The abi1-1 mutation blocks ABA signaling downstream of cADPR action

Authors

  • Yan Wu,

    1. Laboratory of Plant Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA,
    Search for more papers by this author
    • Present address: Torrey Mesa Research Institute, Syngenta, 3115 Merryfield Row, San Diego, CA 92121, USA.

  • Juan Pablo Sanchez,

    1. Laboratory of Plant Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA,
    Search for more papers by this author
  • Luis Lopez-Molina,

    1. Laboratory of Plant Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA,
    2. Biology Department, City College, City University of New York, 138th Street and Convent Avenue New York, NY 10031, USA
    Search for more papers by this author
  • Axel Himmelbach,

    1. TU-München, Lehrstuhl für Botanik, Am Hochanger 4, D-85350, Freising, Germany, and
    Search for more papers by this author
  • Erwin Grill,

    1. TU-München, Lehrstuhl für Botanik, Am Hochanger 4, D-85350, Freising, Germany, and
    Search for more papers by this author
  • Nam-Hai Chua

    Corresponding author
    1. Laboratory of Plant Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA,
      For correspondence (fax +1 212 3278327; e-mail chua@rockvax.rockfeller.edu).
    Search for more papers by this author

For correspondence (fax +1 212 3278327; e-mail chua@rockvax.rockfeller.edu).

Summary

Arabidopsis thaliana abscisic acid insensitive 1-1 (abi1-1) is a dominant mutant that is insensitive to the inhibition of germination and growth by the plant hormone, abscisic acid (ABA). The mutation severely decreases the catalytic activity of the ABI1 type 2C protein phosphatase (PP2C). However, the site of action of the abi1-1/ABI1 in the ABA signal transduction pathway has not yet been determined. Using single cell assays, we showed that microinjecting mutant abi1-1 protein inhibited the activation of RD29A-GUS and KIN2-GUS in response to ABA, cyclic ADP-ribose (cADPR), and Ca2+. The inhibitory effect of the mutant protein, however, was reversed by co-microinjection of an excess amount of the ABI1 protein. In transgenic Arabidopsis plants, overexpression of abi1-1 rendered the plants insensitive to ABA during germination, whereas overexpression of ABI1 did not have any apparent effect. Moreover, transgenic plants overexpressing abi1-1 were blocked in the induction of ABA-responsive genes; however, overexpression of ABI1 did not affect gene expression. Taken together, our results demonstrate that abi1-1 is likely to be a dominant negative mutation and ABI1 likely acts downstream of cADPR in the ABA-signaling pathway. Our results on ABI1 overexpression in Arabidopsis are not compatible with a negative regulatory role of this phosphatase in ABA responses.

Introduction

The plant hormone abscisic acid (ABA) controls several important aspects of plant development including seed germination and adaptation to adverse environmental conditions. Acclimation of vegetative tissues to cold, drought or hypersalinity elicit physiological responses that require ABA (Chandler and Robertson, 1994; Leung and Giraudat, 1998; Merlot and Giraudat, 1997). Our previous pharmacologic experiments have implicated a cyclic nucleotide, cyclic ADP-ribose (cADPR), in the activation of ABA-responsive genes via an increase in cytosolic calcium. However, little is known regarding protein components that modulate the ABA-response pathway.

ABA-response mutants such as abscisic acid insensitive 3 (abi3), abi4, abi5 (Finkelstein, 1994; Finkelstein and Lynch, 2000; Finkelstein et al., 1998; Koornneef et al., 1984; Lopez-Molina and Chua, 2000; Lopez-Molina et al., 2001; Parcy et al., 1994), and era1 (Culter et al., 1996) are specific for seed maturation, and are not affected in ABA-responsive gene expression in vegetative tissues. By contrast, both the pleiotropic mutants, abi1-1 and abi2-1 (Koornneef et al., 1984), exhibit a germination-related phenotype and also are being impaired in stomatal closure and ABA-mediated gene expression in vegetative tissues (Allen et al., 1999; Chak et al., 2000; Grill and Himmelbach, 1998; Leung and Giraudat, 1998; Rodriguez, 1998). ABI1 and ABI2 encode proteins that share a similar structure with 86% sequence identity in the C-terminal domain containing the type 2C protein phosphatase (PP2C) core. These two PP2Cs have been proposed to function in seed development, stomatal closure, and in the regulation of gene expression (Allen et al., 1999; Bertauche et al., 1996; Leung et al., 1994, 1997; Meyer et al., 1994; Rodriguez et al., 1998a,b; Sheen, 1996, 1998). The N-terminal extensions of ABI1 and ABI2 are less conserved (48% identity). The N-terminal domain of ABI1 contains an EF-hand motif, which is a type of Ca2+-binding domain, but this EF-hand motif is absent from ABI2 (Leung et al., 1994, 1997; Meyer et al., 1994; Rodriguez et al., 1998a,b).

Both abi1-1 and abi2-1 are (semi-)dominant mutations resulting in an equivalent Gly-Asp amino acid substitution in the PP2C domain, and both mutations produce very similar phenotypes (Leung et al., 1994, 1997; Meyer et al., 1994). Because of the dominant nature of both the mutants, it is not clear whether the two ABI phosphatases are indeed involved in ABA signaling or whether the dominant mutations produce novel phenotypes (neomorphic) unrelated to the function of the wild-type (WT) allele. Recently, intragenic revertants of abi1-1 and abi2-1 were shown to be recessive to their WT alleles in enhancing sensitivity to ABA with respect to seed germination and stomatal closure (Gosti et al., 1999; Merlot et al., 2001). As these revertants have greatly reduced or no phosphatase activity, it was concluded that ABI1 and ABI2 PP2C act as negative regulators of the ABA-signaling pathway (Gosti et al., 1999; Merlot et al., 2001). However, it should be pointed out that these loss-of-function or reduction-of-function alleles of abi1-1 and abi2-1 still produce mutant proteins, albeit lacking protein phosphatase activity. Whether the mutant proteins are responsible for the ABA hypersensitive phenotype is not known. Given the dominant nature of abi1-1 mutation, it is plausible that blockage of ABA signaling is because of a protein-mediated sequestration effect rather than a reduction in phosphatase activity. To date, no mutant allele of abi1-1 or abi2-1 has been isolated in which the PP2C protein is not produced.

The site of action of abi1/ABI1 in ABA-signaling pathways has not been clearly defined. In guard cells, there is evidence that abi1-1 acts upstream of calcium release (Allen et al., 1999; Murata et al., 2001). It is not known whether the same site of action applies in the vegetative ABA response, and whether there might be more than one site of action for this phosphatase.

We wished to investigate the relationship of abi1-1 to the biochemical ABA-signaling pathway elucidated previously (Wu et al., 1997). In addition, we asked whether abi1-1 mutation was neomorphic in nature by asking whether the inhibitory effects of the abi1-1 mutant protein could be outcompeted by its WT counterpart in single cell assays. We examined whether: (i) the abi1-1 mutant protein could inhibit the expression of two ABA-responsive genes, RD29A (Yamaguchi-Shinozaki and Shinozaki, 1993) and KIN2 (Kurkela and Borg-Franck, 1992); (ii) an excess amount of ABI1 protein could rescue the inhibitory effect of abi1-1 mutant protein; and (iii) overexpression of the WT ABI1 or abi1-1 mutant protein affected ABA responses in transgenic plants.

Results

ABA-induced activity of RD29A-GUS and KIN2-GUS is blocked by abi1-1 protein

First, we tested the effects of the mutant abi1-1 and WT ABI1 proteins in the expression of ABA-responsive promoters in single cell assays using RD29A-GUS and KIN2-GUS as ABA-responsive reporter genes (Wu et al., 1997). Recombinant WT and mutant proteins were purified from Escherichia coli extracts, assayed to verify their phosphatase activity (Figure 1), and co-injected with a reporter gene into the subepidermal cells of tomato aurea hypocotyls. ABA-induced activation of RD29A-GUS or KIN2-GUS was effectively inhibited by the addition of as few as 500 molecules per cell of the abi1-1 protein, with complete inhibition occurring at about 3000 molecules per cell (Table 1). As a control, microinjection of abi1 protein alone in the absence of ABA had no effect. Bovine serum albumin (BSA) was used as the negative control in protein microinjection studies. At about 3000 molecules per cell, with or without ABA, BSA had no apparent effect on the expression of either RD29A-GUS or KIN2-GUS. Taken together, these results show that with respect to ABA induction of KIN2 and RD29A, the abi1-1 mutant protein was able to block the gene expression in tomato aurea cells as expected from the genetic data (Koornneef et al., 1984; Leung et al., 1994; Meyer et al., 1994; Rodriguez et al., 1998a). By contrast, the ABI1 WT protein, at approximately 3000 molecules per cell, had no apparent effect on ABA-induced RD29A-GUS and KIN2-GUS expression (Table 1).

Figure 1.

In vitro type 2C protein phosphatase (PP2C) activity of the recombinant proteins.

In vitro phosphatase activities of the ABI1 (filled squares), abi1-1 (filled circles), and NAP (filled triangles) proteins were assayed by incubation with 32P-labeled myelin basic protein at 30°C for 10 min. Phosphatase activity is expressed as picomoles of 32P released. Values shown are means of duplicate assays.

Table 1.  abi1-1 inhibits ABA-induced gene expression
Reporter genesCo-injectedExternally added ABAGUS efficiencya in percentage (n)
abi1-1ABI1BSA
  1. abi1-1 or ABI1 protein was co-injected with RD29A-GUS and KIN2-GUS into the subepidermal cells of tomato aurea hypocotyls, respectively. The number of injected molecules of abi1-1 and ABI1 proteins are indicated in the table. Bovine serum albumin (BSA; 3000 molecules) was used as a control. After injection, the hypocotyls were incubated in the medium with (+) or without (−) abscisic acid (ABA; 100 µm).

  2. n = number of injected cells.

  3. (+), co-injected; (−), not co-injected.

  4. a Using the constitutive 35S promoter (CaMV35S-GUS) as a control, we found that 5.0–11.1% of the injected cells expressed GUS activity. Therefore, we regard values that fall within the 5–12% range as the maximum that can be obtained using our microinjection system. No quantitative significance is attached to numbers within this range (Wu et al., 1997).

RD29A0 (218)
RD29A+7.1 (225)
RD29A5000 (223)
RD29A500+0.4 (243)
RD29A15000 (224)
RD29A1500+0.4 (229)
RD29A30000 (208)
RD29A3000+0 (445)
RD29A30000 (209)
RD29A3000+6.2 (227)
RD29A30000 (205)
RD29A3000+10.2 (214)
KIN20 (224)
KIN2+11.3 (222)
KIN25000 (242)
KIN2500+1.7 (298)
KIN215000 (231)
KIN21500+0.8 (252)
KIN23000 0 (221)
KIN23000+0.6 (439)
KIN230000 (215)
KIN23000+8.3 (264)
KIN230000 (242)
KIN23000+7.4 (257)

The observation that the abi1-1 protein could block ABA activation of RD29A and KIN2 in the microinjection assays was consistent with the result that ABA-induced RD29A and KIN2 gene expression was dramatically reduced in abi1-1 mutant plants as compared to WT plants (Hoth et al., 2002).

Overexpression of abi1-1 protein blocks ABA signal transduction in planta

We generated transgenic Arabidopsis thaliana plants expressing 35S-ABI1 or 35S-abi1-1 transgene. An RNA gel blot analysis showed that both the ABI1 and abi1-1 transgenes were constitutively expressed at levels 12–15-folds higher than the endogenous ABI1 in WT plants (Figure 2b). Six transgenic lines carrying the ABI1 and abi1-1 genes each were analyzed. The majority of the transgenic lines showed a similar level of expression by Northern blot analysis (data not shown). Two transgenic lines for each construct were advanced for further analysis. Extracts prepared from the transgenic lines were assayed for total phosphatase activity in the presence of inhibitors of PP1 and PP2A, and the levels were very similar. Figure 2(c) shows a sixfold difference in phosphatase activity between the ABI1 transgenic line in comparison with the abi1-1 transgenic line and WT. These results confirmed the activity of ABI1 and the reduced activity of the abi1-1 mutant protein. Germination assays in the presence of ABA showed that transgenic lines expressing the 35S-abi1-1 transgene were as insensitive to ABA as the original abi1-1 mutant. By contrast, overexpression of ABI1 did not affect the sensitivity to ABA (Figure 2a). All six transgenic lines (ABI1 and abi1-1) displayed similar germination phenotypes in the presence of ABA, and only representative results are shown in Figure 2(a). We noted that in seeds of some 35S-ABI1 transgenic lines, radicle emergence was observed, although the seedlings never developed further in the ABA medium. Expressions of KIN2, RD29A, and COR47 were induced to comparable levels by ABA in 35S-ABI1 transgenic plants as in WT plants, whereas no induced expression was detected in 35S-abi1-1 transgenic plants (Figure 2b).

Figure 2.

Germination and RNA gel blot analysis of wild-type (WT) and transgenic lines expressing 35S-ABI1 or 35S-abi1-1.

(a) Seeds of wild-type Landsdberg (open circles), transgenic 35S-ABI1 (filled squares), and transgenic 35S-abi1-1 (filled circles) were plated on ABA (3.5 µm) medium. The number of germinated seeds (with fully emerged radicle tip) at any given time was expressed as a percent of the total number of seeds plated (n = 100 seeds or more). Mean values ± SD are provided for three independent experiments.

(b) ABA induction of KIN2, RD29A, and COR47 genes in transgenic plants expressing 35S-ABI1 and 35S-abi1-1. Transgenic and control lines for each construct were grown for 2 weeks and the seedlings transferred to a hydroponic growth system (Aoyama and Chua, 1997). After 2 days, the medium was replaced with a fresh medium containing ABA (50 µm). Total RNA was isolated at 0 and 4 h, and RNA gel blots were hybridized with probes specific to KIN2, RD29a, COR47, ABI1, and 18S rDNA. WT Landsdberg plants were used as controls.

(c) In vitro type 2C protein phosphatase (PP2C) activity of protein extracts from WT plants and transgenic lines expressing 35S-ABI1 and 35S-abi1-1.

Plantlets were grown as previously described in (b), except that they were transferred to liquid medium after 3 weeks. Plants were collected and extracts were prepared as described in Experimental procedures. The phosphatase activity (PP2C) was assayed in vitro using a 32P-labeled myelin basic protein as a substrate. Protein extracts from ABI1 transgenic line (filled bars), abi1-1 transgenic line (open bars), and WT (hatched bars) were assayed. Phosphatase activity was expressed in picomoles of 32P released. For more details, see Experimental procedures. Values shown are mean ± SD of triplicate assays.

abi1-1 blocks ABA signaling downstream of cADPR

To define the site of action of abi1-1 with respect to the other components in the pathway, reporter gene activity was monitored upon co-microinjection of cADP-ribose (cADPR; pipette concentration, 160 µm) or Ca2+ (pipette concentration, 160 µm) with the abi1-1 protein. Table 2 shows that expression of the reporter constructs was sensitive to abi1-1, although a higher concentration of the mutant protein was needed to inhibit cADPR- or Ca2+-activated gene expression as compared to that required to block the gene expression induced by ABA. It is possible that higher than physiological concentrations of cADPR or Ca2+ were used, which may have caused an increase in signal amplitude. Irrespective of the agonist used, the negative control protein BSA had no effect on the induction of RD29A and KIN2 gene expression. These results indicate that with respect to ABA-responsive RD29A and KIN2 gene activation, at least one site of abi1-1 protein action is downstream of cADPR.

Table 2.  abi1-1 inhibits cADPR- and Ca2+-induced activity of RD29A and KIN2
Reporter genesCo-injectedGUS efficiency in percentage (n)
cADPRCa2+abi1-1BSA
  1. cADPR (160 µm, pipette concentration), CaCl2 (160 µm, pipette concentration), or abi1-1 protein (7500 molecules) was co-injected with RD29A-GUS and KIN2-GUS into the subepidermal cells of tomato aurea hypocotyls. Bovine serum albumin (BSA; 7500 molecules) was used as a control. After injection, the hypocotyls were incubated in the medium without abscisic acid (ABA).

  2. n = number of injected cells.

  3. (+), co-injected; (−), not co-injected.

RD29A+9.6 (228)
RD29A+75000.8 (461)
RD29A+6.3 (224)
RD29A+75000 (217)
RD29A+75005.4 (241)
RD29A+75006.0 (248)
KIN2+10.4 (212)
KIN2+75000 (232)
KIN2+8.3 (218)
KIN2+75000 (224)
KIN2+75005.7 (264)
KIN2+75006.3 (268)

abi1-1 is a dominant negative mutation

Arabidopsis abi1-1 is a dominant mutation, and no loss-of-function alleles have been isolated so far for this locus. There are two formal possibilities to explain the dominant nature of this mutation. Either the mutation is neomorphic and the corresponding WT protein (ABI1) is actually not involved in ABA signaling, or abi1-1 is a dominant negative mutation that interferes with the ABA-signaling function of its WT counterpart. To discriminate between these possibilities, we investigated whether the inhibitory effects of the abi1-1 mutant protein could be reversed by an excess amount of ABI1. Table 3 shows that this was indeed the case. Restoration of reporter gene activity was obtained when ABI1 was present in about two- to threefold excess of abi1-1, irrespective of the agonists used. We used BSA as a negative control and found no restoration of gene expression with this protein (Table 3). For example, a two- to threefold excess of ABI1 (about 20 000 molecules) was able to overcome the inhibitory effect of abi1-1 (about 7500 molecules) on cADPR-induced RD29A activation; however, the same amount of BSA (about 20 000 molecules) was not able to overcome the same inhibitory effect of abi1-1 (about 7500 molecules). Moreover, in control experiments, as much as 20 000 molecules per cell of ABI1 protein did not have any apparent effect on ABA-responsive gene expression. These results make it unlikely that abi1-1 is a neomorphic mutation; rather, it is consistently considered to be a dominant negative mutation that acts in the ABA-signaling pathway.

Table 3.  ABI1 can reverse the inhibitory effect of abi1
Reporter genesCo-injectedExternally added ABAGUS efficiency in percentage (n)
cADPRCa2+ABI1abi1-1BSA
  1. cADPR (160 µm, pipette concentration), CaCl2 (160 µm, pipette concentration), ABI1, or abi1-1 protein was co-injected with RD29A-GUS and KIN2-GUS, respectively, into the subepidermal cells of tomato aurea hypocotyls. Bovine serum albumin (BSA) was used as a control. The number of injected molecules for each protein is indicated in the table. After injection, the hypocotyls were incubated in the medium with (+) or without (−) ABA (100 µm).

  2. n = number of injected cells.

  3. (+), co-injected; (−), not co-injected.

RD29A+9.6 (208)
RD29A60000 (206)
RD29A6000+9.6 (219)
RD29A60003000+8.5 (213)
RD29A30006000+0 (231)
RD29A200000 (228)
RD29A+200006.6 (227)
RD29A+2000075007.4 (216)
RD29A+7500200001.0 (648)
RD29A+200006.0 (216)
RD29A+2000075008.2 (219)
RD29A+7500200001.0 (664)
KIN2+8.1 (211)
KIN260000 (214)
KIN26000+11.3 (221)
KIN260003000+5.8 (257)
KIN230006000+0 (220)
KIN2200000 (219)
KIN2+200006.8 (221)
KIN2+2000075005.9 (219)
KIN2+7500200000.9 (411)
KIN2+200005.6 (425)
KIN2+2000075006.5 (217)
KIN2+7500200000.2 (416)

PP2C activity is crucial in ABA-responsive gene expression

A null mutation of ABI1, ABI1D177A (NAP), was constructed by replacing Asp177 with Ala. This site was selected because it is highly conserved in all PP2Cs, and as revealed by the crystal structure (Das et al., 1996; Rodriguez, 1998), it constitutes the metal-binding core for Mg2+ or Mn2+, which is essential for PP2C activity. Recombinant ABI1D177A (NAP) was purified from E. coli extracts (Himmelbach et al., 2002) and assayed to verify that mutation destroys the phosphatase activity (see Figure 1). Table 4 shows that NAP blocked gene activation of RD29A and KIN2 induced by ABA, cADPR, or Ca2+. However, this inhibition was reversed by co-injecting an excess amount of the ABI1 protein. Not surprisingly, NAP was not able to rescue the inhibitory effect of the abi1-1 protein (data not shown).

Table 4.  ABI1 can rescue the inhibitory effect of NAP (ABI1D177A) on ABA-responsive gene activity
ReporterCo-injectedExternally added ABAGUS efficiency in percentage (n)
cADPRCa2+ABI1NAP
  1. cADPR (160 µm, pipette concentration), CaCl2 (160 µm, pipette concentration), ABI1, or NAP protein was co-injected with RD29A-GUS and KIN2-GUS, respectively, into the subepidermal cells of tomato aurea hypocotyls. The number of injected molecules for each protein is indicated in the table. After injection, the hypocotyls were incubated in the medium with (+) or without (−) abscisic acid (ABA; 100 µm). n = number of injected cells.

  2. (+), co-injected; (−), not co-injected.

RD29A0 (205)
RD29A+10.4 (226)
RD29A3000+1.4 (218)
RD29A600030000 (234)
RD29A60003000+10.2 (206)
RD29A+7.6 (225)
RD29A+50000 (217)
RD29A+1000050006.5 (214)
RD29A+9.6 (219)
RD29A+50000 (227)
RD29A+1000050005.4 (226)
KIN20 (223)
KIN2+8.0 (232)
KIN23000+0.9 (212)
KIN2600030000 (219)
KIN260003000+7.9 (215)
KIN2+6.4 (236)
KIN2+50000 (212)
KIN2+1000050007.4 (219)
KIN2+9.8 (205)
KIN2+50000 (214)
KIN2+1000050008.5 (247)

Discussion

Arabidopsis abi1-1 blocks ABA signaling

We have previously used a pharmacological approach to show the involvement of cADPR and calcium as downstream intermediates in the primary ABA-signaling cascade (Wu et al., 1997). The single-cell system we used (Wu et al., 1997) provided an assay to locate the site of action of candidate-signaling components with respect to cADPR and calcium. Here, we demonstrate that microinjection of the abi1-1 mutant protein blocked ABA-, cADPR-, and calcium-induced RD29A and KIN2 gene expression (Tables 1 and 2). These results indicate that abi1-1 acts downstream of cADPR. Studies in stomata system have shown that cADPR regulates ABA signaling through calcium release (Leckie et al., 1998). As cADPR regulates calcium release, we hypothesized that abi1-1 acts downstream of calcium release in ABA-induced gene expression. Nonetheless, our results do not rule out the possibility that ABI1 may also act upstream of the calcium signal. Studies done in guard cells show that cytoplasmic calcium increase in response to ABA was reduced by 2.5-folds in the abi1-1 mutant, which suggested that abi1-1 acts upstream of the calcium release (Allen et al., 1999). Another report from the same group (Murata et al., 2001) also showed that in guard cells, abi1-1 impairs ABA signaling in a step between ABA perception and ROS production, which is upstream of plasma membrane Ca2+ channel activation. Two possible explanations may account for the differences between these results and those reported here. First, abi1-1 may act both upstream and downstream of calcium release in different ABA-response pathways, but only the downstream site was revealed in our experiments. In addition, it has been suggested that cADPR may be involved in the regulation of calcium release from internal organelles (Allen et al., 1995; Leckie et al., 1998). There are two reports (Allen et al., 1999; Murata et al., 2001) indicating that abi1-1 affects calcium release at the plasma membrane level, thus suggesting that abi1-1 may act at different steps in ABA signal transduction. Secondly, the ABA-signaling pathway in guard cells, which represents a highly specialized cell type, may be different from hypocotyl cells (present work).

abi1-1 is a dominant negative mutation

Because no loss-of-function alleles of abi1-1 have been recovered (Gosti et al., 1999; Merlot et al., 2001), it is important to ask whether the impairment of ABA signaling associated with abi1-1 is the result of a neomorphic mutation. We have demonstrated that the abi1-1 mutant protein was able to block ABA-, cADPR-, and Ca2+-induced activation of KIN2-GUS and RD29A-GUS. The inhibitory effect of the abi1-1 protein, however, could be reversed by co-injecting excess amounts (two- to threefolds) of WT ABI1 protein, suggesting that abi1-1 is not neomorphic. In addition, overexpression of abi1-1 in transgenic plants resulted in ABA insensitivity and a block in the expression of ABA-responsive genes. Rather, our results were consistent with the fact that the abi1-1 mutant protein acts as a dominant negative regulator that disrupts ABA signaling probably by forming non-productive complexes with signaling intermediates.

Human PP2C contains an aspartic acid at residue 60 (D60), which is crucial for the formation of the complex with Mg2+ (Das et al., 1996). The equivalent amino acid in ABI1 is D177. Leube et al. (1998) reported that D177, a mutation of ABI1, affected Mg2+-dependent PP2C catalytic activity (Leube et al., 1998). Here, we show that besides abi1-1, the ABI1D177A (NAP) mutant protein can also act as a dominant negative regulator of ABA signaling in single cells. Taken together, these results suggest that the two catalytically inactive forms of ABI1 likely interact non-productively with ABA-signaling components, thereby blocking ABA responses.

Two recent papers (Himmelbach et al., 2002; Hoth et al., 2002) provide insight into the function of ABI1. First, Himmelbach et al. isolated a transcription factor (ATHB6) that interacts specifically with ABI1. Overexpression of ATHB6 specifically affects the ABA pathway. This report shows that the abi1-1 protein binds to ATHB6 with a lower efficiency (8–9%) than ABI1, its WT counterpart. One possible explanation for the dominant negative effect of abi1-1 is that an excess of abi1-1 over ABI1 (10-folds or greater) may outcompete ABI1 for binding to ATHB6, thereby blocking the ABA pathway. The specificity of ATHB6 to ABA signaling suggests that an excess of ABI1 would reactivate the pathway by dissociating non-productive abi1-1/ATHB6 complexes, thereby overcoming the dominant negative phenotype. Taken together, our results and those of Himmelbach et al. (2002) suggest that abi1-1 is not a neomorphic mutation. Secondly, Hoth et al. (2002) have shown that about 90% of the ABA-responsive genes are affected by the abi1-1 mutation.

Is ABI1 a negative regulator of ABA-response pathways?

Gosti et al. (1999) isolated seven alleles of intragenic revertants of abi1-1, which turned out to be more sensitive than WT to ABA with respect to seed germination and seedling root growth. As these revertant alleles were recessive and the revertant proteins lacked any detectable PP2C phosphatase activity, it was logical for the authors to conclude that ABI1 is a negative regulator of ABA responses. It is a known fact that ABA treatment induces a 5–10-fold increase in ABI1 transcript levels. Gosti et al. (1999) proposed two models to explain their observations. The first model assumes that the perception of ABA signal leads to the activation of physiological responses and to an increase in ABI1 transcript and protein levels, which in turn attenuate the ABA signal transduction pathway. If this was the case, overexpression of ABI1 (10–12-folds) would block ABA-induced gene expression, but this was not seen in our 35S-ABI1 transgenic plants. The second model suggests that ABA inhibits the catalytic activity of ABI1. In the absence of ABA, ABI1 represses ABA-induced gene expression. In the presence of the hormone on the other hand, the ABI1 phosphatase activity is inhibited, therefore, allowing ABA responses to occur. However, Merlot et al. (2001) subsequently showed that the total PP2C phosphatase activity is increased by only 30% upon ABA treatment.

At present, it is difficult to reconcile our results with those of Gosti et al. (1999). Additional experiments, for example, using RNAi to suppress ABI1 expression, are needed to finally identify the real role of this PP2C phosphatase in ABA signaling.

Experimental procedures

Microinjection

The injection solutions and techniques were as described previously (Wu et al., 1996, 1997). Seven- to ten-day-old etiolated seedlings of the tomato mutant aurea were used for microinjection. The ABA-responsive promoters of RD29A and KIN2 were fused to the β-glucuronidase (GUS) coding sequence, and the fusion genes were used as the reporter constructs (Wu et al., 1997). After injection, the seedlings were incubated in KNOP medium for 2 days (Wu et al., 1997) before analysis of GUS activity (Jefferson, 1987). Unless otherwise indicated, the pipette concentrations of cyclic ADP-ribose (Molecular Probes, Inc., Eugene, Oregon, USA) and CaCl2 were both 160 µm. ABA was added to the incubation medium at a final concentration of 100 µm.

Expression and purification of the ABI proteins

Sequence-verified cDNA clones encoding ABI1, abi1-1, and NAP with a C-terminal protein kinase A recognition tag (RRASV) and a His6-tag (pQE60, Qiagen, Valencia, CA, USA) were expressed in E. coli strain M15. Recombinant proteins were affinity-purified under native conditions on Ni2+-NTA resin (Qiagen) as indicated by the supplier. Purified proteins were dialyzed against 50 mm NaH2PO4, 200 mm NaCl, 0.1% Tween-20, 10 mmβ-mercaptoethanol, and 30% glycerol (pH 8.0).

Phosphatase assay

Proteins were also assayed using the protein serine/threonine phosphatase (PSP) assay system (BioLabs, Berely, MA) according to the manufacturer's specifications. Extracts were prepared as described by Sanchez and Chua (2001). Phosphatase assays were performed in the presence of 5 µm okadaic acid (an inhibitor of PP1 and PP2A) and in the presence or absence of 20 mm magnesium acetate. The PP2C activity was calculated as the Mg-dependent phosphatase activity (McKintosh, 1993).

RNA blot analysis of ABI1 and abi1-1 transgenic plants

ABI1 cDNA was kindly provided by Dr Fumiaki Katagiri and KIN2 cDNAs were obtained as EST clones (ID# 103216T7, ATT53026, Arabidopsis Biological Resource Center (ABRC), Ohio State University). The 18S rDNA probe was obtained from Dr Taku Takahashi. The authenticity of the probes was verified by sequencing. The abi1-1 mutation was obtained by replacing the GGC (GLY) codon by GAC (ASP) using the quick-change site-directed mutagenesis kit (Stratagene, La Jolla, CA). The ABI1 and abi1-1 cDNAs were placed under the control of the constitutive CaMV35S promoter in the binary vector pBA002 (Hajdukiewicz et al., 1994). The constructs were verified by restriction analysis and sequencing. All constructs were introduced into the Agrobacterium tumefaciens strain ABI (Aoyama and Chua, 1997).

Arabidopsis thaliana plants (Landsberg) were transformed with Agrobacterium by vacuum infiltration (Bent et al., 1994). Six to seven T1 lines were selected on plates containing 20 µg ml−1 BASTA (Sigma, St Louis, MO), and confirmed by PCR analysis. Two-week-old transgenic plants were transferred to a hydroponic system (Sanchez and Chua, 2001). After 2 days, the medium was replaced with one containing ABA (50 µm). Samples were harvested at the designated time points, washed, and finally frozen in liquid nitrogen. Total RNA was isolated using the Qiagen RNA purification kit (Qiagen, Valencia, CA). RNA gel blot analysis was performed as described previously (Ausubel et al., 1994). Each lane contained 10 µg of total RNA. ABI1, KIN2, RD29a, COR47, and 18S rDNA fragments were obtained by PCR amplification with Pfu polymerase as described by the manufacturer. Fragments were purified using the Qiaquick Gel Extraction protocol (Qiagen, Valencia, CA). All DNA fragments were labeled with 32P-dCTP and 32P-dATP by random priming (Amersham, Arlington Heigths, IL). Hybridization signals were quantified using the Phosphoimager STORM system (Molecular Dynamics, Amersham Biosciences Corp., Piscataway, NJ, USA), and the data were analyzed with the image quant v1.1 program.

Acknowledgements

We extend our thanks to Dr Funiaki Katagiri for the ABI1 cDNA clone. This work was supported in part by DOE grant DE-FG02–94ERQ20142 to N.-H. C.

Ancillary