ABA induction of miR159 controls transcript levels of two MYB factors during Arabidopsis seed germination

Authors


(fax +1 212 327 8327; e-mail chua@mail.rockefeller.edu).

Summary

Upon seed imbibition, abscisic acid (ABA) levels decrease to allow embryos to germinate and develop into seedlings. However, under abiotic stress conditions, ABA levels remain high, and growth and development are arrested. Several transcription factors, including abscisic acid-insensitive (ABI)3 and ABI5, are known to control this developmental checkpoint. Here, we show that, in germinating Arabidopsis thaliana seeds, ABA induces the accumulation of microRNA 159 (miR159) in an ABI3-dependent fashion, and miRNA159 mediates cleavage of MYB101 and MYB33 transcripts in vitro and in vivo. The two MYB transcription factors function as positive regulators of ABA responses, as null mutants of myb33 and myb101 show hyposensitivity to the hormone. Consistent with this, miR159 over-expression suppresses MYB33 and MYB101 transcript levels and renders plants hyposensitive to ABA, whereas transgenic plants over-expressing cleavage-resistant forms of MYB33 and MYB101 are hypersensitive, as are plants over-expressing the Turnip mosaic virus (TuMV) P1/HC-Pro viral protein that is known to inhibit miRNA function. Our results suggest that ABA-induced accumulation of miR159 is a homeostatic mechanism to direct MYB33 and MYB101 transcript degradation to desensitize hormone signaling during seedling stress responses.

Introduction

A class of small RNA molecules regulates gene expression in eukaryotic organisms ranging from animals to plants. These short RNAs or microRNAs (miRNAs) are 20–22 nucleotides long, and specifically base-pair to target messenger RNAs to repress their translation or induce their cleavage (Bartel, 2004). There are several indications that miRNAs regulate important aspects of plant development (Jones-Rhoades et al., 2006). First, mutations in genes encoding enzymes critical for miRNA biogenesis, such as DCL1, HYL1, HEN1 and AGO1, result in developmental defects (Jacobsen et al., 1999; Lu and Fedoroff, 2000; Park et al., 2002; Vaucheret et al., 2004). Second, target mRNAs for many known miRNAs in Arabidopsis thaliana encode putative transcription factors (Kidner and Martienssen, 2005). Third, several mRNAs targeted by plant miRNAs are involved in various developmental processes. For example, miR319 controls the expression of TCP factors during leaf morphogenesis (Palatnik et al., 2003), miR165/166 control HD-ZIP transcription factor mRNAs involved in the determination of the adaxial/abaxial leaf axis (Emery et al., 2003; Juarez et al., 2004; Kidner and Martienssen, 2004) and vascular development (Kim et al., 2005), miR164 regulates CUC1/CUC2 and NAC1 mRNAs that function in flower development and lateral root initiation, respectively (Baker et al., 2005; Guo et al., 2005; Mallory et al., 2004), and miR172-targeted mRNAs have been implicated in setting the flowering time of Arabidopsis (Aukerman and Sakai, 2003; Chen, 2004).

In addition to their developmental role, recent evidence has implicated plant hormone signaling in the regulation of miRNA gene expression. Gibberellic acid (GA) has been shown to modulate miR159 levels during anther development (Achard et al., 2004), and miR164 is induced by auxin to clearNAC1 mRNA to reset auxin signals (Guo et al., 2005). Furthermore, miR160 has been shown to regulate expression of ARF17, an auxin response transcription factor (Mallory et al., 2005), and the hyl1-1 mutant displays impaired responses to auxin, ABA and cytokinin (Lu and Fedoroff, 2000). Finally, the predicted targets for miRNAs include several mRNAs involved in hormone responses, such as those encoding other auxin response factors and TIR1 (Jones-Rhoades and Bartel, 2004; Rhoades et al., 2002; Sunkar and Zhu, 2004; Wang et al., 2004).

These observations, and the finding that the hyl1 mutant is hypersensitive to ABA during germination (Lu and Fedoroff, 2000), prompted us to hypothesize that components of the ABA signaling pathway might be regulated by miRNAs. By isolating and cloning small RNA molecules present in A. thaliana during early germination stages, we identified miR159 as being induced by ABA and drought treatments. We analyzed the function of its target MYB101 and MYB33 transcripts and showed that the two encoded MYB factors are positive regulators of ABA responses. The transcription factors ABI3 and ABI5 are known to be important regulators of ABA responses during germination (Lopez-Molina et al., 2002; Zhang et al., 2005). We used abi3-1 and abi5-4 mutants to relate miR159 to known components of the ABA signaling pathway, and found that ABA-induced miR159 accumulation requires ABI3 but is only partially dependent on ABI5. Our results indicate that MYB33 and MYB101 are positive regulators of ABA responses during germination and are subject to ABA-dependent miR159 regulation.

Results

miR159 accumulates in response to ABA and drought during seed germination

To broaden our understanding of the molecular events surrounding germination and its control by ABA, we sought to identify small RNA molecules present in early stages of seed germination. As the hyl1 mutant displays ABA hypersensitivity (Lu and Fedoroff, 2000), in principle we considered it likely that miRNAs are involved in ABA responses during seed germination. To address this issue, we germinated A. thaliana seeds (Col-0 ecotype) and treated 1-day-old seedlings with ABA for 1 or 2 days. Total RNA was isolated and pooled and used for small RNA cloning according to published procedures (Elbashir et al., 2001). We identified miR159a among the first sequences obtained, and confirmed its increased accumulation in ABA-treated seedlings by Northern blot analysis. We found that miR159 accumulation was maximal at 4–8 h after ABA addition (approximately fivefold), and its levels remained high relative to untreated control seedlings even after 48 h (Figure 1a). Moreover, when seedlings germinated for 48 h on MS medium were subjected to drought treatment for 8 h, miR159 levels were also increased (data not shown, and Figure 7).

Figure 1.

Figure 1.

 Regulation of miR159 expression by abscisic acid.
(a) miR159 expression is induced by ABA. Untreated (-ABA) WT (Col-0) seedlings were collected 1 day post-imbibition without further treatment (0) or 24 and 48 h later. Alternatively, 1-day-old seedlings were treated with 3.0 μm ABA for various time periods (in hours, +ABA). Each lane contained 20 μg total RNA. Numbers below the top panel indicate the fold induction over the untreated level, normalized to the 5S rRNA loading control signal as determined by Phosphorimager analysis of the Northern blots.
(b) Expression of individual MIR159 genes as determined by RT-PCR. RNA samples were extracted from 2-day-old untreated seedlings (C), 1-day-old seedlings treated with 3.0 μm ABA for 1 day (ABA), or 2-day-old seedlings after 8 h of drought treatment (Dr) and used for reverse transcription (+RT lanes). Control reactions lacking RT enzyme are also shown (−RT). The lane designated ‘DNA’ indicates the genomic DNA positive control used for PCR amplification, whereas ‘–’ indicates the PCR negative control. Pre-a, pre-b and pre-c refer to the detection of pre-miR159a, b and c transcripts, respectively.
(c) Proposed secondary structures for miR159a, b and c precursors. The vertical line indicates the position of the mature miRNA. miR159b and c differ from miR159a in the last one and two nucleotides at their 3′ ends, respectively (black circles). The loop lengths of pre-miR159a, pre-miR159b and pre-miR159c not shown in the diagram are indicated.

Figure 1.

Figure 1.

 Regulation of miR159 expression by abscisic acid.
(a) miR159 expression is induced by ABA. Untreated (-ABA) WT (Col-0) seedlings were collected 1 day post-imbibition without further treatment (0) or 24 and 48 h later. Alternatively, 1-day-old seedlings were treated with 3.0 μm ABA for various time periods (in hours, +ABA). Each lane contained 20 μg total RNA. Numbers below the top panel indicate the fold induction over the untreated level, normalized to the 5S rRNA loading control signal as determined by Phosphorimager analysis of the Northern blots.
(b) Expression of individual MIR159 genes as determined by RT-PCR. RNA samples were extracted from 2-day-old untreated seedlings (C), 1-day-old seedlings treated with 3.0 μm ABA for 1 day (ABA), or 2-day-old seedlings after 8 h of drought treatment (Dr) and used for reverse transcription (+RT lanes). Control reactions lacking RT enzyme are also shown (−RT). The lane designated ‘DNA’ indicates the genomic DNA positive control used for PCR amplification, whereas ‘–’ indicates the PCR negative control. Pre-a, pre-b and pre-c refer to the detection of pre-miR159a, b and c transcripts, respectively.
(c) Proposed secondary structures for miR159a, b and c precursors. The vertical line indicates the position of the mature miRNA. miR159b and c differ from miR159a in the last one and two nucleotides at their 3′ ends, respectively (black circles). The loop lengths of pre-miR159a, pre-miR159b and pre-miR159c not shown in the diagram are indicated.

Figure 7.

 ABI3 regulates miR159 accumulation during seedling development.
(a) Total RNA (20 μg) from WT (Ler) or abi3-1 mutant plants was extracted from 1-day-old seedlings subjected to no treatment for 24 h (C), 3.0 μm ABA for 24 h (ABA) or 8 h drought treatment (Dr). Northern blots were used to analyze miR159 or 5S RNA levels, as indicated. Numbers below the top panel indicate the fold induction over the untreated control level, normalized to the 5S rRNA signal as determined by Phosphorimager analysis.
(b) Total RNA (20 μg) from WT (Ws) or abi5-4 mutant plants was extracted from 1-day-old seedlings subjected to no treatment for 24 h (C), 3.0 μm ABA for 24 h (ABA) or 8 h drought treatment (Dr). Northern blots were used to analyze miR159 or 5S RNA levels, as indicated. Numbers below the top panel indicate the fold induction over the untreated control level, normalized to the 5S rRNA signal as determined by Phosphorimager analysis.
(c) A schematic diagram showing the proposed regulation of miR159 expression by ABI3 during seedling response to ABA. The model proposes that ABA signal is transduced through MYB33 and MYB101 (dashed arrow) as they act as positive regulators of ABA responses. However, their site of action relative to the other positive regulators such as ABI3 and ABI5 is not known.

miR159 is encoded by three different loci, MIR159a, MIR159b and MIR159c. Mature miR159b and miR159c differ from miR159a by one and two nucleotides, respectively, at the 3′ end (circled in Figure 1c). Because the oligonucleotide probe used cannot discriminate amongst these three transcripts, we used gene-specific RT-PCR to analyze pre-miR159 expression. Figure 1(b) shows that the three members of the MIR159 gene family are expressed during germination as we could detect the amplification products corresponding to each of the three longer RNA precursors. However, we did not detect an increased expression level for any of the pre-miRNAs upon ABA treatment or under drought conditions. Thus, under our assay conditions, we could not distinguish whether a particular locus is induced by ABA or whether a subsequent biogenesis step limits accumulation of mature miRNA from a particular pre-miRNA.

miR159 mediates cleavage of MYB33 and MYB101 mRNAs

Messenger RNAs encoding a subfamily of MYB transcription factors have been predicted to be targets of miR159 (Rhoades et al., 2002). During leaf morphogenesis, MYB33 was previously confirmed to be an miR159 target (Palatnik et al., 2003), and more recently, this gene was found to respond to GA regulation during flowering (Achard et al., 2004). To determine mRNA targets regulated by miR159 during germination in response to ABA, we analyzed expression of the various MYB transcripts. Northern blot analysis failed to detect any of the transcripts (data not shown), suggesting that they are expressed at levels below our detection limits or absent at this stage of development. When we performed RT-PCR experiments, only MYB101 transcripts were detected (Figure 2a), whereas transcripts for MYB33, MYB65, MYB104 or MYB120 were not detectable (data not shown). Figure 2(a) shows that the MYB101 transcript was present in control samples but not in samples treated with ABA nor subjected to drought treatment, suggesting that this mRNA is negatively regulated by ABA/drought.

Figure 2.

MYB33 and MYB101 transcripts are targets of miR159.
(a) RT-PCR analysis of RNA samples from untreated 2-day-old seedlings (C), 1-day-old seedlings treated with ABA (ABA; 3.0 μm, for 1 day), or 2-day-old seedlings subjected to drought (Dr; 8 h) (see Figure 1). As a control, RT-PCR was performed to amplify the ACT2 mRNA. The difference in size obtained with the RNA and the DNA samples is due to the presence of an intron in the genomic ACT2 sequence. This difference serves as a convenient control for possible DNA contamination in the RNA samples. The lane designated ‘DNA’ contained genomic DNA which was used as a positive control, whereas the lane designated ‘–’ was a sample without DNA that was used as a negative control for PCR reactions.
(b) [32P]-5′-end-labeled RNA fragments derived from MYB101 and MYB33 mRNAs were incubated in wheat germ extracts under processing conditions for the different times (min) indicated. RNA samples were separated on a 6% polyacrylamide gel and subjected to autoradiography. The diagrams on the right show two RNA species: the RNA substrate with a [32P]-labeled cap (*G), and the 5′ fragment after cleavage at the miRNA recognition site (marked with an asterisk).
(c) Base-pairing interaction between miR159a and MYB33 or MYB101 transcripts. G:U wobble base pairs are indicated by a dot. Vertical arrows indicate the 5′ positions of the cleaved mRNA fragments identified by 5′ RACE, and the numbers refer to the number of independent clones analyzed.

We used wheat germ extracts to determine whether MYB33 and MYB101 transcripts are substrates of miR159 in vitro. miR159 was detected previously in Triticum aestivum (Axtell and Bartel, 2005), and we have confirmed its presence in wheat germ extract by Northern blot analysis (data not shown). Both MYB101- and MYB33-derived transcripts could be efficiently processed in wheat germ extracts (Figure 2b). To confirm regulation of MYB transcripts by miR159 in vivo, we used 5′ RACE to identify their cleavage products in RNA samples obtained from ABA-treated seedlings. Figure 2(c) shows the detection of cleavage products for both MYB101 and MYB33 transcripts. Out of eight independent 5′ RACE clones analyzed for MYB101 transcript cleavage, four were located three nucleotides downstream of the predicted cleavage site. This result suggests that the 3′ half mRNA product is unstable, and its degradation proceeds rapidly after miRNA-directed cleavage. The same RNA sample was used for amplification of the MYB33-derived 5′ RACE products, but shorter products could not be identified, indicating that degradation of the RNA sample was not responsible for the shorter 5′ RACE products observed for MYB101. The location of the cleavage site in MYB101 and MYB33 is consistent with cleavage directed by miR159 and not by the related miR319 that is one nucleotide shorter at the 5′ end and would direct cleavage one nucleotide upstream in the MYB101 and MYB33 transcripts. miR319 has been shown to target mRNAs encoding members of the TCP family of transcription factors (Palatnik et al., 2003). We failed to obtain equivalent 5′ RACE products for the related MYB104 and MYB120 transcripts from seedling RNA samples. From the above experiments, we conclude that both MYB33 and MYB101 mRNAs are substrates for miR159-directed cleavage during seed germination in the presence of ABA.

miR159 regulates ABA signaling

Next, we analyzed the effect of altering miR159 expression levels on ABA responses during seed germination. As an alternative to producing a triple mutant plant devoid of miR159a, miR159b and miR159c, we used a transgenic plant over-expressing the viral silencing suppressor protein P1/HC-Pro from turnip mosaic virus (TuMV), which interferes with miRNA function and accumulates presumably inactive miRNAs (Kasschau et al., 2003; Mallory et al., 2002), including miR159 (Palatnik et al., 2003). Consistent with this result, we determined by RT-PCR that HC-Pro-over-expressing plants also accumulated miRNA-regulated transcripts, including MYB101, MYB33 and ATHB15, an unrelated transcript targeted by miR165/166 (Figure 3b) (Kim et al., 2005; Williams et al., 2005). Therefore, we reasoned that over-expression of HC-Pro should phenocopy a lack of miR159 expression.

Figure 3.

 miR159 regulates ABA-induced growth arrest during germination.
(a) Total RNA (20 μg) from 2-week-old WT plants (Col-0) or TuMV P1/HC-Pro-over-expressing plants (HC-Pro) was resolved on a 15% polyacrylamide denaturing gel. Northern blot was used to analyze miR159 and 5S rRNA levels.
(b) Similar RNA preparations from 2-week-old WT (Col-0) or HC-Pro-over-expressing plants were used in RT-PCR experiments to amplify transcripts corresponding to MYB101, MYB33, ATHB15 (an miR165/166 target) or ACT2 as an internal loading control. ‘DNA’ refers to genomic DNA used as a positive amplification control for MYB101 and ACT2 and cDNA in the case of ATHB15 amplification, while ‘–’ indicates the PCR negative control. The difference in size obtained with the RNA and the DNA samples is due to the presence of an intron in the genomic ACT2 sequence.
(c) Seeds from WT (Col-0, n = 200 seeds, minimum, per treatment) or TuMV P1/HC-Pro-over-expressing plants (n = 100 seeds, minimum, per treatment) were germinated on MS medium containing increasing concentrations of ABA. Germination efficiencies were determined 7 days after transferring plates to continuous light at 23°C. Values are the mean of three independent experiments and error bars indicate standard deviations.
(d) Seeds from WT (Ler, n = 100 seeds, minimum, per treatment) or 35S:miR159a-1- and 35S:miR159a-3-over-expressing plant lines (n = 50 seeds, minimum, per treatment) (Achard et al., 2004) were germinated in the presence of 0.25, 0.5 or 1.0 μm ABA. Germination efficiencies were determined 7 days after transferring plates to continuous light at 23°C. Values are the mean of three independent experiments and error bars indicate standard deviations.

We found that, under normal conditions, germination of HC-Pro transgenic seeds was indistinguishable from that of wild-type (WT, Col-0) seeds after 7 days (Figure 3c), indicating that the presence of the HC-Pro viral suppressor protein does not interfere with normal seed germination. By contrast, in the presence of various ABA concentrations, there was a reduced germination efficiency compared to WT seeds, with a marked difference seen at 1.0 μm ABA (Figure 3c). These results are consistent with the previous finding that a mutation in the HYL1 gene known to be involved in miRNA biogenesis also results in ABA hypersensitivity during germination, whereas germination under normal conditions is not affected (Han et al., 2004; Lu and Fedoroff, 2000; Vazquez et al., 2004). Taken together, our data suggest that reduced levels of active miR159 (due to the inhibitory action of a viral silencing suppressor or the failure to accumulate miRNAs in the absence of the HYL1 protein) result in ABA hypersensitivity during seed germination. Under normal conditions, germination efficiencies were not affected in the HC-Pro-over-expressing plants nor in the hyl1-1 mutant (Lu and Fedoroff, 2000), indicating that reduced miRNA functions are not detrimental at this early developmental stage. However, we cannot rule out a pleiotropic effect due to reduced levels of other unrelated miRNAs in the presence of ABA.

To further clarify the proposed role of miR159, we analyzed previously reported transgenic plant lines over-expressing miR159a (35S:miR159a-1 and 35S:miR159a-3; Achard et al., 2004). At low ABA concentrations (0.25 μm), we observed only a small difference in germination efficiency between both 35S:miR159a-over-expressing lines and WT (Ler). By contrast, at higher ABA concentrations (0.5 and 1.0 μm), 35S:miR159a-1- and 35S:miR159a-3-over-expressing seeds clearly showed higher germination efficiencies compared with WT seeds under the same conditions (Figure 3d). These data show that miR159a over-expression results in hyposensitivity to ABA, in direct contrast to the hypersensitive phenotype observed for the HC-Pro-over-expressing plants and the hyl1-1 mutant. Together, these results support the notion that miR159 negatively regulates ABA responses during seed germination.

MYB101 and MYB33 are positive regulators of ABA signaling during germination

To further explore the role of miR159 regulation of ABA signaling, we altered the expression levels of MYB33 and MYB101 that have been identified as targets of miR159 during early seedling development (see above). From the SALK collection of T-DNA insertion lines (Alonso et al., 2003), we identified knockout lines for MYB101 (myb101, SALK_061355) and MYB33 (myb33, SALK_065473). Both lines contain a T-DNA inserted in the open reading frame (Figure 4a). Homozygous lines for the insertion were assayed for germination efficiency in the absence or presence of ABA. Under control conditions, WT (Col-0) and myb101 and myb33 seeds germinated with similar efficiencies (Figure 4b). By contrast, in the presence of 1.0 μm ABA, WT seeds showed a reduced germination efficiency of 47% after 7 days, whereas both myb101 and myb33 mutants displayed a higher germination efficiency (72% and 79%, respectively, Figure 4b). However, we failed to detect germination differences between the knockout lines and WT at higher or lower ABA concentrations (0.25, 0.5 or 3.0 μm ABA, data not shown). As a negative control for these experiments, a SALK line containing a T-DNA insertion in the related MYB65 gene (myb65, SALK_063552) showed germination efficiencies comparable to those of WT under the same conditions (Figure 4b). The ABA hyposensitivity observed in mutants blocked in MYB33 or MYB101 expression correlates well with the phenotype of plants over-expressing miR159a (Figure 3). These results suggest that both MYB101 and MYB33 participate as positive regulators of growth arrest induced by ABA during seed germination.

Figure 4.

myb33 and myb101 mutants are hyposensitive to ABA.
(a) Schematic representations of MYB101, MYB33 and MYB65 genes. White blocks represent exons, gray boxes show the locations of the R2 and R3 MYB domains interrupted by an intron in all three genes. A vertical black rectangle indicates the recognition sequence for miR159. A vertical arrowhead indicates the position of the inserted T-DNA in the SALK lines used.
(b) Seeds from wild-type (Col-0, n = 100 seeds) or T-DNA insertion mutants for MYB33 (myb33, SALK_065473; n = 50 seeds, minimum, per experiment), MYB101 (myb101, SALK_061355; n = 50 seeds, minimum, per experiment) or MYB65 (myb65, SALK_063552; n = 50 seeds, minimum, per experiment) were germinated on MS medium with or without 1.0 μm ABA. Germination efficiencies were determined 7 days after transferring plates to continuous light at 23°C. Values represent the mean of three independent experiments and error bars indicate standard deviations.

To study the effects of MYB101 and MYB33 over-expression, we used transgenic plants expressing a mutated version of these genes (mMYB101 and mMYB33, respectively) in which the sequence recognized by miR159 has been altered to abolish miRNA base-pairing but maintaining the encoded amino acid sequences (Figure 5a). For mMYB33, we used plants previously shown to generate an MYB33 transcript resistant to miR159-directed cleavage (Achard et al., 2004). In the case of MYB101, we constructed an MYB101 mRNA containing four point mutations in the middle of the miR159 recognition sequence and a myc tag at the C terminus of the protein. Resistance of the mutant MYB101 mRNA to miRNA-directed cleavage was confirmed using the wheat germ in vitro assay (Figure 5b) and a transient expression system comprising Agrobacterium infiltration of Nicotiana benthamiana leaves (Figure 5c,d). In both cases we detected reduced cleavage of mMYB101 mRNA (Figure 5b,c), and corresponding expression of the tagged mMYB101 protein in N. benthamiana leaves (Figure 5d), whilst the WT MYB101 mRNA was efficiently cleaved in vitro (Figure 5b) and accumulated reduced transcript levels with no detectable protein in the transient expression assay (Figure 5c,d).

Figure 5.

 A mutant MYB101 is resistant to miR159-directed cleavage.
(a) Schematic representations of the base-pairing interaction between miR159a and mutated versions of MYB101 or MYB33 mRNAs. Changed nucleotide positions are underlined. The sequence in mMYB33 is as reported by Achard et al. (2004). The unchanged amino acid sequence is shown above the mRNA sequences.
(b) In vitro assay using wheat germ extract and a [32P]-5′-end-labeled RNA fragment corresponding to WT MYB101 (MYB101) or mutant MYB101 (mMYB101) transcripts. RNA samples were treated as in Figure 2. In the mMYB101 samples, the band migrating closely to the corresponding cleavage product is present at time 0 (t, min), and probably represents a degradation product of the input RNA.
(c) Nicotiana benthamiana leaves (untreated, ‘–’) were infiltrated with Agrobacterium cells carrying the empty vector (C), a vector containing 35S:MYB101 (MYB101) or a vector containing 35S:mMYB101 (mMYB101). After 2 days, total RNA was isolated and used for Northern blot analysis of MYB101 expression levels.
(d) Tissue samples of leaves infiltrated with the same constructs as in (c) were used to purify total protein, which was then resolved by 12% SDS–PAGE. MYB101 protein expression was analyzed by Western blot using an antibody against the myc epitope present at the C terminus of the protein. The band labeled with an asterisk represents a non-specific signal and served as a loading control.

We used the 35S:MYB101 (lines 35S:MYB101-1, -2 and -3) and 35S:mMYB101 (lines 35S:mMYB101-7, -8 and -9) constructs described above to transform Arabidopsis Col-0 plants and test their germination efficiencies in the presence of ABA. In the absence of ABA or at a concentration of 0.25 μm, all transgenic plants, including those transformed with the empty vector, and Col-0 seeds germinated with similar efficiencies (Figure 6a). At the intermediate concentration of 0.5 μm, Col-0 seeds and transgenic lines with the empty vector or expressing WT MYB101 germinated with similar efficiencies, whilst transgenic lines expressing 35S:mMYB101 showed slightly reduced germination efficiencies. When the ABA concentration was increased to 1.0 μm, transgenic seeds expressing 35S:MYB101 showed germination levels similar to those obtained with WT (Col-0) plants or plants carrying the empty vector (Figure 6a). By contrast, independent transgenic plant lines expressing 35S:mMYB101 germinated with a significantly lower efficiency under the same conditions (Figure 6a). We confirmed that the observed phenotypes were not the result of potentially different expression levels of the transgenes. Using RT-PCR, we estimated that representative transgenic plants expressed comparable levels of MYB101 and mMYB101 transcripts (Figure 6b). Similar to the results obtained with mMYB101, transgenic plants carrying an mMYB33 construct, 35S:mMYB33-3 and 35S:mMYB33-4 (Achard et al., 2004), also showed a reduced germination efficiency in the presence of 0.5, 1.0 or 3.0 μm ABA when compared to WT (Ler) seeds (Figure 6c). The ABA hypersensitivity caused by the over-expression of mutant MYB101 or MYB33 transcripts correlates well with the hypersensitivity observed in plants lacking miR159 function, as observed in transgenic plants over-expressing the TuMV protein P1/HC-Pro (Figure 3a) or in the hyl1-1 mutant affected in miRNA biogenesis (Han et al., 2004; Lu and Fedoroff, 2000). Together, these results indicate that both MYB33 and MYB101 factors are positive regulators of ABA responses during germination. In their absence, either because of mutations in the MYB genes or owing to miR159 over-expression, seedlings fail to mount a full ABA response, resulting in higher germination efficiencies in the presence of otherwise growth-inhibiting concentrations of ABA.

Figure 6.

 Transgenic plants expressing cleavage-resistant mutant mRNAs of MYB33 and MYB101 are hypersensitive to ABA.
(a) Seeds from WT plants (Col-0, n = 100) or transgenic plants over-expressing 35S:MYB101 (n = 100) or 35S:mMYB101 (n = 100) were germinated on MS medium containing various ABA concentrations as indicated. Germination efficiency (defined as radicle emergence) was determined 4 days after transferring plates to continuous light at 23°C. In the absence (not shown) or at 0.25 μm ABA, germination efficiencies were greater than 95% for all seed lines used. Values represent the mean of at least three independent experiments and error bars indicate standard deviations.
(b) MYB101 mRNA abundance in transgenic over-expressing lines. Total RNA from 1-week-old plants was extracted and used in RT-PCR experiments to detect mRNA from plants over-expressing WT (MYB101) and mutant versions of MYB101 (mMYB101). A representative transgenic line for each construct is shown; similar results were obtained for the other two lines in each case. Amplification of ACT2 mRNA was used as a control (see legend to Figure 2). ‘vec’, empty vector control; ‘DNA’ and ‘–’, see legend to Figure 2.
(c) Seeds from WT plants (Ler, n = 100) or transgenic plants over-expressing a mutant MYB33 coding sequence under the control of a CaMV 35S promoter (35S:mMYB33-3 and 35S:mMYB33-4, n = 40; Achard et al., 2004) were germinated on MS medium containing various ABA concentrations as indicated. Germination efficiency (defined as radicle emergence) was determined 4 days after transferring plates to continuous light at 23°C. Values represent the mean of at least three independent experiments and error bars indicate standard deviations.

ABI3 regulates miR159 expression

Two transcription factors, ABI3 and ABI5, have been implicated in the regulation of ABA responses in germinating seedlings (Lopez-Molina et al., 2003; Zhang et al., 2005). To explore their relationship with ABA-dependent miR159 accumulation, abi3-1 and abi5-4 mutant seeds were germinated for 1 day and then transferred to plates containing 3.0 μm ABA for 24 h, or germinated for 2 days and then subjected to drought treatment for 8 h, and the expression levels of miR159 were analyzed. In the abi3-1 mutant, we did not observe the increase in the accumulation of miR159 upon drought treatment that was observed for WT (Ler) seeds (Figure 7a). In the presence of ABA, accumulation of miR159 was only slightly increased (1.5-fold) when compared with untreated abi3-1 seeds (Figure 7a). In addition, the basal accumulation of miR159 in abi3-1 seedlings was reduced 2- to 3-fold with respect to that of WT (Ler) seedlings. By contrast, in the abi5-4 mutant, we observed increased miR159 accumulation in response to both ABA and drought treatments, although the accumulation level in abi5-4 seedlings was slightly reduced relative to that of the WT (Ws) seedlings (Figure 7b). These results indicate that miR159 accumulation upon ABA treatment depends predominantly on ABA signaling mediated by ABI3. ABI5 has been shown to function downstream of ABI3 (Lopez-Molina et al., 2001); however, its disruption only slightly affects ABA and drought induction of miR159, suggesting that ABI5 might not be directly involved in miR159 gene expression in response to ABA (indicated by the short dashed arrow in Figure 7c).

Discussion

miR159 is induced by ABA during germination

By cloning small RNA molecules present in germinating seedlings treated with ABA, we sought to identify the most abundant miRNAs under these conditions. We found miR159a among the first identified sequences, as well as other sequences likely to correspond to other small non-coding RNAs (data not shown). Northern blot analysis confirmed the increased accumulation of miR159 under both ABA treatment and drought conditions. Whereas direct cloning of small RNA molecules is considered to be an incomplete means to characterize the majority of the sequences present in a given RNA sample, it can be used to analyze the most abundant species. The fact that we could recover miR159 under these conditions validates the approach. In a similar manner, other miRNAs involved in stress responses have been previously identified (Sunkar and Zhu, 2004), although these authors focused on stress responses in adult plants. Consistently, we failed to detect any increase in miR159 levels in 4-week-old plants treated with ABA or by drought (data not shown). Although we recovered the mature miR159a sequence by cloning, three different genes encode this miRNA: MIR159a, MIR159b and MIR159c, and these are indistinguishable by Northern blot analysis. In the multi-genic miR164 family, the expression of a single gene affects petal (Baker et al., 2005) or lateral root development (miR164a; Guo et al., 2005). However, during early germination, we could not determine whether a single gene is responsible for the accumulation of miR159 in response to ABA because all three genes are expressed and are thus likely to have redundant functions (Figure 1).

Regulation of the interaction between miR159 and MYB33/MYB101 by GA and ABA

Regulation of miRNA expression by hormones has been suggested from the predicted targets for a few miRNAs. Two subfamilies of auxin response factors have been predicted and in a few cases proven to be targets of miR160 and miR167 (Jones-Rhoades and Bartel, 2004; Mallory et al., 2005), and TIR1, an miR393 target (Jones-Rhoades and Bartel, 2004; Wang et al., 2004), is an auxin receptor (Dharmasiri et al., 2005; Kepinski and Leyser, 2005). However, there are only two other documented examples of direct hormonal regulation of miRNA expression. Achard et al. (2004) reported that miR159 levels are affected in gai and ga1-2 mutants and that GA levels regulate the accumulation of MYB33 mRNA. Moreover, the authors reported defects associated with anther development in transgenic plants over-expressing miR159a. Another target of miRNA regulation is NAC1 mRNA, which encodes a transcription activator for lateral root initiation (Xie et al., 2000 and 2002). During lateral root development, auxins induce miR164 accumulation, which in turn results in increased cleavage of NAC1 mRNA to reset auxin signaling (Guo et al., 2005; Xie et al., 2000).

Our results here, along with those of Achard et al. (2004), show that miR159 expression is controlled by at least two different hormones, GA and ABA. We have confirmed that neither MYB101- and MYB33-over-expressing plants nor myb33 and myb101 insertion mutants exhibit any altered response to GA during germination (data not shown). Moreover, germination of the various mutants analyzed in this work is not affected under control conditions (Figures 3–6). Support for a direct role of ABA in miR159 accumulation during germination comes from two independent sources. First, ABA-induced accumulation of miR159 is blocked in the ABA-insensitive abi3-1 mutant (Figure 7). Second, the myb33 and myb101 mutants and the MYB33- and MYB101-over-expressing lines show opposing phenotypes in response to ABA (Figures 4 and 6). Thus, these data suggest that the ABA phenotypes observed in this work are not the consequence of an unintended alteration in endogenous GA levels and/or signaling.

We envisage an interacting regulatory network in which ABA and GA signaling pathways can trigger miR159 accumulation depending on the developmental context. During seed germination, ABA regulates miR159 accumulation, whilst during flower development, miR159 expression is controlled by GA. Another possibility is that individual MIR159 genes are responsive to different hormone signals. Indeed, Harberd and colleagues have identified potential GA-responsive elements in the putative promoter region of MIR159 genes (Achard et al., 2004). Using RT-PCR, we detected expression of all three MIR159 genes during germination (see Figure 1). Analysis of the upstream region of the three MIR159 genes shows putative binding sites for ABA-responsive and stress-related factors such as AtMYC2 (Abe et al., 2003), and, in particular, putative ABRE-like elements are found in the upstream region of MIR159a. In this work, we found that ABI3 function is necessary for miR159 accumulation; however, it has been proposed that ABI3 requires interaction with other protein(s) to be tethered to ABA-responsive promoters (Suzuki et al., 1997). Therefore, further studies will be required to determine whether ABI3 and/or other factors directly regulate transcription of any of the three MIR159 genes. It is also possible that tissue-specific expression of individual miRNA genes may play an important role in the regulation of their target mRNAs. The recent finding that Arabidopsis miR164c (but not miR164a or b) is responsible for the control of petal number by regulating CUC1 and CUC2 transcripts supports this possibility (Baker et al., 2005).

miR159 regulates the abundance of MYB33 and MYB101 mRNAs during germination in response to ABA

In addition to MYB33 and MYB101, other possible targets of miR159 include MYB65 and MYB104 (Rhoades et al., 2002); however, we could not detect the presence of the latter mRNAs at the early developmental stages analyzed here (data not shown). In germinating seeds, MYB33 was previously shown to be expressed in root tissues, whilst MYB101 was expressed in hypocotyl hooks (Gocal et al., 2001). Accordingly, using 5′ RACE, we identified partial mRNA products consistent with miRNA-directed cleavage for MYB33 and MYB101. A role for MYB33 during flowering and leaf morphogenesis has been previously proposed (Gocal et al., 2001; Palatnik et al., 2003), and its involvement in GA signaling has been examined recently (Achard et al., 2004). However, MYB101 and its regulation by miR159 had not been investigated before. The fact that expression of cleavage-resistant mRNAs of MYB33 and MYB101 results in an ABA hypersensitivity phenotype, whilst the myb33 and myb101 mutants are hyposensitive to ABA during germination, argues that both MYB factors act as positive regulators of ABA signaling during germination. Mutation of each of these genes confers a hyposensitive response at 1 μm ABA, but no differences from WT were detected at higher or lower ABA concentrations (Figure 4b). These results suggest that these two MYB factors may have a redundant function, and therefore knockout of both genes may be needed to uncover a stronger ABA phenotype. Consistent with this notion, a greater effect on ABA response is observed by the reduction of functional miR159 caused by over-expression of the viral protein P1/HC-Pro than by the complementary experiment of removing the miR159 recognition sequence in either MYB33 or MYB101 mRNAs. The presence of HC-Pro increases sensitivity at a broader range of ABA concentrations tested than any of the other mutations analyzed. This is expected if miR159 regulation of both genes is required to reduce their activity. However, we cannot rule out the possibility that another as yet unidentified target of miR159 might cooperate with MYB33 and MYB101 to act in response to ABA. Furthermore, loss of an unrelated miRNA function in the presence of HC-Pro could indirectly affect ABA sensitivity. Finally, the contribution of an independent mechanism of response to ABA at this early stage of development should be considered. ABI5 has been shown to participate in ABA responses during early germination (Lopez-Molina et al., 2001, 2002). Here, we show that accumulation of miR159 is slightly affected in the abi5-1 mutant, suggesting that regulation of ABA responses by MYB33/MYB101 may occur independently of ABI5 signaling. It is still possible, however, that ABI5 and MYB33/MYB101 function coordinately to induce ABA-dependent gene expression at certain loci.

MYB33 is an interesting target of miR159: our results show that it has a role in ABA responses during seedling development. In addition, it has been suggested to play a role in GA control of flower development, during anther formation and during leaf morphogenesis (Achard et al., 2004; Millar and Gubler, 2005; Palatnik et al., 2003). Another R2R3 MYB factor, MYB65, has been shown to be highly similar to, and at least in one study shown to have redundant functions with, MYB33 (Millar and Gubler, 2005). During germination, and in response to ABA, we were unable to detect the expression of MYB65, consistent with previous results using in situ hybridization (Gocal et al., 2001). Thus, it appears that MYB33 could work independently of other factors such as MYB65 for certain functions. It is also possible that, during germination, it works in coordination with MYB101 to regulate ABA responses, whilst other functions require its association with other MYB proteins such as MYB65 during anther development.

The role of miR159 in resetting ABA responses

The addition of exogenous ABA results in an accumulation of miR159, a negative regulator of the ABA positive regulators MYB33 and MYB101. This observation is counter-intuitive at a first glance. However, in order to sense a decrease in ABA levels in the environment and to establish the appropriate conditions to resume growth, cells need to degrade positive factors of ABA signaling to reset the developmental program. Thus, the role of miR159 would be to continually degrade mRNAs of positive factors so as to allow a fast recovery from high ABA levels when the signal disappears, effectively resulting in a negative feedback of abscisic acid signaling. This concept would have a similar logic to that of protein ubiquitination as a label for protein degradation of signaling molecules. We propose that short-lived proteins are likely to be encoded by short-lived mRNAs, a requirement for a cellular system that necessitates clearing of a signal to resume non-induced conditions. This hypothesis proposes that MYB33 and MYB101 proteins are unstable, and possibly regulated by ubiquitin-induced degradation as has been demonstrated for the miRNA-regulated NAC1 transcription factor (Guo et al., 2005; Xie et al., 2002), and the ABI5 and ABI3 transcription factors involved in ABA signaling during early seedling development (Lopez-Molina et al., 2003; Zhang et al., 2005). Experiments are underway to test this idea.

Experimental procedures

Plant materials and growth conditions

Arabidopsis thaliana seeds were germinated on solid Murashige & Skoog (MS) medium without sucrose at 23°C under continuous light. Abscisic acid (ABA; Sigma, St Louis, MO, USA) dissolved in methanol was added at different concentrations (see text for details). For drought treatments, 2-day-old seedlings were incubated in a chamber for 8 h at 60% relative humidity. Seed germination efficiencies (as determined by radicle emergence) were scored 7 days after transferring plates to continuous light at 23°C. Seeds used for all germination experiments were age-matched and stored under similar conditions for at least 1 month before the experiments were performed. SALK-generated T-DNA insertion lines (Alonso et al., 2003) for MYB101 (SALK_061355), MYB33 (SALK_065473) and MYB65 (SALK_063552) were obtained from the Arabidopsis Biological Resource Center (ABRC, Ohio State University, Columbus, OH, USA). The TuMV P1/HC-Pro-over-expressing line (in the Col-0 background) (Mallory et al., 2002) was kindly provided by Dr V. Vance (University of South Carolina, USA). Transgenic lines over-expressing mMYB33 and miR159a were previously characterized (Achard et al., 2004) and kindly provided by Dr N. Harberd (John Innes Centre, Norwich, UK).

MYB101 cloning and expression

The ORF encoding MYB101 (At2g32460) was PCR-amplified from genomic DNA and directionally cloned using the pENTR/SD/TOPO vector (Invitrogen, Carlsbad, CA, USA). To generate Arabidopsis transgenic plants, the MYB101 ORF was transferred to the binary vector pBA-DC (derived from pBA0002; Zhang et al., 2005) under the control of the CaMV 35S promoter using Gateway technology procedures (Invitrogen).

Northern blot hybridizations

Total RNA was extracted from seedlings using Trizol reagent (Invitrogen). Samples of 20 μg total RNA were resolved on a 15% polyacrylamide/1x TBE/7 m urea gel and blotted to a Hybond-N+ membrane (Amersham, Piscataway, NJ, USA). DNA oligonucleotides with the exact complementary sequence to miR159a were end-labeled with [γ-32P]-ATP (Perkin-Elmer, Wellesley, MA, USA) and T4 polynucleotide kinase (New England Biolabs, Ipswich, MA, USA) to generate probes with high specific activity. Hybridization was carried out using the ULTRA-Hyb Oligo solution (Ambion, Austin, TX, USA) according to the manufacturer's instructions, and signals were detected by autoradiography.

Analysis of in vivo and in vitro miRNA-directed cleavage

For analysis of in vitro mRNA cleavage, RNA fragments of various MYB transcripts spanning the miR159 cleavage site were 5′-capped using vaccinia virus guanylyltransferase (Ambion) and [α-32P]-ATP (Perkin-Elmer). The resulting 5′-end-labeled mRNA fragments were incubated in wheat germ extracts (Promega, Madison, WI, USA) under miRNA processing conditions (Tang et al., 2003). Aliquots were removed at various time intervals, and RNA products resolved on a 6% polyacrylamide/1x TBE/8 m urea gel. Signals were analyzed by autoradiography.

To map the MYB101 and MYB33 mRNA cleavage sites in vivo, total RNA was extracted from WT (Col-0) 2-day-old seedlings after ABA treatment. The 5′ end of the cleavage product was determined by a modified RNA ligase-mediated 5′ RACE method (Kasschau et al., 2003; Llave et al., 2002) using RLM-RACE kits (Ambion) according to the manufacturer's instructions, except that total RNA was used for ligation to the 5′ RNA adapter without prior treatment. PCR fragments were cloned using the pCR2.1-TOPO cloning kit (Invitrogen) and independent clones sequenced.

Agrobacterium-mediated expression of MYB101 and related constructs in Nicotiana benthamiana was performed as previously described (Guo et al., 2003). Briefly, A. tumefaciens cultures carrying the pBA-derivative plasmids (Zhang et al., 2005) pBA-35S:MYB101-myc or pBA-35S:mMYB101-myc were used to infiltrate mature leaves of N. benthamiana. After 2 days, leaf tissue was ground in liquid nitrogen and used for RNA extraction using Trizol reagent (Invitrogen) or resuspended in SDS protein loading buffer before boiling for 5 min. For RNA analysis, total RNA was resolved in a formaldehyde-containing agarose gel, transferred to Hybond-N+ membrane (Amersham) and analyzed by Northern blot hybridizations using a fragment of MYB101 as probe. For protein analysis, SDS–PAGE was used, followed by transfer to a Hybond-P membrane (Amersham). For Western blotting, we used an anti-myc antibody (c-Myc (A-14) sc-789, rabbit polyclonal IgG), following the manufacturer's recommendations for its use (Santa Cruz Biotechnology, Santa Cruz, CA, USA).

Sequences for all oligonucleotides used in this work are available from the authors upon request.

Acknowledgements

We thank Drs S. Pfeffer and T. Tuschl for advice on small RNA cloning, Dr P. Zamore for valuable discussions and advice on the wheat germ processing reactions, Drs Vicky Vance and Nick Harberd for transgenic seeds, and the ABRC for the T-DNA insertion mutants. This work was supported by NIH grant GM 44640 to N.-H.C. J.L.R. was supported by a Pew Latin American Fellowship for the Biomedical Sciences.

Ancillary