Ectopic expression of a tomato 9-cis-epoxycarotenoid dioxygenase gene causes over-production of abscisic acid


For correspondence (fax +44 1789 470552; e-mail


The tomato mutant notabilis has a wilty phenotype as a result of abscisic acid (ABA) deficiency. The wild-type allele of notabilis, LeNCED1, encodes a putative 9-cis-epoxycarotenoid dioxygenase (NCED) with a potential regulatory role in ABA biosynthesis. We have created transgenic tobacco plants in which expression of the LeNCED1 coding region is under tetracycline-inducible control. When leaf explants from these plants were treated with tetracycline, NCED mRNA was induced and bulk leaf ABA content increased by up to 10-fold. Transgenic tomato plants were also produced containing the LeNCED1 coding region under the control of one of two strong constitutive promoters, either the doubly enhanced CaMV 35S promoter or the chimaeric ‘Super-Promoter’. Many of these plants were wilty, suggesting co-suppression of endogenous gene activity; however three transformants displayed a common, heritable phenotype that could be due to enhanced ABA biosynthesis, showing increased guttation and seed dormancy. Progeny from two of these transformants were further characterized, and it was shown that they also exhibited reduced stomatal conductance, increased NCED mRNA and elevated seed ABA content. Progeny of one transformant had significantly higher bulk leaf ABA content compared to the wild type. The increased seed dormancy was reversed by addition of the carotenoid biosynthesis inhibitor norflurazon. These data provide strong evidence that NCED is indeed a key regulatory enzyme in ABA biosynthesis in leaves, and demonstrate for the first time that plant ABA content can be increased through manipulating NCED.


The plant hormone abscisic acid (ABA) is involved in mediating responses to environmental stresses, notably stomatal closure, alterations in growth patterns and induction of many stress-related gene products. It also has a role in developmental processes such as seed maturation and dormancy ( Zeevaart & Creelman, 1988). Evidence from 18O2 incorporation experiments ( Creelman & Zeevaart, 1984) and studies of ABA-deficient mutants ( Duckham et al. 1991 ; Rock & Zeevaart, 1991) strongly indicated that ABA is synthesized from C40 epoxycarotenoid precursors which are cleaved to produce a C15 intermediate. The first cloned gene in the ABA biosynthetic pathway encodes a zeaxanthin epoxidase (ZEP) from tobacco ( Marin et al. 1996 ) which catalyses the conversion of zeaxanthin to the epoxycarotenoid, violaxanthin. This can be considered as the first step in ABA biosynthesis ( Taylor, 1991). More recently, the Vp14 gene encoding a 9-cis-epoxycarotenoid dioxygenase (NCED) ( Schwartz et al. 1997b ) was cloned, via transposon tagging, from a mutant of Zea mays with reduced seed dormancy ( Tan et al. 1997 ). Genes homologous to Vp14 were subsequently isolated from Lycopersicon esculentum ( Burbidge et al. 1997a ) and Arabidopsis thaliana ( Neill et al. 1998 ), and are now named LeNCED1 and AtNCED1, respectively. LeNCED1 was subsequently shown to be the wild-type allele of notabilis (Burbidge et al. 1999), a wilty tomato mutant first described by Stubbe (1958).

Recombinant NCED produced in Escherichia coli from the wild-type Vp14 allele was able to selectively cleave, in vitro, 9-cis isomers of a number of epoxycarotenoids to form xanthoxin ( Schwartz et al. 1997b ). Candidate substrates for NCED in planta include 9′-cis-neoxanthin and 9-cis-violaxanthin ( Parry et al. 1990 ), producing the C15 intermediate xanthoxin which can then be oxidized in two further enzyme-catalysed steps to ABA. Mutants have been found affecting the last two steps in the pathway ( Schwartz et al. 1997a ; Taylor, 1991).

The high abundance of some potential 9-cis-epoxycarotenoid precursors in photosynthetic tissues ( Parry et al. 1990 ) suggests that the formation of these precursors is not normally rate-limiting in ABA biosynthesis. In vitro enzymatic conversion of xanthoxin to ABA occurs equally rapidly in cell-free extracts from non-stressed and droughted plants ( Sindhu & Walton, 1987), implying that de novo synthesis of these enzymes may not be required for stress-induced ABA biosynthesis. Inhibitor studies have indicated that de novo transcription and translation are required for stress-induced ABA biosynthesis ( Guerrero & Mullet, 1986; Quarrie & Lister, 1984), allowing the possibility that transcriptional up-regulation of at least one rate-limiting enzyme is a regulatory mechanism. NCED mRNA does increase in response to water-deficit stress in leaves and roots of L. esculentum ( Burbidge et al. 1997a ; Thompson et al. 2000 ) and Phaseolus vulgaris (Qin & Zeevaart, 1999), and in leaves of Z. mays ( Tan et al. 1997 ) and A. thaliana ( Neill et al. 1998 ). All the above evidence leads to the hypothesis that cleavage of 9-cis-epoxycarotenoids by NCED is an important regulatory step in stress-induced ABA biosynthesis, at least in photosynthetic tissue. It has been suggested that overexpression of NCED may not result in ABA accumulation because ‘enhanced biosynthesis will be accompanied by a concomitant increase in ABA degradation’ ( Zeevaart, 1999). The availability of genes encoding NCED allows these hypotheses to be tested in transgenic plants, and potentially allows ABA biosynthesis rates to be increased. We report here the overexpression of LeNCED1 in tobacco using an inducible promoter, and in tomato using two strong, constitutive promoters, and describe the resulting changes in seed and leaf ABA content, seed dormancy and stomatal conductance.


Tetracycline-inducible expression of LeNCED1 results in an increase in ABA content in tobacco leaves

Tobacco plants were produced that allowed induction of LeNCED1 gene expression by the application of tetracycline (Tc). Tobacco explants showing stable expression of tetR, encoding the tetracycline repressor protein ( Gatz et al. 1991 ), were transformed with pBHT-NCED ( Fig. 1). This plasmid contained the ‘Triple-Op’ promoter, a modified CaMV 35S promoter carrying three TetR binding sites. Transgenic plants were then selected, based on regeneration and rooting in the presence of hygromycin. The Triple-Op promoter is repressed by the binding of TetR, but when Tc is added, a Tc–TetR complex is formed and the promoter is de-repressed ( Gatz et al. 1992 ).

Figure 1.

Scale diagram of constructs used in plant transformation.

RB, right T-DNA border; LB, left T-DNA border; D35S(p), doubly enhanced CaMV 35S promoter; SP, ‘Super-Promoter’; Triple-Op, CaMV 35S promoter containing three TetR operator sites; nos(p), nopaline synthase promoter; 35S(a), nos(a), ocs(a) and G7(a), polyadenylation signals from CaMV 35S, nopaline synthase, octopine synthase and Agrobacterium tumefaciens gene 7, respectively; nptII, neomycin phosphotransferase II; hptII, hygromycin phosphotransferase II.

In a preliminary experiment, leaf explants from 27 hygromycin-resistant plants were given 24 h incubation with and without Tc. Twenty of these primary transformants showed at least a twofold increase in ABA in response to Tc (data not shown). Leaf explants from six of these transformants, shown to be from independent transformation events (data not shown), plus the tetR control, were again incubated for 24 h with or without tetracycline, then NCED mRNA and ABA were determined ( Fig. 2). Total RNA from the tetR control plant showed no detectable hybridization to the LeNCED1 probe, and the ABA content was at the same basal level when incubated with or without Tc. This important control demonstrated that application of Tc had no effect on ABA content when the inducible LeNCED1 transgene was not present. When the inducible transgene was present, but no Tc was applied, NCED mRNA was detected at a basal level. This indicates that the TetR protein is not totally effective as a repressor of the Triple-Op promoter. However this basal level was apparently not sufficient to cause an increase in bulk leaf ABA content, at least in transformants 1, 6 and 20 ( Fig. 2). This may be because the basal level was low in comparison to the expression of a presumed endogenous tobacco gene functionally homologous to LeNCED1.

Figure 2.

Induction of ABA and NCED mRNA following tetracycline treatment of transgenic tobacco leaf explants.

Leaf explants of primary transformants were incubated for 24 h in artificial sap alone (black bars) or artificial sap plus 1 mg l−1 tetracycline (white bars). Both NCED mRNA and ABA were measured in each sample. The symbol tetR indicates a transgenic tobacco plant expressing the tetR transgene only. The other numbered transformants contain tetR and were further transformed with the pBHT-NCED construct ( Fig. 1). In the upper chart, NCED mRNA data are from a single assay of a single tissue sample, except for transformant 20 where triplicate leaf samples were taken, and the standard error of these triplicate data is shown. In the lower chart, with the exception of transformant 20, ABA data are from duplicate assays from single tissue samples and the standard error for the duplicate assay is indicated. In the case of transformant 20, duplicate assays were performed on the triplicate tissue samples, and the standard error of the triplicate means is shown.

When Tc was applied to the six transformants containing the inducible transgene, a very dramatic increase in both NCED mRNA and ABA content was observed. Transformant 20 had the largest response to Tc, with the highest induced levels of NCED mRNA and a remarkable 10-fold increase in ABA content.

Long-term constitutive over-expression of LeNCED1 in tomato

Tc-induced expression of LeNCED1 clearly caused an increase in ABA content in tobacco, but as the induction was only over a 24 h period this may have been a transient effect. To allow constitutive high expression of LeNCED1 in tomato, the LeNCED1 coding region was cloned downstream from a doubly enhanced CaMV 35S promoter (D35S) and the chimaeric ‘Super Promoter’ (SP), to produce plasmids pBP-D35B-NCED and pBP-SP-NCED, respectively ( Fig. 1). The T-DNAs were introduced into wild-type tomato yielding 51 and 10 independent regenerated plants from pBP-D35B-NCED and pBP-SP-NCED, respectively. After plants were weaned from tissue culture to compost, it was apparent that many of them had a strong tendency to wilt. To prevent wilting, and to allow continued growth, plants were transferred to a controlled-environment cabinet set to 90% relative humidity. Three transformants were then observed to grow slowly and to guttate, i.e. to produce droplets of liquid at the leaf margins, a phenomenon not observed in any other transformants or wild-type plants in the same environment. These plants are referred to as ‘overguttating’.

After transfer of all transgenic plants to the glasshouse and observation for a few weeks, it was possible to place each transformant into one of three categories: wilty, normal or overguttating. For D35S, plants were scored as 36 wilty, 14 normal and one overguttating. For SP there were three wilty, five normal and two overguttating transformants. Many of the wilty transformants also showed leaf epinasty, swelling of stems and adventitious rooting behaviours characteristic of the three tomato ABA-deficient mutants sitiens, flacca and notabilis ( Tal, 1966). The three overguttating transformants were called D9 (using the D35S promoter), and SP5 and SP6 (using the SP promoter). In the glasshouse the unusual guttation appeared to coincide with periods of high humidity. Excessive guttation could be readily observed by increasing humidity with a polythene cover ( Fig. 3a,b), and was observed in progeny of all three transformants. Overguttation was sometimes associated with chlorotic patches on leaves, which were unusual because they always had discrete edges bordered by leaf veins ( Fig. 3c,d).

Figure 3.

Guttation and leaf-margin chlorosis in tomato plants overexpressing LeNCED1. A progeny plant from a selfing of SP5 is shown to guttate excessively after transfer for 24 h to a sealed polythene chamber in the glasshouse (a,b). Loss of chlorophyll at the leaf margin is shown in a transgenic progeny from a wt × D9 cross (c) and the SP6 primary transformant (d).

Since overguttation could be interpreted as a symptom of a constitutive high ABA content, the remaining work presented here was restricted to the progeny of the overguttating transformants. Southern blot analysis of progeny (data not shown) indicated that D9 and SP5 each contained a single T-DNA locus, whereas SP6 had at least three T-DNA loci that showed segregation from each other. To allow the most simple analysis of progeny, D9 and SP5 were chosen for further study.

Leaf ABA and NCED mRNA content in progeny of overguttating transformants

Progeny from selfing D9 that were positive for T-DNA on Southern analysis had a mean ABA content considerably higher than wild-type plants, representing a 2.8-fold increase ( Table 1). This difference in mean ABA content was significant at the 5% level (P= 0.014). When Northern blots were probed with the LeNCED1 coding region, able to hybridize to mRNA from both the endogenous LeNCED1 and the transgene, D9 progeny were shown to have 17-fold higher NCED mRNA than the wild type, and this was significant at the 1% level ( Table 1; Fig. 4). In order to measure the endogenous NCED mRNA independently of the mRNA derived from the NCED transgene, Northern blots were hybridized to a 3′ untranslated region of LeNCED1 which was not present in the NCED transgene. The ratio of endogenous NCED mRNA between D9 and the wild-type control was 0.8, and this difference was not significant ( Table 1; Fig. 4). Since the endogenous LeNCED1 mRNA level is similar between the D9 progeny and the wild type, regulation of the endogenous gene does not account for the large difference in ABA content. This strongly suggests that it is the expression of the LeNCED1 transgene that is responsible for the higher ABA content.

Table 1.  ABA and NCED mRNA in leaves of D9 and SP5 progeny
 NCED mRNA, Coding region (% maximum)NCED mRNA, 3′-UTR (% maximum) ABA (ng g−1 FW)
nMeanMean of logsMeanMean of logs
  1. For the D9 × self cross, transgenic progeny (+T) were compared to wild-type (wt) plants grown under identical conditions. For the wt × SP5 cross, progeny containing the transgene (+T) were compared to azygous progeny (–T). All plants were sampled from the glasshouse at 11:00 h. Probes for NCED mRNA analysis were either the complete coding region or the 3′ untranslated region (3′-UTR) of LeNCED1, as indicated. SED, standard error of difference; n, number of progeny assayed. Means were significantly different at the 0.1% (***), 1% (**), or 5% (*) levels, or were not significant at 5% (ns).

Wild type32.40.86854.45698
D9 × self (+T)3403.70644.161937
SED  0.52 ** 0.22 (ns)365 *
wt × SP5 (–T)61.20.20253.23382
wt × SP5 (+T)65.11.63413.72424
SED  0.18 *** 0.29 (ns)40 (ns)
Figure 4.

Northern analysis of NCED mRNA in progeny from D9 and SP5.

D9 × self progeny that were positive for T-DNA were compared to wild-type plants. Progeny from a wt × SP5 cross were scored for presence (+) or absence (–) of T-DNA, then the two classes were compared for each cross. Both sets of plants are the same as those used for the quantification presented in Table 1. Probes for NCED mRNA analysis were either the complete coding region or the 3′ untranslated (3′ UTR) of LeNCED1, as indicated.

For SP5, progeny from a back-cross to wild type (wt × SP5) were assayed ( Table 1; Fig. 4). NCED mRNA, as determined with the coding-region probe, was again significantly higher (0.1% level) when the transgene was present, but in this case only by 3.9-fold. However, this was not reflected in a significant difference in bulk leaf ABA content ( Table 1). As in the case of D9, there was no significant difference in endogenous NCED mRNA in the presence or absence of the transgene.

Stomatal conductance in overguttating transformants

Stomatal conductance (gs) was determined for SP5 and D9 homozygous plants relative to wild-type control plants ( Fig. 5). SP5 had a mean gs value that was 51% of wild type (significant at the 1% level); gs in D9 was also lower, at only 59% of the wild-type mean (significant at the 0.1% level).

Figure 5.

Stomatal conductance (gs) of tomato plants overexpressing LeNCED1.

Transgenic plants homozygous for SP5 and D9 (white bars) and wild-type control plants (black bars) were grown in controlled-environment cabinets and gs was determined by gas-exchange analysis. SP5 and D9 were studied in separate experiments under different conditions, so wild-type control data are presented in each case. gs was determined for 10 or five replicate plants for D9 and SP5, respectively; error bars represent standard error.

Enhanced seed dormancy

We observed that seeds from the three overguttating transformants germinated later and to a lower final percentage than wild-type seeds. To investigate further, seeds obtained by selfing D9 and SP5 were sown on an agar medium with or without the addition of 5 mg l−1 norflurazon, an inhibitor of phytoene desaturase ( Tomlin, 1997) that depletes the precursor pool for NCED, and therefore blocks ABA biosynthesis ( Fig. 6). In the absence of norflurazon, wild-type seed germinated to 100% within 13 days, and this was reproducibly advanced by 2 days in the presence of norflurazon. Both D9 and SP5 had a reduced final percentage germination in the absence of norflurazon: 53 and 15%, respectively. Germination was restored to the wild-type level when seed were sown directly onto norflurazon. Seed on media lacking norflurazon that remained ungerminated after 26 days were transferred to the same medium plus norflurazon. Within 1–2 days of transfer, there was a burst of germination in both D9 and SP5 ( Fig. 6).

Figure 6.

Percentage germination of seed from selfing of D9 and SP5 in the presence and absence of norflurazon.

Forty seeds per treatment were imbibed according to protocol (i) on 3MSB5K media (▴) or 3MSB5K media plus 5 mg l−1 norflurazon (●). After 26 days (vertical dashed line), seeds that remained ungerminated on 3MSB5K media alone were transferred to 3MSB5K plus 5 mg l−1 norflurazon and further germination was scored.

Seed dormancy and viability may vary somewhat between seed batches, so germination was also determined in a single batch of back-crossed seed to provide unequivocal evidence of a transgene effect. A population of seeds was obtained by a back-cross of the SP5 primary transformant to wild type, using wild type as the maternal parent. Thus all the seed had wild-type maternal tissues (testa), but 50% of the embryos had lost the transgene and the other 50% were heterozygous for the transgene. SP5 was chosen because it appeared to have a greater dormancy than D9 ( Fig. 6). Progeny from the wt × SP5 cross were germinated on water plus or minus 3 mg l−1 norflurazon ( Fig. 7). A lower dose of norflurazon was used here to allow recovery of viable seedlings so that the presence or absence of T-DNA could be determined later by Southern blot analysis. In addition, seed were germinated in individual compartments to prevent potential diffusion of phytohormones from seed to seed. Taking the plus or minus norflurazon treatments together, there were 42 seedlings with the transgene and 40 without ( Fig. 7b,c), thus confirming the expected 1 : 1 ratio from the cross. Figure 7(a) shows that when seeds were germinated without norflurazon, there was a biphasic germination curve with two bursts of germination centred on days 2–3 and then days 7–9. When the seed were germinated on norflurazon there was a single burst of germination on days 2–3, reaching a maximum on day 5. In Fig. 7(b), germination without norflurazon was plotted separately for seed with or without the transgene. It is clear that the burst of germination on days 2–3 seen in Fig. 7(a) is all due to seeds lacking the transgene, and that the burst between days 7 and 9 is due to those with the transgene. The transgene effectively delays germination by 5 days. In Fig. 7(c), the effect of the transgene was diminished in the presence of norflurazon, with a delay of only 1–2 days. All seeds that had not germinated after 14 days on water alone were transferred to water plus norflurazon. This resulted in germination of a further four seeds within 6 h, all of which carried the transgene ( Fig. 7b).

Figure 7.

Germination of seed from the back-cross wt × SP5.

Fifty seeds from the back-cross wt × SP5 were imbibed on filter paper soaked in water or water plus 3 mg l−1 norflurazon, according to protocol (ii). Each seedling was then scored for presence or absence of the segregating T-DNA by Southern blot analysis. Genomic DNA was cut with BamHI and blots were probed with the LeNCED1 coding region to detect the LeNCED1 transgene.

(a) Germination was scored at the indicated times for the two treatments, irrespective of whether or not individual seedlings had inherited the T-DNA. ●, water plus norflurazon; ▴, water alone. (b) Germination on water alone with data plotted separately for seedlings containing T-DNA (▿), or not containing T-DNA (▪). (c) As (b), but for germination on water plus norflurazon.

To provide further evidence that the delayed germination effect was due to increased ABA biosynthesis, we measured ABA content by GC–MS in imbibed but ungerminated seeds of the wild type, and from selfing of the SP5 and D9 primary tranformants ( Fig. 8). The SP5 × self and D9 × self seed batches had threefold more ABA than wild-type seed (significant at the 0.1% level).

Figure 8.

ABA content of seeds from selfing of D9 and SP5.

ABA content of ungerminated seeds was determined by GC–MS after 48 h imbibition in water. Fifty seeds were sampled in triplicate; error bars, standard error.


Using a transgenic approach we have increased the expression of LeNCED1, a gene believed to encode a key regulatory enzyme in the ABA-biosynthesis pathway. Increased expression led to increased ABA content in both tobacco and tomato, and to phenotypic effects in tomato that can be explained in terms of the known physiological actions of ABA.

Phenotype in tomato – guttation and stomatal conductance

The predominant phenotype associated with introduction of the constitutively expressed LeNCED1 transgene was the wiltyness associated with ABA deficiency. This could have resulted from a co-suppression effect ( Flavell, 1994) in which expression of both the transgene and the endogenous LeNCED1 gene was reduced, resulting in an ABA-deficient plant phenotypically similar to the notabilis mutant. However, three further independent transgenic plants were identified by their non-wilty, overguttating phenotype (SP5, SP6 and D9). Guttation is the extrusion of xylem sap through hydathodes at the leaf margins. The guttation could be attributed to the presence of the transgene, because two different constructs resulted in the phenotype and because the effect was heritable. Root pressure is the hydrostatic pressure in the stele that induces a volume flow of solute in the xylem vessels. Increased guttation could be explained by an increase in root pressure as a result of increased root hydraulic conductivity, previously observed as an action of ABA ( Glinka, 1980; Zhang et al. 1995 ). Thus the guttation phenotype could be a direct effect of increased ABA biosynthesis resulting from expression of the transgene. High root pressure combined with low transpiration will result in guttation and, at high humidity (low vapour pressure deficit), transpiration would be very low in all plants irrespective of stomatal aperture. Therefore the increased guttation exhibited by the transgenic plants under high humidity is likely to be caused by increased root pressure rather than stomatal closure.

When plants were grown in unstressed conditions at low-to-moderate humidity there was an observed reduction in stomatal conductance in SP5 and D9 compared to the wild type ( Fig. 5). This is consistent with an increase in ABA signal in these transgenic plants, as ABA is well known to cause stomatal closure ( Zeevaart & Creelman, 1988).

Tomato phenotype – seed dormancy

Overguttating transformants showed prolonged seed dormancy ( Fig. 6), and this was clearly demonstrated to be linked to the presence of the LeNCED1 transgene in embryonic tissues by analysis of a back-cross population from SP5 ( Fig. 7). Imbibed seeds from both SP5 and D9 contained more ABA than wild-type seeds, and an inhibitor which blocks carotenoid biosynthesis, norflurazon, was able to restore seed germination to wild-type behaviour. Together these data strongly suggest that enhanced expression of LeNCED1 in the embryo is able to increase ABA synthesis, and that the ABA produced causes prolonged dormancy. Although primary seed dormancy in tomato is weak, it is well known that exogenous ABA prevents germination in tomato seeds ( Hilhorst et al. 1998 ). Our work indicates that enhanced endogenous ABA biosynthesis is also able to prolong dormancy. Previous work has shown that in situ ABA biosynthesis is required to impose and maintain embryo dormancy ( Le Page-Degivry & Garello, 1992) and so ABA biosynthesis in the imbibed seed is a natural mechanism of dormancy in some species. Ectopic expression of ZEP in Nicotiana plumbaginifolia led to increased seed dormancy and ABA content of mature seeds ( Frey et al. 1999 ). Since we have shown in L. esculentum that ectopic expression of LeNCED1 also increased ABA-mediated seed dormancy, it is likely that in embryos the rate of ABA biosynthesis can be influenced by abundance of either ZEP or NCED.

Increased leaf ABA content through overexpression of NCED

We have shown that transcriptional activation of an LeNCED1 transgene, using a Tc-inducible system, results in a large increase in bulk leaf ABA in tobacco. Since we demonstrated that Tc does not lead to ABA accumulation in the absence of the inducible transgene, this experiment demonstrates that it is the induction of the NCED mRNA that causes the increase in leaf ABA content. This is strong evidence that NCED is indeed a key rate-limiting enzyme in leaf ABA biosynthesis. The increase in NCED mRNA during water deficit-induced ABA biosynthesis in both leaves and roots ( Burbidge et al. 1997a ; Neill et al. 1998 ; Tan et al. 1997 ; Thompson et al. 2000 ) further supports the idea that transcriptional regulation of NCED is a major control mechanism for ABA biosynthesis. Further, the observation that ABA accumulation follows transcriptional activation of a single gene ( Fig. 2) indicates that the normal increase in the endogenous NCED mRNA observed upon water deficit is sufficient to produce the subsequent accumulation of ABA, although regulation of other enzymes in the pathway may also occur.

ZEP mRNA does not accumulate in dehydrating leaves ( Burbidge et al. 1997b ), but it is induced in dehydrating roots where there are comparatively low levels of epoxycarotenoids ( Audran et al. 1998 ; Thompson et al. 2000 ). Although constitutive overexpression of ZEP in N. plumbaginifolia resulted in extended seed dormancy and ABA accumulation in seeds, effects on vegetative tissues were not reported ( Frey et al. 1999 ). Because of the large epoxycarotenoid pools in leaves, it is unlikely that overexpression of ZEP would result in increased ABA biosynthesis in this tissue.

Since we measured ABA content only after a 24 h Tc treatment in tobacco, it remained possible that in the longer term the ABA concentration might return to basal levels, despite the high-level expression of LeNCED1. It is known that ABA stimulates its own oxidation to phaseic acid, presumably because of increased activity of abscisic acid 8′ hydroxylase, in both seed tissues ( Babiano, 1995; Uknes & Ho, 1984) and suspension-culture cells ( Cutler et al. 1997 ; Windsor & Zeevaart, 1997). Based on these reports, it has been suggested ( Zeevaart, 1999) that attempts to engineer increased ABA biosynthesis in transgenic plants would be balanced by increased ABA catabolism, and may not cause additional ABA accumulation. However, we have demonstrated that long-term constitutive expression of the LeNCED1 transgene resulted in a heritable phenotype in three independent tomato transformants, D9, SP5 and SP6, that could be explained by overproduction of ABA. We further observed a substantial and significant increase in bulk leaf ABA content in progeny of one transformant (D9; Table 1), and a significant threefold increase in seed ABA content in progeny of two transformants (SP5 and D9; Fig. 8). Transformant SP5 was notable because, although progeny showed elevated ABA in imbibed seeds, increased seed dormancy, elevated leaf LeNCED1 mRNA and reduction in stomatal conductance, there was no increase in bulk leaf ABA. It is possible that there was an increase in ABA biosynthesis that was balanced by increased catabolism in the leaves; however the resulting increased flux of ABA in the leaf, or a possible change in compartmentation of the ABA, may still have stimulated stomatal closure. Alternatively, in SP5 there may be no increase in ABA biosynthesis in the leaves, and the stomatal closure may result from additional root-sourced ABA ( Davies et al. 1994 ). In D9, where leaf NCED mRNA was detected at a higher level than in SP5 ( Table 1), and where an increase in ABA content was observed, the rate of biosynthesis may have exceeded the rate of catabolism. Thus the catabolic regulatory mechanisms known to be induced by high ABA concentrations may have had a role in limiting ABA accumulation in plants overexpressing LeNCED1, but they were not always sufficient to prevent either detectable ABA accumulation in leaves or seeds, or phenotypic responses in the plant.


We have provided the first direct evidence that NCED is a regulatory enzyme in ABA biosynthesis in seeds and leaves. We have shown that constitutive high-level expression of an NCED transgene can lead to accumulation of ABA, despite the fact that increased ABA biosynthesis is known to stimulate ABA degradation ( Zeevaart, 1999). We have identified NCED as a suitable target for the engineering of ABA-controlled phenotypes such as seed dormancy. Finally, we have described a system that allows manipulation of ABA biosynthesis and ABA accumulation by application of the chemical inducer, tetracycline. This will provide an alternative approach, other than exogenous ABA application, to studying the physiological role of ABA, and may provide additional information about how ABA signals are generated, transmitted and perceived.

Experimental procedures

Plant material

Lycopersicon esculentum Mill. L. cv. Ailsa Craig was used as wild-type cultivar, and as the host for tomato Agrobacterium-mediated transformations. Nicotiana tabacum L. var. Wisconsin 38, transformed with plasmid pTet1 and so containing the tetR transgene, was described previously ( Gatz et al. 1991 ) and was obtained from C. Gatz, University of Bielefeld, Germany.

DNA constructs

The coding region of LeNCED1 was amplified from a genomic DNA clone using the polymerase chain reaction (PCR) with primers 5′-cgTCTAGAggtagctATGgcaactactac-3′ (forward primer) and 5′-tGGATCCTGCAGtaatagTCAcaatctagcctg-3′ (reverse). The start codon and the complement of the stop codon are indicated in capitals, as are the XbaI, BamHI and PstI sites used in cloning. The PCR product was cloned into pGEM3Zf(+) (Promega, Southampton, UK) using the XbaI and BamHI sites to give pNCEaa.3. The coding region within pNCEaa.3 was sequenced using an automated sequencer (ABI) to confirm that no sequence errors were introduced by the PCR.

The KpnI–SphI fragment from pJIT60 ( Guerineau & Mullineaux, 1993), containing a doubly enhanced CaMV 35S promoter and CaMV 35S 3′ untranslated region, was cloned into the KpnI and SphI sites in pUCAP ( van Engelen et al. 1995 ). The resulting plasmid was named pUCJIT and contained PacI and AscI sites to allow transfer of subsequent constructions into pBINPLUS ( van Engelen et al. 1995 ). The LeNCED1 coding region was excised from pNCEaa.3 with PstI, and cloned into the PstI site of pUCJIT to give pD35-NCE. The entire expression cassette was excised from pD35-NCE with PacI and AscI and cloned into pBINPLUS to produce pBP-D35B-NCED ( Fig. 1).

The doubly enhanced CaMV 35S promoter was removed and additional cloning sites were introduced into pUCJIT by excising a PacI–EcoRI fragment and ligating an annealed oligonucleotide pair into the remaining vector. The resulting plasmid was named pUCJIT-OL. Plasmid E1068 containing the previously described ( Ni et al. 1995 ) chimaeric promoter, the Super Promoter (SP), was obtained from S. Gelvin, Purdue University, USA. SP consists of a trimer of the octopine synthase upstream activator element linked to the mannopine synthase promoter. An E1068 SalI–XbaI fragment containing SP was cloned into the XhoI and XbaI sites of pUCJIT-OL to give pUCJIT-OL-SP. The LeNCED1 coding region from pNCEaa.3 was excised with XbaI and PstI and cloned into these same sites in pUCJIT-OL-SP to give pNCE-SP. The resulting expression cassette was transferred to pBINPLUS using the PacI and AscI sites to give pBP-SP-NCED ( Fig. 1).

The pBinHygTX vector is a binary plant transformation vector described by Gatz et al. (1992) containing the Triple-Op promoter upstream of a multiple cloning site, and was obtained from C. Gatz, University of Bielefeld, Germany. In order to introduce a SalI site required for cloning into the pBinHygTX multiple-cloning site, the LeNCED1 coding region was excised from pNCEaa.3 using XbaI and EcoRI and cloned into pBluescript II KS (+/–) using the same two sites. The resulting plasmid was named pNCE-blue. The coding region was then excised from pNCE-blue using XbaI and SalI and cloned into these same two sites in pBinHygTX. This final construct was named pBHT-NCED ( Fig. 1).

Plant transformation

Foreign DNA was transferred to L. esculentum by Agrobacterium-mediated transformation according to the method of Bird et al. (1988) . Leaf discs from tobacco plants previously transformed with tetR were transformed with pBHT-NCED according to the method of Horsch et al. (1985) , but using 40 mg l−1 hygromycin as selective agent. Both tomato and tobacco methods utilized Agrobacterium tumefaciens strain LBA4404.

Tetracycline induction treatments

Tobacco plants were propagated by rooting in 2MS agar medium and grown on until they reached approximately 10–15 cm in height. The fifth, sixth and seventh leaves were taken, counting from the first visible leaf at the apical meristem, and the mid-ribs were excised with a scalpel. For each leaf the two remaining pieces of leaf lamella, from either side of the mid-rib, were used as paired samples for treatment plus or minus tetracycline. For each transformant, three leaf explants, each from a different leaf, were incubated in 30 ml artificial sap, pH 6.0 ( Wilkinson & Davies, 1997) with 1 mg l−1 tetracycline. The other three leaf explants, from the same three leaves, were incubated in the artificial sap without the addition of tetracycline. All incubations were in 9 cm Petri dishes, with three leaf explants per dish, adaxial side uppermost, for 24 h at 22°C with a 16 : 8 light : dark cycle and illumination at 70 μmol m−2 sec−1. At the end of the incubation, tissue was frozen in liquid N and ground to a fine powder under liquid N before removing approximately 0.5 and 0.2 g for extraction of total RNA and ABA analysis, respectively.

Northern blotting and hybridization

Total RNA was extracted from leaf samples and Northern blots were prepared as described by Thompson & Corlett 1995) . Blots were prehybridized in 50% formamide, 2 × SSPE, 0.5% SDS, 5 × Denhartd's solution, 50 μg ml−1 denatured herring sperm DNA, 50 μg ml−1 yeast tRNA at 53°C for 4 h. RNA probes were purified by Sephadex G-50 gel-filtration spun columns, boiled for 1 min, cooled on ice and then added to the prehybridization solution. Hybridization was for 16–48 h at 53°C. All blots were washed twice in 2 × SSPE, 0.5% SDS, and twice in 0.1 × SSPE, 0.5% SDS, at 60°C. Quantification of mRNA was as described by Corlett et al. (1998 ).

Synthesis of antisense RNA probes

The RNA probe for the coding region of LeNCED1 was transcribed from plasmid pNCEaa.3 (described above) linearized with XbaI. The antisense probe representing the 3′ untranslated region of LeNCED1 was transcribed from plasmid pNCED1.7 linearized with EcoRI. To create plasmid pNCED1.7, a 770-nucleotide 3′ region of LeNCED1 was amplified by PCR from a sequenced genomic clone using primers 5′-cgaattcatgattgTGActattactgag-3′ and 5′-ggtaagcttaacaaacatcatcgagtgtc-3′. The LeNCED1 stop codon is capitalized in the upstream primer. The resulting PCR product contains 265 bases between this stop codon and the polyadenylation site of the cDNA described by Burbidge et al. (1999) , and a further 505 bases downstream of the polyadenylation site. The PCR product was cloned into the EcoRI and HindIII sites of vector pDP18 (Ambion Inc., TX, USA). RNA probes were labelled by transcription with T7 RNA polymerase in the presence of α-[32P]UTP, as described by Melton et al. (1984) .

Southern blotting and hybridisation

Genomic DNA was prepared from 0.2–0.3 g expanding tomato leaves using the method described by Burbidge et al. (1995) , but scaled down to allow extraction in 1.5 ml microcentrifuge tubes. Tobacco genomic DNA was prepared as described by Dellaporta et al. (1983) . DNA (5–10 μg) was restricted with the indicated restriction enzymes and then agarose electrophoresis, Southern blotting and hybridization to randomly primed DNA probes were carried out as previously described ( Corlett et al. 1996 ). Final washing of blots was at 60°C in 0.1 × SSPE, 0.5% SDS.

Sampling of tomato leaf tissue for mRNA and ABA analysis

Tomato plants were grown in a glasshouse until six to 10 fully expanded leaves were produced. The three terminal leaflets of the fully expanded leaf nearest to the apical meristem were removed and immediately frozen in liquid N, and then stored at −70°C. Leaflets were ground to a powder in a pestle and mortar under liquid N, then samples were removed to precooled vials for both extraction of total RNA and ABA radioimmunoassay.

Determination of ABA content by radioimmunoassay and GC–MS

Powdered leaf samples (see above) were extracted overnight in deionized water at 4°C, using 0.5 ml water per 100 mg tissue. 50 μl samples, at appropriate dilutions in water, were assayed using the monoclonal antibody AFRC MAC252 as described by Quarrie et al. (1988) . The assay was previously validated for tomato leaf tissue by GC–MS ( Mulholland, 1994).

Seeds were imbibed on water-soaked filter paper in the dark at 25°C. After 48 h (prior to radicle emergence), samples of 50 seeds were blotted dry, weighed and frozen in liquid N. Each sample was homogenized in 7.5 ml cold (4°C) 80% methanol containing 20 mg l−1 butylated hydroxy toluene, then 25 ng [2H6]ABA ( Rivier et al. 1977 ) was added before shaking overnight at 4°C. Samples were then centrifuged at 3000 g for 10 min at 4°C and the supernatant was recovered. The pellet was resuspended in 10 ml 80% methanol, shaken for a further 4 h, then recentrifuged. The supernatants were combined and the methanol was removed under vacuum at 30°C. An equal volume of 0.5 m potassium phosphate buffer pH 3.5 was then added to the aqueous residue, and the samples were partitioned against diethyl ether (3 × 10 ml). Water (2 ml) was added to the combined organic phases. The ether was removed under vacuum and the aqueous residue adjusted to pH 8.0 with potassium hydroxide. This extract was loaded onto a column (5 ml bed volume) of QAE Sephadex-A25 resin (formate form) then washed with 10 ml water. ABA was eluted from the column in 0.2 m formic acid and loaded directly onto a C18 Sep-Pak cartridge (Waters Corporation, Milford, MA, USA). The cartridge was washed with 2 ml 40% methanol and ABA recovered from the cartridge with 5 ml 70% ethanol. The sample was evaporated to dryness, redissolved in 100 μl methanol, then methylated with excess ethereal diazomethane. Samples were transferred to vials, again evaporated to dryness, and redissolved in 10 μl dry pyridine before quantitative GC–MS analysis. Samples for qualitative GC–MS analysis were purified as above, but [2H6]ABA was omitted.

Methyl-ABA was analysed using a VG TRIO-1 GC–MS system (ThermoQuest, Manchester, UK). The BP1 (SGE (Europe) Ltd, Milton Keynes, UK) capillary column (25 m × 0.25 mm internal diameter × 0.25 μ δf) was coupled directly to the ion source with an interface temperature of 275°C, and helium carrier gas was supplied under electronic pressure control to maintain a linear velocity of 35 cm sec−1. The MS source temperature was 200°C and the electron energy 70 eV. The injector was used in the splitless mode at a temperature of 250°C. After injection of Me-ABA samples (1 μl), the GC oven was maintained at 60°C for 1 min with the splitter closed. The splitter (50 : 1) was then opened and 30 sec later the oven temperature increased at 25°C min−1 to 215°C, and then at 5°C min−1 to 250°C. For quantification by selected ion monitoring, the VG lab- base data system software was used to monitor responses to ions of m/z 162 and 190 for Me-ABA and 166 and 194 for Me-[2H6]ABA. Responses were integrated and the amounts of endogenous ABA computed using the VG lab- base software from the ratio of 190/194, using calibrations relating these ratios to the appropriate molar ratios. Full scan-mass spectra were obtained using the same GC conditions, and the spectra were acquired after 8 min scanning from 50 to 350 amu at 0.9 sec mass decade−1.

Stomatal conductance

Stomatal conductance was determined using a portable IRGA and a Parkinson leaf cuvette with a 2.5 cm2 chamber (CIRAS 1, PP Systems, Hitchin, UK). Wild-type and homozygous SP5 plants were grown in a controlled-environment cabinet with a 14 : 10 h light : dark cycle (400 μmol m−2 sec−1 photosynthetically active radiation), 25/20°C day/night temperature and 65% relative humidity. For five plants of each genotype, stomatal conductance was determined for the youngest fully expanded leaf, between 5 and 6 h into the light period at a saturating irradiance of 1500 μmol m−2 sec−1. In a separate experiment, wild-type and homozygous D9 plants were grown with a 12 : 12 h light : dark cycle (170 μmol m−2 sec−1 photosynthetically active radiation) at 23/18°C day/night temperature and ambient humidity. Stomatal conductance was measured for both the youngest and next-youngest fully expanded leaves for 10 plants of each genotype at ambient cabinet irradiation (170 μmol m−2 sec−1).

Statistical analysis

In Table 1, mRNA data were obtained as the percentage of the maximum value observed. As this results in a ratio scale, data were log-transformed prior to applying a t-test. This also improved the homogeneity of variance. The means were determined by back-transformation of the means of logs. For the ABA and stomatal conductance data, t-tests were performed without log transformation.

Seed germination assays

Tomato seeds were recovered from red-ripe fruit and incubated in 120 m m hydrochloric acid plus 1 g l−1 pectinase (Danisco Ingredients, Bury St Edmunds, UK) for 16 h, washed and dried at ambient humidity on the laboratory bench, then stored at room temperature. Prior to germination seeds were surface-sterilized in 10% household bleach for 30 min and washed in sterile water, then germination was performed by one of two protocols. (i) Seeds were placed onto 3MSB5K medium with or without norflurazon at 5 mg l−1, at a density of 50 seeds per 9-cm-diameter Petri dish. 3MSB5K medium contained Murashige and Skoog micro- and macro-elements with Gamborg B5 Vitamins (Duchefa, Haarlem, The Netherlands) plus 30 g l−1 sucrose, 8 g l−1 agar and 50 mg l−1 kanamycin. Dishes were incubated at 22°C with a 16 : 8 h light : dark cycle, illuminated at 70 μmol m−2 sec−1 with OSRAM cool white fluorescent tubes. To allow recovery of seedlings germinated on norflurazon, germinated seed were transferred to 3MSB5K medium without norflurazon immediately after radicle emergence. After this bleached cotyledons developed, but the first or second true leaf was green. (ii) Seeds were placed, one seed per well, into 25-well Petri dishes with each well containing a square of filter paper soaked in sterile water plus or minus 3 mg l−1 norflurazon. Petri dishes were incubated at 25°C in the dark. To recover viable seedlings from the norflurazon-treated seeds, it was again necessary to transfer germinated seeds to 3MSB5K media, but in the case of SP5 no kanamycin was added.


This work was supported by a BBSRC Competitive Strategic Grant to HRI. A.B. was supported by EC framework IV Contract Bio4CT960443. R.C.S. was funded by a joint studentship between The University of Nottingham, HRI and industrial partners. We thank Annie Marion-Poll for release of proofs prior to publication.