically active ABA is an epoxy-carotenoid in the 9-cis configuration in order for biologically active ABA to be produced. In higher plants, the major 9-cis-epoxycarotenoids are 9#-cis-neoxanthin and 9-cis-violaxanthin. Because 9#-cis-neoxanthin is the most abundant 9-cisepoxy-carotenoid in higher plants, it has been speculated that 9#-cis-neoxanthin is the main endogenous substrate of ABA. However, in vitro enzyme assays with recombinant 9-cis-epoxy-carotenoid dioxygenase (NCED) proteins from several plant species have shown that NCEDs are capable of cleaving both 9#-cisneoxanthin and 9-cis-violaxanthin (Schwartz et al., 1997;Qin and Zeevaart, 1999;Chernys and Zeevaart, 2000), with a higher activity when 9-cis-violaxanthin is used as substrate (Schwartz et al., 2003b). This raised the question whether one or both are the endogenous substrates of ABA biosynthesis.
Cuscuta reflexa is of interest in addressing this question, because it lacks 9#-cis-neoxanthin, one of the major 9-cis-epoxy-carotenoids found in green plants. A study by Bungard et al. (1999) investigated the xanthophyll-carotenoid complement of the main lightharvesting complex involved in light capture of photosynthesis. They found that in most higher plants, the complex is highly conserved and includes, in addition to chlorophyll, the carotenoids neoxanthin, violaxanthin, and lutein. In C. reflexa, neoxanthin is replaced by another xanthophyll, lutein-5,6-epoxide. Therefore, C. reflexa lacks the step that converts violaxanthin to neoxanthin. There are two xanthophyll cycles in C. reflexa: under excess irradiance violaxanthin and lutein-5,6-epoxide are converted to zeaxanthin and lutein, respectively, by de-epoxidation; the cycle is reversed under low irradiance (Bungard et al., 1999).
C. reflexa is parasitic on the above-ground parts of other plants. The leafless stems coil around the host stems and petioles. C. reflexa contains only a small amount of chlorophyll and relies on obtaining photosynthetic assimilates from the phloem of its host. It forms specialized structures, called haustoria, which are used to acquire nutrients from its host plant (Hibberd et al., 1998).
The lack of 9#-cis-neoxanthin in C. reflexa may result in an inability of the plant to produce ABA, and the parasite may acquire it from its host, thereby eliminating the need to synthesize ABA on its own (Bungard et al., 1999). The objective of this study was to determine whether C. reflexa is able to synthesize ABA or acquires it from its host. Increased ABA in C. reflexa grown independently of a host would indicate that it is capable of ABA synthesis. We also searched for the presence of an NCED gene in the Cuscuta genome. Presence of an NCED gene would be evidence for the 9-cis-epoxycarotenoid cleavage reaction, the regulatory step for ABA synthesis. Our results show that despite the lack of 9#-cis-neoxanthin, C. reflexa is capable of ABA synthesis in a manner similar to that in other higher plants.

RESULTS
Cultured Cuscuta Shoot Tips Accumulate ABA ABA present in Cuscuta stems could be synthesized in situ, synthesized in the host plant and imported by the parasite, or acquired by a combination of import and endogenous synthesis. To determine whether Cuscuta is capable of producing ABA, it is necessary to grow the parasite independently of a host, as first reported by Baldev (1962).
When stem tips were cultured in vitro for approximately 20 d, the stems became thicker and much greener than material taken directly from host plants (Fig. 1). In our three experiments, the ABA content increased 4-to 8-fold over that initially present in the shoot tips (Table I). This result demonstrates that ABA was synthesized by the cultured shoot tips. In other words, Cuscuta is capable of producing ABA independently of a host plant.

Dehydration of Cuscuta Shoot Tips Stimulates ABA Accumulation
Leaves of mesophytic plants can usually be dehydrated to 85% to 90% of their initial fresh weight (FW) by a stream of warm air from a hair dryer in a few minutes. The ABA content of such water-stressed leaves increases rapidly for approximately 4 h and then levels off (e.g. Qin and Zeevaart, 1999). C. reflexa produces only stems, which cannot be quickly dehydrated with a hair dryer. Instead, we dehydrated stem tips gradually by placing them in a desiccator in the presence of Drierite. Such water-stressed stems lost 20% to 30% of their FW over a 24-h period and then showed signs of loss of turgidity. The results in Table II show that dehydration caused a large increase in ABA content. Thus, water stress induces ABA accumulation in Cuscuta stems, as it does in leaves of nonparasitic plants. However, ABA content also increased in nonstressed stem tips, albeit much less than in dehydrated stems. This increase in ABA occurred even when the stems were stored in polythene bags in the presence of some water. This observation is in contrast to turgid detached leaves, which maintained turgor and never showed an increase in ABA when kept in polythene bags (Zeevaart, 1980). When Cuscuta stem tips are detached, a brown viscous fluid exudes from the cut surface, perhaps reducing turgor in the stems and inducing ABA biosynthesis (Pierce and Raschke, 1980).

Incorporation of 18 O into ABA during Dehydration
Previous work has established that ABA is a cleavage product of 9#-cis-epoxy-carotenoids (Zeevaart et al., 1989. Whether the carotenoid cleavage reactions are catalyzed by monooxygenases or dioxygenases has been debated (Leuenberger et al., 2001). For the biosynthesis of ABA and related compounds in plants, there is compelling evidence for a dioxygenasecatalyzed reaction (Schmidt et al., 2006). Consequently, when ABA synthesis takes place in an atmosphere enriched in 18 O 2 , one 18 O is incorporated into the carboxyl group of ABA. In long-term experiments, especially with tissues having a small xanthophyll pool, this pool is used up and has to be replenished, so that 18 O is also incorporated into the ring positions of xanthophylls that become the 1#-hydroxyl and 4#-keto oxygens of ABA. Under these conditions, three atoms of 18 O are incorporated into newly synthesized ABA (Zeevaart et al., 1989. To demonstrate that ABA is synthesized de novo in Cuscuta and not released from a conjugated form, stem tips were incubated in an 18 O 2 atmosphere in the presence of Drierite for 24 h. ABA isolated from these shoots was analyzed as the methyl ester by chemical ionization mass spectrometry (MS). Instead of a single M 2 at mass-to-charge ratio (m/z) 278 as in unlabeled ABA, we observed a cluster of ions representing M 2 , M 2 12, M 2 14, and M 2 16 (Table III). Thus, in addition to the carboxyl group of the side chain, the two O-atoms of the ring also became enriched in 18 O. In the two experiments, 7% and 16%, respectively, of the ABA molecules had acquired three 18 O atoms. These results indicate that dehydration resulted in extensive ABA biosynthesis. Furthermore, we can conclude that xanthophylls became depleted during the 24-h incubation period and had to be replenished by oxidation of carotenoid precursors.

Isolation of C. reflexa NCED Genes
All higher plant NCED genes identified to date lack introns, thus providing an approach to isolate C. reflexa NCEDs from genomic DNA. The initial 625-bp partial sequence obtained by degenerate primers showed strong homology to other known NCEDs. A Southern blot hybridized with this partial fragment at high stringency showed two bands when SspI was used to restriction digest C. reflexa genomic DNA. Thus, we suspected that there may be two NCED sequences present in the C. reflexa genome and that they share significant sequence identity in this region. The inverse PCR identified two fragments, both NCED-related se-quences. The final CrNCED clones were amplified from C. reflexa genomic DNA containing only the open reading frames. We assigned CrNCED1 to the clone that contained the exact 625-bp sequence obtained by the initial degenerate PCR and CrNCED2 to the second clone. Neither gene shows introns in its sequence. At the region where the degenerate primers were used to amplify the initial CrNCED1 fragment, CrNCED1 and CrNCED2 share 85.1% identity at the nucleic acid level. This explains why two bands were obtained when the Southern blot was probed with the CrNCED1 fragment. The CrNCED1 and CrNCED2 genes encode proteins of 586 and 636 amino acids, respectively (Supplemental Fig. S1). At the amino acid level, the sequence of CrNCED1 shows 76.1% identity with the tomato (Solanum lycopersicum, formerly Lycopersicon esculentum) LeNCED1, 68.7% identity with the bean (Phaseolus vulgaris) PvNCED1, and 61.0% identity with the maize (Zea mays) VP14. The CrNCED proteins are very similar to other NCEDs (Fig. 2). The presence of NCED genes in C. reflexa demonstrates that this parasitic plant possesses the key enzyme of the ABA biosynthetic pathway in higher plants. This suggests that C. reflexa is able to synthesize ABA via the same pathway as other higher plants, despite its parasitic habit.

Northern-Blot Analysis of the Two CrNCED Genes
We studied expression of CrNCED1 and CrNCED2 in relation to dehydration by northern blotting of RNA isolated from C. reflexa stem tips. Because of the high identity of the two genes at the nucleotide level, we designed gene-specific probes (Supplemental Fig. S2). Figure 3 shows the expression of CrNCED1 and CrNCED2, as well as ABA accumulation, in detached C. reflexa stem tips prior to and during desiccation. Expression of both CrNCEDs was strongly induced after 6 h of dehydration. Weak expression of CrNCED2 in the control after 6 h supports the finding that nonstressed stems in long-term treatments accumulate ABA (Table II). The time course of CrNCED expression during dehydration is shown in Figure 4. Clearly, the level of transcripts was much increased after 2 h of dehydration, reached a maximum around 4 h, and then slowly declined. This pattern of expression is similar to that of PvNCED1 in water-stressed primary bean leaves (Qin and Zeevaart, 1999).

Biochemical Activity of CrNCED1
In higher plants, 9#-cis-neoxanthin is the most abundant epoxy-carotenoid in the cis configuration that can be used as substrate for ABA biosynthesis. However, in vitro assays with NCEDs from a variety of plants showed that recombinant protein has higher catalytic activity with 9-cis-violaxanthin than with 9#-cis-neoxanthin. The absence of neoxanthin in C. reflexa raised the question whether CrNCEDs have lost their ability to cleave neoxanthin. We produced recombinant protein of CrNCED1 and CrNCED2 in Escherichia coli strain BL21 cells. The induction levels were very high; however, both of the proteins remained mostly in the insoluble fraction. Purified CrNCED1 protein was concentrated with a Millipore Ultra-free 0.5 centrifugal filter and used for enzyme assays. The protein was able to cleave both 9-cis-violaxanthin and 9#-cis-neoxanthin. The cleavage product was identified as xanthoxin by gas chromatography (GC)-MS (Supplemental Fig. S3). This result indicates that the function of NCED is conserved in C. reflexa. As reported with other NCEDs, CrNCED1 also had higher activity with 9-cis-violaxanthin as substrate than with 9#-cis-neoxanthin.

DISCUSSION
Elucidation of the ABA biosynthetic pathway has been mainly achieved by characterization of ABAdeficient mutants and cloning of the corresponding genes (Schwartz et al., 2003a). So far, no gene has been isolated for the conversion of violaxanthin to neoxanthin. However, North et al. (2007) isolated a mutant, aba4, which lacks neoxanthin, accumulates violaxanthin, is deficient in ABA, and has a mild phenotype. In seeds and nonstressed vegetative tissue, sufficient ABA was synthesized (from 9-cis-violaxanthin) to give a normal phenotype. But upon dehydration, ABA accumulation was severely limited so that it was concluded that stress-induced ABA is predominantly derived from neoxanthin (North et al., 2007). However, it cannot be ruled out that under water stress, the substrate becomes limiting, considering that in these mutant plants 9-cis-violaxanthin will probably occupy the sites vacated by 9#-cis-neoxanthin in the major light-harvesting complex (Snyder et al., 2004).
The holoparasite C. reflexa also lacks neoxanthin (Bungard et al., 1999;J.A.D. Zeevaart, unpublished data). Despite the absence of neoxanthin in C. reflexa, this plant is capable of synthesizing ABA in the absence of a host, as shown by the following evidence. First, stem tips cultured in vitro separate from a host increase their ABA content (Table I). Second, dehydration of detached stem tips results in a large increase in ABA (Table II). Third, labeling stem tips with 18 O 2 during dehydration yields ABA that is enriched in one to three 18 O atoms (Table III), thus conclusively demonstrating that ABA is synthesized de novo during water stress. The labeling of ABA with more than one 18 O atom shows that the xanthophyll substrate was limiting and that precursor carotenoids were converted to 9-cis-epoxy-carotenoids during the 24-h incubation.
Considering the high level of 9-cis-violaxanthin in C. reflexa (Bungard et al., 1999;Snyder et al., 2005), it is likely that this xanthophyll is the C 40 substrate for ABA. C. reflexa also contains lutein-5,6-epoxide, but because it is in the all-trans form (Bungard et al., 1999), it cannot serve as a precursor for ABA. Bungard et al. (1999) speculated that host-acquired ABA might have eliminated the need for ABA biosyn-  thesis in C. reflexa, which would have resulted in redundancy of neoxanthin. However, our results demonstrate that despite the lack of neoxanthin, C. reflexa does synthesize its own ABA like other higher plants.
Cuscuta stems are connected via haustoria with their host plants to obtain organic and inorganic nutrients, as well as water, from the host. Symplastic connections also permit transfer of proteins, viruses (Birschwilks et al., 2006(Birschwilks et al., , 2007, and even mRNA (Roney et al., 2007) from host to parasite. ABA is transported in both phloem and xylem (Zeevaart and Boyer, 1984), so that ABA produced in the host is undoubtedly transferred to Cuscuta. Stem tips from plants parasitizing on waterstressed Perilla contained 3 times more ABA than tips from well-watered plants (J.A.D. Zeevaart, unpublished data). However, in this case, ABA synthesized in the host and parasite cannot be distinguished. To study transfer of host ABA to the parasite, a Cuscuta species devoid of chlorophyll, such as C. odorata (Berg et al., 2003), and probably also lacking carotenoids would be suitable material.
Whereas ABA biosynthesis is preserved in C. reflexa, other aspects of ABA metabolism may differ from other plants. For example, rehydration of water-stressed stem tips did not cause a rapid decline in ABA levels. Furthermore, no phaseic acid (PA) or dihydrophaseic acid (DPA) could be detected in Cuscuta stem tips (J.A.D. Zeevaart, unpublished data). This is surprising, because even if Cuscuta does not convert ABA to PA and DPA, it would at least acquire these metabolites from the host plant. It is possible that Cuscuta rapidly conjugates or degrades both PA and DPA, so that the steady-state levels are below the level of detection. Thus, how ABA is catabolized by Cuscuta and the possible presence of genes encoding ABA 8#-hydroxylase (Kushiro et al., 2004;Yang and Zeevaart, 2006) remain for future studies.

Plant Material
Seeds of Cuscuta reflexa were obtained from Dr. B. Baldev (Baldev, 1962). The seeds were germinated in a petri dish, and developing seedlings were clamped in cut stems of Ricinus communis plants. Later, Perilla ocymoides (green perilla) was used as host species for Cuscuta. The plants were grown in a greenhouse in which the natural daylength was extended with light from incandescent bulbs to prevent flowering in Cuscuta and Perilla, which are both short-day plants.
For in vitro culture, shoot tips (approximately 2 cm long) were cut from Cuscuta stems. The tips were surface sterilized in a 5% bleach solution for 5 min followed by rinsing in sterile distilled water. The tips were then placed in culture tubes containing 10 mL culture medium and with their basal ends inserted into the medium. The medium consisted of Murashige and Skoog basal medium with Gamborg vitamins, 5% Suc, and 0.8% agar. The cultures were maintained at 23°C for 3 weeks with continuous light from fluorescent lamps at a photo flux density of 40 mmol m 22 s 21 . Upon harvest, the tips were frozen in liquid N 2 and lyophilized.
For desiccation experiments, 3-cm stem tips (15-20 tips per treatment) were placed in a desiccator in the presence of Drierite (anhydrous calcium sulfate, 8 mesh). For labeling with 18 O 2 , shoot tips were placed in the presence of Drierite in a 250-mL Erlenmeyer flask sealed with a serum stopper. The procedure for incubation with 18 O 2 (95%; Cambridge Isotope Labs) was as described by Creelman and Zeevaart (1984). The 18 O 2 in the flask was replenished after 12 h.

ABA Extraction, Purification, and Measurement
These procedures were as described by Tian et al. (2004). ABA was expressed in micrograms per gram dry weight (DW) or in micrograms per  gram FW. The latter was used for desiccated shoot tips, which retained some moisture following lyophilization. DW was approximately 20% of FW.
GC-NCI-MS of Me-ABA was performed on a Waters GCT Premier Micromass mass spectrometer equipped with a 6890N HP gas chromatograph (Agilent). The column used was an SLB-5ms fused silica capillary column (10-m 3 0.1-mm 3 0.1-mm film thickness; Supelco) with helium as the carrier gas (flow rate 0.5 mL min 21 ). Gas-liquid chromatography conditions were as follows: the oven temperature was kept at 100°C for 1 min, then programmed from 100°C to 200°C at 25°C min 21 , and finally from 200°C to 250°C at 12°C min 21 . Methane was used as the reagent gas. The spectra were recorded at 10 scans per second at a collision energy of 70 eV in the mass range m/z 50 to 320.

Isolation of C. reflexa Genomic DNA
One gram of C. reflexa tips was pulverized in liquid N 2 and extracted in 5 mL of extraction buffer (2.1 g urea, 0.5 M NaCl, 50 mM Tris-HCl at pH 8.0, 20 mM EDTA at pH 8.0, and 1% Sarkosyl detergent). An equal volume of phenol:chloroform (1:1, v/v) was added to the mixture and the sample was centrifuged for 5 min at 5,000g to separate the aqueous phase from the organic solvents. DNA in the supernatant was precipitated with 10 mL ethanol. The DNA pellet was then resuspended in water.
A Southern blot was performed with the partial CrNCED1 as probe to identify a suitable restriction enzyme that would generate a genomic fragment with the full-length CrNCED1 gene. Digestion of genomic DNA with SspI produced two bands at approximately 3.2 and 3.4 kb. This restriction enzyme was, therefore, selected for use in inverse PCR. One microgram of genomic DNA was digested with SspI for more than 6 h. The DNA fragments were ligated in a total volume of 600 mL at 4°C overnight in favor of self-ligation. The DNA was then precipitated with 2 volumes of ethanol and resuspended in 10 mL of distilled water. One microliter of this DNA was used in the first round PCR reaction with primers JZ1152 (forward, 5#-CGTCGTCGTCCCCGACC-AGC-3#) and JZ1154 (reverse, 5#-CAGGTCGCCGGACGGTGTTAC-3#). The resulting product was diluted 100-fold with distilled water, and 1 mL of this diluted product was used as a template for a second round nested PCR amplification with primers JZ1151 (forward, 5#-CCACGGCGGCTCCCC-GGTGG-3#) and JZ1153 (reverse, 5#-TAGACCAGACCGGCGTTGGC-3#). Two fragments were visible on agarose gel at sizes of 2.9 and 3.1 kb, respectively, which match the calculated size based on the Southern blot. The two fragments were cloned into the pGEM-T Easy vector (Promega) for sequencing at the MSU Research Technology Support Facility.

Cloning of CrNCED1 and CrNCED2 and the Construction of Bacterial Expression Vectors
The sequences of the two C. reflexa genomic fragments indicated that both are putative NCED sequences. The starting and stop codons were predicted according to other available NCED sequences. To amplify the CrNCED1 and CrNCED2 clones for construction of expression vectors, the primers used were: JZ1300 (CrNCED1 forward: 5#-ATGGCGAATTCTTTGTATAAACCC-3#) and JZ1301 (CrNCED1 reverse: 5#-CTAGACTTGGGTGGCCAAGTC-3#); and JZ1302 (CrNCED2 forward: 5#-TCCATGTTGCAACACGTTGG-3#) and JZ1303 (CrNCED2 reverse: 5#-TCACATGACTTCAGTTAATAAATC-3#). The PCR was carried out with high-fidelity Taq enzyme (Pfu Turbo polymerase; Stratagene) to reduce amplification errors. The amplification with C. reflexa genomic DNA as template yielded the full-length ORF sequences of CrNCED1 (1,761 bp) and CrNCED2 (1,911 bp), respectively. The products were cloned into the pGEM-T Easy vector (Promega), generating clones pGEM102 (CrNCED1) and pGEM103 (CrNCED2). To construct bacterial expression vectors, the NotI fragments of pGEM102 or pGEM103 were inserted into the GST-fusion vector pGEX-5X-3 digested with NotI. The sense-orientation clones were identified and named 5X102 and 5X103.

RNA Isolation and Northern Hybridization
Stem tips (2 cm) of Cuscuta were frozen in liquid N 2 and ground with a mortar and pestle. Total RNA was isolated according to the CTAB (hexadecyltrimethylammonium bromide) method (Milligan, 1992) or by the acid AGPC (guanidinium thiocyanate-phenol-chloroform) method (Chomczynski and Sacchi, 1987). Total RNA (20 mg each) was separated by electrophoresis in formaldehyde-1.2% agarose gel and blotted onto a Hybond N 1 membrane (Amersham-Pharmacia) in 203 SSC by capillary blotting. After crosslinking the RNA by UV irradiation, hybridization was carried out at 42°C for 16 h with gentle shaking in hybridization buffer containing a 32 P-labeled cDNA probe (CrNCED1-3#UTR or CrNCED2-3#UTR; Supplemental Fig. S2 Gene-specific probes for CrNCED1 and CrNCED2 were designed as follows. 3# RACE (Invitrogen) was performed following the manufacturer's protocol using poly(A 1 ) RNA isolated from C. reflexa. Following RT with primers supplied by CLONTECH, the first-strand cDNA was used directly in 3#-RACE PCR reactions. Primary PCR amplification reactions were achieved using a high-fidelity enzyme (Pfu Turbo polymerase; Stratagene) and gene-specific primers JZ1167 (GTGAAGTTACCATCAAGAGTTC) and poly dT (CrNCED1), or JZ1169 (GGCATCTGTGAAATTTACCATC) and poly dT (CrNCED2) to generate the 3#-cDNA RACE fragments, respectively. A schematic representation of the CrNCED-3#UTRs structures is presented in Supplemental Figure S2. The PCR reaction consisted of the first denaturation for 3 min at 94°C, a series of 30 cycles (1 min at 94°C, 1 min at 52°C or 55°C, 1 min at 72°C), and a final extension for 7 min at 72°C. A 5-mL aliquot of the RACE reaction solution was analyzed by 1.2% agarose gel electrophoresis.

Recombinant Protein Purification
The plasmid 5X102 was transformed into Escherichia coli bacterial strain BL21 cells. An overnight culture of 5 mL was diluted to 100 mL with 23 YTA medium (16 g L 21 tryptone, 10 g L 21 yeast extract, 5 g L 21 NaCl, 100 mg L 21 ampicillin) and was grown at 28°C for 4 h. The culture was induced for protein expression for 2 h at 28°C with isopropyl b-D-1-thiogalactopyranoside at a final concentration of 0.1 mM. The cells were harvested, washed, and resuspended in 10 mL of PBS buffer containing 10 mg mL 21 lysozyme and frozen at 280°C. The frozen cells were thawed and sonicated in the presence of 0.5 mM dithiothreitol. Triton X-100 was added to the mixture to give a final concentration of 0.1%. The mixture was incubated in iced water for 30 min and then centrifuged at 10,000g for 10 min. The supernatant was applied to 1 mL of 50% slurry of glutathione Sepharose 4B (Amersham-Pharmacia) equilibrated in PBS buffer. After binding, the beads were washed with PBS buffer and were treated with 25 units of Factor Xa protease (Amersham-Pharmacia) for 4 h at 22°C. The purified CrNCED1 protein was eluted in the presence of 0.1% Triton X-100 and concentrated to 50 mL with a Millipore Ultrafree 0.5 centrifugal filter device equipped with BIOMAX 10 membrane.

Substrate Preparation and Enzyme Assay of CrNCED1
The C 40 -epoxycarotenoids all-trans-violaxanthin and 9#-cis-neoxanthin were isolated from spinach (Spinacia oleracea) leaves as described . After isomerization with iodine, all-trans-violaxanthin, 9-cisviolaxanthin, all-trans-neoxanthin, and 9#-cis-neoxanthin were separated by normal phase HPLC with a 0.78-3 30-cm mPorasil semipreparative column (Waters) using a linear gradient from 10% to 100% ethyl acetate in hexane for 65 min at a flow rate of 2.5 mL min 21 . The isomers of violaxanthin and neoxanthin were verified by their absorption spectra in ethanol and were quantified by spectrophotometry. The enzyme assay was performed as described by Schwartz et al. (1997). Appropriate amounts of protein and substrate were added in a total volume of 100 mL. Assays were incubated for 15 min at 22°C and then partitioned with ethyl acetate. The products were analyzed by HPLC with a 0.4-3 30-cm mPorasil column (Waters). The peaks eluting at the retention time of xanthoxin were collected and analyzed by GC-MS.
Sequence data for this article can be found in the GenBank/EMBL data libraries under accession numbers AY974807 and AY974808.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Amino acids sequence alignment of NCEDs.
Supplemental Figure S3. HPLC chromatograms of the cleavage reaction product xanthoxin.