Nuclear localization of the mutant protein phosphatase abi1 is required for insensitivity towards ABA responses in Arabidopsis


For correspondence (fax +49 8161 715432; e-mail


ABI1, a protein phosphatase 2C, is a key component of ABA signal transduction in Arabidopsis that regulates numerous ABA responses, such as stomatal closure, seed germination and inhibition of vegetative growth. The abi1-1 mutation, so far the only characterized dominant allele for ABI1, impairs ABA responsitivity in both seeds and vegetative tissues. The site of action of ABI1 is unknown. We show that there is an essential requirement for nuclear localization of abi1 to confer insensitivity towards ABA responses. Transient analyses in protoplasts revealed a strict dependence of wild-type ABI1 and mutant abi1 on a functional nuclear localization sequence (NLS) for regulating ABA-dependent gene expression. Arabidopsis lines with ectopic expression of various abi1 forms corroborated the necessity of a functional NLS to control ABA sensitivity. Disruption of the NLS function in abi1 rescued ABA-controlled gene transcription to wild-type levels, but also attenuated abi1-conferred insensitivity towards ABA during seed germination, root growth and stomatal movement. The mutation in the PP2C resulted in a preferential accumulation of the protein in the nucleus. Application of a proteosomal inhibitor led to both a preferential nuclear accumulation of ABI1 and an enhancement of PP2C-dependent inhibitory action on the ABA response. Thus, abi1-1 acts as a hypermorphic allele, and ABI1 reprograms sensitivity towards ABA in the nucleus.


The plant hormone ABA plays a crucial role in plant adaptive responses to environmental challenges such as drought and osmotic stress (Zhu, 2002). ABA is, however, also required under non-stress conditions as an endogenous signal that fine-tunes growth and development in plants. The roles of the phytohormone are manifested by numerous ABA-controlled physiological processes, such as seed dormancy and germination, seedling and root development, and stomatal closure (De Smet et al., 2006; Finkelstein et al., 2002; LeNoble et al., 2004; Schroeder et al., 2001a; Zhang et al., 2007). Whereas the stomatal aperture is regulated through rapid ABA-triggered alteration of ion fluxes in guard cells (Li et al., 2006; Schroeder et al., 2001b), most of the ABA-mediated processes emanate from massive ABA-regulated changes in gene expression. Genome-wide expression analyses in Arabidopsis seedlings and guard cells have led to the identification of a large number of genes regulated by ABA (Hoth et al., 2002; Leonhardt et al., 2004; Nemhauser et al., 2006; Seki et al., 2002; Takahashi et al., 2004).

ABA-induced transcriptional upregulation has been reported for genes encoding Mg2+-dependent serine/threonine phosphatase type 2C (PP2C) that act as negative regulators of ABA responses in Arabidopsis (Kuhn et al., 2006; Nishimura et al., 2007; Robert et al., 2006; Rodriguez, 1998; Saez et al., 2004; Yoshida et al., 2006a). Among these PP2Cs are ABI1 and its closest structural homologue ABI2. Both act as negative key regulators with partly overlapping roles in ABA-controlled processes (Merlot et al., 2001). The involvement of ABI1 and ABI2 in ABA signalling was revealed by the characterization of the ABA-insensitive Arabidopsis mutants abi1-1 and abi2-1. The mutant proteins abi1 and abi2 are characterized by a single amino acid exchange in the catalytic domain, ABI1G180D and ABI2G168D, conferring a dominant ABA-insensitive phenotype in seed germination and seedling development, as well as attenuation of seed dormancy and stomatal closure (Koornneef et al., 1984). The mutation impairs magnesium binding and results in a strongly reduced protein phosphatase activity at cellular magnesium levels (Leube et al., 1998).

Intragenic loss-of-function revertants of the abi1-1 mutant and functional inactivation of the PP2C genes ABI1 and HAB1 by T-DNA insertions resulted in an ABA-hypersensitive phenotype (Gosti et al., 1999; Saez et al., 2004; Yoshida et al., 2006a). ABI1 is compartmentalized to the nucleus, and is also associated with the plasma membrane (Himmelbach et al., 2002; Hoth et al., 2002; Zhang et al., 2004). Whether the protein acts at the plasma membrane and/or in the nucleus, or elsewhere, is unclear. Tethering of ABI1 to the plasma membrane by binding to phosphatidic acid (PA) is required for ABA-induced stomatal closure (Mishra et al., 2006; Zhang et al., 2004). PA is generated by the stress-activated phospholipase D1α (PLD1α). PLD1α action is dependent on the α-subunit of a trimeric G protein GPA1. GPA1 interacts with the plasmalemma-localized G protein coupled receptor GCR2 (Liu et al., 2007), which is controversially discussed as an ABA receptor (Gao et al., 2007). PA binding to ABI1 is considered to release the negative control exerted by the PP2C on ABA responses, both by PA-mediated catalytic inhibition and membrane recruitment of ABI1 (Mishra et al., 2006; Zhang et al., 2004). H2O2-triggered catalytic inhibition of the ABI1 response regulator provides an additional post-translational control mechanism (Meinhard and Grill, 2001). ABI1 physically interacts with glutathione peroxidase 3 (AtGPX3) that functions as a H2O2-sensitive redoxtransducer (Miao et al., 2006). In addition, ABI1 binds to a calcineurin B-like calcium sensor and its associated protein kinase in the cytosol (D’Angelo et al., 2006; Guo et al., 2002). and to the ABA response regulator OST1 (Yoshida et al., 2006b). Microinjection experiments of ABI1 and abi1 revealed a rescue of ABA-inducible transcription by possibly out-competing the response-inhibitory mutant protein with ABI1 (Wu et al., 2003). ABA-induced stomatal closure and ABA-dependent gene expression might thus reflect different modes of ABI1 action.

Several studies have shown an effect of ABI1 on transcriptional regulation. Overexpression of ABI1 inhibits ABA-inducible gene transcription in transient transfection analyses via its protein phosphatases activity (Sheen, 1998). The catalytically diminished abi1, however, fully blocks ABA-inducible gene expression, and might impede ABA responses via a dominant negative effect (Sheen, 1998; Yang et al., 2006). In the abi1-1 mutant, ABA-dependent transcriptional regulation is attenuated for more than 90% of the 1400 identified ABA-responsive genes (Hoth et al., 2002).

In this study, we show an essential requirement for nuclear localization of ABI1 and abi1 to mediate insensitivity towards ABA responses. Deletion of or point mutations within the nuclear localization sequence (NLS) domain abrogated the PP2C action as a negative regulator of ABA responses. Replacing the ABI1-NLS by the NLS of the simian virus 40 large T-antigen fully rescued nuclear localization and the PP2C-mediated control of ABA-dependent reporter expression. The mutant abi1 showed a preferential nuclear accumulation compared with ABI1, consistent with a gain-of-function mutation. Our analyses reveal a reprogramming of ABA sensitivity in the nucleus by ABI1 and abi1.


Subcellular localization of ABI1 and abi1 in Arabidopsis protoplasts is dependent on the integrity of the nuclear localization sequence

ABI1 interaction with the transcription factor ATHB6 implied a nuclear localization of the PP2C (Himmelbach et al., 2002). Sequence analysis of ABI1 revealed a short region enriched in basic amino acids sharing similarity to the monopartite NLS of simian virus 40 large T antigen, which has known functionality in plants (Hicks and Raikhel, 1993; Kanneganti et al., 2007; Merkle, 2001). The predicted NLS of ABI1 is located at the very end of the carboxyl-terminal domain of the PP2C (Figure 1a). Amino-terminal GFP fusions of ABI1 and abi1, and of NLS-deleted versions (dNLS), were expressed in Arabidopsis protoplasts to examine a possible NLS function. ABI1-GFP and abi1-GFP predominantly accumulated in the nucleus (Figure 1b). Deletion of the NLS led to a redistribution of the proteins between the nuclear and cytosolic compartments, similar to the GFP control. To examine whether disruption of NLS function rather than mere carboxyl-terminal truncation of the PP2C yielded the change in subcellular distribution, four lysine and arginine residues of the NLS were mutated into asparagine and glutamine residues (ABI1mtNLS; Figure 1a) with the aim of inactivating the NLS of both wild-type and mutant PP2Cs. The basic amino acid residues of NLS are critical for function, and substitution for neutral amino acid residues abrogates nuclear import (Merkle, 2003). The cellular localization of abi1mtNLS (Figure 1b) and ABI1mtNLS (data not shown) was indistinguishable from the GFP control. In addition, we examined whether the NLS of simian virus 40 large T antigen is capable of rescuing the nuclear compartmentation of abi1mtNLS. The fluorescent signal of abi1dNLS-SV40NLS-GFP clearly accumulated in the nucleus of Arabidopsis protoplasts (Figure 1c). The analysis provides strong evidence for a NLS-dependent import of the PP2C into the nucleus.

Figure 1.

 Subcellular localization of ABI1 and abi1, and their dependence on the nuclear localization sequence (NLS) motif.
(a) The carboxyl-terminal sequence of ABI1 has similarities to the NLS of the simian virus 40 (SV40) large T antigen. The ABI1 carboxyl-terminal sequence shown was deleted in ABI1dNLS. The NLS-like motif of ABI1 is underlined, and the amino acid substitutions created in ABI1mtNLS and abi1mtNLS are indicated.
(b) Transient expression of ABI1 variants fused to GFP in protoplasts. (A) The subcellular localization of the fusion proteins was visualized by confocal laser scanning microscopy or epifluorescence (ABI1-GFP). (B) Protoplasts were counterstained with bisbenzimide H33342 or DAPI (ABI-GFP) to reveal the position of nuclei.
(c) Subcellular distribution of abi1dNLS-SV40NLS-GFP. Scale bar: 10 μm.

The regulation of ABA-mediated gene expression by ABI1 and abi1 depends on a functional NLS

Next, we addressed whether the capacity of ABI1 to enter the nucleus affects ABA signal transduction. ABI1 and abi1 are known to block ABA-inducible gene expression when they are ectopically expressed in protoplasts of maize or Arabidopsis (Sheen, 1998; Yang et al., 2006). To monitor ABI1 action on the ABA signal pathway, Arabidopsis protoplasts were transfected with various ABI1 effector genes and an ABA-dependent reporter construct (Figure 2a). Aliquots of the transfected sample were analysed for their ABA response after incubation in the presence and absence of ABA, by normalizing the reporter expression to GUS activity that was generated by co-transfection of a constitutively expressed GUS cassette. In the absence of effectors, ABA exposure (30 μm) of the mesophyll protoplasts resulted in an approximately 20-fold increase of reporter activity (Figure 2b). This increase was not affected by transfection of the empty effector cassette (10 μg DNA, data not shown). However, effector cassettes that provide expression of either ABI1 or abi1 led to a clear and dose-dependent inhibition of the ABA response (Figure 2b). The response to ABA was essentially eradicated by transfection with 10 μg of the abi1-expressing plasmid, resulting in a residual activity of 3% above control levels. Expression of ABI1 was less effective in downregulating ABA-mediated reporter expression. A twofold inhibition of the response required about 200 ng of the ABI1 effector construct, and approximately threefold lower quantities of the corresponding abi1 plasmid. These differences in effectiveness were consistently observed in the analyses (n > 5) carried out independently and with different plasmid preparations, and hence reflect a functional consequence of the single nucleotide exchange within the PP2C gene. Functional inactivation of the NLS completely abolished the negative regulation exerted by abi1 (abi1mtNLS; Figure 2b) and ABI1 (not shown) on ABA signalling. A complete loss of PP2C action was also observed by expression of a phosphatase non-active form of ABI1 (ABINAP; Figure 2b). We conclude that the nuclear localization of the PP2C seems to be required to negatively regulate ABA signalling in protoplasts. Alternatively, protein stability of the NLS-mutated PP2Cs is severely compromised.

Figure 2.

 Effect of ABI1 and abi1 expression on ABA-mediated promoter activation in Arabidopsis protoplasts.
(a) Schematic presentation of the reporter and effector constructs. The reporter construct consists of the ABA-regulated pRAB18 promoter controlling the expression of the firefly luciferase (LUC).
(b) Control of ABA response by different PP2C effectors. The induction of reporter activity by ABA is presented relative to induction levels without effector expression. Protoplasts were incubated with or without 30 μm ABA for 20 h after transfection, and the ABA response was subsequently determined and normalized to GUS activity. The effector constructs permit the expression of ABI1 (bsl00043 and △), the catalytically non-active ABI1NAP (bsl00041), abi1 (bsl00000) and abi1mtNLS (bsl00001) under the control of the CaMV p35S promoter. The dotted line indicates the response of ABI1-expressing protoplasts without ABA challenge. The data represent the mean values of independent transfections (n = 5).

ABI1 and abi1 fusions differentially regulate ABA-induced gene expression

We then sought to establish whether the NLS-mutated PP2Cs protein was unstable. Unfortunately, polyclonal antibodies that were raised against ABI1 crossreacted with the homologous ABI2 (Yang et al., 2006), and might even have recognized other related PP2Cs (Saez et al., 2006). Therefore, we decided to tag the different PP2Cs with GUS (Figure 3a), and determine whether the stability of the fusion proteins was affected by the carboxyl-terminal modifications. Arabidopsis mesophyll protoplasts were transfected with the corresponding effector plasmids. Subsequently, GUS activity was determined and normalized to the expression of a transfected p35S-driven LUC control gene. There was no major difference in reporter activity, irrespective of the fused PP2C version (Figure 3b). The fusion proteins were still functional to block ABA signalling, albeit with a reduced efficacy compared with the non-fused forms (Figure 3c). Again, the ABI1 fusion was less inhibitory than the abi1 version. Taken together, PP2C stability was not detectably affected by the NLS modification. Similarly to the results obtained with GUS fusions, ABI1-GFP and abi1-GFP effectors inhibited ABA-dependent reporter expression (Figure 4a). The reporter fusion impaired the efficiency of the effectors to act on ABA signalling to some extent. Deletion of the NLS abrogated the action of abi1 on ABA signalling, whereas substitution of the endogenous NLS with the NLS of simian virus 40 large T antigen rescued the response (Figure 4b). Furthermore, regulation of the ABA response by ectopic expression of abi1-GFP was similar, irrespective of expressing the effector under the control of the p35S promoter or pABI1 promoter (Figure 4b). We conclude from these experiments that the NLS modification does not impact protein stability and that the tagged ABI1 and abi1 versions are still functional, albeit at a reduced level.

Figure 3.

 ABI1 and abi1 fusions to GUS affect ABA signalling.
(a) The reporter and effector constructs are represented schematically.
(b) The GUS activity of fusion proteins was determined in cell-free extracts of transfected protoplasts, which was normalized to the expression of a co-transfected p35S::LUC reporter construct and expressed as the ratio of relative fluorescence units to light units (RFU/RLU).
(c) The effector constructs provide the expression of GUS fusions to ABI1 (⋄), abi1 (bsl00043), ABI1dNLS (◆) and abi1dNLS (bsl00041) fusions. The ABA response for expression of non-fused abi1 (bsl00000) is included for comparison. The analysis was performed as described in Figure 2, and data represent the mean values of independent transfections (n = 5).

Figure 4.

 Inhibitory action of ABI1 and abi1 for dependence on tag fusion, nuclear localization sequence (NLS) function and promoter.
(a) The differential effect of GFP- and GUS-tagged forms of the PP2C on ABA-responsive pRD29B::LUC expression. Arabidopsis protoplasts were transfected with effector plasmid or empty vector (1 μg) as a control, and exposed to 30 μm ABA. The ABA-responsive reporter activity is expressed relative to the control. Data are given as means of four independent transfections.
(b) The expression of abi1-GFP under the control of the pABI1 promoter fragment (bsl00041) yielded a downregulation of the ABA response similar to p35S::abi1-GFP expression (bsl00043). The impaired action of abi1dNLS-GFP (bsl00066) was rescued by carboxyl-terminal fusion of the simian virus 40 (SV40) large T antigen NLS (bsl00001). Conditions of (a) and (b) were as described in Figure 2.

ABA insensitivity mediated by abi1 in Arabidopsis plants depends on a functional NLS

Tagging the PP2C with GUS or GFP reduced its ability to inhibit ABA signalling. The ABI1 fusions still blocked the ABA-mediated response in protoplasts, although less so than the abi1 versions, whereas deletion or mutation of the NLS abolished the inhibitory action. To evaluate the contribution of the ABI1 NLS domain on other ABA responses in plants, transgenic Arabidopsis lines were generated expressing either GUS or GFP fused to ABI1, abi1, ABI1dNLS and abi1dNLS. Transgenic plants expressing GUS fusions were recovered, and seeds of homozygous lines were analysed with respect to ABA-mediated inhibition of germination, by scoring radicle emergence after 4 days of incubation in the presence of exogenous ABA. The expression of GUS-abi1 resulted in a strong change in ABA sensitivity of the seed material. Control seeds transformed with the empty GUS cassette were fully inhibited from germinating in the presence of 1 μm ABA (Figure 5a). In contrast, GUS-abi1 seeds revealed a germination efficiency that was greater than 90%, even in the presence of 10 μm ABA, which is consistent with an inhibitory action of the effector protein on the ABA response. The marked ABA insensitivity mediated by abi1 was entirely lost, however, in seeds expressing the NLS-deficient version. The germination rate was indistinguishable from that of control seeds. Expression of both ABI1-GUS and the fusion protein with deleted NLS had no significant effect on ABA-dependent seed germination.

Figure 5.

 Analysis of ABA responses in Arabidopsis plants expressing variants of ABI1.
(a) Germination assay of seeds (n = 80–100, SD ± 10%) after 4 days of incubation.
(b) Root growth assay in the presence of ABA. Root growth was measured within 4 days of the transfer of 5-day-old seedlings (n = 25, SD ± 16%). In the absence of ABA, root growth equalled 16 ± 2, 16 ± 2, 17 ± 2, 17 ± 2 and 16 ± 2 mm for unfused GUS, GUS-ABI1, GUS-ABI1dNLS, GUS-abi1 and GUS-abi1dNLS plants, respectively.
(c) The stomatal control was analysed in GUS-expressing lines by determining water loss of excised leaves. Leaves (n = 8) of comparable size and age (3-weeks old) were excised and the loss of fresh weight was measured in ambient conditions (SD ± 12%).
(a–c) Control plants expressing GUS △) or fusions of GUS to abi1 (bsl00041) and ABI1 (bsl00001), as well as ABI1dNLS (bsl00000) and abi1dNLS (bsl00043) were analyzed. The data of one transgenic line for each PP2C version is shown. Two other independent lines for the same version yielded comparable results.
(d) Analysis of ABA-dependent germination in Arabidopsis plants expressing abi1-GFP under the control of the pABI1 promoter. Germination assay of seeds expressing non-fused GFP (bsl00000) or fusions of GFP to abi1 under the control of p35S (bsl00001) and pABI1 (bsl00041), respectively. The experimental procedures were as described in (a).

The requirement for a functional NLS domain for conferring ABA insensitivity by abi1-GUS was corroborated by analysis of ABA-mediated inhibition of root expansion (Figure 5b). In the presence of high levels of ABA (100 μm), root elongation of abi1-GUS expressing seedlings was reduced by approximately 35% compared with non-exposed seedlings. At the same ABA level, the transgenic line expressing the NLS-deleted abi1 was inhibited by >80%, and was comparable with the GUS control and the ABI1-GUS line. ABA insensitivity of stomatal responses leads to a wilting phenotype characterized by increased water loss of detached leaves (Himmelbach et al., 2002; Meyer et al., 1994; Pei et al., 1998). To analyse stomatal responses, water loss was determined from leaves of comparable size and age that were obtained from the different transgenic Arabidopsis lines. Expression of ABI1 and its carboxyl-terminally truncated version did not affect stomatal closure. Leaves expressing abi1-GUS revealed a twofold enhanced water loss relative to the leaves of the control line or the line expressing the NLS-deficient abi1 (Figure 5c). The increased water loss of the abi1 line resulted in a clearly visible wilting phenotype (data not shown). Arabidopsis lines expressing the corresponding PP2C-GFP fusions yielded similar results. ABA-insensitive germination was observed in seed material expressing abi1-GFP under the control of either the p35S or the pABI1 promoter (Figure 5d), whereas expression of abi1dNLS-GFP had no effect on ABA responsiveness during germination (Figure 6a).

Figure 6.

 Differential nuclear localization of abi1 and ABI1.
(a) The germination rate of seeds (n = 90–120) expressing GFP fusions to ABI1, abi1 and abi1dNLS was assessed after 4 days of incubation in the absence (open bars) or presence of 15 μm ABA (filled bars). The experiment was performed in three replicates.
(b) The signal intensity of GFP fluorescence in the nuclear versus cell plane of root and guard cells (open columns, n = 30; filled columns, n = 50, respectively) is presented in the upper panel. Confocal imaging of representative cells expressing GFP-abi1, GFP-ABI1 or GFP (lower panel). The signal intensity was determined by using Simple PCI imaging software. Scale bars: 10 μm.

The phenotypes of the transgenic lines are consistent with the previous analysis in protoplasts, which revealed a requirement for a functional NLS domain in the PP2C to mediate insensitivity towards ABA responses. In variance to the analysis in protoplasts, expression of ABI1-GFP or ABI1-GUS proteins did not significantly alter the ABA response of Arabidopsis plantlets. The necessity for nuclear localization of the PP2C to mediate the inhibitory effect prompted us to analyse in more detail the intracellular localization of the fusion proteins in the transgenic lines.

The cytological analysis of guard and root cells revealed a predominance of abi1-GFP in the nucleus compared with the non-fused GFP control, which was similar to the previous analyses in protoplasts (Figure 6b,c). In contrast, the fluorescent signal of the non-fused GFP marker was clearly shifted to the cytosolic compartment, and the shift was even more pronounced with ABI1-GFP. The PP2C-tagged proteins were detected throughout the seedling, similar to GFP expression under control of the p35S promoter, by confocal laser scanning microscopy. There was a single difference in the root tip. The PP2C-GFP signal was clearly detected in the lateral root cap and columella cells, whereas the meristematic zone of the root tip yielded a much lower signal (Figure 7). The GFP control under the same promoter showed an even expression pattern in both the cap and meristematic zone of the root. Close-ups of the samples revealed a subcellular distribution of ABI1-GFP, consistent with a nuclear and cytosolic compartmentation (Figure 7c). The abi1-GFP preferentially accumulated in the nucleus (Figure 7f), whereas abi1 with deleted NLS yielded an intracellular distribution resembling ABI1 (Figure 7i). An intracellular relocation of ABI-GFP and abi1-GFP was not observed in response to ABA (data not shown).

Figure 7.

 Expression and subcellular localization of ABI1 variants in the root tip.
Representative confocal scans of plants expressing ABI1-GFP (a–c), abi1-GFP (d–f), abi1dNLS-GFP (g–i) or GFP (j–l) are presented. The GFP signal (a, d, g and j) and the fluorescence obtained by plasma membrane staining with FM 4-64 (b, e, h and k) were merged (c, f, i and l). The inset shows a close-up of the root cells. Scale bars: 20 μm, except in the insets (5 μm).

The analyses indicate an intracellular distribution of tagged ABI1 that is different from abi1, in that the mutant protein was primarily detected in the nucleus. Comparison of the fluorescent signal obtained with the PP2C-GFP fusion and the GFP control indicate an efficient degradation or export of the fusion proteins from the meristematic zone (Figure 7c,i,l). To address whether a proteosomal pathway operates in the turnover of the PP2C, the proteosomal inhibitor MG132 was administered to Arabidopsis seedlings. Inhibitor application did not stabilize the GFP signal of the fusion proteins in the meristem. It did, however, affect the nuclear compartmentation of ABI1 (Figure 8a). ABI1-GFP now localized preferentially to the nucleus, and the subcellular distribution pattern resembled more that of abi1-GFP, whereas the proteosomal inhibitor had little effect on the subcellular localization of abi1. We examined the effect of MG132 on both ABI1-controlled reporter expression and PP2C stability in transient expression analyses. Arabidopsis protoplasts transfected with effector constructs revealed a marked stimulation of ABI1 and abi1 inhibitory action on ABA signalling in the presence of the proteosomal inhibitor (Figure 8b). The 11.3-fold induction of reporter expression by ABA was little affected by the proteosomal inhibitor (9.8-fold increase). In the presence of MG132, protoplasts expressing ABI1-GFP showed an approximately twofold enhanced reduction of ABA-responsive reporter expression, resulting in expression levels of between one- and twofold of the non-induced control. Western blot analysis revealed no significant change in the protein abundance of differently tagged PP2C proteins (Figure 8c).

Figure 8.

 Impact of proteosomal inhibitor MG132 on nuclear compartmentation of ABI1 and abi1.
(a) Localization of ABI1-GFP and abi1-GFP in mock- and MG132-treated Arabidopsis root tips was visualized by confocal laser scanning microscopy after 3 h of inhibitor exposure (50 μm). Scale bars: 20 μm. The insets shows close-ups of the root cells, with scale bars that correspond to 5 μm.
(b) Increased inhibitory action of ABI1 and abi1 by MG132. Arabidopsis protoplasts were transfected with 1 or 3 μg of effector plasmid, as described in Figure 2 in the absence (open bars) and presence of MG132 (filled bars). The ABA induction of pRAB18-driven reporter activity is presented relative to protoplasts transfected with the empty vector, which yielded an ABA-dependent increase by a factor of 11.3 ± 3.9 and 9.9 ± 2.8 in the absence and presence of MG132, respectively. Data are given as the means of four independent transfections. (c) Protein levels of abi1-GFP (∼71 kDa) and ABI1-c-myc protein (∼49 kDa) were analysed by western blot analysis (arrows), and the relative quantities are indicated below in MG132- and mock-treated protoplasts.

The control of abi1 on the ABA-induced gene expression depends on nuclear localization

ABA signalling re-addresses gene expression, and ABA-induced transcripts are deregulated in the abi1-1 mutant (Hoth et al., 2002; Seki et al., 2002). We wanted to know whether the NLS function of abi1 affects ABA-induced gene transcription. Therefore, we isolated RNA from seedlings expressing abi1 and abi1dNLS fused to either GUS or GFP, in addition to seedlings of the wild type and the abi1-1 mutant. Genes such as the dehydrin RAB18 (At5g66400), the desiccation-responsive gene RD29B (At5g52300) and the homeobox-leucine zipper protein AtHB12 (At3g61890) failed to respond towards ABA in abi1-1 (Gosti et al., 1995; Hoth et al., 2002; Lang et al., 1994; Uno et al., 2000), and were selected for analysis. Transcript abundance of these ABA-inducible genes was analysed in response to 3 μm ABA. An impaired ABA-mediated induction in the abi1-1 mutant, and also in the abi1 fusion protein expressing plants, could be observed by semiquantitative RT-PCR analysis (Figure 9a). Real-time RT-PCR analysis of ABA-induced RAB18 transcripts confirmed the finding (Figure 9b). Inhibition of ABA-dependent gene induction was less effective in the abi1 fusion expressing lines compared with abi1 expressing plants, similar to the previous transient and stable expression analyses. No significant differences in ABA-dependent gene regulation were observed between the abi1dNLS-GUS expressing seedlings and the wild type.

Figure 9.

 Differential regulation of ABA-responsive genes.
(a) The transcript abundance of ABA-regulated genes was determined by semiquantitative RT-PCR analysis. RNA was extracted from seedlings expressing abi1, abi1-GFP, abi1dNLS-GFP, abi1-GUS or abi1dNLS-GUS that were exposed for 4 h to ABA (3 μm) or were mock treated. The cDNA of actin (ACT1, At2g37620), RAB18 (At5g66400), RD29B (At5g52300) and ATHB12 (At3g61890) were amplified, and the products were separated by electrophoresis and stained with ethidium bromide. The actin product served as a control.
(b) Real-time RT-PCR analysis of the RAB18 transcripts isolated in (a). The RAB18 transcripts were normalized to the abundance of actin transcripts. The fold induction by ABA is presented, including the standard deviations from two independent runs.


The phytohormone ABA plays a crucial role in seed developmental processes such as maturation, dormancy or germination (Finkelstein et al., 2002), and regulates a wide spectrum of vegetative responses including growth (Cheng et al., 2002; LeNoble et al., 2004; Lin et al., 2007), stomatal movements (Schroeder et al., 2001b) and the integration of signals resulting from drought, high salinity and low temperature (Christmann et al., 2005; Yamaguchi-Shinozaki and Shinozaki, 2006). In this context, a plethora of genes is regulated by ABA (Hoth et al., 2002). Among the regulators of ABA signal transduction, a role for a number of protein kinases and protein phosphatases has been assigned to the regulation of ABA-dependent gene expression, including the ABI1-type PP2C phosphatases (Christmann et al., 2006; Fujii et al., 2007; Yamaguchi-Shinozaki and Shinozaki, 2006; Yoshida et al., 2006b).

In general, eukaryotic PP2Cs are considered to be cytosolic proteins. However, some members are also nuclear localized (Gilbert et al., 2007; Reyes et al., 2006). In this study, we analysed a putative monopartite NLS of ABI1, and its function in mediating ABA responses. Cytological analyses of GFP-labelled ABI1 revealed the requirement of a functional NLS for nuclear accumulation in Arabidopsis. Deletion or mutation of the NLS domain abrogated the control of ABI1 and of the point-mutated abi1 (ABI1G180D) to regulate ABA-dependent reporter expression. Fusion of a functional NLS to the NLS-deleted abi1 or ABI1 rescued the response. Furthermore, ABA insensitivity conferred by expression of tagged abi1 in seedlings required a functional NLS. The negative regulation of ABA responses by PP2C also depended on an active phosphatase domain, as was reported previously (Sheen, 1998).

In transient analyses of protoplasts, expression of abi1 consistently proved to be more efficient in blocking ABA-dependent gene regulation than the wild-type protein. The inhibitory action of the PP2Cs was gradually reduced by carboxyl-terminal fusion of GFP and GUS to ABI1 and abi1. The GFP and GUS tags have a reported tendency to form dimeric and tetrameric aggregates, respectively, that might interfere with nuclear import (Jefferson et al., 1987; Palm et al., 1997; Yang et al., 1996). In transgenic plants, only reporter fusions to abi1 generated aberrant ABA responses, possibly reflecting a reduced expression level in stably transformed cells compared with the transiently transformed protoplasts. In these analyses, expression of abi1 yielded similar results, irrespective of expression by the endogenous promoter or the viral p35S promoter.

Ectopic expression of abi1-GFP and abi1-GUS in Arabidopsis mimicked the dominant ABA insensitivity observed in the abi1-1 mutant. The responses affected included ABA-dependent gene expression and stomatal control, as well as seed germination and vegetative growth. Similar to the analyses in protoplasts with ABI1 and abi1, the ABA insensitivity conferred to the transgenic plants was strictly dependent on a functional NLS of the PP2C. The cytological analysis revealed a preferential nuclear compartmentation of abi1. Nuclear import of ABI1 could involve SAD2, an importin β-like protein that is specifically involved in ABA responses (Verslues et al., 2006). Importins are components of the nuclear transport machinery, and functional inactivation of SAD2, but not of its closest structural homologue, generated an ABA-hypersensitive phenotype.

We know that there are several levels at which ABA sensitivity of plants is controlled within the nuclear compartment. The control of transcriptional processes, including the regulation of the C-terminal domain phosphatase of RNA polymerase II (Koiwa et al., 2002), and post-transcriptional events such as RNA splicing and capping (Hugouvieux et al., 2001; Papp et al., 2004; Xiong et al., 2001), has been reported to affect ABA sensitivity. Transcriptome analyses revealed a dramatic change in ABA-dependent transcripts in abi1-1, indicating a prominent role of the PP2C in the regulation of gene expression (Hoth et al., 2002). The homeodomain transcription factor AtHB6 is a nuclear interactor of ABI1 that negatively controls ABA-dependent stomatal closure and seed germination in a similar way to ABI1 (Himmelbach et al., 2002). Downregulation of AtHB6 expression diminished the abi1-mediated ABA insensitivity (Himmelbach, Moes and Grill, unpublished data), consistent with an AtHB6-relayed pathway to signal ABA insensitivity. At this stage, it remains elusive how nuclear ABI1/abi1 action leads to a desensitization of ABA responses. It could involve more than one mechanism, but evidence is supporting a control mechanism via the regulation of gene expression.

Besides the nuclear compartmentation, we observed a cytosolic localization and a plasma membrane association of ABI1. ABI1 promotes an increase in PA from phosphatidylcholine by PLDα1, which stimulates recruitment of ABI1 to the cellular membrane (Mishra et al., 2006; Zhang et al., 2004). PLDα1 deficiency resulted in the detection of ABI1 at the perinuclear region (Zhang et al., 2004), indicating an ABA-dependent mobilization of ABI1. In our experiments, we could not observe such a signal-induced relocation of ABI1. In seedlings, the intracellular partitioning of GFP-tagged ABI1 differed from abi1 in that the mutant protein was primarily detected in the nucleus, whereas ABI1 was more evenly distributed between the cytosol and the nucleus.

In the presence of the proteosomal inhibitor MG132, however, the nuclear compartmentation of ABI1 was enhanced, and was more similar to the pattern of abi1, which was not detectably affected by the inhibitor. The inhibitory action of ABI1 and abi1 was clearly enhanced by MG132 administration to protoplasts, and might reflect a more efficient recruitment of the PP2C into the nuclear compartment, or a reduced export from it in the presence of the inhibitor. The finding that cellular ABI1 levels were not detectably altered by MG132 argues for an inhibitor-dependent redistribution of the PP2C into the nucleus, which is possibly controlled by a ubiquitin-dependent process.

The requirement for a nuclear compartmentation of ABI1 and abi1 to impose ABA-insensitivie plant responses, and the more efficient nuclear accumulation of abi1, provide an explanation for the dominant nature of the ABA insensitivity conferred by the mutant protein. The G180D mutation in abi1 seems to generate a protein that is more efficiently recruited to the site of action for negative regulation of the ABA signal pathway. The conclusion of a hypermorphic mutant is consistent with the functional analysis of the PP2C HAB1 (Robert et al., 2006). Introducing the abi1-equivalent mutation into HAB1 or AtPP2CA also conferred ABA insensitivity to Arabidopsis (Robert et al., 2006; Yoshida et al., 2006a). Robert et al. argued for a mutant protein that may have hyperactive phosphatase activity with the endogenous substrate. Enzymatic characterization of ABI1 with artificial substrates in vitro revealed an impaired efficiency of Mg2+ binding in abi1 (Leube et al., 1998). At optimal Mg2+ levels, there was no difference in phosphatase activity between the wild type and the mutant, supporting the notion that the preferential nuclear localization of abi1 mediates the phenotype, despite a reduced phosphatase activity at the ABI1-limiting cellular Mg2+ levels. Besides the ABA insensitivity conferred by abi1-1-like gain-of-function mutations, loss-of-function alleles of PP2Cs resulted in a recessive ABA-hypersensitive phenotype (Nishimura et al., 2007; Robert et al., 2006; Saez et al., 2006; Yoshida et al., 2006a). Whether these PP2Cs also require nuclear localization to desensitize ABA signalling is an open question. Only ABI2 and AtPP2C reveal promising motifs for a potential carboxyl-terminal NLS, consisting of the residues KGIRKFK and RKRR, respectively. However, NLS-independent mechanisms for nuclear protein import also seem to operate in plants (Cole et al., 2006; Rout et al., 2003; Zhou et al., 2006).

Based on structural alignment with human PP2Cs (Almo et al., 2007; Das et al., 1996; Robert et al., 2006), the abi1 mutation is localized at the periphery of the protein in the proximity of amino acid residues DGH179 and GST243. Both motifs are evolutionarily conserved among PP2Cs, and are part of two adjacent anti-parallel β sheets. The abi1 mutation introduces a negative charge to the protein and may mimic a phosphorylated ABI1. A putative phosphorylation site directly facing G180 in ABI1 is T239, which is localized at the protein periphery, and which is conserved among PP2Cs involved in ABA responses. The phosphorylation mimic could escape a phosphorylation-dependent mechanism to regulate subcellular compartmentation. Previous analyses have revealed that the abi1 mutation is located in a protein interaction domain that is highly conserved among ABI1 and ABI2. In ABI2, the corresponding G168D mutation leads to the loss of the interaction between ABI2 and sucrose non-fermenting kinase-like PKS3/CIPK15 (Ohta et al., 2003) and prefibrillin (Yang et al., 2006) in the cytosolic compartment. Similarly, abi1 is impaired in the interaction with AtHB6 (Himmelbach et al., 2002), with the ABA response kinase SRK2E/OST1 (Yoshida et al., 2006b) and is likely to be impaired in the interaction with two other cytosolic CIPKs (Gong et al., 2002; Guo et al., 2002). An impaired interaction of abi1 with a presumed cytosolic anchoring protein is expected to result in a preferential nuclear accumulation.

The cellular interaction of ABI1 involves signalling components at the plasma membrane, cytosol and nucleus. Thus, the desensitization of ABA responses by the PP2C in the nucleus is likely to reflect just one function of the ABA response regulator. The exact role of ABI1 and the position of its action in the signal pathway is still a matter of debate (Christmann et al., 2006; Merlot et al., 2001; Murata et al., 2001; Pei et al., 1997; Saez et al., 2006; Wu et al., 2003), but now a specific role of ABI1 emerges in the nuclear compartment for adjusting the ABA sensitivity of a plant.

Experimental procedures

Plant material and chemicals

Plants of Arabidopsis thaliana Heynh. ecotype Columbia (Col-0) were grown in pots on a perlite/soil mixture at 23°C under long-day conditions with 16-h light (250 μE m−2 sec−1). These plants were used for stable transformation, protoplast preparation and DNA extraction. For microscopy, Arabidopsis seedlings were grown on agar plates as previously described (Christmann et al., 2005). For RNA extraction, transgenic seeds of accession RLD and abi1-1 seeds (accession Landsberg) were grown on agarose plates containing 50% MS salt base (pH 5.7) and 0.5% sucrose. Plates were kept for 2 days at 4°C and thereafter were maintained at 22°C under continuous illumination (60 μE m−2 sec−1). All chemicals used were of analytic grade or of highest purity, and were purchased from Fluka (, Sigma-Aldrich ( and Merck ( ABA (Sigma-Aldrich) was dissolved in 10 mm 2-(N-morpholine)-ethanesulphonic acid (MES, pH 7.0) to yield a 5 mm stock solution.

Plasmid constructs

The pRAB18::LUC reporter plasmid was constructed by cloning the LUC gene from pGEM-Luc (Promega, into the BamHI-EcoRI site of pSK (Stratagene, The pRAB18 promoter (from −524 to +67 bp) was PCR amplified from Arabidopsis (Col-0) genomic DNA using the primers 5′-TCCCCGCGGATCTAAACGCGGCGTTTGG-3′ and 5′-CGCGGATCCACTTTGAGCTAAGCTAG-3′. These primers introduced single BamHI and SacII sites into the promoter fragment, which was fused to the LUC gene by using the SacII and BamHI sites of pSK. Construction of pRDB29::LUC has been described previously (Christmann et al., 2005).

The effector plasmids used for transient expression in protoplasts are all derivatives of the pBI221 vector (Jefferson et al., 1987). The plasmids pBI221-p35S::ABI1 and pBI221-p35S::abi1 are created by replacing the glucuronidase gene of pBI221 by a BamHI-Eco147I fragment of ABI1 and abi1, respectively. The p35S::ABI1NAP (ABI1D177A) construct corresponds to the non-active protein phosphatase sequence previously described (Himmelbach et al., 2002).

To create the p35S::GFP fusion cassettes of ABI1, abi1, abi1mtNLS, abi1dNLS and abi1dNLS-SV40NLS, the GUS gene of pBI221 was replaced by a BamHI-SacI fragment of the smRS-GFP gene (Davis and Vierstra, 1998). The STOP codon of GFP was removed via PCR using the primer 5′-CGAGCTGCAGTTGTATAGTTCATCCATGC-3′, thereby generating an XhoI site at the 3′ end of GFP. The ABI1 and abi1 sequences were both amplified with the primers 5′-AATTCCTCGAGGGAAGTATCTCCGGC-3′ (forward) and 5′-GTACTCGAGTCAGTTCAAGGGTTT-3′ (reverse). The reverse primers used for the amplification of abi1mtNLS, abi1dNLS and abi1dNLS-SV40 were 5′-CCCCTCGAGTCAGTTCAAGGGTTTGCTCTGCAGATTATTCTGAGGCTTC-3′ (abi1mtNLS), 5′-GTACTCGAGCTCTCACACACTTATGTTGTCTTTGC-3′ (abi1dNLS) and 5′-TTGCTCGAGTCAATTTACCTTTCTCTTCTTTTTTGGAGGAGTCACACTTATGTTGTCTTTGC-3′ (abi1dNLS-SV40NLS). All ABI1 and abi1 PCR products were inserted into the XhoI site at the 3′ end of smRS-GFP. The pABI1::GFP-abi1 effector construct was generated by replacing the p35S promoter of the previously described pBI221p35S::GFP-abi1 plasmid with the pABI1 promoter (from −1999 to +456 bp) amplified from genomic Arabidopsis DNA with SphI- and NheI-overhang primers 5′-TTTGGCGCGCCGCATGCTTGAATATATACAAGATTT-3′ and 5′-CCTAGCTAGCTAACGGTAAAGATTTGATC-3′. To generate the p35S::GUS fusion constructs in pBI221, the STOP codon of the GUS gene was removed via PCR with the reverse primer 5′-ACTGCTCGAGCATTGTTTGCCTCCCTG-3′, thereby creating an XhoI site at the 3′ end of GUS. This XhoI site was used to fuse the ABI1, abi1, ABI1dNLS and abi1dNLS PCR products generated with the aforementioned XhoI-overhang primers to the GUS gene in pBI221.

RT-PCR analysis

Total RNA was extracted from 4-week-old leaves using the AurumTM Total RNA Mini Kit (Bio-Rad, A 600-ng sample of total RNA was used for first-strand cDNA synthesis via oligo dT primer following the supplier’s instructions (First Strand cDNA Synthesis Kit, Fermentas, The primer pairs used were as follows: 5′-GCGTCTTACCAGAACCGT-3′ and 5′-GAAGCATTCCTCCCAAGC-3′, 5′-CCCTGTAAAAGATGAAACTCCGAG-3′ and 5′-CCCAATCTCTTTTTCACACAAAG-3′, 5′-CAGCGAGATTAGTAGTGGCA-3′ and 5′-GTAATTGCTGCTAGATTGGTC-3′, 5′-CAATACATGATTAGCCCGAG-3′ and 5′-ACTGTTTGGAGTGATGACTCC-3′. Amplification of the ACT1 transcript (At2g37620) as an internal standard was performed with the primers 5′-TGGGATGACATGGAGAAGAT-3′ and 5′-ATACCAATCATAGATGGCTGG-3′. To determine the RAB18 transcript quantities by real-time RT-PCR, the aforementioned gene-specific primers were used. Actin primers were designed as follows: 5′-TGGAACTGGAATGGTTAAGGCTGG-3′ and 5′-TCTCCAGAGTCGAGCACAATACCG-3′. The experiment was performed on the LightCycler instrument with the FastStart DNA Master SYBR Green I kit (Roche Molecular Biochemicals, RNA from two independent ABA-induction experiments was analysed. Plasmids containing the RAB18 and ACT1 cDNA, respectively, were used as quantification standards (Frey et al., 2000).

Protoplast analysis

Isolation of protoplasts from rosette leaves of 3-week-old Arabidopsis plants and polyethyleneglycol mediated protoplast transfection were performed as described previously (Himmelbach et al., 2002). For transfection experiments, approximately 5 × 104 protoplasts (0.1 ml) were transfected with 15 μg DNA of the reporter plasmid (pRAB18::LUC or pRD29B::LUC, respectively) and 0.1–10 μg DNA of the effector plasmid. In addition, 5 μg of p35S::GUS plasmid was included in each transfection as a control. The analysis of ABI1- and abi1-regulated reporter expression in protoplasts was performed as previously described (Yang et al., 2006). For transient expression of GFP fusion protein in Arabidopsis protoplasts, 10 μg of the pBI221 p35S::GFP fusion plasmids mentioned above were introduced into protoplasts, and GFP localization was monitored 14–18 h after transfection.

Transgenic Arabidopsis plants and physiological analysis

To generate Arabidopsis plants overexpressing GFP fusions to ABI1, abi1 and abi1dNLS, the p35S::GUS cassette of binary vector pBI121 (Jefferson et al., 1987) was exchanged by the respective pBI221 GFP fusion cassettes mentioned above.

The transformation of Arabidopsis plants, ecotype RLD, using Agrobacterium tumefaciens strains GV3101pMP90 (for GFP fusions) and C58pGV3850 (for GUS fusions) was carried out as previously described (Ferrando et al., 2000; Meyer et al., 1994).

Analyses of germination, root growth and stomatal closure in Arabidopsis plants overexpressing either p35S::GUS fusions or p35S::GFP fusions were carried out as previously described (Himmelbach et al., 2002). For GFP analysis in transgenic plants, 6-day-old seedlings were examined using a confocal laser scanning microscope (Fluoview FV1000; Olympus, For inhibitor studies on whole Arabidopsis plants, transgenic seedlings were vacuum infiltrated for 5 min with either 50 μm MG132 or the corresponding inhibitor solvent (2% dimethyl sulfoxide), and were left in culture for 4 h before analysis by confocal laser scanning microscopy (Lagrange et al., 2003). The intensity of GFP fluorescence in guard and root cell nuclei was quantified using confocal microscope pictures and Simple PCI 5.2 software (Compix,

Immunochemical analysis

Western blot analyses of transiently expressed proteins were performed as previously described (Yang et al., 2006) by using anti-GFP and anti-c-myc antibodies (Santa Cruz Biotechnology, Protoplasts were exposed for 4 h to either 50 μm MG132 or 2% DMSO.

Statistical analysis

Data were analysed using the Mann–Whitney U-test and WinSTAT® software (R. Fitch Software, All experiments were repeated at least twice and yielded similar results.


The financial support of the Deutsche Forschungsgemeinschaft, EU Marie-Curie-Action ADONIS, and the Fonds der Chemischen Industrie is gratefully acknowledged. We thank Christoph Heidersberger for technical assistance, as well as Dr Monika Frey and Thomas Rauhut for help and advice on real-time PCR analysis. Dr Ram Yadav gave us valuable suggestions for the preparation of confocal microscopy samples. We are grateful to Dr Farhah Assaad and to Dr Alexander Christmann for their critical reading of the manuscript.