Abscisic aldehyde oxidase in leaves of Arabidopsis thaliana


  • Mitsunori Seo,

    1. Department of Biological Sciences, Tokyo Metropolitan University, Hachioji-shi, Tokyo 192-0397, Japan,
    Search for more papers by this author
  • Hanae Koiwai,

    1. Department of Biological Sciences, Tokyo Metropolitan University, Hachioji-shi, Tokyo 192-0397, Japan,
    Search for more papers by this author
  • Shuichi Akaba,

    1. Department of Biological Sciences, Tokyo Metropolitan University, Hachioji-shi, Tokyo 192-0397, Japan,
    Search for more papers by this author
    • Present address: Central Research Laboratory, Nissui Pharmaceutical Co. Ltd., Yuki-shi, Ibaraki 307-0036, Japan.

  • Teruya Komano,

    1. Department of Biological Sciences, Tokyo Metropolitan University, Hachioji-shi, Tokyo 192-0397, Japan,
    Search for more papers by this author
  • Takayuki Oritani,

    1. Department of Applied Bioorganic Chemistry, Graduate School of Agricultural Science, Tohoku University, Aoba-ku, Sendai 981-8555, Japan, and
    Search for more papers by this author
  • Yuji Kamiya,

    1. Frontier Research Program, Institute of Physical and Chemical Research (RIKEN), Wako-shi, Saitama 351-0198, Japan
    Search for more papers by this author
  • Tomokazu Koshiba

    Corresponding author
    1. Department of Biological Sciences, Tokyo Metropolitan University, Hachioji-shi, Tokyo 192-0397, Japan,
      *For correspondence (fax +81 426 77 2559; e-mail
    Search for more papers by this author

*For correspondence (fax +81 426 77 2559; e-mail koshiba-tomokazu@c.metro-u.ac.jp).


Abscisic acid (ABA) is a plant hormone involved in seed development and responses to various environmental stresses. Oxidation of abscisic aldehyde is the last step of ABA biosynthesis and is catalysed by aldehyde oxidase (EC We have reported the occurrence of three isoforms of aldehyde oxidase, AOα, AOβ and AOγ, in Arabidopsis thaliana seedlings, but none oxidized abscisic aldehyde. Here we report a new isoform, AOδ, found in rosette leaf extracts, which efficiently oxidizes abscisic aldehyde. AO

δ was specifically recognized by antibodies raised against a recombinant peptide encoded by AAO3, one of four Arabidopsis aldehyde oxidase genes (AAO1, AAO2, AAO3 and AAO4). Functionally expressed AAO3 protein in the yeast Pichia pastoris showed a substrate preference very similar to that of rosette AOδ. These results indicate that AOδ is encoded by AAO3. AOδ produced in P. pastoris exhibited a very low Km value for abscisic aldehyde (0.51 μm), and the oxidation product was determined by gas chromatography–mass spectrometry to be ABA. Northern analysis showed that AAO3 mRNA is highly expressed in rosette leaves. When the rosette leaves were detached and exposed to dehydration, AAO3 mRNA expression increased rapidly within 3 h of the treatment. These results suggest that AOδ, the AAO3 gene product, acts as an abscisic aldehyde oxidase in Arabidopsis rosette leaves.


ABA is a plant hormone that plays an important role in many aspects of plant growth and development, including seed maturation and dormancy as well as adaptation to a variety of environmental stresses ( Zeevaart & Creelman, 1988). The regulation of these processes is, in part, due to de novo synthesis of ABA. Thus elucidation of the ABA biosynthetic pathway and the isolation of the related gene(s) are essential for an appreciation of ABA-mediating plant responses.

Recent studies based on the cloning of genes using ABA-deficient mutants have brought us nearer to clarification and verification of the regulatory mechanisms of ABA biosynthesis in plants ( Cutler & Krochko, 1999; Liotenberg et al. 1999 ; Zeevaart, 1999). The most important advance was the isolation of two genes for zeaxanthin epoxidase (ZEP) using the Nicotiana plumbaginifolia mutant aba2 ( Marin et al. 1996 ) and 9-cis-epoxycarotenoid dioxygenase (NCED) using the maize mutant vp14 ( Schwarz et al. 1997a ; Tan et al. 1997 ). ZEP converts zeaxanthin to violaxanthin by a two-step epoxidation, and NCED catalyses the oxidative cleavage of 9-cis-xanthophylls to xanthoxin. Subsequent studies have shown the presence of these genes in other plant species, and suggested their physiological and regulatory function in ABA biosynthesis in plants ( Cutler & Krochko, 1999; Liotenberg et al. 1999 ; Qin & Zeevaart, 1999). In N. plumbaginifolia, a good correlation between ZEP (ABA2) expression and endogenous ABA levels was observed in non-photosynthetic organs, roots and seeds, but in leaves expression appeared to be linked to photosynthesis and exhibited a diurnal fluctuation ( Frey et al. 1999 ; Liotenberg et al. 1999 ). It is likely that ZEP is one of the key steps for regulating ABA synthesis in roots and seeds of N. plumbaginifolia. In contrast, maize NCED (Vp14) mRNA levels in wild-type seedlings were low, while the mRNA accumulated strongly in detached leaves ( Tan et al. 1997 ), suggesting its regulatory role in ABA synthesis. Phaseolus vulgaris NCED (PvNCED1) showed a much clearer relationship between fluctuations of mRNA, protein and ABA amounts in leaves and roots after dehydration and rehydration treatments, indicating that the 9-cis-epoxycarotenoid cleavage reaction is the key regulatory step of ABA biosynthesis in leaves and roots of P. vulgaris ( Qin &Zeevaart, 1999). It was also assumed that there is an NCED gene family in plants, and that the different NCED genes are responsible for regulation of ABA biosynthesis in different tissues.

Recent findings confirmed that ZEP and NCED are located in plastids, and that the product of the NCED reaction, xanthoxin, is converted to ABA in the cytosol by two oxidation steps via abscisic aldehyde ( Cutler & Krochko, 1999). Several related mutants impaired in the final step of ABA biosynthesis, the oxidation of abscisic aldehyde to form ABA, have been isolated, including Arabidopsis aba3 ( Schwartz et al. 1997b ); N. plumbaginifolia aba1 ( Leydecker et al. 1995 ); and tomato flacca and sitiens ( Taylor et al. 1988 ). These mutants are known to lack an activity of aldehyde oxidase but, except for sitiens, are deficient in the synthesis of the molybdenum cofactor (Moco) which is necessary for aldehyde oxidase activity ( Akaba et al. 1998 ; Leydecker et al. 1995 ; Schwartz et al. 1997b ; Taylor et al. 1988 ). Only the tomato mutant sitiens is thought to have a mutation in a structural gene of aldehyde oxidase specific for abscisic aldehyde ( Marin & Marion-Poll, 1997; Taylor et al. 1988 ). However, the corresponding gene for sitiens has not yet been cloned, and final conclusions cannot be drawn.

In tomato sitiens, activity bands of aldehyde oxidase as well as xanthine dehydrogenase/oxidase, which are Moco-containing enzymes, are detected after native polyacrylamide gel electrophoresis (PAGE). Only a small difference in the intensity of the aldehyde oxidase activity band between the wild type and sitiens was observed ( Marin & Marion-Poll, 1997). This strongly indicates that there is more than one aldehyde oxidase isoform in tomato plants, and that one is highly specific for abscisic aldehyde and is responsible for ABA biosynthesis.

The presence of aldehyde oxidase isoforms has been reported in several plants ( Koshiba et al. 1996 ; Omarov et al. 1999 ; Ori et al. 1997 ; Rothe, 1974; Seo et al. 1998 ). Our previous work demonstrated that three aldehyde oxidase isoforms in extracts of Arabidopsis seedlings, AOα, AOβ and AOγ, were detected by activity staining after native PAGE ( Seo et al. 1998 ). They had relatively wide substrate specificities, but almost no activity for abscisic aldehyde. Subsequently we cloned four Arabidopsis aldehyde oxidase cDNAs (AAO1, AAO2, AAO3 and AAO4, formerly called atAO-1, atAO-2, atAO-3 and atAO-4, respectively; Sekimoto et al. 1998 ). AtAO1, AtAO2 and AtAO3, corresponding to AAO1, AAO4 and AAO2, respectively, were also cloned independently by Hoff et al. (1998) . We also obtained specific antibodies against recombinant peptides of each AAO1 and AAO2 expressed in Escherichia coli ( Akaba et al. 1999 ), and recently succeeded in expressing these products as active forms in Pichia pastoris ( Koiwai et al. 2000 ). These results indicate that AAO1 and AAO2 encode three isoforms in Arabidopsis seedlings. AOα and AOγ were shown to be homodimers of AAO1 and AAO2 products, respectively, and AOβ is a heterodimer of the AAO1 and AAO2 products. For the function of AOα in Arabidopsis, we have suggested that the isoform is a possible candidate of aldehyde oxidase for IAA biosynthesis, because it showed a relatively low Km value for indole-3-acetaldehyde, a precursor of indole-3-acetic acid (IAA), and the activity was much higher in IAA-overproducing sur1 mutant seedlings ( Akaba et al. 1999 ; Seo et al. 1998 ).

In the present study we report a new aldehyde oxidase isoform, AOδ, in Arabidopsis rosette leaves, which can efficiently oxidize abscisic aldehyde to produce ABA. AOδ was shown to be the product of AAO3. Possible involvement of AOδ in the regulation of ABA levels in Arabidopsis leaves is discussed.


Abscisic aldehyde oxidase activity in rosette leaves

Our previous work has demonstrated the existence of three isoforms of aldehyde oxidase in Arabidopsis seedlings, but none had activity against abscisic aldehyde, a precursor of ABA. Here we compare zymogram patterns with the enzyme extracts from seedlings and rosette leaves using three substrates: indole-3-aldehyde, 1-naphthaldehyde, and abscisic aldehyde ( Fig. 1). In the seedlings AOα exhibited an intense band for 1-indole-3-aldehyde, and AOγ showed a strong affinity for naphthaldehyde, but almost no activity was detected at these positions when abscisic aldehyde was used as a substrate ( Fig. 1a). AOβ showed an intermediate preference between AOα and AOγ. In rosette leaves, three activity bands were also detected at the positions corresponding to AOα, AOβ and AOγ ( Fig. 1b). However, the uppermost isoform (AOδ) preferred indole- 3-aldehyde a little less than seedling AOα, but showed a stronger activity for 1-naphthaldehyde than AOα ( Fig. 1a,b). Furthermore, an intense band for abscisic aldehyde was detected at the same position. Because AOα could efficiently oxidize neither 1-naphthaldehyde nor abscisic aldehyde, it appears that the activity is derived not from AOα, but from a new aldehyde oxidase isoform tentatively named AOδ. As the lowest band observed in the rosette extracts showed the same substrate preferences as those of seedling AOγ, and showed the same reactivity against specific antibodies against AOγ (data not shown), we concluded the lowest band was derived from AOγ in the rosette leaves. Although as yet we have no experimental evidence, the isoform of the middle band in the rosette extracts may be a heterodimer of the AOδ and AOγ subunits, in the same way as the seedling AOβ is a heterodimer of the AOα and AOγ subunits ( Akaba et al. 1999 ).

Figure 1.

Aldehyde oxidase activities in seedlings and rosettes of Arabidopsis.

Extracts of seedlings (a) and rosettes (b) were subjected to native PAGE followed by activity staining as described in Experimental procedures. Indole-3-aldehyde (IAld), 1-naphthaldehyde (NAld) and abscisic aldehyde (ABAld) were used as substrates. Approximately 60 μg protein for seedlings, and 80 μg protein for rosettes, was loaded on each gel.

Immunological analysis of AOδ

In order to confirm that AOδ is different from AOα, and to test which of the four AAO genes encodes AOδ, immunological analysis was performed using antibodies against peptides encoded by three AAO cDNAs (AAO1, AAO2 and AAO3). We have previously raised specific antibodies against AAO1 or AAO2 products which could specifically react with AOα and AOβ, or AOγ and AOβ, respectively ( Akaba et al. 1999 ), and here we also obtained antibodies against a polypeptide fragment (411 amino acids) of AAO3 produced in E. coli. Seedling AOα and rosette AOδ were checked for reactivity with these three antibodies. After immunoprecipitation, the remaining aldehyde oxidase activity in the supernatant of seedling and rosette extracts was detected after native PAGE using indole-3-aldehyde and abscisic aldehyde, respectively ( Fig. 2). AOα in the seedlings was specifically reacted with anti-AAO1 antibodies, while anti-AAO3 antibodies removed only AOδ activity. These results indicate that AOδ is a new isoform of aldehyde oxidase encoded by AAO3, and has the ability to oxidize abscisic aldehyde.

Figure 2.

Immunoprecipitation of aldehyde oxidase isoforms by anti-AAO antibodies.

Extracts from seedlings (a) and rosettes (b) of Arabidopsis were treated with anti-AAO1, anti-AAO2 and anti-AAO3 antibodies. The immuno- complexes formed were precipitated by adding protein A sepharose. Aldehyde oxidase activity remaining in the supernatant was determined after native PAGE by staining the gel using indole-3-aldehyde (IAld) or abscisic aldehyde (ABAld) as substrates. Approximately 60 μg protein for seedlings, and 80 μg protein for rosettes, was loaded on each gel.

Expression of AOδ in Pichia pastoris

To investigate substrate specificity of AOδ, the AAO3 protein was expressed in yeast P. pastoris cells as an active enzyme form. AAO3 was immunologically identical to AOδ, as anti-AAO1 and anti-AAO2 antibodies could not recognize AAO3, and only anti-AAO3 antibodies reacted with the protein (data not shown). The mobility of AAO3 on native PAGE was the same as that of rosette AOδ. The intensity of the activity bands for seven aldehydes was compared between the rosette AOδ and AAO3 produced in P. pastoris ( Fig. 3). They showed very similar zymogram patterns: relatively strong activity against abscisic aldehyde, 1-naphthaldehyde and cinnamaldehyde, indicating that AOδ is accurately produced in the yeast system from the introduced AAO3 cDNA, and that the major component of abscisic aldehyde oxidase in rosettes is AOδ. To find the substrate affinity of AAO3 for these substrates, the Km value was determined against five aldehydes: abscisic aldehyde, indole-3-aldehyde, 1-naphthaldehyde, cinnamaldehyde and benzaldehyde ( Table 1). AAO3 showed a low Km value, 0.51 μm, for abscisic aldehyde, indicating the functional role of AOδ in ABA biosynthesis. The oxidation product of abscisic aldehyde produced by AAO3 was subjected to gas chromatography–mass spectrometry (GC–MS) analysis, and was demonstrated to be ABA ( Fig. 4). The enzyme could not discriminate between the (+)- and (–)-enantiomers because the substrate used was a mixture of (+/–)-cis-abscisic aldehyde, and the products contained the same ratio of (+/–)-cis-ABA as that of the substrate (data not shown).

Figure 3.

Substrate preferences of AOδ and AAO3 expressed in Pichia pastoris.

Enzyme solutions were obtained from (a) Arabidopsis rosette leaves and (b) P. pastoris transformed with AAO3 cDNA. After native PAGE, activity bands were developed separately with strips from each lane using abscisic aldehyde (1), indole-3-aldehyde (2), 1-naphthaldehyde (3), cinnamaldehyde (4), benzaldehyde (5), heptaldehyde (6) and acet- aldehyde (7) as substrates. Approximately 80 μg protein for rosettes, and 30 μg protein for P. pastoris extracts, was loaded on each gel.

Table 1. Km value of AAO3 protein produced in Pichia pastoris for five aldehydes
  1. a Enzyme activity was assayed by monitoring the reduction of DCIP as an electron acceptor. On other substrates, enzyme activity was assayed by determining the amount of reaction product formed in the presence of O 2 as an electron acceptor using HPLC.

Abscisic aldehyde0.51
Cinnamaldehyde a700
Figure 4.

Mass spectra of authentic ABA and the oxidation products of abscisic aldehyde produced by the AAO3 protein.

Details are given in Experimental pro- cedures. (a) Fragment patterns with MS of authentic ABA; (b) the oxidation product of abscisic aldehyde by AAO3 was partially purified and subjected to GC–MS.

Organ distribution of AAO mRNA

Organ-specific expressions of four AAO mRNAs were examined in 8-day-old whole seedlings; in roots, rosette leaves, stems, flowers and siliques of 2-month-old mature plants; and in dry seeds. The levels of AAO transcripts were analysed by Northern blotting using four cDNA probes (AAO1, AAO2, AAO3 and AAO4) ( Fig. 5). All probes detected an approximately 4.5 kb RNA that corresponds to the entire open reading frame of these cDNAs. In some cases smaller and smear bands were also detected, and these were thought to be prematurely terminated or breakdown RNA products as discussed previously ( Sekimoto et al. 1998 ). AAO1 was highly expressed in the seedlings, roots and seeds, and AAO2 in seedlings and roots. AAO3 was detected in roots and was relatively high in rosette leaves. AAO4 was highly expressed in siliques. The organ-specific expression of the AAO genes indicates that aldehyde oxidases encoded by the AAO genes play different roles in Arabidopsis.

Figure 5.

Expression of AAO mRNAs in Arabidopsis seedlings and mature plants.

Total RNA was extracted from seedlings, mature plants (roots, rosettes, stems, flowers and siliques), and dry seeds. 10 μg (seedlings and roots, rosettes, stems and flowers of mature plants) and 5 μg (siliques and seeds) of RNA were separated on 1.5% (w/v) agarose gels containing 5% (v/v) formaldehyde. AAO1, AAO2, AAO3 and AAO4 full-length cDNAs were used as probes for hybridization.

AAO gene expression after water stress

In a wide range of plant species, an increase in endogenous ABA levels in plants has been observed after the imposition of a water deficit ( Zeevaart & Creelman, 1988). In detached, water-stressed Arabidopsis rosette leaves, the ABA content was determined to be more than five times higher than that in unstressed leaves ( Rock et al. 1992 ). The response of AAO expression was checked for water stress in rosettes and roots. Rosette leaves detached from approximately 2-month-old Arabidopsis plants were exposed to dehydration for 0, 3, 6 and 9 h, and the changes in AAO mRNA contents were analysed by Northern blotting ( Fig. 6a). While the AAO1 and AAO2 genes showed a constant or slightly decreased expression during treatment, the expression of AAO3 was rapidly induced within 3 h exposure to dehydration conditions. The AAO4 transcript was hardly detected in rosette leaves (see Fig. 5), but the expression was induced after dehydration. Other effects of detachment, such as wounding, were negligible because the signals observed in the wet controls were almost the same as those of the zero time samples. In contrast to changes in rosette leaves, no correlation of AAO expression and water stress was observed in roots ( Fig. 6b).

Figure 6.

AAO gene expression in Arabidopsis leaves (a) and roots (b) after water stress.

(a) Total RNA was prepared from just-detached (0) and detached rosette leaves 3, 6 and 9 h after dehydration, and from a wet control (w). 10 μg RNA from each sample was loaded onto the gels.

(b) Total RNA was extracted from non-dehydrated (0) and 6-h-dehydrated (6) roots. 2 μg (AAO1), 5 μg (AAO2) and 10 μg (AAO3, AAO4) RNA were loaded onto the gels. AAO1, AAO2, AAO3 and AAO4 full-length cDNAs were used as probes for hybridization.


Several studies have shown that the last step of ABA biosynthesis is the oxidation of abscisic aldehyde. We have demonstrated the existence in seedling extracts of at least four Arabidopsis aldehyde oxidase genes, AAO1–4 ( Sekimoto et al. 1998 ), and three aldehyde oxidase isoforms, AOα, AOβ and AOγ ( Seo et al. 1998 ). As these isoforms showed almost no activity for abscisic aldehyde, we have focused on the identification of a new isoform of aldehyde oxidase responsible for ABA biosynthesis. In the present study we have analysed abscisic aldehyde oxidase activity in Arabidopsis rosette leaves, and found a novel aldehyde oxidase isoform, AOδ, which was demonstrated to be the product of AAO3. The AAO3 protein (AOδ) expressed in P. pastoris showed a very low Km value (0.51 μm) for abscisic aldehyde. The AAO3 oxidation product of abscisic aldehyde was determined to be ABA. AAO3 can convert both (+)- and (–)-cis-abscisic aldehyde, as reported previously ( Schwartz et al. 1997b ; Yamamoto & Oritani, 1996). These results indicate that AOδ is an abscisic aldehyde oxidase.

ABA is known to be involved in water-stress responses. Increase in ABA content in roots, xylem sap and leaves of drought-stressed plants has been widely reported ( Davies & Zhang, 1991; Zeevaart, 1999). Stress-induced ABA production in roots might have considerable importance, modifying the plant's water balance before the leaves have received a water stress signal. ABA in water-stressed roots is thought to be a signal which is transported to the leaves, resulting in stomatal closure. However, it is known that the level of ABA in leaves is much higher than that in roots, and even in detached leaves a significant increase of ABA level has been observed after drought treatment ( Audran et al. 1998 ; Qin & Zeevaart, 1999; Rock et al. 1992 ). These observations indicate that the leaf itself may be an organ where ABA biosynthesis takes place during water stress responses. NCED was shown to have an important role in ABA biosynthesis in maize leaves, because Vp14 (ZmNCED) was upregulated by dehydration of leaves ( Tan et al. 1997 ). Recently, P. vulgaris NCED cDNA (PvNCED1) was cloned and the fluctuation of mRNA level was investigated in leaves and roots after water stress and rehydration treatments ( Qin & Zeevaart, 1999). The results showed that PvNCED1 mRNA and protein levels are rapidly increased in water-stressed leaves and roots with a concomitant increase in ABA level in these tissues. In addition, a decrease in ABA level in stressed leaves was also observed after rehydration treatment. This report provides evidence that the 9-cis-epoxycarotenoid cleavage reaction is a key step in ABA biosynthesis in water-stressed P. vulgaris. In Arabidopsis, several putative NCED genes were identified in the database, and one was reported to be upregulated in shoots after water stress ( Neill et al. 1998 ).

The present study has revealed that in Arabidopsis, AAO3 gene expression was rapidly induced in dehydrated rosette leaves. This supports the proposal that the AAO3 (AOδ) functions as an abscisic aldehyde oxidase in vivo. Unlike in maize and P. vulgaris leaves, where NCED mRNA was not detected in intact leaves and only increased after a dehydration treatment, Arabidopsis AAO3 mRNA showed considerable expression even in intact leaves. It is likely that AAO3 is responsible for ABA biosynthesis, even in unstressed leaves. Our recent results with the AAO3 Arabidopsis mutant aao3 revealed that in mutant leaves, not only was the level of ABA significantly lower than in wild-type leaves, but the increase of ABA content after water stress was also impaired ( Seo et al. 2000 ). These results suggest that AAO3 is involved in ABA biosynthesis in Arabidopsis leaves. However, our preliminary experiments showed that no obvious increase in the levels of AAO3 protein or its activity in dehydrated leaves was observed. Further experiments concerning the regulation of protein and the activity level of AAO3 are required to elucidate its physiological role in the regulation of ABA biosynthesis in Arabidopsis leaves.

Although the expression of the tobacco ZEP gene in leaves was not increased upon water stress, its expression in roots was increased during dehydration ( Audran et al. 1998 ). In seeds, tobacco ZEP mRNA reaches a maximum just before mid-development, at approximately the same time as ABA concentrations are maximal. PvNECD expression was induced by water stress not only in leaves, but also in roots of P. vulgaris ( Qin & Zeevaart, 1999). The expression of Arabidopsis AAO3 was upregulated only in water-stressed leaves, but not in roots. These observations can be explained by the existence of a complex controlling system of ABA biosynthesis which has different steps depending on the organ and/or developmental stage. Furthermore, the presence of a gene family of NCED ( Burbidge et al. 1999 ; Tan et al. 1997 ), as well as of aldehyde oxidase genes, means a more complex and highly organized mechanism is involved in the regulation of ABA levels in plants.

The function of the AAO4 product is still unknown, but in the present work it has been shown that AAO4 is mainly expressed in siliques, and that the expression of AAO4 in leaves is induced by dehydration. This presents a possible role of AAO4 in ABA biosynthesis. The aao3 mutant exhibited a wilty phenotype in the leaves, but did not affect seed dormancy ( Seo et al. 2000 ). Thus it is assumed that the AAO3 and AAO4 proteins are responsible for ABA biosynthesis in different organs of Arabidopsis. Further studies to identify the AAO4 gene product will provide ideas for the regulation mechanism of ABA biosynthesis in Arabidopsis. Production of transgenic Arabidopsis with AAO genes, and isolation of mutants from T-DNA insertion lines, are now in progress and will be useful for understanding the functions of these genes in plants.

Experimental procedures

Plant material and growth conditions

Arabidopsis thaliana (Columbia ecotype) wild-type seeds were sown in pots containing vermiculite with nutrient solution under a regime of 16 h light and 8 h darkness at 22°C. For the dehydration experiment, rosette leaves were detached and placed in a laminar flow hood. They were sampled 0, 3, 6 and 9 h after the treatment, and stored at −80°C. For the turgid control, leaves were kept on wet tissue paper wrapped with a plastic film for 9 h. To prepare root samples, plants were removed from the pots. After washing the roots, whole plants were placed under the dehydration conditions for 0 and 6 h.

Expression of recombinant AAO3 protein in Pichia pastoris

A full-length cDNA of AAO3 with an ACC sequence just before the ATG initiation codon was cloned into pPICZC Pichia expression vector (Invitrogen, Carlsbad, USA). Transformation, expression, and enzyme extraction were carried out as described by Koiwai et al. (2000) .

Enzyme extraction and activity staining

Plant tissue was ground to a powder with liquid nitrogen and homogenized in 8 ml g−1 FW extraction buffer (50 m m Tris–HCl pH 7.5; 1 m m ethylenediaminetetra-acetate, EDTA; 1 μm sodium molybdate; 10 μm flavine adenine dinucleotide; 2 m m dithiothreitol; protease inhibitor, Complete Protease inhibitor cocktail tablets, Roche Diagnostics (Mannheim, Germany), one tablet per 100 ml) and Polyclar AT (0.2 g g−1 FW, Wako Pure Chemicals, Osaka, Japan). After the sample was centrifuged at 12 000 g for 15 min, the supernatant was fractionated with ammonium sulfate (0–60% saturation). The precipitate was dialysed against extraction buffer and the dialysed sample was centrifuged at 15 000 g for 10 min. After calculating the protein concentration, the contaminating proteins were removed by heat treatment (3 min at 60°C) and centrifugation, then used for native PAGE. Recombinant AAO3 protein was extracted as described by Koiwai et al. (2000) . Ammonium sulfate-fractionated and heat-treated enzyme samples were used for native PAGE.

Native PAGE was performed with a 7.5% acrylamide gel in a Laemmli buffer system ( Laemmli, 1970) in the absence of sodium dodecyl sulfate (SDS) at 4°C. After electrophoresis, the gel was immersed in 0.1 m potassium phosphate buffer (pH 7.5) for 2–3 min, then enzyme-activity staining was developed in a mixture containing 0.1 m phosphate buffer pH 7.5, 300 μm substrate, 0.1 m m phenazine methosulfate and 0.4 m m 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide at 30°C in the dark for 30–60 min.


Two antibodies against the AAO1 and AAO2 polypeptides were prepared as described by Akaba et al. (1999) . To obtain anti-AAO3 antibody, a polypeptide fragment corresponding to amino-acid residues 347–757 of AAO3 protein was produced in E. coli according to the procedure described previously ( Akaba et al. 1999 ). The expressed polypeptide was injected into a rabbit. Ammonium sulfate-fractionated (0–40%) antiserum was used as anti-AAO3 antibodies. Immunoprecipitation was performed using Protein A Sepharose (Protein A Sepharose CL-4B, Amersham Pharmacia Biotech, Buckinghamshire, UK) as described previously ( Akaba et al. 1999 ).

Aldehyde oxidase assay

Aldehyde oxidase activity of AAO3 expressed by P. pastoris was assayed by determining the amount of the product using HPLC ( Koiwai et al. 2000 ).

GC–MS analysis

The reaction product produced from abscisic aldehyde by AAO3 reaction was partially purified by ether fractionation and HPLC, then methylated with diazomethan. The sample was then subjected to GC–MS. A mass spectrometer (AUTO MASS; JEOL, Tokyo, Japan) was coupled to a gas chromatograph (5890 series II; Hewlett Packard, DE, USA) with a capillary column (DB-1, 0.25 mm i.d. × 15 mm, film thickness 0.25 μm; J&W, CA, USA). The injection temperature was set at 250°C; inlet pressure was 70 kPa. After injection the oven temperature was maintained at 80°C for 1 min, then increased to 245°C at a rate of 30°C min−1 followed by a further increment to 300°C at 5°C min−1. Mass spectra were obtained under the following conditions: electron energy, 70 eV; ion source temperature, 200°C; capillary interface temperature, 250°C; ion range 50–450 (m/z).

The (+)- and (–)-enantiomers of abscisic aldehyde and ABA were analysed using HPLC with a chiral column (Daicel Chiral CEL OD; 250 × 4.6 mm).

Northern analysis

For Northern blotting, total RNA was extracted by the method of Verwoerd et al. (1989) , except that the buffer composition was 100 m m Tris–HCl pH 8.0; 100 m m LiCl; 10 m m EDTA; 1% SDS. Total RNA was electrophoresed on a 1.5% agarose gel containing formaldehyde and 3-(N-morpholino)propanesulfonic acid (MOPS) buffer, and transferred to a nylon membrane. Hybridization was performed using 32P-dCTP labelled probes under highly stringent conditions ( Sambrook et al. 1989 ).


We thank Dr H. Kawaide of RIKEN Frontia for GC–MS analysis. We are grateful to Prof. Dick Kendrick of Wageningen Agricultural University for his critical review of the manuscript. This work was supported in part by a Grant-in-Aid for Scientific Research (B) 10559017 to T.K. from the Ministry of Education, Science, Sports and Culture, Japan, and by the Fund for Research Fellowship of the Japan Society for the Promotion of Science for Young Scientist to M.S.