Transactivation of the Brassica napus napin promoter by ABI3 requires interaction of the conserved B2 and B3 domains of ABI3 with different cis-elements: B2 mediates activation through an ABRE, whereas B3 interacts with an RY/G-box
Department of Plant Biology, Swedish University of Agricultural Sciences, Box 7055, S-750 07 Uppsala, Sweden, and
*For correspondence (fax +46 18 4714975; e-mail email@example.com). †Present address: Plant Gene Expression Center, 800 Buchanan St, Albany, CA 94710, USA.
The transcriptional activator ABI3 is a key regulator of gene expression during embryo maturation in crucifers. In monocots, the related VP1 protein regulates the Em promoter synergistically with abscisic acid (ABA). We identified cis-elements in the Brassica napus napin napA promoter mediating regulation by ABI3 and ABA, by analyzing substitution mutation constructs of napA in transgenic tobacco plantlets ectopically expressing ABI3. In transient analysis using particle bombardment of tobacco leaf sections, a tetramer of the distB ABRE (abscisic acid-responsive element) mediated transactivation by ABI3 and ABI3-dependent response to ABA, whereas a tetramer of the composite RY/G complex, containing RY repeats and a G-box, mediated only ABA-independent transactivation by ABI3. Deletion of the conserved B2 and B3 domains of ABI3 abolished transactivation of napA by ABI3. The two domains of ABI3 interact with different cis-elements: B2 is necessary for ABA-independent and ABA-dependent activations through the distB ABRE, whereas B3 interacts with the RY/G complex. Thus B2 mediates the interaction of ABI3 with the protein complex at the ABRE. The regulation of napA by ABI3 differs from Em regulation by VP1, in that the B3 domain of ABI3 is essential for the ABA-dependent regulation of napA.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The transcriptional activator ABI3 is a major regulator of seed maturation in Arabidopsis thaliana. Seeds of abi3 mutants accumulate reduced amounts of storage proteins, fail to degrade chlorophyll, remain non-dormant, and do not acquire desiccation tolerance ( Nambara et al., 1994 ; Ooms et al., 1993 ). Also, the abi3-4 mutant displays impaired capacity to express various mRNAs regulated by different temporal programs during Arabidopsis seed development ( Parcy et al., 1994 ), several of which are also responsive to abscisic acid (ABA). ABI3 is specifically expressed in seeds, and ectopic expression of ABI3 results in the accumulation of seed-specific mRNAs in transgenic plantlets upon treatment with ABA ( Parcy et al., 1994 ).
Mutants of the related Vp1 gene in maize germinate precociously and fail to accumulate anthocyanins in the embryo and aleurone layer of the maize seed (reviewed by McCarty, 1995). The VP1 protein activates the expression of maturation-specific genes such as the wheat Em gene ( Vasil et al., 1995 ) and the C1 gene regulating anthocyanin biosynthesis ( Hattori et al., 1992 ). Detailed analysis of promoter constructs using transient expression assays identified two elements crucial for transactivation by VP1. One is the sequence CATGCATG, known as the Sph element or RY-repeat, which is essential for the ABA-independent transactivation of the C1 promoter by VP1 ( Hattori et al., 1992 ). The other is the ABA-responsive element (ABRE) Em1a, which mediates synergistic regulation by ABA and VP1 of the wheat Em gene ( Vasil et al., 1995 ).
Giraudat et al. (1992) identified three highly conserved basic domains in the ABI3 and VP1 proteins, which were designated B1, B2 and B3 in order from the amino terminus ( Suzuki et al., 1997 ). A peptide consisting only of the B2 domain of VP1 enhances in vitro the DNA-binding activities of diverse basic leucine zipper (bZIP) type DNA-binding proteins, including plant G-box binding factors ( Hill et al., 1996 ). However, in vivo footprinting studies showed strong protein binding to the ABREs of the rab28 promoter independent of VP1 ( Busk and Pages, 1997) , suggesting that VP1 action does not involve changes in protein–DNA interactions. Hill et al. (1996) also showed that deletion of the B2 domain from VP1 abolished transactivation of the Em promoter. Yet, because B2 contains a putative nuclear localization signal ( Giraudat et al., 1992 ), it is possible that the deleted protein was not targeted to the nucleus, rather than that the B2 domain is necessary for Em regulation. A comparative study of abi3 alleles suggests that the B2 domain is involved in the regulation of a LEA (late embryogenesis-abundant) gene, AtEm6, and two albumins, At2S1 and At2S2 ( Bies-Etheve et al., 1999 ). However, the exact role of the B2 domain has yet to be established. The function of the B3 domain has been extensively characterized by genetic ( McCarty et al., 1989 ) and molecular approaches ( Carson et al., 1997 ; Suzuki et al., 1997 ). Mutations of Vp1 disrupt the B3 domain block expression of the Sph-controlled C1 gene, but do not prevent seed maturation ( McCarty et al., 1989 ). In transient expression assays, mutant VP1 proteins with a truncated B3 domain failed to transactivate both the C1 and Em promoters in the absence of ABA. Additionally, these mutants retained the capacity to enhance synergistically the ABA regulation of the Em promoter ( Carson et al., 1997 ). The isolated B3 peptide binds the Sph/RY element ( Suzuki et al., 1997 ), supporting the notion that B3 is essential for ABA-independent activation mediated through Sph/RY.
Napin 2S albumins accumulate during maturation in crucifer seeds, and their expression is regulated by ABI3 ( Parcy et al., 1994 ) and ABA ( DeLisle and Crouch, 1989 ; Jiang et al., 1996 ). In addition, upon challenge with ABA, napin mRNA is produced in transgenic A. thaliana plantlets ectopically expressing ABI3 ( Parcy et al., 1994 ). Previous studies of the Brassica napus napin napA promoter (−309 to +45) in seeds of transgenic tobacco ( Ellerström et al., 1996 ; Ezcurra et al., 1999 ; Stålberg et al., 1993 ) and rapeseed ( Stålberg et al., 1996 ) identified two major element clusters essential for function: the RY/G complex and the B-box. In the RY/G complex, a G-box interacts synergistically with two adjacent RY-repeats ( Ezcurra et al., 1999 ). The B-box consists of an Em1a-like element, named distB, and an adjacent CA-rich element, named proxB. A tetramer of the B-box mediates strong seed-specific activity and renders moderate ABA responsiveness in vegetative tissues to a construct containing the 35S enhancer ( Ezcurra et al., 1999 ). However, the B-box tetramer alone is not induced by ABA in vegetative tissues, indicating developmental control of the hormone response.
Functional studies of VP1/ABI3 have mainly addressed the regulation by VP1 of the Em promoter in monocots ( Carson et al., 1997 ; Hattori et al., 1995 ; Vasil et al., 1995 ). Here we examine the regulation of the napin napA promoter by ABI3 and ABA. Our study contributes to our understanding of the ABI3-regulation of a gene with an expression pattern temporally different from Em, because napin is accumulated during early and mid-maturation ( Höglund et al., 1991 ), whereas Em is an LEA-type protein. We show that both the B2 and B3 domains of ABI3 are necessary for napin expression during embryo development: B2 is essential for activation mediated by the distB ABRE, whereas B3 is required for regulation through the RY/G complex. Thus the B2 domain appears to be involved in tethering ABI3 to the complex at the ABRE, and may mediate protein–protein interactions. In addition, the regulation of napA by ABI3 is different to the regulation of Em by VP1, because it requires the B3 domain of ABI3. A tetramer of the distB ABRE mediates seed-specific, ABI3-dependent response to ABA, and possible mechanisms of tissue- or development-specific hormone response are discussed.
We have previously analyzed a series of napA promoter constructs fused to the GUS reporter gene in transgenic tobacco ( Ezcurra et al., 1999 ). The observation that napin genes are induced by ABA in transgenic A. thaliana plantlets ectopically expressing ABI3 ( Parcy et al., 1994 ) prompted us to design a transgenic approach to study the regulation of our napA promoter constructs by ABI3 and ABA. Transgenic tobacco plants harboring the A. thaliana ABI3 cDNA driven by the 35S promoter were produced. A series of napA-GUS × 35S–ABI3 genotypes were then generated by crossing a T0 plant ectopically expressing ABI3 in leaves with five different T0 plants harboring each of the napA promoter constructs (for details see Experimental procedures). T1 plantlets derived from crossings were treated with 50 µm ABA and the GUS activity was subsequently analyzed.
Elements in the B-box and RY/G are essential for ABI3 transactivation
Figure 1 shows the structure of the napA−309 promoter and the elements targeted by substitution mutagenesis. The ABA response in T1 plantlets of the different mutant genotypes is shown in Figures 2 and 3. The −309 napA promoter lacks activity and is not induced by ABA, which is expected, since napin genes are strictly seed-specific. However, −309 plantlets ectopically expressing ABI3 showed significant amounts of GUS activity. In addition, an eightfold further increase in activity was obtained on treatment with 50 µm ABA. These results are consistent with the results of Parcy et al. (1994) with transgenic A. thaliana plantlets, demonstrating that tobacco tissues are a competent system for studying the activation of napA by ABI3 in the presence or absence of ABA treatment.
In addition, the substitution mutation analysis showed that both the ABA-dependent and ABA-independent activations of the napA promoter by ectopically expressed ABI3 requires that both the B-box and RY/G are intact. This observation is consistent with earlier studies in transgenic tobacco seeds ( Ezcurra et al., 1999 ), which showed strong synergism between elements within each complex and even between the complexes. In tobacco seeds the effect of mutations was even more drastic, as most mutants lacked activity. The observation that mutation of the proxB element produced a weaker decrease in GUS activity than all other mutations is also consistent with earlier results in seeds.
The tetrameric B-box mediates ABI3-dependent response to ABA
We showed that the construct 4xB containing a tetramer of the B-box fused to the minimal (−46) CaMV 35S promoter drives strong seed-specific expression and confers moderate (2×) ABA responsiveness to the CaMV 35S enhancer in vegetative tissues ( Ezcurra et al., 1999 ). In Figure 4, we show that the tetrameric B-box is activated by ectopically expressed ABI3 and exogenous ABA in a manner similar to the activation obtained with the −309 promoter, indicating that the B-box tetramer is an ABI3-dependent ABRE. In addition, the B-box even mediates ABI3 activation in the absence of exogenous ABA.
Transient analysis of ABI3 deletion mutants
To further characterize the regulation of the napA promoter by ABI3 and ABA, a series of reporter and effector constructs were generated and analyzed in a transient expression system using particle bombardment of leaf sections. Two main issues were addressed in this investigation. First, we wanted to further analyze the response of the RY/G composite element to ABA and ABI3. The mutational analysis in Figure 2 shows that, in the native context, none of the composite elements can function alone. However, the B-box in a multimeric form can compensate for the absence of RY/G, revealing its function as an ABRE ( Figure 4). Thus, to examine the functions mediated by the RY/G complex without the influence of the B-box, a tetramer of this complex was fused to the CaMV 35S minimal promoter as shown in Figure 6a. Secondly, we wanted to investigate the role of the B2 and B3 domains of ABI3 in the ABA-dependent and ABA-independent transactivation by ABI3. For this purpose, two new effector constructs were generated containing deletion mutants of ABI3 driven by the 35S promoter, the structure of which is shown in Figure 5. Since there is a putative nuclear localization signal (NLS) located in the middle of the B2 domain, only the portion upstream of the NLS sequence (18 amino acids) was deleted in the ΔB2 construct. In the ΔB3 construct, most of the B3 domain was removed. Figure 6b shows the response of the −309 promoter and the 4xB and 4xRY/G constructs to ABA and the different ABI3 effectors. Consistent with the results in Figure 4, the wild-type ABI3 effector and ABA synergistically activated the −309 promoter and the 4xB reporter. This shows that the particle bombardment system efficiently reproduces the effects observed in transgenic plantlets. The RY/G tetramer mediated ABA-independent ABI3 activation as effectively as the −309 and 4xB constructs. However, the activation produced by ABA and ABI3 in combination was substantially lower for the 4xRY/G construct.
Deletion of the B2 and B3 domains severely impaired the capacity of ABI3 to activate the napA promoter. Moreover, the two ABI3 deletion constructs showed complementary functions with respect to the activation of 4xB and 4xRY/G. Thus the ΔB3 construct activated 4xB but not 4xRY/G, whereas the ΔB2 construct produced the opposite effect. Detailed examination of the ABA-dependent and ABA-independent activities of the two ABI3 deletions revealed additional features of the functions performed by the B2 and B3 domains. For example, deletion of the B3 domain reduced half of the ABA-independent activation of 4xB, in addition to abolishing activation of 4xRY/G. Moreover, deletion of the Β2 domain completely abolished the ABA-independent activity of 4xB, as well as severely reducing its ABA-dependent activation. However, the activity obtained with ΔB2 upon ABA treatment was fivefold higher than the activity obtained with ΔB2 alone, which is comparable to the increase mediated by wild-type ABI3 upon ABA treatment (fourfold increase). Finally, ΔB2 impaired to some extent the low ABA-dependent activation of 4xRY/G.
In conclusion, deletion of the B3 domain abolished activation through the RY/G complex and reduced the ABA-independent activation through the B-box, whereas deletion of the B2 domain abolished the ABA-independent activation through the B-box and reduced the ABA-dependent activation through RY/G.
The distB tetramer mediates ABI3-dependent response to ABA
Figures 4 and 6 show that the tetrameric B-box is not induced by ABA alone in vegetative tissues. This is surprising, since the distB element is highly homologous to Em1a-type ABREs which, in similar tetramer constructs, mediate ABA induction in the absence of VP1 ( Ono et al., 1996 ; Vasil et al., 1995 ). We thus deduced that the B-box must contain additional information that specifies proper developmental control of its inherent ABA responsiveness.
The developmental specificity in the ABA response of the B-box tetramer could be obtained by different mechanisms. For example, the proxB element might be a coupling element mediating the developmental-specificity of the distB ABRE. Alternatively, the developmental specificity could be specified by features in the ABRE, like its sequence, orientation, or both. The distB element differs from Em1a by one base pair mismatch at the ACGT core ( Table 1), and in that it is inversely oriented with respect to the TATA-box. However, the motif III ABRE in the rab16B promoter also lacks an ACGT core, and still mediates ABI3-independent ABA response ( Ono et al., 1996 ), indicating that the lack of an ACGT core in distB is probably not accountable for its developmental specificity. Thus, to investigate the remaining alternatives we generated three tetrameric B-box reporters, the structures of which are shown in Figure 7a. To test the function of distB without the influence of proxB, a tetramer of the distB element was fused to the CaMV 35S −46 promoter, generating the reporter construct 4xdistB. To test if the orientation of the B-box elements affected their response to ABA or to the ABI3 effectors, the B-box and distB tetramers from 4xB and 4xdistB were cloned in the opposite orientation upstream of the CaMV 35S −46 reporter, thus generating the constructs 4xB-R and 4xdistB-R. The activity of these reporters and their response to ABA and the different ABI3 effectors is shown in Figure 7b. None of the analyzed reporters was induced by exogenous ABA in the absence of ABI3, showing that the developmental specificity of the B-box tetramer is not mediated by the proxB element, nor by the orientation of the ABRE, nor by a combination of both. In addition, deletion of the proxB element in the 4xdistB reporter did not affect the overall activation by ABI3 and ABA of the distB element, showing that distB is the true ABRE. Since we have already shown the involvement of the proxB element during the transactivation of napA by ABI3 and ABA ( Figure 2), we conclude that multimerization of distB compensates for the absence of proxB. Moreover, reversion of the orientation severely affected the overall function of the 4xdistB-R, but not of 4xB-R, showing that, in the reverse orientation, proxB is crucial for transactivation by ABI3 and ABA. This result indicates that the complex at the B-box can promote transcription bidirectionally when fully assembled, and that this functional symmetry is impaired when the proxB element is removed. In addition, deletion of the B2 domain of ABI3 produces loss of activity in the B-box and distB tetramers in both orientations, showing that B2 is involved in the transactivation mediated by the distB ABRE.
Finally, even the reversed B-box tetramer displayed reduced ABA-independent transactivation by ΔB3, confirming our results in Figure 6. Remarkably, the distB tetramer, which lacks the CA-rich proxB element, does not display reduced ABA-independent transactivation by ΔB3, suggesting that the B3 domain interacts with the proxB element. In summary, the distB tetramer functions as an ABI3-dependent ABRE, and the ABI3 complex at the B-box mediates bidirectional transcription activation.
Our results indicate that two composite elements in the napA promoter, the B-box and RY/G, functionally co-operate during ABA-dependent and ABA-independent transactivation by ABI3. In addition, the B2 and B3 domains of ABI3 are involved in transactivation: B2 mediates activation through the B-box, whereas B3 interacts with RY/G. For our studies, we used promoter constructs containing tetramers of napA elements. Multimerization of cis-elements provides a unique way of studying the activity of isolated elements, and is widely used in transcription studies, even though it might generate problems such as artifactual protein–protein interactions. We used both a transgenic and a particle bombardment approach to analyze the ABI3-transactivation of different napA constructs. Our results were reproduced in both systems, and are consistent with previous results in seeds of transgenic tobacco ( Ezcurra et al., 1999 ), showing the biological significance of our data.
Role of the B-box and the RY/G complex
Our data shows that in the RY/G cluster, two RY-repeats interact with a perfect G-box to provide ABA-independent transactivation by ABI3. In addition, the tetrameric RY/G construct displayed a low (twofold) increase in activity upon treatment with ABA ( Figure 6), suggesting that the RY/G complex is moderately responsive to ABA. However, the relative weakness of the response indicates that, in the napA promoter, the main ABRE is within the B-box.
We have earlier shown that the B-box tetramer is a seed-specific ABRE ( Ezcurra et al., 1999 ). Our present analysis provides a molecular basis to this function by demonstrating that ABI3 mediates the ABA responsiveness of the B-box. Moreover, ABI3 activates the B-box even in the absence of ABA. In the composite B-box, the distB element mediates seed specificity by functioning as an ABI3-dependent ABRE, and by providing ABI3 regulation even in the absence of added ABA. In addition, the adjacent proxB element is also involved in napA regulation, because mutation in the native context severely impairs the activation by ABI3 in the presence and absence of ABA treatment ( Figure 2).
Our results showing that the distB tetramer does not respond to ABA in the absence of ABI3 are in contrast with results by others ( Ono et al., 1996 ; Skriver et al., 1991 ; Vasil et al., 1995 ), showing that tetramers of Em1a-type ABREs are responsive to ABA independently of ABI3. However, the type of cell lines used in those studies express Vp1 ( Hollung et al., 1997 ; Nakagawa et al., 1996 ), and thus it is probable that the reported ABI3-independent ABA responses actually reflect endogenous VP1 activity. Another Em1a-type ABRE displaying ABI3-independent response to ABA is Hex3 ( Table 1). Hex3 was reported to mediate seed specificity and ABA response in vegetative tissues of transgenic tobacco ( Lam and Chua, 1991). However, the response was measured in only one transgenic plant, and it was observed when the Hex3 tetramer was fused to the −46 35S minimal promoter, but not when it was fused to the −90 35S truncated promoter. Moreover, the response appeared weak as judged by an S1 protection assay, and it markedly declined after 2 h of ABA treatment, as opposed to Rab 21, which still increased after 12 h treatment ( Mundy and Chua, 1988). In contrast, it was recently reported that a tetramer of an Em1a-type G-box, Box1 ( Table 1), mediates strict seed-specific expression, but is not responsive to ABA in seedlings of transgenic tobacco ( Ishige et al., 1999 ). This suggests that Box1 functions as yet another ABI3-dependent ABRE. Thus, because all results showing VP1-independent response to ABA of Em1a-type tetramers are, in one way or another, questionable, we conclude that tetramers of Em1a-type ABREs are ABI3/VP1-dependent, and, in consequence, seed-specific. It has been established that the promoter unit necessary and sufficient to mediate the ABA response consists of an ABRE and a closely linked sequence, called a coupling element ( Shen et al., 1996 ). Further, a coupling element may be replaced by another ABRE ( Hobo et al., 1999a ; Ono et al., 1996 ), and it was suggested that coupling elements might also mediate tissue specificity to the hormone response ( Rogers and Rogers, 1992). Then, in the tetrameric construct, the additional ABREs might mediate ABI3-dependence to the ABA response. Further, in the context of their corresponding promoters, the ABREs interact with nearby coupling elements and mediate, accordingly, ABI3-dependent (as in napA) or ABI3-independent (as in rab genes) response to ABA (reviewed by Singh, 1998). Interestingly, in the context of napA, the distB ABRE interacts with proxB and the RY/G complex, and mediates the ABI3-dependent ABA response. However, when fused to the 35S enhancer the B-box tetramer mediates ABA response in vegetative tissues ( Ezcurra et al., 1999 ).
Functions of the B2 and B3 domains of ABI3
We analyzed the transactivation of napA constructs by two ABI3 deletion mutants. Because both ABI3 deletion mutants were transcriptionally active, we feel confident that the resulting losses in activity truly reflect specific interactions between ABI3 domains and napA elements, rather than a total inactivation of the mutated effector. Our results show that the B2 domain of ABI3 is crucial for the activation of the distB ABRE in the absence of ABA treatment. In addition, the B2 domain seems to be involved even in the ABA response of distB ( Figure 6). It is not known if the ABA-independent transactivation by ABI3 is a prerequisite of its ABA-dependent activities, or if these two functions are independent. Thus it is difficult to assess whether B2 truly is implicated in the transduction of the ABA signal, or if the decrease observed in the ABA-treated samples is solely a consequence of the decrease produced in the absence of ABA. Still, because B2 is clearly essential for the ABA-independent response, we favor a model in which B2 is involved in tethering ABI3 to the complex at the B-box, rather than in supplying an interface to the ABA signaling pathway. In vitro, a peptide of B2 binds DNA only weakly and unspecifically ( Hill et al., 1996 ), implying that B2 interacts with the ABRE by protein–protein interactions. A bZIP-type transcription factor, TRAB1, was recently isolated in a yeast two-hybrid screen using the B1 and B2 domains of VP1 as bait ( Hobo et al., 1999b ). TRAB1 is a likely candidate for linking ABI3 and the ABRE because, besides interacting with VP1, it binds Em1a-type ABREs in vitro, and mediates ABA-induced transcription in a transient expression assay.
Our results show that the B3 domain is essential for regulation through the RY/G complex ( Figure 6). This observation is in agreement with investigations showing that the B3 domain in VP1 is essential for gene activation mediated by the Sph/RY element ( Carson et al., 1997 ), and that B3 binds the Sph/RY element ( Suzuki et al., 1997 ). In addition, our data show that deletion of B3 reduces the ABA-independent transactivation of the B-box tetramer (in both orientations), but not of the distB tetramer ( Figure 7), implying that B3 interacts with an element present in the B-box but absent in distB. That element should obviously be proxB. Taken together, these observations suggest that the B3 domain is involved even in the ABA-independent response through the B-box, and that this occurs at least partly by interaction with proxB. A fusion protein of the B3 domain and glutathione S-transferase binds the RY-sequence TCCATGCAT ( Suzuki et al., 1997 ). Therefore it is tempting to speculate that ABI3 might also bind the somewhat similar CA-rich element TT CAAA CAC (conserved base pairs underlined) in proxB. This hypothesis is supported by the fact that the motif CAAACAC, in analogy with the RY/Sph element, is highly conserved in seed-specific promoters ( Stålberg et al., 1996 ).
Regulation of napA by ABI3
Our findings on the regulation of napA by ABI3 are summarized in the model shown in Figure 8. Based on this, and previous, studies ( Hobo et al., 1999b ; Suzuki et al., 1997 ), it is logical to assume that ABI3 binds the napA RY-repeats, and that it interacts with a TRAB1-type bZIP factor binding distB. In addition, the B3 domain might also bind the CA-rich proxB element. Our results with inversely oriented B-box tetramers reveal that the ABI3/B-box complex mediates bidirectional transcriptional activation ( Figure 7), suggesting that in the B-box complex at least two ABI3 surfaces mediate transcriptional activation in each direction. Moreover, the distB tetramer mediates transactivation by ABI3 in only one orientation ( Figure 7). Thus one can tentatively envision that at least two ABI3 molecules might target the B-box, each interacting differently with distB and proxB. The detailed structure of this complex remains to be resolved.
Role of the B3 domain in napA versus Em regulation
We have shown that the B3 domain is essential for regulation of napA, a representative member of the 2S gene family in crucifers. Napin 2S proteins are key components of the maturation program in crucifers. In contrast, Carson et al. (1997) showed that the B3 domain of VP1 is not necessary for ABA-regulated gene expression during seed maturation. The molecular explanation of this difference lies in that the RY-repeats are essential for napA regulation, but not for monocot Em regulation ( Vasil et al., 1995 ). It remains to be established if ABA regulation involving RY repeats and the B3 domain is an exceptional property of napin genes, or if it is a common feature of a larger group of maturation-specific genes.
Plasmids used as starting material for recombinant DNA constructs were as follows: pcabi3-4F, containing the ABI3 cDNA insert in the EcoRI site of the vector pBluescript SK ( Giraudat et al., 1992 ), kindly donated by Dr Jérôme Giraudat; pBI221 containing the CaMV 35S promoter fused to the glucuronidase (GUS) gene (uidA from Escherichia coli) and with a NOS termination signal, from Clontech (Palo Alto, CA, USA); pCF101, containing the minimal CaMV 35S promoter (−46 to +8) fused to the GUS gene with the NOS termination signal ( Ezcurra et al., 1999 ); pCF102, as pCF101 with a tetramer of the B-box inserted upstream of the minimal 35S promoter. All constructs were made using standard cloning techniques according to Ausubel et al. (1996) and Sambrook et al. (1989) .
The constructs mut distB, mut proxB, mut B-box, mut RY, mut G-box, mut RY/G and 4xB have been described elsewhere ( Ezcurra et al., 1999 ). The construction of the 35S–ABI3 effector plasmid was performed in two steps. First, a PstI/EcoRI fragment from pBI221, containing the 35S promoter with the GUS gene and the NOS termination signal, was subcloned into the KpnI site of pUC19 by blunt-end ligation. The resulting plasmid, denoted pBI221x, contained additional restriction enzyme sites flanking the 35S:GUS:NOS insert. Further, the 35S:GUS:NOS cassette was in the non-coding orientation with respect to the lacZ promoter in pUC19. This facilitated the subcloning of the ABI3 cDNA, since ABI3 is lethal to E. coli (unpublished observation) and, under our conditions, the lacZ promoter appeared to be active during culture of E. coli. Secondly, pBI221x was digested with SmaI and partially with SacI, thus removing the GUS gene, and pcabi3-4F was digested with HpaI and EcoRV. The excised pBI221 vector was fused to the ABI3-encoding fragment by blunt-end ligation, thus producing the 35S–ABI3 construct, named pCF114. The ΔB3 construct was created by digesting pCF114 with KpnI and partially with AatII. A 5800 bp DNA fragment lacking the B3 domain was purified and ligated in the presence of synthetic linkers Labi31-C (5′-CAGATCGCCGGCAAGACGT-3′) and Labi31-N (5′-CTTGCCGGCGAT CTGGTAC-3′). The resulting ΔB3 construct was deleted between positions 2085 and 2382 in the ABI3 cDNA (clone pcabi3-4F, GenBank accession number X68141). For construction of the ΔB2 plasmid, an ABI3-encoding fragment spanning positions 1780–2094 was produced by PCR, using Pfu polymerase (Stratagene, La Jolla, CA, USA). The oligonucleotide primers for the PCR reaction generated a Sse8387I site 5′ of the PCR product. The PCR product was digested with Sse8387I and KpnI and cloned into the corresponding sites in the ABI3 portion of pCF114. Since there is an additional Sse8387I site in pUC19, pCF114 was digested partially with this enzyme. The resulting ΔB2 construct was deleted between positions 1725 and 1779 in the ABI3 cDNA.
The reporter construct 4xRY/G was created by ligating a synthetic double-stranded dimer of the RY/G cluster, TG CATGCATTATTA CACGTGATCGC CATGCAA (RY-repeats and G-box underlined), supplemented with HindIII and KpnI sticky ends, into the corresponding sites of pCF101. A tetramer was produced by a subsequent cloning step. The reporter construct 4xdistB was made by ligating a synthetic tetramer of the distB element, CTTC GCCACTTGTCACTCC (ABRE motif underlined), supplemented with HindIII and KpnI sticky ends, into the corresponding sites of pCF101. The resulting plasmid was named pIE118. The reporter constructs 4xB-rev and p4xdistB-rev were produced by excising pCF102 and pIE118 with HindIII and KpnI. The tetramer fragments were fused to HindIII-digested pCF101 by blunt-end ligation. All constructs were sequenced using established techniques.
For analysis in transgenic plants, the 35S–ABI3 insert from pCF114 was excised with SacI and SalI and subsequently cloned in the HindIII/EcoRI sites of the binary vector pGA581 by blunt-end ligation.
T0 plant material and crossings
Transgenic T0 plants harboring the 35S–ABI3 construct were produced by transformation of the Agrobacterium tumefaciens strain C58C1 with the 35S–ABI3 construct, and the subsequent transformation of the tobacco cultivar SR1 as described by Stålberg et al. (1993) . Plants were screened by Northern analysis using an ABI3-derived DNA probe and total RNA extracts from leaf tissues (not shown). The transgenic plants harboring the different napA constructs have been described elsewhere ( Ezcurra et al., 1999 ). To obtain the different napA-GUS×35S–ABI3 genotypes, five different T0 plants of each napA construct were selected and crossed to a T0 plant ectopically expressing ABI3 mRNA in leaves. The selected napA construct plants displayed seed GUS activities close to the mean value of the actual construct. Crosses were performed using the 35S–ABI3 plants as pollen donors. As a negative control, T0 plants harboring the napA constructs were self-pollinated. After 30 d, T1 seeds were harvested and frozen at −70°C.
T1 plant material and ABA treatments
For each individual genotype, approximately 20 T1 seeds derived from crossings were surface-sterilized and vernalized for 4 days at 4°C. Seeds were subsequently cultured in vitro in 30 mL liquid 2/3 × MS medium without sucrose and provided with 200 mg L−1 kanamycin, in 10 cm diameter Petri dishes in a growth chamber (25°C, 16 h photoperiod). For faster growth, Petri dishes were not sealed. After 3 weeks plantlets were treated with 50 µm ABA for 48 h.
GUS activity was assayed by luminometry utilizing the GUS-Light Assay kit (Tropix, Bedford, MA, USA) and a luminometer (MicroLumat LB 96 from E&G Berthold, Bad Wildbad, Germany) following the manufacturer's instructions. Plant extracts were produced by homogenization of approximately 10 plantlets in 50 mL Falcon tubes, using a nylon pestle driven by a drill. Leaf sections from bombardments were homogenized in Eppendorf tubes using a nylon pestle driven by a drill. The protein concentration of the extracts was determined using the Protein Assay dye from BioRad (Hercules, CA, USA), according to the manufacturer's instructions.
Leaf sections for bombardment were obtained from young (3–4-week-old) plants of Nicotiana tabacum SR1, grown in vitro in solid MS with 3% sucrose. Larger, expanding leaves (4–5 cm long) were collected and placed on a moistened paper filter. Approximately 2 cm2 rectangular pieces were excised, avoiding major veins. The leaf pieces were immediately placed right-side up on MS agar plates with 2 mg L−1 NAA and 0.5 mg L−1 BAP, and incubated for 48 h in a growth chamber (25°C, 16 h photoperiod). Particle bombardments were performed using a PDS-1000/He Biolistic Particle Delivery System from BioRad. Plasmids were purified using the Wizard Maxiprep kit (Promega, Madison, WI, USA). Coating of gold microcarriers (1.5–3.0 µm, Aldrich, Milwaukee, WI, USA) with plasmid DNA was performed by resuspending 10 mg gold in 125 µL water, and then adding 10 or 20 µg plasmid DNA in 100 µL water, 25 µL 3 m sodium acetate and 625 µL 99% ethanol. The gold particles were resuspended by vortexing between each addition. All coatings contained 10 µg reporter plasmid and, when indicated, also 10 µg effector plasmid. The DNA/gold mixtures were incubated at −20°C for 30 min, then briefly centrifuged and resuspended in 400 µL ethanol. The particles were resuspended by vortexing and a 20 µL aliquot was immediately spotted onto a macrocarrier. After drying, the microcarriers were bombarded onto leaf pieces at 1100 psi. Six leaf pieces were bombarded for every coating. After bombardment, leaf sections were cut into halves; one half was placed in 3 mL MS, and the other half was placed in 3 mL MS with 100 m m ABA. The leaf sections were harvested after 40 h for measurement of GUS activity.
We thank Dr Jérôme Giraudat for generously providing the pcabi3-f plasmid, and Drs Sheila McCormick and Ute Hoecker for critically reading this manuscript. This work was supported by grants from the Swedish Natural Science Research Council, The Foundation for Strategic Research, Carl Trygger's Foundation and The Nilsson–Ehle Fund.