Triple loss of function of protein phosphatases type 2C leads to partial constitutive response to endogenous abscisic acid.

The phytohormone abscisic acid (ABA) is a key regulator of plant growth and development as well as plant responses to situations of decreased water availability. Protein phosphatases type 2C (PP2Cs) from group A, which includes the ABI1/HAB1 and PP2CA branches, are key negative regulators of ABA signaling. Specifically, HAB1, ABI1, ABI2, and PP2CA have been shown to affect both seed and vegetative responses to ABA. To further understand their contribution to ABA signaling and to unravel possible genetic interactions and functional redundancy among them, we have generated different combinations of double and triple mutants impaired in these PP2Cs. Interestingly, hab1-1pp2ca-1 and abi1-2pp2ca-1 double mutants showed reduced water loss and enhanced resistance to drought stress, which further supports the role of PP2CA in vegetative responses to ABA. Two triple hab1-1abi1-2abi2-2 and hab1-1abi1-2pp2ca-1 mutants were generated, which showed an extreme response to exogenous ABA, impaired growth, and partial constitutive response to endogenous ABA. Thus, transcriptomic analysis revealed a partial up-regulation/down-regulation of a subset of ABA-responsive genes in both triple mutants in the absence of exogenous ABA. Comparison of ABA responses in the different pp2c mutants showed that a progressive increase in ABA sensitivity could be obtained through combined inactivation of these PP2Cs. These results indicate that ABA response is finely tuned by the integrated action of these genes, which is required to prevent a constitutive response to endogenous ABA that might have a deleterious effect on growth and development in the absence of environmental stress.

The phytohormone abscisic acid (ABA) is a key regulator of plant growth and development as well as plant responses to situations of decreased water availability. Protein phosphatases type 2C (PP2Cs) from group A, which includes the ABI1/ HAB1 and PP2CA branches, are key negative regulators of ABA signaling. Specifically, HAB1, ABI1, ABI2, and PP2CA have been shown to affect both seed and vegetative responses to ABA. To further understand their contribution to ABA signaling and to unravel possible genetic interactions and functional redundancy among them, we have generated different combinations of double and triple mutants impaired in these PP2Cs. Interestingly, hab1-1pp2ca-1 and abi1-2pp2ca-1 double mutants showed reduced water loss and enhanced resistance to drought stress, which further supports the role of PP2CA in vegetative responses to ABA. Two triple hab1-1abi1-2abi2-2 and hab1-1abi1-2pp2ca-1 mutants were generated, which showed an extreme response to exogenous ABA, impaired growth, and partial constitutive response to endogenous ABA. Thus, transcriptomic analysis revealed a partial up-regulation/down-regulation of a subset of ABA-responsive genes in both triple mutants in the absence of exogenous ABA. Comparison of ABA responses in the different pp2c mutants showed that a progressive increase in ABA sensitivity could be obtained through combined inactivation of these PP2Cs. These results indicate that ABA response is finely tuned by the integrated action of these genes, which is required to prevent a constitutive response to endogenous ABA that might have a deleterious effect on growth and development in the absence of environmental stress.
On the other hand, protein phosphatases type 2A (PP2As) and type 2C (PP2Cs) are involved in the regulation of ABA signaling. Disruption of the PP2A regulatory subunit RCN1 confers ABA insensitivity in Arabidopsis, which suggests that RCN1 functions as a positive transducer of ABA signaling (Kwak et al., 2002). Recent results have reported that a catalytic subunit of PP2A functions as a negative regulator of ABA signaling (Pernas et al., 2007). According to the classification of Schweighofer et al. (2004), 76 Arabidopsis genes qualify as members of the big PP2C family. Among them, group A contains most of the PP2Cs that are associated with ABA signaling, which are separated in two subgroups (i.e. the ABI1 and PP2CA branches). The pioneering evidence for the involvement of PP2Cs in the regulation of ABA signaling process was provided by the identification of the ABA-insensitive abi1-1 and abi2-1 mutants and the cloning of the corresponding loci (Koornneef et al., 1984;Leung et al., 1994Leung et al., , 1997Meyer et al., 1994;Rodriguez, et al., 1998). Currently, genetic evidence indicates that at least six Arabidopsis PP2Cs, namely, ABI1, ABI2, PP2CA/AHG3, AHG1, HAB1, and HAB2, act as negative regulators of ABA signaling (Gosti et al., 1999;Merlot et al., 2001;Tahtiharju and Palva, 2001;Gonzalez-Garcia et al., 2003;Leonhardt et al., 2004;Saez et al., 2004Saez et al., , 2006Kuhn et al., 2006;Yoshida et al., 2006;Nishimura et al., 2007). AHG1 and PP2CA/ AHG3 appear to play an essential role for ABA signaling in germination and postgermination growth (Kuhn et al., 2006;Yoshida et al., 2006;Nishimura et al., 2007), but the ahg1-1 mutant has no ABA-related phenotype in adult plants (Nishimura et al., 2007). The analysis of a double ahg1-1ahg3-1 mutant suggests that AHG1 has specific functions in seed development and germination, shared partially with PP2CA/AHG3 (Nishimura et al., 2007). Kuhn et al. (2006) showed that PP2CA/AHG3 plays an essential role for ABA signaling both in seed and vegetative tissue, as the pp2ca mutant alleles showed ABA hypersensitivity in germination, growth, and stomatal closure assays. Conversely, 35S:PP2CA expression caused ABA insensitivity in seed germination and ABA-induced stomatal closure assays (Kuhn et al., 2006). Previous work of Sheen (1998) showed that PP2CA can block ABAmediated gene induction when transiently overexpressed in protoplasts. These results suggest that PP2CA function might be related to other PP2Cs from the ABI1 branch; however, genetic interactions among both branches have not been investigated.
Single reduction/loss-of-function alleles from ABI1, ABI2, and HAB1 produced phenotypic effects on ABA signaling to a different extent and it was apparent from double mutant analyses that some functional redundancy occurs among them (Merlot et al., 2001;Saez et al., 2006). For instance, inactivation of both HAB1 and ABI1 led to a stronger response to ABA than that found in either hab1-1 or abi1-2 monogenic mutants. A similar trend was obtained by combination of the recessive abi1-1R4 and abi2-1R1 alleles (Merlot et al., 2001). Combined inactivation of close members of a gene family is usually required to unravel possible functional genetic redundancy and to establish a functional hierarchy among them. This fact has been particularly evident in hormonal signaling pathways (Hua and Meyerowitz, 1998;Kwak et al., 2003;Higuchi et al., 2004;Achard et al., 2006;Iuchi et al., 2007).
In this article, we have combined loss-of-function mutations in the ABI1, ABI2, HAB1, and PP2CA genes to unravel their contributions to ABA signaling. The hab1-1, abi1-2, and pp2ca-1 alleles have been described previously (Leonhardt et al., 2004;Saez et al., 2004Saez et al., , 2006Kuhn et al., 2006) and in this article we have identified a loss-of-function allele for ABI2, named abi2-2. Different combinations of mutations were generated (Table I) and, as a result, two triple mutants, hab1-1abi1-2abi2-2 and hab1-1abi1-2pp2ca-1, were generated, which showed an extreme response to exogenous ABA and a partial constitutive response to endogenous ABA. Additionally, hab1-1pp2ca-1 and abi1-2 pp2ca-1 double mutants showed stronger responses to ABA than single parental mutants, which reveals a genetic interaction between both branches of PP2Cs for the regulation of ABA signaling.

RESULTS
Generation and Analysis of Double hab1-1pp2ca-1 and abi1-2pp2ca-1 Mutants Seeds of pp2ca-1 are strongly ABA-hypersensitive, whereas pp2ca-1 seedlings show only a moderated enhanced sensitivity to ABA-mediated inhibition of root growth and loss of fresh weight of detached rosette leaves in pp2ca-1 was similar to wild type (Kuhn et al., 2006). We wondered whether combination of the pp2ca-1 mutation with either hab1-1 or abi1-2 might reinforce ABA responses as it was described previously for the double hab1-1abi1-2 mutant (Saez et al., 2006). Therefore, the corresponding double mutants were generated and their ABA response analyzed. ABA-mediated inhibition of seed germination was stronger in the pp2ca-1 single mutant than in hab1-1 and abi1-2 single mutants (Fig. 1A), which is in agreement with previous results from Yoshida et al. (2006). Interestingly, both double hab1-1pp2ca-1 and Table I. List of T-DNA mutants used in this work The hab1-1, abi1-2, and pp2ca-1 single and double hab1-1abi1-2 mutants have been described previously (Saez et al., 2004(Saez et al., , 2006Kuhn et al., 2006). The abi2-2 mutant was obtained from the NASC. Different crosses were performed to generate the double and triple mutants mentioned in the table.

Single Mutants
Double Mutants Triple Mutants hab1-1 SALK2104 hab1-1abi1-2 hab1-1abi1-2pp2ca-1 abi1-2 SALK72009 abi1-2abi2-2 hab1-1abi1-2abi2-2 pp2ca-1 SALK28132 hab1-1pp2ca-1 abi2-2 SALK15166 abi1-2pp2ca-1 abi1-2pp2ca-1 mutants showed enhanced sensitivity to ABA-mediated inhibition of seed germination and growth inhibition than single parental mutants ( Fig.  1, A and B). Likewise, short-term water loss assays showed that combined inactivation of either PP2CA and HAB1 or PP2CA and ABI1 resulted in a phenotype of reduced water loss compared to Columbia wild type (Col) or single parental mutants (Fig. 1C). Additionally, water-loss data were obtained under greenhouse conditions after exposing 21-d-old plants to drought stress by stopping irrigation and minimizing soil evaporation. Figure 1D shows that, after 8 d, the double mutants showed reduced water loss as compared to wild type. In agreement with these data, the double mutants did not show symptoms of wilting and they had turgid green rosette leaves, whereas wild-type plants wilted and rosette leaves yellowed (Fig. 1E).

Generation and Analysis of Double abi1-2abi2-2 Mutants
A T-DNA mutant of ABI2 was identified in the Salk collection, corresponding to donor stock number SALK_015166, and it was named abi2-2. Sequencing of the T-DNA flanking region in abi2-2 showed that the insertion was localized 19 nucleotides upstream of the TGA stop codon. The T-DNA insertion severely impaired ABI2 expression, based on quantitative realtime (qRT)-PCR analyses (see Fig. 3); however, the Figure 1. ABA-hypersensitive germination and growth inhibition of hab1-1, abi1-2, pp2ca-1, and double hab1-1abi1-2, hab1-1pp2ca-1, abi1-2pp2ca-1 mutants compared to wild type. A, Percentage of seeds that germinated and developed green cotyledons in the presence of the indicated concentrations of ABA. Approximately 200 seeds of each genotype were sowed on each plate and scored 10 d later. The assays were done in the period encompassed between 1 to 2 months after seed harvesting. Values are averages 6 SD for three independent experiments (not indicated in the figure); P , 0.05 (Student's t test) when comparing data from each genotype and wild type in ABA medium. B, Percentage of fresh weight from the different mutants as compared to wild type. The percentage was calculated with respect to the fresh weight of wild type in Murashige and Skoog medium either lacking or containing 10 mM ABA. Fresh weight of wild type was reduced by 39% in plates supplemented with ABA as compared to medium lacking ABA. Values are averages from three independent experiments 6 SD (n = 15). * indicates P , 0.01 (Student's t test) when comparing data from each genotype and wild type in the same assay conditions. C, Reduced water loss measured in detached leaves of double hab1-1abi1-2, hab1-1pp2ca-1, abi1-2pp2ca-1 mutants as compared to wild type or single parental mutants. Values are averages from two independent experiments (n = 5). SD values were lower than 8%. D, Quantification of water loss in 30-d-old plants after 8 d without watering. Values are averages from two independent experiments 6 SD (n = 10). * as in B. FW, Fresh weight. E, Enhanced drought resistance of the indicated double mutants as compared to wild type. Photograph was taken 10 d after withholding water. Shoot was cut to better show the effect of drought on rosette leaves. abi2-2 allele showed wild-type responses in ABAmediated inhibition of seed germination and growth assays, as well as ABA-induced stomatal closing (Fig.  2). In contrast, the double abi1-2abi2-2 mutant showed higher sensitivity to ABA-mediated inhibition of germination and growth by ABA than single parental mutants (Fig. 2, A and B). Additionally, it showed enhanced ABA-induced stomatal closing (Fig. 2C), reduced water loss (Fig. 1D), and enhanced resistance to drought stress (Fig. 1E) as compared to wild-type plants.
Generation and Analysis of ABA Response in Triple hab1-1abi1-2pp2ca-1 and hab1-1abi1-2abi2-2 Mutants To further study the effect on ABA signaling of the combined inactivation of negative regulators of the pathway, we generated two triple mutants impaired in three of the above-described PP2Cs. Thus, hab1-1abi1-2pp2ca-1 and hab1-1abi1-2abi2-2 triple mutants were generated and qRT-PCR analyses confirmed that expression of HAB1, ABI1, and either PP2CA or ABI2, respectively, was severely impaired (Fig. 3A). The phenotype of the triple mutants in the absence and presence of exogenous ABA was analyzed. Interestingly, both mutants showed impaired growth in the absence of exogenous ABA, either under in vitro conditions or when plants were grown in soil under greenhouse conditions (Fig. 3B). The impaired growth of these mutants suggests that combined inactivation of three negative regulators of ABA signaling might lead to a partial constitutive response to endogenous ABA levels. For instance, roots of these mutants were shorter than those of wild type and root growth was extremely hypersensitive to ABA-mediated inhibition of growth (Fig. 3C). Indeed, root growth of the triple mutants in the absence of exogenous ABA was approximately 60% to 65% of wild type. This decrease was very similar to that obtained in wild-type seedlings grown in 10 mM ABA (Fig. 3C). Additionally, a comparison of root growth for different ABA-hypersensitive mutants reveals a progressive increase in their sensitivity to ABA-mediated inhibition of growth ( Fig. 3C), which suggests that PP2Cs act in concert to modulate root growth sensitivity to ABA.
Germination was slower in triple mutants than in wild type (Fig. 4A) and, indeed, the germination percentage of triple hab1-1abi1-2pp2ca-1 mutants was 40% of wild type 80 h after sowing (Fig. 4A). Seeds of the triple mutants were hypersensitive to very low concentrations (nM) of ABA in germination assays (Fig.  4B). Thus, the inhibitory ABA concentration to achieve 50% inhibition (ABA-IC 50 ) of seed germination was 10 and 40 nM for the triple hab1-1abi1-2pp2ca-1 and hab1-1abi1-2abi2-2 mutants, respectively, whereas it was 380 nM for wild-type seeds. Interestingly, the germination percentage of the triple hab1-1abi1-2pp2ca-1 mutant in the absence of exogenous ABA was similar to that found in wild-type seeds germinated in 300 nM ABA. Double mutants that contain the pp2ca-1 allele were more sensitive to ABA-mediated inhibition of germination than hab1-1abi1-2. Thus, the double hab1-1abi1-2 mutant had an ABA-IC 50 of 120 nM, whereas the ABA-IC 50 for the double hab1-1pp2ca-1 and abi1-2pp2ca-1 mutants was 50 and 70 nM, respectively. Early seedling growth was more inhibited by ABA in the triple mutants than in the double mutants, and double . ABA-hypersensitive response of double abi1-2abi2-2 mutant compared to wild type and single parental mutants. A, Percentage of seeds that germinated and developed green cotyledons in the presence of the indicated concentrations of ABA. Approximately 200 seeds of each genotype were sowed on each plate and scored 10 d later. The assays were done in the period encompassed between 1 to 2 months after seed harvesting. Values are averages 6 SD for three independent experiments; *, P , 0.05 (Student's t test) when comparing data from each genotype and wild type in the same assay conditions. B, Percentage of fresh weight from the different mutants as compared to wild type. The percentage was calculated with respect to the fresh weight of wild type in Murashige and Skoog medium either lacking or containing 10 mM ABA. Fresh weight of wild type was reduced by 22% in plates supplemented with ABA as compared to medium lacking ABA. Values are averages from three independent experiments 6 SD (n = 15); *, P , 0.01 (Student's t test) when comparing data from each genotype and wild type in the same assay conditions. C, ABA-hypersensitive stomatal closing in double abi1-2abi2-2 mutant. Stomatal apertures were measured 2 h and 30 min after addition of 0.01 or 0.1 mM ABA. Data represent average aperture ratio (width/length) of three independent experiments 6 SE (n = 30-40 stomata per experiment); *, P , 0.01 (Student's t test) when comparing data from the mutant and wild type in the same assay conditions. mutants containing the pp2ca-1 allele were more ABAhypersensitive than hab1-1abi1-2 or single pp2ca-1 mutants ( Fig. 4C; Supplemental Fig. S1). Therefore, taking into account the sensitivity to ABA-mediated inhibition of germination and a dose-response (25-500 nM ABA) analysis of ABA-mediated inhibition of early growth (Supplemental Fig. S1), the following hierarchy of ABA hypersensitivity can be established: hab1- Finally, measurements of both stomatal aperture and aperture ratio (width/length) reveal that stomata of the triple hab1-1abi1-2abi2-2 mutant were more closed than wild type in the absence of exogenous ABA (Fig. 5, A and B). Stomatal aperture measurements also showed that triple hab1-1abi1-2abi2-2 mutant stomata were extremely hypersensitive to ABA-induced stomatal closing in the range of 10 to 100 nM ABA (Fig. 5). Interestingly, stomatal apertures in the triple hab1-1abi1-2abi2-2 mutant in the absence of Figure 3. Growth impairment and ABA-hypersensitive growth inhibition of triple hab1-1abi1-2abi2-2 (triple abi2-2) and hab1-1abi1-2pp2ca-1 (triple pp2ca-1) mutants as compared to wild type. A, qRT-PCR analysis of HAB1, PP2CA, ABI1, and ABI2 transcript levels in wild type and triple hab1-1abi1-2abi2-2 and hab1-1abi1-2pp2ca-1 mutants. Gene expression was analyzed in mRNAs extracted from 14-d-old seedlings treated with 10 mM ABA for 3 h. Data are averages 6 SD from three independent experiments. B, Reduced growth in both Murashige and Skoog (MS) medium and soil of triple mutants compared to wild type. The photographs were taken after 3 weeks of plant growth in Murashige and Skoog medium or soil. The graphics show the quantification of fresh weight (n = 30) and leaf area (n = 10) in triple mutants with respect to wild type. Data are averages 6 SD from two independent experiments. *, P , 0.01 (Student's t test) when comparing data from the mutant and wild type in the same assay conditions. C, ABAhypersensitive root growth inhibition of double and triple mutants compared to wild type and single parental mutants. Representative seedlings were selected and a photograph was taken after 7 d of growth in medium lacking or supplemented with 10 mM ABA. Data are averages 6 SD from three independent experiments (n = 15). *, P , 0.01 (Student's t test) when comparing data from the mutant and wild type in the same assay conditions. [See online article for color version of this figure.] exogenous ABA were similar to those found in wild type treated with 10 nM ABA. Additionally, we measured stomatal conductance in the absence of exogenous ABA for both triple mutants. As a result, we found decreased stomatal conductance for water vapor (g s ) and decreased transpiration (E) in both triple mutants as compared to wild type (Fig. 5,C and D). This result indicates that, in the absence of exogenous ABA, stomata of triple mutants have a lower aperture than wild type.

Transcriptomic Profiling Suggests That Triple Mutants Show a Constitutive Response to Endogenous ABA
The above data suggested that the phenotype of triple hab1-1abi1-2pp2ca-1 and hab1-1abi1-2abi2-2 mutants might reflect a partial constitutive response to endogenous ABA levels. To challenge this hypothesis, we have compared transcriptomic profiles of wild type and triple mutants using whole-genome long-oligonucleotide microarrays. First, we have identified the total number of genes up-regulated (.2-fold) and down-regulated (,0.5) in wild type by treatment with 10 mM ABA for 3 h (false discovery rate P , 0.05), which qualify as ABA-responsive genes according to the ratio of expression Col + 10 mM ABA/Col no exogenous ABA (Fig. 6A). In addition, we have compared whole-genome expression of triple mutants with respect to wild type, in the absence of exogenous ABA (ratio of expression for triple no exogenous ABA/Col no exogenous ABA; Fig. 6A). The number of genes that (1) showed enhanced or diminished expression in the triple mutants with respect to the wild type, in the absence of exogenous ABA and (2) were ABA-responsive are indicated in Figure 6A and a complete list of these genes is provided as Supplemental Table S1. Thus, 194 or 266 ABA-responsive genes were up-regulated in the triple abi2-2 or pp2ca-1 mutants, respectively, in the absence of exogenous ABA. Conversely, 247 and 188 ABA-responsive genes were down-regulated in the triple abi2-2 or pp2ca-1 mutants, respectively, in the absence of exogenous ABA. Interestingly, 133 or 140 ABA-responsive genes were up-regulated or down-regulated, respectively, both in triple abi2-2 and pp2ca-1 mutants in the absence of exogenous ABA with respect to wild type. Independent confirmation of these results was obtained by qRT-PCR for two genes, RAB18 and RD29B, which are up-regulated by ABA treatment (Fig. 6B). In the absence of exogenous ABA, these gene markers showed a 5-to 7-fold higher expression in the triple mutants than in wild type. Interestingly, expression of these genes in the triple mutants (in the absence of exogenous ABA) was approximately 50% to 60% of that found in wild type treated with 10 mM ABA. Finally, both triple mutants showed enhanced up-regulation by ABA of RAB18 and RD29B, as their expression level upon ABA treatment was between 2-and 8-fold higher than in wild type (Fig. 6B). . ABA-hypersensitive germination and early seedling growth inhibition of different mutants compared to wild type. A, Seed germination time course of triple hab1-1abi1-2abi2-2 (triple abi2-2) and hab1-1abi1-2pp2ca-1 (triple pp2ca-1) mutants compared to wild type. Data are averages 6 SD from three independent experiments. B, Percentage of seeds that germinated and developed green cotyledons in the presence of the indicated concentrations of ABA. Approximately 200 seeds of each genotype were sowed on each plate and scored 5 d later. The assays were done in the period encompassed between 1 to 2 months after seed harvesting. Data are averages 6 SD from three independent experiments. C, ABA-hypersensitive early seedling growth inhibition of different mutants compared to wild type. Approximately 80 seeds of each genotype were sowed on each plate and the photograph was taken 12 d later. MS, Murashige and Skoog medium.
In summary, a partial constitutive up-regulation and down-regulation of ABA-responsive genes was found in both triple mutants in the absence of exogenous ABA (Supplemental Table S1). Furthermore, gene ontology (GO) analysis performed at The Arabidopsis Information Resource (TAIR; http://www.Arabidopsis. org/tools/bulk/go/index.jsp) reveals that these genes belong to the same categories that are overrepresented in ABA-treated wild-type seedlings, as, for instance, genes related to plant response to abiotic/biotic stimulus and stress response (oxidative stress, osmotic, salt, heat shock, and cold; Fig. 6C). Additionally, we identified ABA-responsive genes that were differentially up-regulated/down-regulated (.2-fold) between both triple mutants when compared with each other (Supplemental Table S2). For instance, genes encoding ABA-responsive seed storage proteins and AAA-type ATPases showed enhanced up-regulation in triple abi2-2, whereas genes encoding defensin-like proteins showed enhanced down-regulation compared to triple pp2ca-1 (Supplemental Table S2). Interestingly, some genes encoding late embryogenesis-abundant proteins showed enhanced up-regulation in triple pp2ca-1 compared to triple abi2-2 (Supplemental Table S2).
Finally, to clarify whether the phenotypes described for the triple mutants reflect either enhanced synthesis of ABA or enhanced signaling, we have measured endogenous ABA levels in 12-d-old seedlings (Supplemental Fig. S2). Interestingly, endogenous ABA levels did not significantly differ when compared with data measured in triple mutants with respect to wild type (Supplemental Fig. S2). Moreover, after osmotic stress treatment (350 mM mannitol), both triple mutants did not produce more ABA than wild type (Supplemental Fig. S2). Therefore, these results indicate that the phenotypes observed in both triple mutants reflect a genuine effect of enhanced ABA signaling.

DISCUSSION
In this article, we show that different degrees of ABA hypersensitivity, ranging from mild to extreme, can be engineered through progressive inactivation of PP2Cs from group A. Thus, a fine tuning of ABA signaling can be accomplished in Arabidopsis through genetic inactivation of these genes, which might be of biotechnological interest once the corresponding orthologous genes are identified in crops. For instance, from this article and previous results of Saez et al. (2006), we conclude that inactivation of two major PP2Cs (HAB1/ABI1, ABI1/ABI2, ABI1/PP2CA, HAB1/PP2CA) is enough in Arabidopsis to generate drought-avoidant plants. Taking into account the important advance in RNA-mediated gene-silencing Figure 5. Reduced stomatal aperture in the absence of exogenous ABA for triple hab1-1abi1-2abi2-2 (triple abi2-2) and hab1-1abi1-2pp2ca-1 (triple pp2ca-1) mutants. A and B, ABA-hypersensitive stomatal closing in triple hab1-1abi1-2abi2-2 mutant as compared to wild type. Plants were kept overnight in high humidity and then leaves were preincubated for 2 h in opening solution. Stomatal apertures were measured 2 h and 30 min after addition of 0, 0.01, or 0.1 mM ABA. Data represent the average of three independent experiments 6 SEM (n = 30-40 stomata per experiment). C and D, Leaf gas-exchange measurements reveal both reduced stomatal conductance and transpiration in triple pp2ca-1 and triple abi2-2 mutants as compared to wild type. Data represent the average of two independent experiments 6 SEM (n = 10 plants/experiment). *, P , 0.01 (Student's t test) when comparing data from each genotype and Col in the same assay conditions. technology, either through the use of hairpin RNAi or artificial microRNAs (Ossowski et al., 2008), it is reasonable to suggest that combined inactivation of these PP2Cs might be attained in crop plants where genome information is available. However, this work also shows that overactivation of ABA signaling is detrimental for plant growth in the absence of water stress because triple mutants that show a partial constitutive response to ABA are impaired in growth. Therefore, the multiplicity of PP2Cs that regulate ABA signaling appears to be a versatile mechanism to adequately control ABA response both in the absence or presence or stress.
The analysis of combined mutations in PP2Cs from group A has revealed a genetic interaction between the ABI1/HAB1 and PP2CA branches, which extends our knowledge on these PP2Cs. For instance, although it was known that PP2CA plays a major role in regulating ABA signaling in seeds, the phenotypes of double mutants that combine the pp2ca-1 allele with either hab1-1 or abi1-2 reveal that PP2CA also has a key role in regulating water loss in vegetative tissue. This evidence could not be obtained from the analysis of single pp2ca mutants, which show similar water loss than wild type due to a slight ABA-hypersensitive response, as discussed by Kuhn et al. (2006). Additionally, enhanced sensitivity to ABA-mediated inhibition of seed germination was found in the double hab1-1pp2ca-1 and abi1-2pp2ca-1 mutants, whose ABA-IC 50 was 50 and 70 nM, respectively, compared to pp2ca-1 (ABA-IC 50 = 110 nM). These double mutants were also more sensitive to ABA-mediated inhibition of root growth than pp2ca-1. Therefore, taken together, these results confirm that PP2CA, as shown previously for Figure 6. Triple hab1-1abi1-2abi2-2 (triple abi2-2) and hab1-1abi1-2pp2ca-1 (triple pp2ca-1) mutants show partial constitutive up-regulation and down-regulation of ABAresponsive genes in the absence of exogenous ABA. A, Venn diagrams. Number of ABA-responsive genes that show up-regulation and downregulation in triple mutants in the absence of exogenous ABA. B, Expression of RAB18 and RD29B in triple mutants with respect to wild type. qRT-PCR analyses were made in triplicate on RNA samples of 2-week-old seedlings that were either mock or 10 mM ABA-treated for 3 h. Numbers indicate the expression level of the RAB18 and RD29B genes under mock (2) or ABA treatment (+) in each mutant genotype with respect to the wild type (value 1). C, Comparison of GO categories for ABA-responsive genes in ABA-treated wild type versus triple mutants in the absence of exogenous ABA. HAB1 and ABI1, is a key regulator of ABA signaling both in seeds and vegetative tissues (Kuhn et al., 2006). This conclusion is consistent with gene expression data found at public databases (Supplemental Fig. S3). In addition, the phenotypes of the different mutants hereby reported suggest that PP2Cs provide a threshold of negative regulation required for a normal response to ABA. If ABI1/HAB1 and PP2CA branches regulate independently ABA signaling, the loss of one branch would be enough to overcome this threshold and, therefore, lead to ABA hypersensitivity. For instance, even though HAB1 and ABI1 are expressed in pp2ca-1, this mutant shows a strong ABA-hypersensitive phenotype in seeds. Alternatively, it is possible that ABA signaling is negatively regulated through a mechanism that involves an accurate PP2C dosage effect. For instance, such a mechanism is supported for the progressive increase in root growth sensitivity to ABA observed in single, double, and triple pp2c mutants (Fig. 4C).
Finally, we show that triple pp2c mutants are extremely sensitive to exogenous ABA. Particularly noticeable is the nanomolar sensitivity of the triple mutants for ABA-mediated inhibition of seed germination, which is particularly appealing, taking into account that other PP2Cs that regulate ABA signaling in seed are still active in such mutants. For instance, the triple hab1-1abi1-2abi2-2 mutant shows an ABA-IC 50 of 40 nM, even though two PP2Cs (PP2CA/AHG3 and AHG1) that are essential components of ABA signaling in seed are active. This fact suggests the existence of at least two branches of PP2Cs to negatively regulate ABA signaling in seed. Additionally, triple mutants show a partial constitutive response to ABA in different assays. For instance, root growth of triple mutants in the absence of exogenous ABA was very similar to wild type in the presence of 10 mM ABA. Partial stomatal closure was found in triple mutants as compared to wild type and, interestingly, in the absence of exogenous ABA the triple mutants showed a stomatal aperture only slightly higher than wild type treated with 10 nM ABA. This partial constitutive response to ABA was also found by gene expression analysis. Thus, expression of RAB18 and RD29B in triple mutants, in the absence of exogenous ABA, was only 50% to 60% of that found in wild type treated with 10 mM ABA. Whole transcriptome analysis reveals that approximately 15% or approximately 25% of genes up-regulated or down-regulated by ABA in wild type, respectively, were constitutively up-regulated or down-regulated in the triple mutants in the absence of exogenous ABA. Even though this effect represents only a fraction of the whole set of ABA-responsive genes, it leads to impaired plant growth. This phenotype is reminiscent of that found in plants that overexpress DRE-binding protein DREB1A under the control of a 35S promoter, which show constitutive expression of many stress tolerance genes, but also show severe growth retardation under normal growing conditions (Kasuga et al., 1999). Similarly, consti-tutive expression of ABRE-binding factors ABF3 and ABF4 also led to constitutive expression of ABAresponsive genes and growth retardation (Kang et al., 2002). Therefore, PP2Cs play a major role in regulating ABA signaling both under stress as well as normal growth conditions. Additionally, the phenotype of triple pp2c mutants serves to illustrate the importance of ABA in stress response as well as growth regulation.

Plant Material and Growth Conditions
Arabidopsis (Arabidopsis thaliana) plants were routinely grown under greenhouse conditions in pots containing a 1:3 vermiculite:soil mixture. For in vitro culture, seeds were surface sterilized by treatment with 70% ethanol for 20 min, followed by commercial bleach (2.5% sodium hypochlorite) containing 0.05% Triton X-100 for 10 min and, finally, four washes with sterile distilled water. Stratification of the seeds was conducted in the dark at 4°C for 3 d. Then seeds were sowed on plates composed of Murashige and Skoog basal salts, 0.1% MES acid, 1% agar, and 1% Suc. The pH was adjusted to 5.7 with KOH before autoclaving. Plates were sealed and incubated in a controlled-environment growth chamber at 22°C under a 16-h-light/8-hdark photoperiod at 80 to 100 mE m 22 s 21 .

Germination, Growth, and Stomatal Aperture Assays
To measure sensitivity to ABA-mediated inhibition of germination, seeds (approximately 200 seeds/experiment) were plated on solid medium composed of Murashige and Skoog basal salts, 1% Suc, and increasing concentrations of ABA. To score seed germination, either radicle emergence or the percentage of seeds that had germinated and developed fully green expanded cotyledons was determined. The ABA-resistant growth was scored as described by Saez et al. (2006). Data were obtained for three independent experiments, each done with 15 plants. To quantify root growth inhibition, 5-d-old seedlings grown vertically onto Murashige and Skoog plates were transferred to either Murashige and Skoog plates or Murashige and Skoog plates supplemented with 10 mM ABA. After 7 d, the plates were scanned on a flatbed scanner to produce image files suitable for quantitative analysis using National Institutes of Health Image software (ImageJ version 1.37). Data were obtained for three independent experiments, each done with 15 plants. Assays of ABA-induced stomatal closing were performed as described by Saez et al. (2006). Data were expressed as the average of three experiments where 30 to 40 stomata were measured for each one.

Leaf Gas-Exchange Measurements
For leaf gas-exchange measurements, plants were grown in hydroponics (Hoagland solution at 50% in 250-mL nontransparent pots) inside a growth chamber (12 h at 150 mmol photons m 22 s 21 , approximately 60% humidity, and 26°C day/20°C night air temperature). Nutrition solution was changed every 2 to 3 d to prevent nutrition and water limitations. From preliminary light response curves, photosynthesis was proved to be saturating at photon flux densities above 500 mmol photons m 22 s 21 . Light-saturating net photosynthesis (A N ), leaf conductance for water vapor (g s ), transpiration rate (E), and substomatal CO 2 concentration (C i ) were therefore measured under steadystate conditions at 800 mmol photons m 22 s 21 with an open gas-exchange infrared gas analyzer (Li-6400; LI-COR) on the youngest fully expanded leaf of 38-to 45-d-old plants for Col and triple mutants. CO 2 concentration in the leaf chamber was maintained at 400 mmol CO 2 mol 21 air, whereas humidity and vapor pressure deficit were kept around 45% and 2 kPa, respectively.

Drought Stress and Water Loss Assays
Two different water-loss assays were performed. Short-term assays were performed basically as described by Saez et al. (2006) in detached leaves at the same developmental stage and size from 21-d-old plants. Four samples of three leaves per genotype were excised and fresh weight was determined by submitting the leaves to the drying atmosphere of a flow laminar hood for 6 h. Data are averages 6 SE from three independent experiments (n = 5). Long-term assays were performed after withholding water in plants maintained under greenhouse conditions basically as described by Saez et al. (2006). Eight plants of each genotype (two independent experiments) were grown under normal watering conditions for 21 d and then subjected to drought stress for 8 d by completely terminating irrigation and minimizing soil evaporation by covering pots with plastic wrap film. Eight leaves from each plant were removed, weighed, incubated in demineralized water for 3 h, and weighed again. The difference in weight was considered as water loss.
qRT-PCR amplifications were monitored using Eva-Green fluorescent stain (Biotium). Relative quantification of gene expression data was carried out using the 2 2DDC T or comparative C T method (Livak and Schmittgen, 2001). Expression levels were normalized using the C T values obtained for the b-actin-8 gene. The presence of a single PCR product was further verified by dissociation analysis in all amplifications. All quantifications were made in triplicate on RNA samples obtained from three independent biological replicas.

RNA Amplification, Labeling, and Microarray Hybridization
These assays were performed as described by Rodrigues et al. (2009), according to MIAME guidelines (Brazma et al., 2001.

Identification of Differentially Expressed Genes and GO Analysis
Significance analysis of microarrays (Tusher et al., 2001) was performed on the three normalized datasets to identify differentially expressed genes. The parameters for significance analysis of microarrays were adjusted so that the false discovery rate probability for every experiment was below 0.05. A 2-fold expression cutoff was considered. A functional category analysis of the genes simultaneously up-regulated or down-regulated in the three genotypes (Col, triple abi2-2, and triple pp2ca-1) was carried out by the Munich Information Center for Protein Sequences (MIPS; http://mips.gsf.de/proj/funcatDB/ search_main_frame.html). Only overrepresented categories with a P value smaller than 0.05 were further considered. Functional categorization of upregulated genes in each of the genotypes, based on the high-level terms in the GO hierarchy, was conducted using TAIR (http://www.Arabidopsis.org/ tools/bulk/go/index.jsp). Venn diagrams were generated to illustrate differences in expression.

ABA Content Measurement
Three independent biological samples of 12-d-old seedlings grown in Murashige and Skoog plates were used for ABA content determination. Seedlings were either mock-or 350 mM mannitol treated for 8 h. After measuring fresh weight, plants were ground in liquid nitrogen and then extracted with 1 mL extraction buffer (80% methanol, 100 mg/L butylated hydroxytoluene, and 0.5 g/L citric acid monohydrate) overnight at 4°C (Xiong et al., 2001). After centrifugation at 1,000g for 20 min at 4°C, supernatants were collected and dried. Samples were resuspended in buffer containing 10% methanol, 25 mM Tris, pH 7.5, 100 mM NaCl, and 1 mM MgCl 2 , and they were subjected to ABA measurement using the Phytodetek ABA test kit (Agdia).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. ABA-hypersensitive early seedling growth inhibition of different mutants compared to wild type at different concentrations of ABA ranging from 25 to 500 nM ABA.
Supplemental Figure S2. Comparison of the endogenous ABA levels in wild type and triple mutants under nonstressed conditions or after treatment with 350 mM mannitol for 8 h.
Supplemental Figure S3. Cladogram and gene expression data from clade A PP2Cs.
Supplemental Table S1. ABA-responsive genes constitutively up-regulated (threshold of 2-fold, ratio .2) or down-regulated (threshold of 2-fold, ratio ,0.5) in the triple mutants in the absence of exogenous ABA.
Supplemental Table S2. Differential expression of ABA-responsive genes constitutively up-regulated or down-regulated in the triple abi2-2 and pp2ca-1 mutants compared each other (genes selected previously from Supplemental Table S1).