Use of the glucosyltransferase UGT71B6 to disturb abscisic acid homeostasis in Arabidopsis thaliana

Authors


*(fax +44 1904 328772; e-mail djb32@york.ac.uk).

Summary

A glucosyltransferase (GT) of Arabidopsis, UGT71B6, recognizing the naturally occurring enantiomer of abscisic acid (ABA) in vitro, has been used to disturb ABA homeostasis in planta. Transgenic plants constitutively overexpressing UGT71B6 (71B6-OE) have been analysed for changes in ABA and the related ABA metabolites abscisic acid glucose ester (ABA-GE), phaseic acid (PA), dihydrophaseic acid (DPA), 7′-hydroxyABA and neo-phaseic acid. Overexpression of the GT led to massive accumulation of ABA-GE and reduced levels of the oxidative metabolites PA and DPA, but had marginal effect on levels of free ABA. The control of ABA homeostasis, as reflected in levels of the different metabolites, differed in the 71B6-OEs whether the plants were grown under standard conditions or subjected to wilt stress. The impact of increased glucosylation of ABA on ABA-related phenotypes has also been assessed. Increased glucosylation of ABA led to phenotypic changes in post-germinative growth. The use of two structural analogues of ABA, known to have biological activity but to differ in their capacity to act as substrates for 71B6 in vitro, confirmed that the phenotypic changes arose specifically from the increased glucosylation caused by overexpression of 71B6. The phenotype and profile of ABA and related metabolites in a knockout line of 71B6, relative to wild type, has been assessed during Arabidopsis development and following stress treatments. The lack of major changes in these parameters is discussed in the context of functional redundancy of the multigene family of GTs in Arabidopsis.

Introduction

Abscisic acid (ABA) plays an important role in the growth, development and stress tolerance of plants (reviewed by Finkelstein and Rock, 2002). The level of the hormone is regulated by the relative rates of biosynthesis, catabolism, conjugation and redistribution throughout the plant. The control of these processes enables plants to increase or decrease the level of ABA in different cells at specific times during growth and development, as well as in response to biotic and abiotic stresses (reviewed by Nambara and Marion-Poll, 2005). While different homeostatic mechanisms are likely to operate in different conditions, it will be possible to understand how these are regulated only when genes encoding proteins involved in each of the processes have been identified and analysed.

In this context, genetic screens and analyses of mutants compromised in growth, dormancy and stomatal control have revealed a number of genes encoding enzymes involved in the biosynthesis of the hormone, from initial modification and cleavage of carotenoids through to steps leading from the oxidation of xanthoxin to ABA (Finkelstein and Rock, 2002; Nambara and Marion-Poll, 2005; Schwartz et al., 2003). In each biosynthetic mutant the phenotype could be linked to a major deficiency in ABA, for example aba3-2 accumulated only 2–20% compared with wild type (Leon-Kloosterziel et al., 1996a,b; Rock and Zeevaart, 1991; Seo et al., 2000). Other mutants have been identified using genetic screens that reveal greater sensitivity to ABA levels, such as those involving post-germinative growth (Arenas-Huertero et al., 2000; Lopez-Molina et al., 2001; Quesada et al., 2000). For example, while wild-type seedlings grown on high glucose will arrest and not green due to increased ABA levels, mutants impaired in hormone synthesis are not compromised (Arenas-Huertero et al., 2000; Brocard-Gifford et al., 2004; Jang et al., 1997; Rook et al., 2001). Using these screens that revealed greater sensitivity to ABA levels, gin5 was identified which accumulated up to 40% of normal levels of ABA (Arenas-Huertero et al., 2000).

In contrast to their successful application to ABA biosynthesis, genetic screens have not identified any genes involved in the catabolism or conjugation of ABA. The biochemistry of these processes has been well characterized, with many of the enzymic steps defined (Zaharia et al., 2005a; Zeevaart, 1999; Figure 1). It is considered that the principal oxidation pathway occurs through hydroxylation of the 8′-methyl to give 8′-hydroxy ABA (8′OH-ABA), which rearranges to form phaseic acid (PA), which is then further reduced to dihydrophaseic acid (DPA; Cutler and Krochko, 1999; Zaharia et al., 2004). Other oxidation products of ABA have also been reported, including 7′-hydroxy ABA (Hampson et al., 1992), 9′-hydroxy ABA, and neo-phaseic acid (neo-PA; Zhou et al., 2004). Further oxidation products of 8′-hydroxy ABA have been identified (Zaharia et al., 2004). The main conjugation pathway occurs through glucosylation of the carboxyl group to form ABA-glucose ester (ABA-GE; Koshimizu et al., 1968). While it has been suggested that ABA-GE is a storage and/or transport form of the hormone (Sauter et al., 2002), there is no evidence that endogenous ABA-GE is cleaved in plants to release free ABA. Further, it has been shown that the bulk of ABA-GE is found in vacuoles, and that levels of the conjugate do not decrease in response to stress (Lehmann and Schutte, 1984; Milborrow, 1978; Neill et al., 1983; Zeevaart, 1980, 1983; Zeevaart and Boyer, 1984). These observations suggest that ABA-GE is an inactive end product of ABA metabolism (Zeevaart, 1999).

Figure 1.

Catabolic pathways for ABA.
ABA, abscisic acid; 8′OH-ABA, 8′-hydroxyABA; PA, phaseic acid; DPA, dihydrophaseic acid; 9′OH-ABA, 9′-hydroxyABA; neo-PA, neo-phaseic acid; 7′OH-ABA, 7′-hydroxyABA; ABA-GE, abscisic acid-glucose ester.

Recently, several publications have reported progress in identifying the genes encoding enzymes that may be involved in ABA catabolism and conjugation. Two laboratories used phylogenetic and expression analyses combined with biochemical studies to define enzymes of the cytochrome P450 family able to catalyse the 8′-hydroxylation of ABA in vitro (Kushiro et al., 2004; Saito et al., 2004). Only one of these, CYP707A2, has been confirmed to have this function in planta (Kushiro et al., 2004). With respect to glucosylation, Xu et al. (2002) purified an enzyme fraction from Adzuki bean (Vigna angularis) capable of glucosylating ABA in vitro, and cloned the corresponding glucosyltransferase (GT) gene. However, the ability of this GT to glucosylate ABA in planta was not studied. The recombinant enzyme displayed preferential activity towards 2-trans-ABA, which has been shown to be biologically inactive (Finkelstein and Rock, 2002; Kriedemann et al., 1972).

Our studies have used an alternative approach to identify GTs of interesting small molecules, including ABA. The members of a large multigene family of GTs in the Arabidopsis genome were identified through the presence of a signature motif (Li et al., 2001; Ross et al., 2001). The GTs, a subset of Family 1, were cloned and recombinant enzymes were screened for their activity in vitro towards hormones and secondary metabolites (reviewed by Bowles et al., 2005; Lim and Bowles, 2004). In the context of ABA glucosylation, the screening strategy discovered eight enzymes that glucosylated ABA in vitro (Lim et al., 2005). One of the eight GTs, UGT71B6 (hereafter called 71B6), displaying enantio-selectivity, recognized only the naturally occurring (+)-ABA enantiomer and did not glucosylate the unnatural (−)-ABA, nor the inactive 2-trans-ABA (Priest et al., 2005). Interestingly, structural analogues of ABA were identified against which 71B6 displayed either greater activity or reduced activity relative to the hormone (Priest et al., 2005).

Given that 71B6 glucosylates the natural enantiomer of ABA in vitro, it was of interest to determine whether the overexpression of 71B6in planta would disturb the homeostasis of ABA and, in consequence, ABA-related responses. The availability of ABA structural analogues, and the known activities of 71B6 towards the analogues, provided a means of confirming whether any phenotypes observed were causally related to the glucosylation event. This study describes the impact of overexpressing 71B6 and the profile of ABA and related metabolites in the plant. To investigate the role of the endogenous 71B6 gene, the effect of knocking out the GT was also analysed. The results are discussed in relation to the utility of GTs in agricultural biotechnology and the role of the GT multigene family in Arabidopsis.

Results

Molecular characterization of transgenic plants

Homozygous transgenic lines were generated that overexpressed 71B6 under control of the CaMV 35S promoter (71B6-OE). Steady-state levels of 71B6 transcripts were assessed using Northern analyses of RNA extracted from 4-week-old rosette leaves harvested from 71B6-OE transgenic lines, and compared with that harvested from wild type, from plants expressing an empty vector (Ri), and from a transposon insertion line for 71B6 (71B6-KO). The results shown in Figure 2(a) indicate that no transcripts were detectable in wild type, Ri or 71B6-KO. Two 71B6-OE lines (7 and 14) did not accumulate the transgene transcripts, whereas six did, with the greatest expression level displayed in line 17.

Figure 2.

Characterization of 71B6-OE transgenic lines.
(a) Northern blot analysis of steady-state levels of UGT71B6 mRNA. RNA was isolated from 4-week-old rosette tissue.
(b) UDP-glucosyltranferase activity towards ABA from plant protein extracts. Data represent the mean of three replicates (±SD). Wild-type (WT), empty vector control (Ri), 71B6-KO transposon-insertion line (KO) and independent 71B6-OE lines.

To determine whether steady-state levels of 71B6 transcripts correlated with increased GT activity, soluble protein extracts from the corresponding leaf tissue were assayed in vitro with the substrates ABA and UDP-glucose, and accumulation of ABA-GE was measured. The results shown in Figure 2(b) compare levels of ABA-GE at 2- and 6-h reaction times. The highest level of product was observed in reactions containing protein extract from 71B6-OE line 17. Lower levels of ABA-GE were observed in reactions containing extracts from 71B6-OE lines 12 and 23. These results correlate with the level of expression of the transgene. 71B6-OE line 14, which did not accumulate the transgene transcript, displayed the same levels of ABA-GE as the WT and Ri controls, and the 71B6-KO line. These results show that overexpression of 71B6in planta correlate with increased enzyme activity in leaf extracts that is capable of glucosylating ABA in vitro.

The lack of detectable 71B6 mRNA in Northern analyses of wild type indicated that the gene was not abundantly expressed in rosette leaves of 4-week old plants under standard growth conditions. Therefore we analysed expression of the endogenous 71B6 gene during a developmental time course and in response to abiotic challenges. The results (Figure S1) indicated that expression was very low (<5% of actin) at all stages of development, and was induced to only a low level by salt, mannitol or application of ABA. Therefore, in order to assess the effectiveness of the 71B6-KO line, hydroponically grown plants were subjected to salt stress. Northern analyses showed that no transcripts were detectable in 71B6-KO plants (Figure 3); but in 71B6-RE-complemented lines (71B6-KO transformed with a genomic fragment containing the endogenous 71B6 gene including its own promoter and terminator regions), levels of 71B6 transcripts were elevated.

Figure 3.

Steady-state levels of 71B6 mRNA accumulation.
Hydroponically grown plants were stressed by submersing the roots in 150 mm NaCl for 24 h. Plants were harvested and transcript levels were determined by Northern analysis using the 500-bp 3′ end of the ORF as a probe. Wild type (WT); 71B6-KO (KO); 71B6-RE (3 and 12).

Phenotypic analysis of transgenic plants

The glucosylated form of ABA is considered to be biologically inactive (Zeevaart, 1999). Therefore we reasoned that, if 71B6 was capable of glucosylating ABA in planta, overexpression of the gene might lead to phenotypes typical of ABA deficiency. However, when a range of parameters was analysed, 71B6-OE plants were found to be indistinguishable from wild type. For example, seeds germinated as wild type under standard growing conditions (with and without stratification), and on media supplemented with exogenous ABA or with the gibberellin biosynthesis inhibitor paclobutrazol (data not shown). 71B6-OE plants developed normally and showed no wiltiness typical of substantial ABA deficiency. Similarly, no phenotypic difference was observed in 71B6-KO (data not shown).

Post-germinative growth on ABA.  It was possible that only a low level of ABA deficiency was caused by overexpression of GT. As wild-type seedlings are known to be highly sensitive to developmental arrest by exogenous ABA (Lopez-Molina et al., 2001), post-germinative growth on ABA was used as a potential means to discriminate between 71B6-OE plants and wild type. Cotyledon emergence of both wild type and 71B6-OE was 100% emergence after 3 days in the absence of ABA (data not shown). In the presence of 0.5 μm (+)-ABA, a phenotype was revealed that discriminated between wild type and 71B6-OE plants. The cotyledons of the transgenic lines emerged more rapidly (Figure 4a). This observation suggested that the increased level of GT activity in the overexpressors led to enhanced deactivation of ABA, and therefore a reduced impact of ABA on the seedlings. Use of the ABA analogues PBI-514 and PBI-413 enabled confirmation that the phenotype was caused by increased glucosylation of ABA by 71B6. While both analogues are known to cause ABA-related responses (Cutler et al., 2000; Nyangulu et al., 2005; Zaharia et al., 2005a), PBI-413 was a better substrate than ABA for recombinant 71B6 in vitro, whereas PBI-514 was found to be a poor substrate (Priest et al., 2005).

Figure 4.

Post-germinative growth on ABA and analogues in transgenic lines.
Post-germinative growth, as scored by percentage of cotyledons emerged from the seed coat, was determined for 71B6-OE13 (bsl00072) and wild type (bsl00043). Seeds were sown on medium supplemented with 0.5 μm (+)-ABA (a); PBI-413 (b); or PBI-514 (c) and stratified for 4 days before transfer to warm, light conditions. Experiment also included 71B6-OE 12 and 17 with similar results. Data are based on three replicates (±SE).

Thus, as shown in Figure 4(b), when PBI-413 was included in the media, a much greater difference in cotyledon emergence between 71B6-OEs and wild-type seedlings was observed. The overexpressors displayed 100% cotyledon emergence by 5 days compared with wild type, with <5% emergence. In contrast, when PBI-514 was included in the media, minimal difference was observed between the 71B6-OEs and wild type (Figure 4c). These results are consistent with the in vitro data for glucosylation of the analogues by 71B6, and demonstrate that the phenotype arises from glucosylation of ABA/ABA analogues.

Post-germinative growth on glucose.  It has been shown previously that several glucose-insensitive mutants are allelic to ABA-deficient or ABA-insensitive mutants (Arenas-Huertero et al., 2000; Brocard-Gifford et al., 2004; Rook et al., 2001). Therefore we reasoned that overexpression of 71B6 could lead to glucose insensitivity. To assess this possibility, seeds were germinated on 6% glucose and greening of cotyledons was scored (Figure 5). Under these conditions, wild-type seedlings were developmentally arrested and the cotyledons did not green. As a positive control the ABA-deficient mutant aba3-2 (Leon-Kloosterziel et al., 1996a) was assessed, and seedlings both greened and continued to develop and grow. Interestingly, 71B6-OE lines displayed an intermediate phenotype with respect to both greening and growth. As glucose stress leads to elevated ABA and the phenotypes observed, the results suggest that overexpression of GT prevented the increase in endogenous ABA and thereby reduced the associated phenotype.

Figure 5.

Post-germinative growth on high glucose.
Seeds were sown on medium supplemented with 6% glucose. The glucose-insensitive phenotype was scored as the percentage of seedlings showing green cotyledons after 10 days. Data are based on three replicates (±SE). Wild type (WT); 71B6-OE (12, 13 and 17); ABA-deficient mutant aba3-2 (aba3). Pictures of representative seedlings are shown.

Water loss from detached leaves.  A standard assay for ABA deficiency is the loss of water from detached leaves. As anticipated, detached leaves of the ABA-deficient aba3-2 mutant lost weight much more rapidly than wild type (Figure 6). There was a slight but consistent increase in the rate of water loss from 71B6-OE leaves. Unexpectedly, water loss from detached leaves of the NCED3-OX transgenic line, known to accumulate higher levels of ABA in turgid rosettes (Iuchi et al., 2001), was not significantly different from wild type. These data suggest that increased glucosylation of ABA in the overexpressors leads to partial impairment in stomatal response. 71B6-KO plants were also analysed in the three assays described above and no difference was observed compared with wild type (data not shown).

Figure 6.

Detached-leaf water loss.
Assays were carried out on NCED3-OX, wild-type (WT), 71B6-OE 13 and aba3-2 mutant plants. Five individual leaves were detached from independent plants and placed upside down on a weighing boat. Fresh weight was determined immediately and weight loss was measured hourly. Data are based on three independent replicates of five leaves (±SE). Experiment also included 71B6-OE 12 and 17 with similar results (data not shown).

Endogenous levels of ABA and ABA metabolites

To determine the extent to which changing the expression of 71B6 affects levels of endogenous ABA and ABA-GE, the hormone and its metabolites were measured in planta. A method using liquid chromatography tandem mass spectrometry has been developed previously that allows quantification of levels of not only ABA and ABA-GE, but also the acidic catabolites PA, DPA, 7′OH-ABA and neo-PA in a single sample (Zhou et al., 2003, 2004). Given that ABA increases in response to stress (Leon-Kloosterziel et al., 1996a,b; Walton and Li, 1995), it was possible that the relative flux through the different catabolic pathways in wild type compared with the overexpressors and knockout would change when levels of the hormone were elevated. Therefore the compounds were measured in both turgid and wilted rosette tissue. The data are illustrated in Figure 7 and provided in numerical form in Table S1.

Figure 7.

Endogenous levels of ABA and catabolites.
Analysis of endogenous ABA, ABA-GE, PA, DPA, 7′OHABA and neo-PA in 4-week-old turgid (a) and wilted rosettes (b) of: wild type (WT); empty vector control (Ri); transposon insertion mutant (71B6-KO); three independent 71B6-OE lines (left to right, 12, 13 and 17); aba3-2; and NCED3-OX. Data are based on three independent replicates (±SE). Note different scales.

In turgid rosettes of wild-type and Ri plants, levels of free ABA, ABA-GE and PA were found to be similar at 150–200 pmol g−1 DW, whereas the catabolite DPA was much higher at approximately 1000 pmol g−1 DW. As anticipated, under wilted conditions levels of free ABA in the rosette leaves increased substantially to approximately 4000 pmol g−1 DW. Phaseic acid also increased approximately 20-fold, whereas levels of DPA and ABA-GE only rose twofold. In both turgid and wilted conditions, 7′OH-ABA was detectable at only approximately 22–32 pmol g−1 DW. Neo-PA was present at 6–9 pmol g−1 DW in turgid tissue, increasing at least 12-fold in wilted tissue to 110–119 pmol g−1 DW.

The aba3-2 mutant was used as a control for analysing the metabolite profile of an ABA-deficient plant. Negligible levels of ABA and its metabolites were observed in this mutant under turgid or wilted conditions, with the exception of DPA, present at approximately 170 pmol g−1 DW in turgid leaves and approximately 270 pmol g−1 DW in wilted leaves. The NCED3-OX transgenic line (Iuchi et al., 2001) was used as a further control for analysing the metabolite profile of a plant with elevated ABA. In turgid leaves, levels of free ABA in the transgenic line were approximately double those found in wild type. Interestingly, under wilt stress both wild-type and NCED3-OX plants contained near-identical levels of free ABA. This pattern was very similar for the metabolites PA and DPA, which were again at higher levels in transgenic compared with wild-type plants under turgid conditions, but were very similar under wilt stress. In NCED3-OX levels of ABA-GE were found to be very similar to those in wild type under both conditions assayed.

When 71B6-KO was analysed, no significant change was observed in any of the metabolites compared with wild-type and Ri control plants. The level of ABA-GE in turgid leaves was below the detection limit of the analytical technique. In contrast, substantial changes were observed in the 71B6-OEs. Levels of ABA-GE in the 71B6-OEs were extremely high, whether the plants were turgid or wilted (ninefold or 19-fold higher, respectively, relative to wild type). Despite the accumulation of ABA-GE, levels of free ABA in turgid conditions were unaffected. When the rosettes were wilted, a decrease in free ABA in the 71B6-OEs was observed. Interestingly, levels of the ABA catabolites PA, DPA and neo-PA were reduced significantly in the 71B6-OEs compared with wild type in both turgid and wilted rosettes.

Metabolite profiling was also carried out on different developmental stages of Arabidopsis (Table 1; and Table S2) and following salt stress (Table S3). The purpose was to compare changes in level of ABA-GE in wild-type and 71B6-KO plants under these different conditions. In wild type, levels of ABA-GE were highest in 5-day seedlings; 14-day and 4-week aerial tissue; flowers; siliques; and salt-stressed rosette leaves (Table 1; Table S3). For these stages, ABA-GE levels decreased only slightly in 71B6-KO, but the significance of this is uncertain due to biological variation (Tables S2 and S3). Concerning the ABA catabolites, it can be observed that high relative levels of PA, DPA, 7′OH-ABA and neo-PA correlated with high levels of ABA in flowers and siliques, and in salt-stressed rosette leaves. Interestingly, in dry seeds and 1-day-stratified seeds, ABA accumulated at a high level but the catabolites did not.

Table 1.  Accumulation of ABA/catabolites at different developmental stages in wild-type plants
 Wild typeABA (pmol g−1 DW)ABA-GE (pmol g−1 DW)PA pmol  g−1 DW)DPA (pmol g−1 DW)7′OH-ABA (pmol g−1 DW)neoPA (pmol g−1 DW)
  1. Levels of free ABA, ABA-GE, PA, DPA, 7′OH-ABA and neo-PA were measured in wild-type plants. Data in rows 2–7 correspond to tissue incubated/grown on filter paper soaked in water; 8–11, grown on 0.5 × MS (0.8% agar); 12–15, grown on soil. Data are based on three independent replicates (±SE). Absent values indicate no replicates above limit of quantification. Absent SE indicates only one or two replicates were measurable. Neo-PA is estimated based on d4 7′OH-ABA as internal standard.

 10-day (dry seed)902 ± 891 ± 3120 ± 29418 ± 5729 ± 0 
 21-day stratified948 ± 3307140 ± 3532 ± 126  
 34-day stratified141 ± 1059 ± 7 304 ± 6524 
 41-day seedling73 ± 4  443 ± 200  
 52-day seedling84 ± 398 692 ± 127  
 63-day seedling285 ± 15273 ± 8 611 ± 86  
 75-day seedling (paper)175 ± 31172 ± 40 2069 ± 155  
 85-day seedling (MS)105 ± 1948 ± 5 278 ± 1  
 97-day seedling (MS)154 ± 3291 ± 929 ± 2671 ± 183  
1014-day root (MS)497 ± 26393 ± 141214 ± 1263351 ± 115  
1114-day aerial (MS)334 ± 30314 ± 19277 ± 131212 ± 5217 
124-week rosette395 ± 15142 ± 24820 ± 1292617 ± 37614 ± 124
13Stem87 ± 152295 ± 112713 ± 462  
14Flower1004 ± 101153 ± 12891 ± 13411111 ± 96522 ± 1165 ± 6
15Silique860 ± 17189 ± 8357 ± 1411655 ± 104456 ± 1258 ± 20

Discussion

This study focuses on conjugation of the hormone ABA in planta and investigates the impact of changing the level of glucosylation on ABA metabolism and ABA-related responses. The conjugation of plant hormones to glucose is a well studied phenomenon at the biochemical level, and is thought to lead to changed bioactivity (reviewed by Bowles et al., 2006). In the context of ABA, there is considerable early literature to suggest that the endogenous glucose ester of ABA is inactive, is not hydrolyzed, and represents an end-point of ABA metabolism that is stored in the vacuole (Bray and Zeevaart, 1985; Lehmann, 1983; Lehmann and Glund, 1986; Lehmann and Schutte, 1984; Milborrow, 1978; Neill et al., 1983; Zeevaart, 1980, 1983; Zeevaart and Boyer, 1984). Some recent papers suggest that, in contrast, ABA-GE is a long-distance-transport form of the hormone, providing a source of ABA on subsequent hydrolysis (Sauter et al., 2002). In order to investigate any of these possibilities and to determine whether ABA homeostasis can be disturbed by changing the level of glucosylation in the plant, it is essential to have the relevant genetic tools. Our finding that an Arabidopsis GT, 71B6, recognized only the naturally occurring enantiomer of ABA has provided a new foundation for undertaking studies such as these (Lim et al., 2005; Priest et al., 2005).

A major finding from this study is that the plant can accommodate substantial disturbance in ABA glucosylation, with very little impact on absolute levels of free ABA and, in consequence, only restricted impact on plant growth and development. Thus constitutive overexpression of 71B6 led to massive quantities of ABA-GE relative to wild type, whereas levels of free ABA remained very similar. This indicates that the plant can regulate ABA homeostasis against a background of massively increased glucosylation. The metabolite profiles suggest that, in turgid rosette leaves, this is regulated mainly by increased synthesis of the hormone as well as some decreased catabolism via PA and DPA.

In wild type and the empty vector control, the total amount of ABA/metabolites following wilt treatment of the leaves increased eightfold from 1300 to 10 500 pmol g−1 DW, indicating that a substantial rise in ABA synthesis was induced by the wilt stress. This effect has been well documented (Hiron and Wright, 1973; Walton and Li, 1995). Interestingly, following wilt treatment the total amount of ABA/metabolites in leaves of the 71B6-OEs was very similar to that observed in wild type, but the relative pattern of metabolites was different. Again, massive quantities of ABA-GE were produced, but levels of free ABA, PA and DPA were reduced in the transgenic plants relative to wild type. This suggests that ABA homeostasis in response to imposed glucosylation is regulated in wilted leaves principally by decreased catabolism rather than increased synthesis.

Two other means of disturbing ABA homeostasis were analysed, through use of the ABA biosynthesis mutant aba3-2 and the transgenic line NCED3-OX. Under conditions of extreme ABA deficiency in the aba3-2 mutant, DPA remained at significant levels. This indicates that, even under conditions in which negligible ABA accumulates, the 8′-hydroxylation pathway of catabolism continues to operate. There is evidence that the catabolites may have important bioactivities (Zhou et al., 2004; Zou et al., 1995). The enzymic step catalyzed by NCED3 is considered to be rate-limiting (Nambara and Marion-Poll, 2005). ABA levels in the transgenic line constitutively overexpressing NCED3 were found to be double those of wild type (Iuchi et al., 2001). This has been confirmed in the present study, and the analysis has been extended to include other ABA metabolites and the consequences of wilt treatment on metabolite profiles of the NCED3-OX plants. There are two principal observations. First, in unstressed plants, high levels of DPA and PA were detected, indicating that the impact of NCED3 overexpression was greater than that suggested from only free-ABA levels. In this case, ABA homeostasis was regulated by greatly increased catabolism through 8′-hydroxylation. Second, under wilt treatment, the NCED3-OX plants and wild type had very similar levels of free ABA. It is possible that, under wilt stress, the step catalyzed by NCED3 was no longer rate-limiting.

The metabolite profiles of the 71B6-OEs provided an explanation for the restricted phenotypic changes observed in the transgenic plants compared with wild type. No major changes were observed in germination, dormancy and stomatal control. However, a change was observed in post-germinative growth in relation to both cotyledon emergence and cotyledon greening. Use of the ABA structural analogues confirmed that the change was due to glucosylation of ABA/ABA analogues by 71B6. A substantial number of ABA analogues have been designed for use in exploring the structural basis of the ABA molecule in relation to its bioactivity and its inactivation and catabolism (reviewed by Zaharia et al., 2005a). Interestingly, several analogues have been identified that retain bioactivity, but are no longer catabolized through the main pathway initiated by 8′-hydroxylation (Cutler et al., 2000; Zaharia et al., 2005a). These offer opportunities for ‘long-lasting’ ABA-like molecules for field applications, such as the possibility of using analogues that retain bioactivity but lose the capacity for inactivation/catabolism by either pathway, whether glucosylation or hydroxylation (Priest et al., 2005). In relation to this study, overexpression of 71B6 provided resistance to the growth inhibition induced by the analogue PBI-413. Therefore it would be of interest to determine whether transgenic crops overexpressing 71B6 could be used together with an application of PBI-413 as a means of enabling the crop to grow under conditions in which weeds would be arrested by the ABA activity of the analogue.

This study contributes to the continuing discussion of the role of the glucose ester of ABA. In wild-type plants, whether stressed by wilt treatment or grown under standard conditions, very little ABA-GE accumulated. This suggests that under the conditions of the experiments used, either the conversion of ABA to ABA-GE is likely not to be a principal pathway of inactivation; or ABA-GE is continuously hydrolyzed. Results from early feeding experiments suggest that ABA-GE is not hydrolyzed (Lehmann and Schutte, 1984; Milborrow, 1978), but it is interesting that glucose esters of many other small molecules, such as sinapates, represent high-energy intermediates in transfer reactions (Baumert et al., 2005; Lehfeldt et al., 2000; Mock and Strack, 1993). In transgenic plants overexpressing 71B6, the very high levels of ABA-GE that accumulated apparently had little impact on phenotype.

An alternative means of exploring the role of ABA-GE is to analyse plants incapable of its accumulation. One method of achieving this is to knock out the GTs involved in ABA-GE synthesis. In this study, a knockout in 71B6 was analysed, but no significant changes in the profile of ABA/metabolites were observed, and no changes within the phenotypic parameters analysed could be detected. These data can be interpreted in several ways. First, it is possible that the endogenous gene 71B6 does not function in ABA metabolism in planta, despite recognizing the hormone stereoselectively in vitro. Second, the expression and function of 71B6 may be highly specific to particular cells or developmental/stress events, and the impact of losing its activity was not observed under the conditions assayed in this study. Third, there may be a number of genes encoding GTs that glucosylate ABA in planta, and knocking out a single gene will not produce a phenotype because of compensation by other GTs that are co-ordinately regulated.

The expression of 71B6 was analysed and found to be very low, both during development and in response to challenges. As yet, the cell specificity of 71B6 expression is unknown and the range of challenges applied to plants has been limited. Results from affymetrix microarray experiments, publicly available as Genevestigator (http://www.genevestigator.ethz.ch), confirmed low expression of 71B6, with only one challenge, application of cycloheximide, producing high levels of transcripts. ABA is known to be synthesized by pathogens as a virulence factor (Audenaert et al., 2002; Kettner and Dorffling, 1995). As conjugation is also considered to be a form of detoxification of exogenous/xenobiotic compounds (Bowles et al., 2005), it is possible that the glucosylation of ABA is primarily used as a defence-related process in plant–pathogen interactions.

From in vitro screening data, eight GTs of Arabidopsis were identified that were capable of glucosylating ABA (Lim et al., 2005). It is possible that these GTs also function in planta to glucosylate ABA and could, in principle, compensate for the lack of 71B6 in the knockout. It is increasingly clear that individual GTs can recognize multiple substrates and, from study of the multigene family of GTs in Arabidopsis, that a single substrate may be recognized by many individuals of the family (Bowles et al., 2006; Lim et al., 2003). This multiplicity potentially provides a highly flexible system for homeostatic adaptation to environmental challenges. In relation to ABA glucosylation, it may well prove essential to knock down multiple GTs to determine the consequences for a plant of a lack of ABA-GE.

Experimental procedures

Chemicals

The majority of the chemicals were obtained from Sigma-Aldrich (Poole, Dorset, UK). (+)-ABA for growth experiments was purchased from OlChemIm Ltd (Olomouc, Czech Republic). ABA analogues, standards and deuterated standards for mass spectrometry were synthesized at the Plant Biotechnology Institute (Saskatoon, Canada) as described: PBI-514 (Cutler et al., 2000); PBI-413 (S.R.A., unpublished data); (−)-PA (Balsevich et al., 1994); (−)-DPA, (+)-ABA-GE and (±)-7′OH ABA (Nelson et al., 1991); (−)-5,8′,8′,8′-d4-ABA (Abrams et al., 2003); (−)-7′,7′,7′-d3-PA, (−)-7′,7′,7′-d3-DPA, (−)-5,8′,8′,8′d4-7′OH ABA and (+)-4,5,8′,8′,8′-d5-ABA-GE (Zaharia et al., 2005b).

Plants and growth conditions

UGT71B6 was overexpressed in Arabidopsis thaliana plants (ecotype Columbia) under control of the CaMV 35S promoter from binary vector pJR1Ri. 71B6-KO is a transposon-insertion line obtained from the Sainsbury Laboratory Arabidopsis Transposants (SLAT) collection (02_26_08). The 71B6-KO line was complemented by transformation with pART27 vector carrying the genomic fragment containing the UGT71B6 gene including its own promoter (440 bp) and terminator (1000 bp) to give rise to the 71B6-RE lines.

NCED3-OX (Iuchi et al., 2001) was kindly provided by Kazuo Shinozaki; aba3-2 was previously described by Leon-Kloosterziel et al. (1996a).

Plants were grown on Levington's seed and modular compost in a controlled environment of 16:8 h light–dark cycle (22°C day, 18°C night, 150 μmol m−2 sec−1). In some figures, for clarity, only one representative 71B6-OE line is shown.

Salt treatment was performed as follows: 4-week-old plants, grown hydroponically in 0.5 × MS medium, were incubated with their roots submerged in media supplemented with 150 mm NaCl for 24 h.

For post-germinative growth and glucose-sensitivity experiments, seeds of the same age were surface sterilized with chlorine gas and plated on 1 × MS medium containing 0.8% (w/v) agar and, where relevant, 0.5 μm (+)-ABA, PBI-514, PBI-413 or 6% (w/v) glucose. Following stratification at 4°C in the dark for 4 days, seeds were transferred to a controlled environment of 16:8 h light–dark cycle (22°C day, 18°C night, 80 μmol m−2 sec−1). The percentage of seedlings with cotyledons to have emerged from the seed coat or to have greened was scored. Experiments were performed with three replicates of at least 150 seeds.

RNA analysis and enzymatic activity

Total RNA was extracted from plant tissue using TRI reagent (Sigma-Aldrich) according to the manufacturer's instructions. Standard procedures were used for Northern blot analyses. Aerial tissue (1 g FW) from 4-week-old plants was ground to a fine powder in liquid nitrogen using a pestle and mortar. 2.5 ml extraction buffer (100 mm Tris pH 6.95, 10% (v/v) glycerol, 20 mm dithiothreitol, 1 mm phenylmethylsulfonyl fluoride, 1% (w/v) polyvinylpolypyrrolindone) was added to the powder and the slurry thawed on ice. Subsequently the slurry was mixed vigorously, filtered through two layers of miracloth, and centrifuged at 10 500 g at 4°C for 20 min. The protein concentration of the supernatant was determined with Bio-Rad Protein Assay Dye (Hemel Hempstead, UK) and the crude extract assayed for glycosyltransferase enzyme activity. 50 μl crude extract (20–50 μg total protein) was mixed with 1 mm (±)-ABA, 5 mm UDP-glucose, in a final volume of 100 μl. The reaction was incubated at 30°C for 0, 2 or 6 h and stopped by the addition of 10 μl trichloroacetic acid (240 mg ml−1). The reaction mix was analysed by reverse-phase HPLC (Waters Alliance 2690 and Waters Tuneable Absorbance Detector 486, Waters Ltd, Elstree, UK) and a 5-μm C18 column (250 × 4.60 mm; Phenomenex, Torrance, CA, USA). A linear gradient with increasing methanol against 0.1 m acetic acid (pH 3.5 triethylamine) from 10% to 80% over 30 min was used to separate ABA-GE from ABA, and the eluate was monitored at 270 nm.

Detached leaf water-loss assay

Plants were grown to 4 weeks as described, and for the last 2 days the humidity was increased to 80%. Five leaves from independent plants were detached and placed upside down on a weighing boat. Fresh weight was determined immediately and weight loss measured hourly. Data are based on three independent replicates (±SE).

Extraction and mass spectrometric analysis of ABA and ABA metabolites

For turgid tissues, 4-week-old rosettes were harvested and frozen immediately in liquid N. The wilting treatment was carried out according to Leon-Kloosterziel et al. (1996a). Detached rosettes were dehydrated in dry air at 20°C until 15% of the fresh weight was lost. The stressed material was kept in polythene bags at 22°C for 6 h and then frozen in liquid N. The frozen material was freeze-dried and the dry weight determined.

ABA, ABA-GE, PA, DPA, 7′hydroxyABA and neoPA were extracted for analysis as described. Freeze-dried tissue (50–100 mg) was ground to powder in 2-ml screw-cap microcentrifuge tubes (Starstedt, Leicester, UK) using 6.35 mm ceramic beads (Q-biogene, Kent, UK) with a Fast Prep FP 120 machine (Q-biogene; 10 sec at 5 m sec−1). Internal standards [20 ng: (−)-5, 8′,8′,8′-d4 ABA; (−)-7′,7′,7′-d3 PA; (−)-5,8′,8′,8′-d4 7′OH-ABA; (−)-7′,7′,7′-d3 DPA; (−)-8′,8′,8′-d3 neoPA; (+)-5,4,8′,8′,8′-d5 ABA-GE] and 1 ml extraction solvent [80% isopropanol, 19% water, 1% glacial acetic acid (v/v)] were added. Samples were mixed by vortexing then incubated, with shaking, overnight at 20°C. Following centrifugation (16 000 g for 2 min, 20°C) the supernatant was collected and pellets rinsed with a further 0.5-ml extraction solvent. The supernatants were pooled in a fresh tube and dried by centrifugation under vacuum. Following resuspension in 1 ml 99% isopropanol:1% acetic acid (v/v) by vortexing and sonication, samples were centrifuged (16 000 g for 2 min, 20°C), and the supernatant was transferred to a fresh tube then dried again. Samples were dissolved in 50 μl 99% methanol:1% acetic acid (v/v) and a further 450 μl 99% water:1% acetic acid (v/v) was added. Oils in the samples were removed by partitioning using 1 ml hexane and the remaining aqueous extracts (bottom layer) were again removed to a fresh tube and dried by centrifugation under vacuum. Extracts were dissolved in 100 μl 99% methanol:1% acetic acid (v/v) topped up to 1 ml with 99% water:1% acetic acid (v/v). Oasis HLB 1-ml solid-phase extraction cartridges (Waters) were conditioned with 1 ml 99% acetonitrile:1% acetic acid (v/v) followed by 1 ml 99% methanol:1% acetic acid (v/v) and equilibrated with 1 ml 99% water:1% acetic acid (v/v). Samples were centrifuged (16 000 g for 2 min) to remove any remaining particulate material prior to loading. Samples were loaded under a vacuum of 0.16–0.19 kPa below atmosphere, followed by a wash with 1 ml 99% water:1% acetic acid (v/v).

Analytes were eluted using 1 ml 50% acetonitrile, 49% water:1% acetic acid (v/v) before samples were dried by centrifugation under vacuum and stored at 4°C. ABA and ABA metabolites were quantified by mass spectrometry as described by Feurtado et al. (2004) with the addition of the analyte neo-PA (cone voltage = 20 V, collision cell = 13 V; transition 279 > 205 and d3 neo-PA 282 > 208).

Acknowledgements

The authors would like to thank Rosamond Jackson and Steve Penfield for helpful discussions; Allan Feurtado for help with extraction of ABA metabolites for analysis; Kazuo Shinozaki for kindly providing the NCED3-OX line. The Gatsby Charitable Foundation is thanked for providing a studentship to D.M.P. The Biotechnology and Biological Sciences Research Council and The Garfield–Weston Foundation are thanked for funding other aspects of this work.

Ancillary