Sequence characterization of the RAP2.1 gene
The DNA sequence of RAP2.1 gene was first identified by Okamuro et al. [25]. The 836 bp of the full-length cDNA contains an open reading frame encoding a protein of 153 amino acids, with a predicted molecular mass of 17.2 kDa and a calculated pI of 9.82. Examination of the RAP2.1 protein sequence, using programs PROSITE [26] and PredictNLS [27], identified a basic amino acid stretch (10MRKRRQ15) in its N-terminus that resembles a classical nuclear localization signal (NLS) [27]. RAP2.1 nuclear import could be mediated by its NLS, as is the case for many transcription factors, such as RAP2.4 from Arabidopsis [11], OsWRKY31 from rice [28], and GhDBP1 from cotton [29]. In addition to the NLS sequence, RAP2.1 also contains a typical AP2 DNA-binding domain and an acidic region in its C-terminus, which might act as a transcriptional regulatory domain (Figure 1A, see Additional file 1: Figure S1). The AP2 domain contains conserved valine (V) in the 14th position and glutamic acid (E) in the 19th position, both of which have been reported as conserved in the DREB subfamily [29]. Alignment of RAP2.1 against various AP2/ERF proteins revealed that RAP2.1 also contains another conserved domain, DLNxxP (Figure 1A, see Additional file 1: Figure S1). This domain is very similar to the EAR motif [L/FDLNL/F(x)P], which has been identified in many transcriptional repressors of various species [13], suggesting that the RAP2.1 might function as a DREB-type transcriptional repressor in Arabidopsis.
RAP2.1 binds to the DRE element and acts as a transcriptional repressor
To examine whether RAP2.1 could interact specifically with the DRE motif, we expressed the N-terminal 120 aa of the RAP2.1 protein (containing the AP2 DNA-binding domain) as a GST fusion in E. coli, and the purified recombinant proteins were then used for gel mobility shift assays. As shown in Figure 1B, the wild-type DRE (wDRE) interacted with the GST-RAP2.1 fusion protein and was retarded on the gel (lanes 4-6). In contrast, no retardation band was detected for the oligonucleotide harboring the mutant version of the DRE element (mDRE, lanes 7-9). As a control, GST was shown not to bind with wDRE (lanes 1-3). These results suggested that the RAP2.1 protein could bind specifically to the DRE element in vitro. However, it is more important to determine whether DRE-binding activity correlates with the transcriptional activity of RAP2.1 in vivo.
As mentioned above, RAP2.1 contains a conserved sequence (KPDLNQIP) similar to the EAR-motif, which has been reported as a transcriptional repression domain [13, 29]. To determine whether RAP2.1 was capable of repressing DRE-mediated transcription, we performed transient expression assay in Arabidopsis leaves using a reporter gene containing three copies of the DRE sequence from the RD29A promoter, 3×DRE-FLUC (Figure 1C). As shown in Figure 1D, expression of RAP2.1 resulted in a substantial reduction of the expression of the reporter gene FLUC. Further, DREB1A, a well-known Arabidopsis transcriptional activator [15], induced activation of FLUC by about 7-fold, but co-expression of RAP2.1 prevented this activation (Figure 1D). To determine whether the conserved EAR-like-motif was important for the RAP2.1-mediated repression, site-specific mutations were made to convert four conserved amino acids (D143L144N145QIP148) to alanines (AAAQIA) (Figure 1A). As expected, the ability of RAP2.1 to repress transcription was abolished when the EAR-motif was mutated (Figure 1D). Together, these results suggest that RAP2.1 may function as a transcriptional repressor, and an intrinsic repression domain exists in the C-terminal EAR-motif, which contains four conserved amino acids (D, L, N, and P) important for the repression activity of RAP2.1.
RAP2.1 expression is greatly induced by cold and drought stresses
Fowler and Thomashow (2002) showed that transcript levels of RAP2.1 exhibited up-regulation at low temperatures by microarray analysis [30]. To investigate RAP2.1 expression patterns in response to different abiotic stresses, northern blot analysis was conducted using a gene-specific probe for RAP2.1. As shown in Figure 2A, the expression level of RAP2.1 was greatly induced by cold and drought stresses, and slightly increased by high salinity stress. In contrast, RAP2.1 expression was not influenced by ABA treatment. Similar results were also obtained in the ABA-deficient mutant aba4-1 [31], as shown in Figure 2B, indicating that the expression of RAP2.1 was governed via an ABA-independent pathway under drought and cold conditions. Interestingly, in all tested Arabidopsis plants, the elevated expression level of RAP2.1 resulting from 12-h of drought or cold treatment was reduced by 3-h of rehydration (Figure 2A and 2B).
The promoter sequence of the RAP2.1 gene, with a length of 1.5-kb (containing the 5'-UTR), was isolated from the Arabidopsis genome. Histochemical analysis of the RAP2.1 promoter-driven β-glucuronidase (RAP2.1p:GUS) expression assay is shown in Figure 2C (a-f). The RAP2.1 promoter was only responsive to cold (b) and drought (c) stresses, but not to normal conditions (a), high salt stress (d), PEG8000 (e), or ABA (f) treatments. Combining the results from the northern blot and histochemical GUS assays, we conclude that expression of the RAP2.1 gene was greatly induced by both cold and drought stresses through an ABA-independent regulatory pathway. This conclusion provides the insight that RAP2.1 may play a critical role in modulating plant responses to drought and cold stresses.
RAP2.1 negatively regulates drought and cold stresses in Arabidopsis
To investigate the in vivo role of RAP2.1 in modulating plant responses to drought and cold stresses, "loss of function" and "gain of function" phenotypes of the RAP2.1 protein were identified. For loss of function analysis, we used two Arabidopsis T-DNA insertion mutant alleles of RAP2.1, rap2.1-1 (SALK_092889) and rap2.1-2 (SALK_097874), in which the T-DNAs were inserted into the promoter and 5'-UTR regions of the RAP2.1 gene, respectively (Figure 3A). Both rap2.1-1 and rap2.1-2 were RAP2.1 null alleles, showing no detectable RAP2.1 transcript in either allele by northern analysis, even after 12-h of cold treatment (Figure 3B). For gain of function analysis, RAP2.1-overexpressing transgenic lines (35S:myc:RAP2.1) were generated using wild-type plants as background. To perform functional characterization of the EAR-motif of RAP2.1, we also generated a transgenic line expressing a variant of RAP2.1 in the rap2.1-2 mutant background (rap2.1-2/35S:myc:RAP2.1m). This transgenic contained a site-specific mutation that converted the DLNQIP EAR-motif at positions 143-148 to AAAQIA at the same position (as shown in Figure 1A). As a positive control, the transgenic line rap2.1-2/35S:myc:RAP2.1 was also generated by expressing the wild type RAP2.1 gene in the rap2.1-2 mutant background. For each transgenic, at least five independent homozygous lines with high levels of transgene expression (assayed by western blot analysis with anti-myc antibody, data not shown) were identified, and two of these transgenics were randomly selected for subsequent stress tolerance assays.
Firstly, wild-type, mutant and transgenic plants were subjected to cold stress. RAP2.1 expression caused increased cold sensitivity, based on growth phenotype (Figure 3C) and relative electrolyte leakage assay (Figure 3D). After 3 weeks of chilling, leaf chlorosis and necrosis were visible in 355S:myc:RAP2.1 plants (line 5), but could not be detected in wild type plants (Figure 3C). A similar phenotype was also detected in another 355S:myc:RAP2.1 line (line 8, data not shown). The RAP2.1 mutants, rap2.1-1 and rap2.1-2, displayed significantly better growth than the wild type plants (Figure 3C). The phenotype results were confirmed by a relative electrolyte leakage assay. Electrolyte leakage from 35S:myc:RAP2.1 plants was approximately 1.5-fold greater than that of wild type plants under either 4°C or 0°C treatment. In contrast, leakage from rap2.1 mutants, rap2.1-1 and rap2.1-2, was only about 70% of that from wild type, even though leakage was similar at the 22°C control temperature (Figure 3D). Expression of the wild-type allele 35S:myc:RAP2.1 suppressed the cold tolerance of rap2.1-2 plants, while the EAR-motif mutated allele 35S:myc:RAP2.1m could not (Figure 3C and 3D), further confirming that RAP2.1 can function as a negative regulator in plant responses to cold stress and that the EAR-motif of RAP2.1 is directly involved in this process.
Next, the wild-type, mutant and RAP2.1-overexpressing plants were further subjected to drought stress. For the 2-week-old seedlings, wild type and RAP2.1 mutant plants began wilting 30 min after putting them on dry paper, while the transgenic plants overexpressing RAP2.1 could speed up the process, displaying wilt within several minutes. After withholding water for 8 h and rehydration for 48 h, the mutant plants recovered much better (survival ratio of 75.4 ± 6.9% for rap2.1-1 and 76.2 ± 8.9% for rap2.1-2) compared to the wild type plants (51.7 ± 6.9%), while only about a quarter of the 35S:myc:RAP2.1 plants survived (25.9 ± 3.7%, for line 5) (Figure 3E). To test whether the altered drought tolerance of the RAP2.1-overexpressing plants and rap2.1 mutants might be due to leaf transpiration, water-loss rates were measured. As shown in Figure 3F, no significant differences were found between the plants of the three genotypes. A similar phenotype was also detected for another 355S:myc:RAP2.1 line (line 8) (data not shown). Together, these results suggest that enhanced or reduced drought tolerance of RAP2.1-overexpressing or rap2.1 mutant plants likely resulted from altered expression of drought-specific responsive genes via an ABA-independent pathway. This would be consistent with the notion that the expression of RAP2.1 is up-regulated under drought conditions by an ABA-independent pathway (Figure 2).
RAP2.1 binds in vivo to the promoters of RD/COR genes and regulates their expression
The transcriptional repression activity of RAP2.1, and the effect of altering RAP2.1 expression levels on plant tolerance to cold and drought stresses, suggested that stress responsive genes may be the major targets of RAP2.1 in vivo. Previous studies have revealed the presence of DRE/CRTs in the promoters of RD/COR/KIN (responsive to dehydration/cold-responsive/cold-inducible) genes, a class of genes up-regulated by cold, water deprivation, salt stress and ABA stimulus [3, 9]. We included three genes in our analysis, RD29A/COR78, COR15A, and KIN1. The distribution of sites and the core sequences of the DRE/CRT elements in the promoters of these three genes, as identified with a plant cis-elements database (PLACE, http://www.dna.affrc.go.jp/PLACE/) search, are illustrated in Figure 4A (also see Additional file 1: Table S1).
To determine whether these genes behaved as direct targets of RAP2.1 in vivo, we used a chromatin immunoprecipition (ChIP) approach, taking advantage of the cold-treated overexpressing transgenic plants, 35S:myc:RAP2.1 (line 5), which express a myc-tagged version of RAP2.1. Wild type plants with same treatment were used as a control. Specific immunoprecipition was conducted with an anti-myc antibody and an anti-His antibody was used as a non-specific IgG control. Actin was used as a control for the non-DRE fragment. As shown in Figure 4A (left panel), both of the promoter fragments of RD29A (DR) and COR15A (DC), which contained more than one tandem DRE/CRT, were specifically amplified from the anti-myc immunoprecipitates of 35S:myc:RAP2.1 extracts (Figure 4A, right panel). However, the KIN1 promoter fragment (DK), which contained only one DRE, could not be recovered from the immunoprecipitates with either the anti-myc or the anti-His antibodies. Similar cases were also detected for the Actin control fragment. While in the wild type seedlings (WT), there was no myc-tagged protein expressed, and no DNA fragment could be detected from neither anti-myc nor anti-His immunoprecipitates. Additionally, RAP2.1 binding was quantitatively determined using real-time PCR of immunoprecipitates with either anti-HA or anti-myc antibodies. The results fully corroborated the specific binding of RAP2.1 to these promoters in vivo (Figure 4B). Both DR and DC fragments included in this analysis showed detectable binding to RAP2.1, while no binding could be detected for the DK and Actin fragments.
Next, we carried out a transient expression assay to determine whether RAP2.1 could repress the transcription of the reporter gene driven by the DRE/CRT fragments identified in the ChIP assay. As shown in Figure 4C, the expression results were consistent with the ChIP results. For the LUC reporters driven by the DRE fragments of the RD/COR gene promoters, RAP2.1 was able to repress the basal activity of the reporter, as well as expression activity in the presence of an additional transcriptional activator, DREB1A. However, no obvious repression was detected in the reporter driven by the DRE fragment from the KIN promoter. These data demonstrate that RD/COR genes are likely direct targets of RAP2.1 in vivo.
The RAP2.1 promoter contains three DRE/CRTs arranged in tandem (Figure 4A and Additional file 1: Table S1). To test whether RAP2.1 could bind to its own promoter in vivo, specific primers were used to amplify the DRE fragments of the RAP2.1 promoter (D1 and D2) from the ChIP immunoprecipitates. As shown in Figure 4A, both the fragments were detected in the anti-myc immunoprecipitates. The D1 fragment, which contained two DRE/CRTs, was detected at particularly high levels. Furthermore, we determined the binding efficiency of RAP2.1 protein to D1 and D2 fragments by real-time PCR assay. Consistent with the above semi-quantitative PCR results, RAP2.1 was more enriched at the D1 fragment (about 9.25% of input) than D2 (about 3.16% of input), indicating that RAP2.1 binds to D1 fragment with higher efficiency than D2 fragment (Figure 4B). The transient expression assay also showed that the LUC reporter, driven by the D1 fragment, was repressed by RAP2.1 (Figure 4C). This result suggests that RAP2.1 can bind to the DRE elements present in its own promoter and repress its own expression, indicating a negative feedback control in the regulation of expression of RAP2.1.
Since RD/COR genes were found to be direct targets of RAP2.1, we used quantitative real-time PCR to determine transcriptional levels of these genes in seedlings of wild-type, rap2.1-2 and 35S:myc:RAP2.1 plants under cold (Figure 5A) or drought (Figure 5B) stresses. In wild-type plants, RAP2.1 mRNA accumulation began 6 to 12 h after exposure of the plants to cold (4°C) and reached a maximum expression level at 12 h, after which levels of the transcript were maintained (Figure 5A). Transcript abundance of the RD29A and COR15A genes slowly and gradually increased over 12 h, reaching a maximum abundance at 24 h after cold treatment. Low temperature-induced transcripts were accumulated to a lesser extent in RAP2.1-overexpressing plants than in wild-type. In contrast, transcripts accumulated to greater levels in rap2.1-2 seedlings (Figure 5A). The expression of DREB1/CBF genes, the upstream regulators of RD/COR genes, were induced rapidly (within 15 min) by low temperature in wild-type plants, and transcript accumulation increased with cold treatment [8]. The expression of DREB1B/CBF1 preceded that of DREB1A/CBF3 (Figure 5A). Furthermore, cold-induced DREB1/CBF transcript accumulation was similar in RAP2.1-overexpressing and rap2.1-2 plants, relative to the control wild type over a 24-h time frame (Figure 5A). A positive regulator of DREB1/CBF expression, ICE1 (Inducer of CBF Expression 1) [32], was also detected. ICE1 transcript abundance was not affected by cold and was similar in the plants of all three genotypes (Figure 5A). Together, these results indicate that RAP2.1 negatively regulates expression of the RD/COR genes and the DREB1/CBF regulons, but does not alter the transcript levels of DRAB1/CBFs or ICE1 during cold stress.
Similar results were also detected under drought stress. As shown in Figure 5B, drought-induced transcript accumulation of the RD29 and COR15A genes to a lesser extent in RAP2.1-overexpressing plants and to a greater extent in rap2.1-2 mutants, relative to the wild type control. The expression of DREB2A and DREB2B were up-regulated with drought treatment in wild type seedlings, as reported [3, 9], and similar expression levels were detected in both RAP2.1-overexpressing and rap2.1-2 plants (Figure 5B). AtEm6 (Early methionine-labelled 6), which is regulated by desiccation through a DREB-independent pathway [33], exhibited similar expression patterns in wild type, RAP2.1 overexpression and rap2.1 mutant plants (Figure 5B). These data indicate that RAP2.1 represses the expression of RD/COR genes, but not DREB2 genes, under drought stress.