Present address: Division of Biological Sciences, Cell and Developmental Biology Section, Center for Molecular Genetics, University of California, San Diego, 9500 Gilman Drive La Jolla, CA 92093-0116, USA.
Integrated Graduate School of Art and Science, Yokohama City University, 1-7-29 Suehiro, Tsurumi, Yokohama 230-0045, Japan,
Gene Discovery Research Group, RIKEN Plant Science Center, 1-7-22 Tsurumi, Yokohama 230-0045, Japan, and
The phytohormone abscisic acid (ABA) regulates physiologically important stress and developmental responses in plants. To reveal the mechanism of response to ABA, we isolated several novel ABA-hypersensitive Arabidopsis thaliana mutants, named ahg (ABA-hypersensitive germination). ahg1-1 mutants showed hypersensitivity to ABA, NaCl, KCl, mannitol, glucose and sucrose during germination and post-germination growth, but did not display any significant phenotypes in adult plants. ahg1-1 seeds accumulated slightly more ABA before stratification and showed increased seed dormancy. Map-based cloning of AHG1 revealed that ahg1-1 has a nonsense mutation in a gene encoding a novel protein phosphatase 2C (PP2C). We previously showed that the ahg3-1 mutant has a point mutation in the AtPP2CA gene, which encodes another PP2C that has a major role in the ABA response in seeds (Yoshida et al., 2006b). The levels of AHG1 mRNA were higher in dry seeds and increased during late seed maturation – an expression pattern similar to that of ABI5. Transcriptome analysis revealed that, in ABA-treated germinating seeds, many seed-specific genes and ABA-inducible genes were highly expressed in ahg1-1 and ahg3-1 mutants compared with the wild-type. Detailed analysis suggested differences between the functions of AHG1 and AHG3. Dozens of genes were expressed more strongly in the ahg1-1 mutant than in ahg3-1. Promoter–GUS analyses demonstrated both overlapping and distinct expression patterns in seed. In addition, the ahg1-1 ahg3-1 double mutant was more hypersensitive than either monogenic mutant. These results suggest that AHG1 has specific functions in seed development and germination, shared partly with AHG3.
The plant hormone abscisic acid (ABA) is involved in important plant physiological phenomena, including seed dormancy, germination, growth regulation, developmental processes and stomatal closure. To understand the ABA response mechanisms, many approaches, including forward and reverse genetic analyses and biochemical analysis, have been utilized, and have allowed the identification of many components involved in ABA synthesis, in ABA signaling from perception to the regulation of gene expression, and in ABA catabolism (Finkelstein et al., 2002; Nambara and Marion-Poll, 2005; Yamaguchi-Shinozaki and Shinozaki, 2006). The results obtained from the studies on these components facilitate our understanding of the ABA response mechanisms.
It has been predicted that protein kinases play crucial roles in the ABA signaling pathway, as in other cellular response phenomena. Although there are many reports of protein kinase genes whose expression is modulated by ABA, the physiological relevance of most of those kinases remains unclear. Genetic and biochemical approaches have revealed that the ABA-dependent activation of OST1/SRK2e, an SNF1-related kinase 2, is required for the ABA response of guard cells (Mustilli et al., 2002; Yoshida et al., 2002). Recent studies of SnRK2 kinases have indicated the presence of a complex phosphorylation network or cascade in the ABA signaling pathway (Belin et al., 2006; Kobayashi et al., 2004; Yoshida et al., 2006a). Another type of SNF1-related kinase, SnRK3/PKS/CIPK, is also implicated in the ABA response. The impaired function of some of SnRK3s, such as SOS2, PKS3, CIPK3 and PKS18, has been demonstrated to affect the ABA response (Gong et al., 2002; Guo et al., 2002; Kim et al., 2003; Ohta et al., 2003). In addition, calcium-dependant protein kinases play pivotal roles in the ABA response in guard cells (Mori et al., 2006).
Previously, we described an ABA-hypersensitive locus, AHG3, encoding AtPP2CA, and demonstrated that disruption mutations of the five ABA-related PP2Cs affected ABA sensitivity in germination, and further that AHG3/AtPP2CA had the strongest effect among them (Yoshida et al., 2006b). Based on expression analysis of PP2C genes, we hypothesized that expression level is a major determinant of the contribution to the ABA response in germination. Our analyses of the mis-sense allele of AHG3/AtPP2CA and transgenic lines with an abi1-1-type mutation of AHG3/AtPP2CA implied that the abi1-1-type dominant mutations are not a dominant-negative mutation. Robert et al. also reported similar results by analyzing HAB1 (Robert et al., 2006). However, to draw further conclusions, information on the native substrates and detailed biochemical analysis are necessary.
In this study, we describe another ABA-hypersensitive locus, AHG1 (Nishimura et al., 2004). The ahg1-1 mutant showed ABA hypersensitivity as strong as that of ahg3-1 in germination and post-germination growth, but not in adult plants. We identified the AHG1 gene by map-based cloning. Interestingly, AHG1 was At5 g51760, encoding a PP2C related to AHG3/AtPP2CA. According to public microarray databases, AHG1 has the strongest expression in seeds among ABI1-related PP2Cs. Our data support the view that PP2Cs have negative regulatory functions in the ABA response, and that there is a correlation between expression levels and the contribution to the ABA response in germination. Detailed characterization of the mutants and gene expression analysis revealed both overlapping and distinct functions of AHG1 and AHG3/AtPP2CA, suggesting elaborate regulation of the ABA response in seed.
The ahg1-1 mutant is hypersensitive to ABA
We have previously shown that ahg1-1 is an ABA-hypersensitive mutant that germinates and grows poorly in the presence of ABA (Nishimura et al., 2004). To assess further the function of AHG1 in the ABA response, we examined the efficiencies of germination (radicle emergence) and post-germination growth (seedling with expanded green cotyledons) in the presence of various concentrations of ABA. These efficiencies in the ahg1-1 mutant were apparently reduced in the presence of exogenous ABA (Figure 1a,b). Time-course assays of germination and post-germination growth revealed delayed germination and post-germination growth of the ahg1-1 mutant in the presence of 0.3 μm ABA (Figure 1c,d). To determine whether the ahg1-1 phenotype is specific to exogenously applied ABA, we examined the effect of stratification length on germination efficiency in the absence of exogenous ABA (Figure 1e,f). The ahg1-1 mutant showed a dramatic inhibition of germination and post-germination growth in the absence of stratification, but no effect when stratified for 2 or 4 days. ahg1-1 seedlings and adult plants showed no abnormal responses to ABA (Nishimura et al., 2004). We checked the effect on drought stress tolerance in adult plants grown on soil for 4 weeks and then left unwatered. After 10 days, ahg1-1 and wild-type plants showed a wilted phenotype (Figure 1 g), whereas era1-2 plants seemed healthier than the wild-type, consistent with previous results (Pei et al., 1998). A water loss assay showed no significant difference between ahg1-1 and wild-type (Figure 1h). These results indicate that AHG1 functions mainly during germination or seed development.
To determine whether or not the ABA hypersensitivity and increased seed dormancy of ahg1-1 were due to a higher level of accumulation of ABA, we measured the endogenous ABA contents of seeds and stressed plants. Dry ahg1-1 seeds accumulated 30% more endogenous ABA than the wild-type. After stratification, the ABA level of ahg1-1 seeds was reduced to almost the same as that of the wild-type (Figure 2a). Two-week-old rosette plants of ahg1-1 had almost the same amount of endogenous ABA as the wild-type. Upon osmotic stress treatment with mannitol, the endogenous ABA contents of both lines were increased dramatically to the same level, indicating that the regulation of ABA production by ahg1-1 is normal in adult plants (Figure 2b).
AHG1 encodes a novel protein phosphatase 2C
To identify how the ahg1-1 mutation conferred the enhanced sensitivity to ABA, we identified the AHG1 gene by map-based cloning. The AHG1 locus was localized between markers AtSO191 and MBK5 on chromosome 5 (Nishimura et al., 2004). We analyzed 1964 chromosomes of F2 plants obtained from an ahg1-1 × Ler test cross, and narrowed the ahg1-1 region down to a segment of approximately 80 kb spanning three BAC clones, K17 N15, K10D11 and MIO24 (Figure 3a). By determining the nucleotide sequences of candidate genes in this region, we found a base substitution from C to T in the first exon of At5 g51760, which caused a nonsense mutation, changing Gln155 to a stop codon. Interestingly, At5 g51760 has been assigned as encoding a novel PP2C.
To confirm that this gene was AHG1, we examined the phenotype of disruption mutants of At5g51760. For this purpose, we obtained Ds transposon insertion (Ds 54-0076-3) and T-DNA insertion (SALK_095052) mutants. The insertion sites of Ds 54-0076-3 and SALK_095052 were in the first exon and the third intron, respectively. We examined their phenotypes and found that both had the same ABA-hypersensitive phenotype in germination and post-germination growth as ahg1-1 (Figure 3b, Table 1). We named these lines ahg1-2 and ahg1-3, respectively. We also constructed ahg1-1 transgenic plants expressing the wild-type At5g51760 cDNA fused to the CaMV 35S promoter. A transgenic line harboring At5g51760 (line L1) had normal post-germination growth, and a control transgenic line with the empty vector showed ABA hypersensitivity (Figure 3c,d). These results confirmed that AHG1 is At5 g51760, which encodes a novel PP2C. Lines with higher expression of At5 g51760 in the wild-type background (WT/L1) and ahg1-1 background (ahg1-1/L1) germinated and grew normally in the presence of 0.8 μm ABA, whereas a control transgenic line with the empty vector grew poorly (Figure 3d). These results suggest that a novel PP2C, AHG1, functions as a negative regulator in the ABA signaling pathway.
Table 1. Germination efficiencies of T-DNA and Ds transposon insertion alleles of AHG1
Percentage of germinated and grown seedsa
0.3 μm ABAb
aApproximately 50 seeds of each line were sown on MS plates containing 0.3 μm ABA or control medium. The post-germination growth was scored after 7 days of incubation.
bData represent means ± standard deviations of three independent experiments.
97.4 ± 3.2
89.0 ± 7.6
100 ± 0
96.8 ± 1.8
96.9 ± 0.6
0 ± 0
93.6 ± 4.5
1.5 ± 1.5
100 ± 0
5.07 ± 2.8
AHG1 is expressed during seed development and germination
As shown in Figure 3(e), AHG1 belongs to a group consisting of ABI1-related PP2Cs. We previously demonstrated that, among such PP2Cs, AHG3/AtPP2CA had a dominant role in seeds, and predicted that its mRNA level is the major determinant of its contribution to the ABA response, as AHG3/AtPP2CA had the highest mRNA level in seeds among the eight PP2C genes we examined (Yoshida et al., 2006b). Notably, AHG1 is the PP2C gene most highly expressed in dry seed among members of the clade according to an extensive transcriptome study of seeds (Nakabayashi et al., 2005) and public microarray data (Figure S1, Table S1). We examined the expression levels of AHG1 mRNA using semi-quantitative RT-PCR analysis (Figure 4a). The mRNA levels of AHG1 were higher in dry seeds and were reduced to undetectable levels after stratification. However, the mRNA levels of AHG1 apparently increased in the presence of ABA in seedlings. To investigate the seed-specific expression of AHG1, we next examined the expression pattern of AHG1 during seed development and maturation. Expression of AHG1 was first detected at 8 days after flowering (DAF), and increased continuously until 16 DAF, whereas the expression of AHG3/AtPP2CA remained steady. Interestingly, this development-dependent expression pattern of AHG1 is similar to that of ABI5 (Figure 4b).
This observation implied that the expression levels and patterns of these genes might be regulated by ABI3, ABI4 or ABI5, which are implicated in the control of seed development and germination. To investigate this, we extracted total RNA from abi3-1, abi4-1, abi5-1 and wild-type seedlings grown on MS plates with or without 3.0 μm ABA for 3 days after stratification. The results of semi-quantitative RT-PCR experiments showed that the levels of AHG1 mRNA were slightly but not significantly decreased in the mutant lines compared with the wild-type lines treated with ABA, whereas the mRNA level of AtEm6, which is regulated by ABI3 and ABI5, was dramatically decreased in abi3-1 and abi5-1 mutants (Figure 4c). The levels of AHG3/AtPP2CA mRNA were similar in the mutant lines and wild-type. These results indicate that expression of AHG1 and AHG3/AtPP2CA is not regulated by ABI3, ABI4 or ABI5, despite their increased expression caused by ABA treatment of germinating seeds.
To confirm this idea, we characterized the genetic relationship between ahg1-1 and the ABA-insensitive loci abi3-1, abi4-1 and abi5-1. We prepared double mutant lines and examined their post-germination growth for 7 days on plates containing ABA (Table 2). The ahg1-1 abi3-1 and ahg1-1 abi5-1 double mutants showed an ABA-insensitive phenotype similar to that of abi3-1 or abi5-1. These results indicate that AHG1 functions upstream of ABI3 and ABI5 in the ABA signaling pathway. Interestingly, the ahg1-1 abi4-1 double mutant showed an ABA-insensitive phenotype in the presence of 0.3 μm ABA but not in the presence of 3.0 μm ABA. The relationship between ABI4 and AHG1 in the ABA signaling pathway is not clear but their relationship seems weaker.
Table 2. Germination efficiencies of single and double mutants
Percentage of germinated seedsa
0.3 μm ABAb
3.0 μm ABAb
aApproximately 50 seeds of each line were sown on MS plates containing 0.3 or 3.0 μm ABA or control medium. The post-germination growth was scored after 7 days of incubation.
bData represent means ± standard deviations of three independent experiments.
98. 0 ± 1.0
90.7 ± 2.9
0 ± 0
96.7 ± 2.3
86.1 ± 0.2
0.8 ± 1.3
99.0 ± 0.8
89.0 ± 3.9
0 ± 0
92.4 ± 3.5
89.3 ± 6.0
92.5 ± 4.0
95.6 ± 2.1
95.4 ± 3.3
73.0 ± 9.7
98.1 ± 2.0
96.6 ± 1.5
96.1 ± 2.5
98.5 ± 1.4
4.5 ± 1.5
0 ± 0
91.6 ± 2.6
95.7 ± 2.1
87.4 ± 2.1
100 ± 0
95.6 ± 3.4
0 ± 0
98.1 ± 1.6
99.4 ± 1.1
89.2 ± 3.5
To examine the spatial expression patterns of AHG1 and AHG3/AtPP2CA in more detail, we constructed transgenic plants with recombinant genes possessing the promoter region of AHG1 or AHG3/AtPP2CA fused to a glucuronidase (GUS) reporter gene (PAHG1::GUS and PAHG3::GUS, respectively). We examined GUS histochemical staining patterns of seeds obtained from transgenic lines. GUS staining was observed in the seeds of both transgenic lines, but the staining patterns were different. Dry PAHG1::GUS transgenic seeds showed strong GUS activity in the whole embryo (Figure 5a). The activity decreased to some extent after imbibition, although the spatial pattern was the same (Figure 5b,c). By contrast, PAHG3::GUS seedlings showed strong GUS activity in apical and root meristems and vascular bundles (Figure 5e–g). Four days after imbibition, the staining was decreased but maintained the same spatial pattern. Before germination, the staining patterns in the seed coat/endosperm of both lines were very similar (Figure 5i–n). After germination, the GUS activities of both lines decreased; the effect was greater in the PAHG1::GUS line, (Figure 5o,p). Interestingly, at this stage, GUS activity showed spatial differentiation in the PAHG1::GUS line; GUS activity was stronger in vascular bundles and the root cap than in other parts (Figure 5d). It should be noted that the staining in seed coat/endosperm of the PAHG1::GUS line almost disappeared, while that of the PAHG3::GUS line was clearly detectable (Figure 5o,p). In 6-day-old seedlings, the PAHG1::GUS line showed GUS activity only at the bases of the lateral root buds (Figure 5q,r), but the PAHG3::GUS line showed GUS activity in vascular bundles and the whole of the lateral root bud (Figure 5s,t), consistent with a previous report (Cherel et al., 2002). These results suggest that AHG1 and AHG3 are controlled differently, and further imply their distinct functions. In contrast to the higher expression in seed, we could not detect AHG1 expression in adult plant organs, including rosette and cauline leaves, stem, flower and root, even in the presence of ABA (data not shown). This is consistent with our observation that the ahg1-1 plant had a normal drought stress response (Figure 1g,h). These results indicate that AHG1 has seed-specific functions.
Microarray analysis during germination in the presence of ABA
Next we tried to identify the downstream pathways regulated by AHG1 or AHG3 by analyzing genes whose expression was affected by the ahg1-1 or ahg3-1 mutations. Wild-type, ahg1-1 and ahg3-1 seeds were stratified for 4 days on normal MS plates and then incubated for 2 days on MS plates containing 0.5 μm ABA. RNA samples were extracted and then subjected to microarray experiments using Affymetrix ATH1 gene chips . Analysis of the expression profile showed that most genes up- and downregulated by ABA in the wild-type (> 93% and > 94%, respectively) were up- and downregulated in the ahg1-1 and ahg3-1 mutants as well (Figure 6a–c, Table S2). Some genes that were upregulated by ABA, such as AtEm6 and ABI5, were more highly expressed in ahg1-1 and ahg3-1 than in the wild-type (Table 3), consistent with our previous results (Nishimura et al., 2004). Interestingly, the expression levels of several genes was differentially affected in the ahg1-1 and ahg3-1 mutants when seeds were treated with ABA (Table 3, Figure S2, Table S3). Most genes that were upregulated to a greater extent in ahg1-1 than in ahg3-1 were seed-specific genes, such as those for seed storage proteins. Genes that were downregulated in ahg1-1 were more divergent, and correlations in their expression patterns were not clear when analyzed in the ATTED-II database (http://www.atted.bio.titech.ac.jp/). However, the common feature of more than half of these genes was root-preferential expression. These results are consistent with our view that ahg1-1 is more dormant than ahg3-1. More importantly, they suggest that ahg1-1 and ahg3-1 affect germination differently. The expression of AHG1 seems to be affected by both the ahg1-1 and ahg3-1 mutations. However, as ahg1-1 is a nonsense mutation, it will affect the mRNA stability. Therefore the expression data for AHG1 in ahg1-1 should be considered carefully. The expression of other ABA-related PP2C genes was not affected by ahg1-1 and ahg3-1 (Table 3).
Table 3. Genes affected more strongly by ahg1-1 than ahg3-1 in germination in the presence of ABAa
Relative signal intensity
ahg1 versus ahg3b
ahg1 versus WTb
ahg3 versus WTb
ABA versus controlc
aTop 15 genes expressed to higher and lower levels in ahg1-1 than in ahg3-1 are listed. For genes expressed at a lower level in ahg1-1, data with less consistency between two independent experiments were removed. Signal values of microarray experiments were Robust Multi-Array normalized and compared using affylmGUI. The values are the means of two independent experiments and are shown in log2. The microarray data were deposited in the NCBI GEO database with accession number GSE6638.
bRelative signal intensities between ahg1-1 and ahg3-1 (ahg1-1 versus ahg3-1), ahg1-1 and wild-type (ahg1-1 versus WT), ahg3-1 and wild-type (ahg3-1 versus WT). RNA was extracted from germinating seeds treated with 0.5 μm ABA for 2 days (see Experimental procedures).
cRelative signal intensities between wild-type seeds treated with (ABA) or without 0.5 μm ABA (control) for 2 days.
dThese values were deduced using two data sets of ABA-treated wild-type seeds that were very different each other (see Table S2).
Genes expressed at a higher level in ahg1-1
2S albumin storage protein 2
Lipid transfer protein
Reductase (SDR) family protein
2S seed storage protein 1
12S seed storage protein
2S seed storage protein 3
Mother of FT and TF1 protein
Plant EC metallothionein-like protein
Putatively similar to manganese superoxide dismutase (MSD1)
Putative (PDF2.1) plant defensin protein
Gibberellin-regulated protein 2 (GASA2)
Amidase family protein
LEA protein in group 3
Short-chain dehydrogenase/reductase (SDR) family protein
Genes expressed at a lower level in ahg1-1
Multi-copper oxidase type I
Photosystem I reaction center subunit V
Chlorophyll a/b binding protein
Homeobox leucine zipper protein 13
Pectinesterase family protein
Terpene synthase/cyclase family protein
CER1 protein identical to maize gl1 homolog
Jacalin lectin family protein
Late embryogenesis abundant protein
Invertase/pectin methylesterase inhibitor family protein
Relationship between AHG1 and AHG3/AtPP2CA
To examine the genetic and physiological relationships between AHG1 and AHG3/AtPP2CA in ABA signaling in more detail, we crossed ahg1-1 with ahg3-1 and obtained a double mutant line from F2 progeny. We examined the post-germination growth of this line for 7 days on plates containing ABA (Figure 7a,b). The efficiency of growth of the double mutant was remarkably reduced in the presence of as little as 0.1 μm ABA, showing stronger ABA hypersensitivity than the parental monogenetic mutants. Even in the absence of ABA, the efficiency of post-germination growth was decreased more drastically than that of the parental monogenetic mutants, implying that this line had deeper seed dormancy. These results indicate an additive effect of ahg1-1 and ahg3-1 on the ABA sensitivity of seed, and further suggest that AHG1 and AHG3 have overlapping functions but also some distinct functions. The ahg1-1 ahg3-1 double mutant did not have any detectable phenotypes in the adult stages (data not shown).
Isolation of the ahg1-1 mutant
Through analysis of ahg1-1, which shows increased sensitivity to ABA in germination and post-germination growth, we showed that a novel PP2C, AHG1, is an important component of the ABA signaling pathway in seed (Nishimura et al., 2004). The phenotype of ahg1-1 is similar to that of ahg3-1: accumulation of higher endogenous ABA in dry seed but not in imbibed seed or adult plants, and clear ABA hypersensitivity in germination but not in adult plants. Map-based cloning of AHG1 revealed that AHG1 is At5g51760, which encodes a PP2C. The identification of AHG1 as a PP2C gene provides concrete confirmation of the negative regulatory function of PP2Cs in the ABA response. More importantly, it unveils the crucial role of a novel PP2C in the ABA response in germination.
According to a genome-wide similarity analysis, Arabidopsis has at least 76 genes for PP2Cs (Schweighofer et al., 2004). So far, only five PP2Cs, namely ABI1, ABI2, HAB1, HAB2 and AHG3/AtPP2CA, have been shown to be involved in the ABA response. AHG1 is closely related to these PP2Cs, and belongs to the same clade in the phylogenetic tree, although at the farthest distance (Figure 3e) (Schweighofer et al., 2004). In this clade, ABI1, ABI2, HAB1 and HAB2 form one group, and AHG1, AHG3, At5g59220, At2g29380 and At1g07430 form the other. It is of interest that AHG1 and AHG3 have functions related to the ABA response in seeds and germination, but the other three PP2Cs have not, although they are expressed in seed (Yoshida et al., 2006b). So far, we cannot find any signatures in the amino acid sequences that would explain the difference in their ABA-related functions. To determine the physiological function of each PP2C will require detailed examination one by one.
As germination and post-germination growth are affected by many physiological abnormalities such as defects in growth, developmental control and metabolic control, it was expected to be difficult to identify mutants specific to the ABA response. However, disruption of a considerable number of genes related to the ABA response turned out to cause an ABA-hypersensitive phenotype in germination, indicating the value of isolation and characterization of mutants with this phenotype. The identification here of a novel PP2C involved in the ABA response in seed, by means of analyzing such a mutant, clearly demonstrates its effectiveness. As neither ahg1-1 nor ahg3-1 had a clear adult phenotype, only such screening would allow us to identify those two components, confirming the importance of the isolation and analysis of ABA-hypersensitive mutants.
Physiological function of AHG1 in seed development and germination
The ahg1-1 mutation, presumably a loss-of-function mutation, causes recessive ABA hypersensitivity during germination. By contrast, over-expression of AHG1 conferred clear ABA insensitivity during germination, even in an ahg1-1 background. These observations strongly suggest that the AHG1 PP2C functions as a negative regulator of the ABA response, as do AHG3/AtPP2CA and other ABI1-related PP2Cs. ahg1-1 seeds had 30% higher endogenous ABA, as did ahg3-1 seeds. This slight increase of the endogenous ABA level in seed does not seem enough to explain the dramatic elevation of the ABA sensitivity of this mutant in seed germination and post-germination growth. This was confirmed by the fact that an ABA-defective mutant, aba2-1, did not suppress the ABA-hypersensitive phenotype of ahg1-1 (data not shown). Therefore, it is more likely that ahg1-1 is an ABA-hypersensitive mutant rather than a simple ABA over-production mutant. The enhanced sensitivity of ahg1-1 to ABA in seeds might activate positive feedback of ABA production, which, in turn, increases seed dormancy further and the sensitivity to ABA.
Extensive analysis of the ahg1-1 mutant and the AHG1 gene revealed the specific function of AHG1 in seeds and germination. First, although ahg1-1 had clear ABA hypersensitivity in germination and post-germination growth, we could not observe any detectable phenotype at adult stages. Second, AHG1 was barely expressed in adult plants (data not shown). Third, our experiments and public microarray data showed strong expression of AHG1 in developing seed and dry seed (Figures 4 and 5, Figure S1). Previously, we postulated that the expression level of PP2C mRNAs is one of the major determinants of the role of PP2C in ABA signaling in seed germination (Yoshida et al., 2006b). The clear correlation between the seed-specific expression of AHG1 and the ABA-hypersensitive phenotype of the ahg1-1 mutant is consistent with this idea.
The expression of AHG1 seems to be positively regulated by ABA, although ABI3 and ABI5 were not involved. The promoter region of AHG1 has several sequences similar to RY or ABRE (ABA responsive element) cis elements, and AHG1 was activated by ABA in germinating seeds (Figure 4a, Table S4). In addition, we found that the expression pattern of AHG1 was similar to that of ABI5 during seed development. Presumably, the functions of positive ABA regulators and of AHG1, a negative regulator, constitute a negative feedback loop to fine-tune the ABA action during seed development and to maintain mature seed. The important roles of AHG1 in seed development and dormancy are evidenced by the deeper dormancy of ahg1-1 (Figure 1). The higher level of accumulation of endogenous ABA in dry seeds and the higher level of expression of seed-specific genes in ABA-treated seeds of the ahg1-1 mutant line support this idea. Several recent reports show that endosperm has an important function in the ABA response in seeds and during germination (Lefebvre et al., 2006; Muller et al., 2006; Penfield et al., 2006). In this respect, the strong expression of AHG1 and AHG3/AtPP2CA in the endosperm is interesting (Figure 5).
Distinct or redundant functions of AHG1 and AHG3/AtPP2CA
The AHG1 gene shares many features with AHG3/AtPP2CA. Both encode closely related PP2Cs and are expressed strongly in seed. Loss-of-function mutations of both genes cause strong ABA hypersensitivity in germination and post-germination growth. However, detailed characterization of these mutants revealed differences in the functions of these PP2Cs, and the mutants have slightly different phenotypes. When compared with ahg3-1, the ahg1-1 mutant had stronger ABA hypersensitivity in radicle emergence (germination) and deeper seed dormancy. Without stratification, the germination efficiency of ahg1-1 remained low (Figure 1f), whereas that of ahg3-1 eventually rose to the normal level (Yoshida et al., 2006b). These mutants also had different sensitivities to sugar in germination (Nishimura et al., 2004) (Table S5). The microarray experiments supported the differences in these two mutants, showing that some genes were affected more specifically by ahg1-1 or ahg3-1. In addition, the ahg1-1 ahg3-1 double mutant showed a stronger phenotype than either monogenic parental mutant line, indicating their additive effect. These results strongly suggest that AHG1 and AHG3 have both overlapping and distinct functions in the ABA response in seeds.
How this difference comes about is intriguing, because it has been postulated that PP2Cs are not regulated at the post-translational level (Schweighofer et al., 2004). The most plausible explanation is the difference in the temporal and spatial expression patterns and levels between AHG1 and AHG3/AtPP2CA during seed development and germination. AHG3/AtPP2CA is constitutively expressed during this period. By contrast, AHG1 was gradually activated as seed development progressed (Figure 4b). The GUS activity of the promoter–GUS transgenic plants indicated that AHG1 is expressed uniformly in the embryo, whereas AHG3 is expressed more strongly in specific parts such as meristems and vascular bundles (Figure 5). These differential expressions must contribute, at least partly, to the distinct functions of AHG1 and AHG3/AtPP2CA.
Protein phosphatases are usually considered to be non-specific. However, interactions with other components can give the specificity (for example, Meskiene et al., 2003; Mapes and Ota, 2004). Recent studies of ABI1 and ABI2 revealed their physical interactions with ABA-related components. ABI1 binds to kinases such as OST1/SRK2e, PKS3, PKS18 and the transcriptional regulator ATHB6, whereas ABI2 binds to SOS2, PKS3, PKS11, PKS24, fibrillin and glutathione peroxidase (Guo et al., 2002; Himmelbach et al., 2002; Miao et al., 2006; Ohta et al., 2003; Yang et al., 2006; Yoshida et al., 2006a). In addition, ABI1 has been shown to be regulated by phosphatidic acid produced by phospholipase D (Mishra et al., 2006; Zhang et al., 2004). Although these differences in interaction have not clearly explained their distinct functions yet, these results suggest that the distinct functions are due to the ability to interact with different components. For instance, AHG3/AtPP2CA interacts with AKT2, a potassium transporter (Cherel et al., 2002). Analysis of the proteins interacting with AHG1 and AHG3 might offer some clues to their distinct functions. Even though they have different substrate preferences, they presumably have an ability to share the targets based on the relatively high similarity in amino acid sequences. Taken together, similar biochemical properties but different spatio-temporal expression patterns and different preferences for interacting proteins would give rise to the distinct and overlapping functions of these PP2Cs. These arguments can be extended to the relationships among other ABA-related PP2Cs. When one PP2C malfunctions, other PP2Cs can compensate for the defect but only partly because of different expression patterns and target preferences.
Our studies show the importance of elucidating the precise roles of each PP2C in order to understand the ABA response. For this purpose, substrates must be determined. Proteomic approaches have led to great successes in analyzing the phosphorylation status of cellular proteins. Characterization of phosphorylated proteins in ahg1-1, ahg3-1 and the ahg1-1 ahg3-1 double mutant would be a worthwhile approach. In addition, despite the strong ABA hypersensitivity, the ahg1-1 ahg3-1 double mutant was still able to germinate, suggesting the important functions of other PP2Cs. It would be of interest to construct triple or quadruple PP2C-defective mutants to reveal the contribution of other PP2Cs and to understand the functions of PP2Cs and ABA in germination.
Plant materials and growth conditions
Arabidopsis thaliana (L.) Heynh. ecotypes Columbia (Col), Landsberg erecta (Ler), Wassilewskija (Ws) and Nossen (Nos) were used. Plant growth conditions have been described previously (Nishimura et al., 2004). The ahg1-2 line (54-0076-3, Nos background) and the ahg1-3 line (SALK_095052, Col background) were obtained from RIKEN BioResource Center and the Arabidopsis Biological Resource Center, respectively.
Germination assays and ABA measurement
For germination assays, approximately 50 seeds were sown on plates containing 1× Murashige & Skoog salt mix and various concentrations of (±)-cis,trans-abscisic acid (ABA, Sigma-Aldrich; http://www.sigmaaldrich.com/). Germination (emergence of radicles) and post-germination growth (green and expanded cotyledons) were scored daily for 7 days. Extraction and quantitative analyses of ABA were performed as described previously (Nishimura et al., 2005).
Map-based cloning of the AHG1 locus
The ahg1-1 mutant was crossed with Ler, and F2 progeny were obtained. ABA-hypersensitive individuals were selected on medium containing 0.2 μm ABA and grown on normal medium. Isolation of genomic DNA and PCR conditions have been described elsewhere (Hirayama et al., 1999). Additional SSLP and CAPS markers were developed and used (Table S6).
Complementation analysis of ahg1-1
The full-length AHG1 cDNA fragment was obtained by PCR using high-fidelity DNA polymerase (KOD-Plus, Toyobo; http://www.toyobo.co.jp/e/index.htm) with the BamHI linker primer AHG1-BamHI-F1, the KpnI/Acc65I linker primer AHG1-KpnI-R1, and first-strand cDNA mixture (Table S7). It was subcloned into pBluescript SK- (Stratagene; http://www.stratagene.com/), and the nucleotide sequence was confirmed. This AHG1 cDNA was re-introduced into the binary vector pROK2. Agrobacterium tumefaciens strain GV3101 was transformed with the resultant plasmid and used for infection of Arabidopsis plants by the flower dipping method (Clough and Bent, 1998). Transgenic lines were screened by kanamycin tolerance in the next generation.
GUS histochemical staining
For construction of PAHG1::GUS transgenic plants, the genomic DNA fragment spanning from 2137 bp upstream to 117 bp downstream of the translation initiation site of AHG1 was introduced into pBI101. For construction of PAHG3::GUS transgenic plants, the genomic DNA fragment spanning from 1529 bp upstream to 1 bp upstream of the translation initiation site of AHG3 was introduced into pBI101. The resultant plasmids were introduced into Col wild-type plants. Before GUS staining, seed coat/endosperm and embryo were separated by forceps. The samples were treated with 90% acetone on ice for 15 min. After washing with 100 mm NaPO4 buffer (pH 7.0), plant materials were incubated in 100 mm NaPO4 buffer, 10 mm EDTA, 2 mm potassium ferricyanide, 2 mm potassium ferrocyanide, 0.1% Triton X-100 and 0.5 mg ml−1 X-Gluc for 14 h at 37°C.
DNA microarray analysis
Total RNA was isolated with Trizol reagent (Invitrogen; http://www.invitrogen.com/) and RNA was purified using an RNeasy purification kit (Qiagen; http://www.qiagen.com/). cDNA synthesis, cRNA synthesis and hybridization to the Affymetrix ATH1 genome array were performed according to the manufacturer’s instructions (Affymetrix; http://www.affymetrix.com/). The experiments were duplicated using different lots of seeds. Microarray data were processed using the affylmGUI package running on the R program (Smyth, 2004). Microarray data were deposited in National Center for Biotechnology Information (NCBI) GEO database with accession number GSE6638.
Total RNA was isolated with a RNeasy plant mini kit (Qiagen) or as described previously (Nishimura et al., 2004). After treatment with RNase-free DNase I (Qiagen), first-strand cDNA was synthesized from 1 μg total RNA using the ReverTra Plus RT-PCR kit (Toyobo) with random hexamers, according to the manufacturer’s instructions. To perform semi-quantitative RT-PCR, 1/40th of the first-strand reaction mixture was used for PCR reaction with gene-specific primers (Table S7). PCR conditions were 95°C for 90 sec, then 20, 28 or 29 cycles of 95°C for 15 sec, 58°C for 10 sec and 72°C for 60 sec, followed by 72°C for 5 min. DNA fragments for AHG1, AHG3/AtPP2CA, ABI3, ABI5 and AtEm6 were amplified for 28 PCR cycles, and those for the genes identified in microarray experiments were amplified for 29 PCR cycles. The 18s rRNA gene transcripts were amplified for 20 PCR cycles.
Genetic analysis of double mutants
The ahg1-1 mutant (Col) was crossed with abi3-1 (Ler), abi4-1 (Col), abi5-1 (Ws), aba2-1 (Col) or ahg3-1 (Col). Double mutants were obtained from F2 progeny by use of mutant-specific CAPS markers (Table S7). The presence of ahg1-1 was monitored by PCR-based methods; corresponding genomic DNA was amplified using primers AHG1-SacI-F1 and AHG1-R3 and digested with the SacI restriction enzyme. The ahg3-1 fragment was digested with BspHI, the abi3-1 fragment with Acll, the abi4-1 and abi5-1 fragments with Avall, and the aba2-1 fragment with AflII.
We thank Drs Julian Schroeder and Bearice Rose for critical reading and comments on the manuscript. We are grateful to Drs M. Okamoto and E. Nambara, RIKEN PSC, for providing various cDNA mixtures, and to the Arabidopsis Biological Resource Center and RIKEN BioResource Center for providing the T-DNA insertion line and the Ds-transposon tagging line, respectively. This work was partly supported by a Grant-in-Aid from the Ministry of Education, Sports, Culture, Science and Technology of Japan, and the RIKEN president’s Special Research Grant (to T.H.), and partly by the Program for Promotion of Basic Research Activities for Innovative Biosciences (K.S.).