SERRATE is a novel nuclear regulator in primary microRNA processing in Arabidopsis

Authors

  • Li Yang,

    1. National Laboratory of Plant Molecular Genetics, Shanghai Institute of Plant Physiology and Ecology, Shanghai Institute for Biological Sciences, Chinese Academy of Sciences, 300 Fenglin Road, Shanghai 200032, China
    Search for more papers by this author
    • Both authors contributed equally to this paper.

  • Ziqiang Liu,

    1. State Key Laboratory of Genetic Engineering, Department of Biochemistry, School of Life Sciences, Fudan University, Shanghai 200433, China
    Search for more papers by this author
    • Both authors contributed equally to this paper.

  • Feng Lu,

    1. National Laboratory of Plant Molecular Genetics, Shanghai Institute of Plant Physiology and Ecology, Shanghai Institute for Biological Sciences, Chinese Academy of Sciences, 300 Fenglin Road, Shanghai 200032, China
    Search for more papers by this author
  • Aiwu Dong,

    Corresponding author
    1. State Key Laboratory of Genetic Engineering, Department of Biochemistry, School of Life Sciences, Fudan University, Shanghai 200433, China
    Search for more papers by this author
  • Hai Huang

    Corresponding author
    1. National Laboratory of Plant Molecular Genetics, Shanghai Institute of Plant Physiology and Ecology, Shanghai Institute for Biological Sciences, Chinese Academy of Sciences, 300 Fenglin Road, Shanghai 200032, China
    Search for more papers by this author

(fax +86 21 5492 4015; e-mail hhuang@sippe.ac.cn or fax +86 21 6565 0149; e-mail awdong001@yahoo.com.cn).

Summary

The Arabidopsis gene SERRATE (SE) controls leaf development, meristem activity, inflorescence architecture and developmental phase transition. It has been suggested that SE, which encodes a C2H2 zinc finger protein, may change gene expression via chromatin modification. Recently, SE has also been shown to regulate specific microRNAs (miRNAs), miR165/166, and thus control shoot meristem function and leaf polarity. However, it remains unclear whether and how SE modulates specific miRNA processing. Here we show that the se mutant exhibits some similar developmental abnormalities as the hyponastic leaves1 (hyl1) mutant. Since HYL1 is a nuclear double-stranded RNA-binding protein acting in the DICER-LIKE1 (DCL1) complex to regulate the first step of primary miRNA transcript (pri-miRNA) processing, we hypothesized that SE could play a previously unrecognized and general role in miRNA processing. Genetic analysis supports that SE and HYL1 act in the same pathway to regulate plant development. Consistently, SE is critical for the accumulation of multiple miRNAs and the trans-acting small interfering RNA (ta-siRNA), but is not required for sense post-transcriptional gene silencing. We further demonstrate that SE is localized in the nucleus and interacts physically with HYL1. Finally, we provide evidence that SE and HYL1 probably act with DCL1 in processing pri-miRNAs before HEN1 in miRNA biogenesis. In plants and animals, miRNAs are known to be processed in a stepwise manner from pri-miRNA. Our data strongly suggest that SE plays an important and general role in pri-miRNA processing, and it would be interesting to determine whether animal SE homologues may play similar roles in vivo.

Introduction

MicroRNAs (miRNAs) constitute a large family of small endogenous non-coding RNAs found in many eukaryotic organisms. Although mature miRNAs are only 20–24 nucleotides (nt) in length, their precursors, primary miRNA transcripts (pri-miRNAs), are generally thought to be much longer, and undergo folding into stem–loop structures (for review, see Bartel, 2004). The first step in miRNA maturation is nuclear cleavage of the pri-miRNA, which releases an approximately100-nt intermediate, called pre-miRNA. The pre-miRNA is further processed into a mature miRNA, which acts to repress expression of the cognate gene through cleavage of mRNA and inhibition of translation (Bartel, 2004), or induction of DNA methylation (Bao et al., 2004).

Biogenesis of miRNA in animals and plants has been studied extensively in the past few years, and recent advances have led to a more detailed understanding of the functions of key components in miRNA processing. In animals, initial cleavage of pri-miRNAs is catalysed by Drosha, a nuclease of the RNase III family (Gregory et al., 2004; Han et al., 2004a; Lee et al., 2003; Zeng et al., 2005). Drosha exists in a multiprotein complex, termed Microprocessor. Along with Drosha, the Microprocessor also contains Pasha (partner of Drosha) or DGCR8, a double-stranded RNA-binding protein (Gregory et al., 2004; Han et al., 2004a; Lee et al., 2003; Zeng et al., 2005). Suppression of Pasha expression in Drosophila or Caenorhabditis elegans caused an accumulation of pri-miRNAs and a reduction in mature miRNAs (Denli et al., 2004). In animals, cleavage of pre-miRNAs into mature forms is mediated by a cytoplasmic protein complex including Dicer, another enzyme in the RNase III family (Chendrimada et al., 2005; Hutvagner et al., 2001).

In plants, several protein components have been found to be essential in miRNA accumulation, including DICER-LIKE1 (DCL1), HUA ENHANCER1 (HEN1), ARGONAUTE1 (AGO1), HASTY (HST) and HYPONASTIC LEAVES1 (HYL1; for review, see Chen, 2005). DCL1 is an Arabidopsis homologue of Dicer, and was found to cleave both pri- and pre-miRNAs. HEN1 is a methyltransferase that is required for protection of miRNAs from extra uridylation. HST is required for transporting mature miRNAs from the nucleus to the cytoplasm (Kurihara et al., 2006; Park et al., 2005; Saito et al., 2005; Yu et al., 2005a). In addition, AGO1 acts as a slicer to cleave the miRNA–mRNA double-stranded structures (Baumberger and Baulcombe, 2005), whereas its function in miRNA biogenesis is unknown. Recent studies also demonstrated that HYL1 and DCL1 can form a protein complex that is required for efficient processing of pri-miRNAs (Kurihara et al., 2006). SERRATE (SE) was previously thought to be a chromatin-remodelling factor in repressing KNOTTED1-like homeobox (KNOX) genes for normal leaf development (Ori et al., 2000). Recent data have also shown that SE regulates shoot meristem and axial patterning of lateral organs by modulating the accumulation of miR165/166 (Grigg et al., 2005). However, how SE modulates these two miRNAs and whether it regulates other miRNA processing is unknown.

In this study, we report that SE is a critical component generally required for miRNA biogenesis, and SE and HYL1 can form a protein complex that may function in pri-miRNA processing. The roles of SE, HYL1 and DCL1 in planta and a possible requirement for SE-like proteins in animal pri-miRNA processing are also discussed.

Results and discussion

In the course of characterizing Arabidopsis mutants with leaf developmental defects, we noted that two previously identified mutants, se and hyl1, both exhibited abnormal leaves with up-curled margins (Grigg et al., 2005; Lu and Fedoroff, 2000). Upon further investigation, we found that these two mutants shared several other abnormal morphological phenotypes in addition to the similar leaf phenotypes. Notably, whereas in wild-type plants a secondary inflorescence branch is usually associated with a cauline leaf at its proximal end (Figure 1a), in se-1 (Figure 1b) and hyl1-1 (Figure 1c) plants many secondary inflorescences lacked an associated cauline leaf. In addition, while wild-type inflorescences produced siliques that developed in a spiral phyllotactic pattern (Figure 1d), the se-1 (Figure 1e) and hyl1-1 (Figure 1f) inflorescences often produced siliques that emerged from the same node of a stem.

Figure 1.

 SE may share functions with HYL1.
(a–c) Architecture of the secondary inflorescence in wild-type Col (a), se-1 (b) and hyl1-1 (c). Note that a cauline leaf is usually associated with the secondary inflorescence at its proximal end, but was often absent in the se-1 and hyl1-1 mutant plants (arrowheads).
(d)–(f) Silique arrangements in wild-type Col (d), se-1 (e) and hyl1-1 (f). Note that unlike the silique arrangement in wild-type, some siliques in se-1 and hyl1-1 (arrowheads) were often associated together at the same node of an inflorescence stem.
(g, h) The se-1 hen1-2 double mutant (right) exhibited a very small plant stature as compared with the hen1-2 (left) and se-1 (centre) single mutant plants. Images were taken with 20-day-old (g) and 40-day-old (h) plants. Bars = 1 cm.

Additional similar phenotypic effects of the se and hyl1 mutations became evident during the vegetative phase transition of plants from a juvenile to an adult phase. While juvenile (early appearing) leaves of wild-type plants produce trichomes only on the adaxial leaf surface, adult leaves have trichomes on both the adaxial and abaxial sides. Unlike the wild-type leaves, the early appearing leaves of both se and hyl1 mutant plants produce abaxial trichomes (Clarke et al., 1999; Jover-Gil et al., 2005), reflecting a similarly altered vegetative phase transition in the two mutants. Previous studies have shown that plants carrying both the hyl1 and hen1 mutations are very small in stature and are infertile (Vazquez et al., 2004). The se-1 hen1-2 double mutant plants were also very small in size and completely infertile (Figure 1g,h). Because of the similar single mutant phenotypes of se and hyl1, and their synergistic interactions with hen1, we hypothesized that SE and HYL1 may act in the same pathway in regulating plant development.

To test for epistasis between SE and HYL1, we crossed se-1 to hyl1-1 and analysed the F2 progeny. Although the segregating F2 progeny contained plants with the expected se-1- and hyl1-1-like phenotypes (45 se-like plants, 43 hyl1-like plants and 137 wild-type-like plants in an F2 population), plants with novel phenotypes were not observed. Since se-1 and hyl1-1 mutations are caused by a 7-base deletion and a T-DNA insertion, respectively, homozygous se-1 and hyl1-1 mutations can be determined by PCR. We analysed 103 individual F2 plants, including 32 wild-type-, 37 se-1- and 34 hyl1-1-like plants for the identification of possible double mutants, but failed to find one. In addition, siliques of the F1 progeny of a cross between se-1 and hyl1-1 contained abortive seeds (22 from 264 F1 seeds versus 0 from 322, 476 and 157 wild-type, se-1 and hyl1-1 seeds, respectively; Figure 2; see Figure S1). All these results indicate that the se-1 hyl1-1 double mutants are embryonically lethal, and SE and HYL1 may act synergistically to control the same genetic pathway.

Figure 2.

se-1 hyl1-1 double mutant is embryonically lethal.
(a) A wild-type silique showed the well-developed seeds.
(b, c) Siliques from F1 progeny of a cross between se-1 and hyl1-1 contained the abortive seeds (arrowheads), shown at early (b) and mature (c) seed stages.

HYL1 is required for the biogenesis of a number of miRNAs (Han et al., 2004b; Vazquez et al., 2004); however, miR165/166 are the only Arabidopsis miRNAs whose normal accumulation is known to require SE (Grigg et al., 2005). In addition, while production of trans-acting small interfering RNA (ta-siRNA) is known to involve HYL1 (Allen et al., 2005; Yoshikawa et al., 2005), it is not yet known whether SE also participates in the ta-siRNA biogenesis pathway. Because SE and HYL1 seem to have related functions in development, we hypothesized that SE also affects the accumulation of other miRNAs and ta-siRNA. To test this hypothesis, we analysed the expression of several miRNAs that are known to be expressed in leaves, including miR157, miR159, miR164, miR165, miR167 and miR168. Similar to those in hyl1-1, the levels of all of these miRNAs were reduced to varying extents in the se-1 leaves (Figure 3a). Meanwhile, our real-time RT-PCR analyses revealed that mRNA levels of the target genes of these miRNAs were generally elevated in both se-1 and hyl1-1 mutant leaves: SQUAMOSA PROMOTER-BINDING PROTEIN-LIKE10 (SPL10; miR157), MYB33 (miR159), CUP-SHAPED COTYLEDON1 (CUC1; miR164), REVOLUTA (REV; miR165), AUXIN RESPONSE FACTOR8 (ARF8; miR167) and AGO1 (miR168; Figure 3b). These results suggest that SE should be considered an additional general component of miRNA accumulation.

Figure 3.

 SE is required for miRNA and ta-siRNA accumulation.
(a) Levels of several mature miRNAs. All miRNA levels analysed were reduced in se-1 and hyl1-1 leaves as compared to those in the wild type. In each group, the RNA blot was first probed for a miRNA, and the same filter was then analysed by a 5S RNA probe.
(b) Analyses of mRNA levels by quantitative real-time RT-PCR. Levels of target mRNAs of the miRNAs analysed in (a) were generally enhanced in the se-1 and hyl1-1 leaves.
(c) mir173 and miR390 levels and ta-siRNA levels in the se-1 seedlings. Note that the mir173 and miR390 contents were both reduced to an undetectable level, and TAS1c-derived ta-siR850 and TAS3-derived ta-siR2141 were both absent.
(d) Levels of At5g18040 and ARF3 mRNAs, which are the targets of ta-siR850 and ta-siR2141, respectively, were elevated markedly in se-1 seedlings, as compared to those in the wild type.
In (a) and (c), the 5S RNA was used as a loading control. In (b) and (d), results were normalized to that for ACTIN, and then to the value of the wild-type plants, whose value was arbitrarily fixed at 1. Leaves from about 10 plants were used for the RNA preparation. Quantifications of each cDNA sample in (b) and (d) were made in triplicate, and the consistent results were obtained from two independent experiments. Bars show standard error.

It has been reported that the production of certain ta-siRNAs requires the presence of miR173 and miR390 (Allen et al., 2005; Yoshikawa et al., 2005). We therefore measured the levels of miR390 and miR173 as well as the presence of two ta-siRNAs, TAS3-derived ta-siR2142 and TAS1c-derived ta-siR850, the accumulation of which depends on the actions of miR390 and miR173, respectively. Relative to the content of miR390 and miR173 in wild-type seedlings, both were dramatically reduced to undetectable levels; likewise, ta-siR2142 and ta-siR850 were not detected in se-1 seedlings (Figure 3c). In contrast, the mRNA levels of AUXIN RESPONSE FACTOR3 (a target of ta-siR2142) and At5g18040 (a target of ta-siR850) were both elevated (Figure 3d). These results indicate that SE is also involved in the ta-siRNA biogenesis pathway.

Several previously characterized Arabidopsis mutants, including RNA dependent RNA polymerase6 (rdr6), suppressor of gene siliencing3 (sgs3), zippy (zip), dcl4 and hyl1-1, have phenotypes characterized by a shortened juvenile leaf phase length (Peragine et al., 2004; Xie et al., 2005; Yoshikawa et al., 2005), resembling that in se (Clarke et al., 1999). The affected gene in each of these mutants has previously been shown to be involved in ta-siRNA accumulation. Therefore, it is possible that the altered vegetative phase change in the se mutant may also be caused by a dysfunctional ta-siRNA pathway.

Genes that are involved in ta-siRNA biogenesis can be grouped into three categories: (i) those that participate in miRNA biogenesis (HYL1 and DCL1); (ii) those that are required for sense post-transcriptional gene silencing (sense-PTGS), but not for miRNA biogenesis (RDR6 and SGS3); and (iii) those that are involved in both miRNA biogenesis and sense-PTGS (HEN1 and AGO1; Boutet et al., 2003; Finnegan et al., 2003; Morel et al., 2002; Mourrain et al., 2000; Vaucheret et al., 2004; Vazquez et al., 2004). Sense-PTGS activity can be determined by examining GUS activity in the L1 line that carries a silenced 35S::GUS fusion, such that enhanced GUS activity in the L1 line is indicative of defective sense-PTGS (Mourrain et al., 2000). To determine whether SE is also involved in sense-PTGS, we introduced by cross the se-1 mutation into the L1 background, and analysed GUS activities of plants homozygous for se in the F3 progeny. Unlike the sense-PTGS-deficient mutant sgs3-11 (Figure 4c), L1 line plants carrying the se-1 mutation (Figure 4b) showed weak GUS staining, similar to that seen in the L1 line itself (Figure 4a). Furthermore, RNA filter hybridization analyses revealed that the GUS mRNA level in the se-1/L1 plants was much lower than that observed in the sgs3-11/L1 plants (Figure 4d). These results indicate that SE should be grouped with HYL1 and DCL1, which participate in miRNA biogenesis but are not required for sense-PTGS.

Figure 4.

 SE is not required for sense-PTGS.
GUS staining of L1 (a), se-1/L1 (b) and sgs3-11/L1 (c) seedlings. Note that se-1 does not interfere with the silencing of L1, as compared with the deeply stained sgs3-11/L1 seedlings.
(d) RNA filter hybridization analyses of GUS transcripts. Consistent with the GUS staining, GUS transcript levels in L1 and se-1/L1 plants were much lower than that in sgs3-11/L1 plants. RNA was extracted from leaves of 20-day-old plants. Bars = 1 cm in (a–c).

It was previously reported that the HYL1 protein is located in the nucleus (Han et al., 2004b). Subcellular distribution of the SE protein, however, has not yet been reported. To investigate the localization of SE in plant cells, we constructed an inducible yellow fluorescent protein (YFP)–SE translational fusion and introduced it into tobacco BY2 cells. A YFP–HYL1 fusion was also constructed and used to transform tobacco BY2 cells, which then served as a positive control for nuclear localization. Our confocal microscopic analyses showed that the fluorescence signals of YFP–SE were detected exclusively in the nucleus (Figure 5a), similar to those of YFP–HYL1, whereas no detectable fluorescence was found in the untransformed BY2 cells or in the uninduced transgenic cells (data not shown). In the nucleus, these two proteins had their preferentially concentrated distribution patterns: HYL1 formed some nuclear bodies similar to those in Arabidopsis as described by Han et al. (2004b) and SE was highly accumulated inside the nuclear membrane. However, they showed clearly overlapped portions in the nucleus.

Figure 5.

 SE is localized in the nucleus and interacts with HYL1.
(a) Confocal microscopy analyses of YFP–SE and YFP–HYL1 fluorescence in transgenic tobacco BY2 cells. Fluorescence from both YFP–SE cells (upper panels) and YFP–HYL1 cells (lower panels) was concentrated in the nucleus. The YFP fluorescence (green) images are shown together with bright-field differential interference contrast images and their combined pictures. Bars = 5 μm.
(b) Yeast two-hybrid analyses. Co-expression of SE and HYL1 in the yeast cells (AH109) turned on the activities of reporter genes HIS3 and ADE2. The pGADT7-AS1 and pGBKT7-AS2 pair served as a positive control (Xu et al., 2003), while empty vector, pGADT7–HYL1/pGBKT7 and pGADT7/pGBKT7–SE pairs were used as negative controls. In (b): 1, pGADT7/pGBKT7; 2, pGADT7-AS2/pGBKT7-AS1; 3, pGADT7–HYL1/pGBKT7; 4, pGADT7–HYL1/pGBKT7–SE; 5, pGADT7/pGBKT7–SE.
(c, d) Pull-down assays. The pull-down fractions were analysed by protein blotting with a polyclonal anti-GFP antibody. The YFP was used as a negative control, which showed no interaction with GST–HYL1 or GST–SE. The input fraction represents 5% of the total proteins used in the pull-down assays. Arrowheads mark the positions of the protein molecular weights.

The similar plant phenotypes of the single and double mutants (se versus hyl1 and se hen1 versus hyl1 hen1), the similar reductions of miRNA and ta-siRNA accumulation in se and hyl1, and the similar SE and HYL1 subcellular localizations led us to hypothesize that the SE and HYL1 proteins may interact physically. To test this hypothesis, we first carried out a yeast two-hybrid assay. Yeast cells that co-expressed the SE bait and HYL1 prey fusion proteins were able to grow on media lacking histidine and adenine, indicating that these proteins promoted the expression of the HIS3 and ADE2 reporter genes (Figure 5b). These results suggest that SE and HYL1 bind each other in yeast cells. To further confirm the SE and HYL1 protein–protein interaction, we performed glutathione S-transferase (GST) pull-down assays using the fusion proteins GST–HYL1, GST–SE, YFP–HYL1 and YFP–SE. YFP–SE or YFP–HYL1 fusion proteins were expressed in tobacco BY2 cells and subjected to pull-down analyses with GST–HYL1 and GST–SE, respectively. As shown in the input lane, Western blotting using anti-GFP antibody clearly detected the YFP, YFP–SE (Figure 5c) and YFP–HYL1 (Figure 5d) bands. Nonetheless, only YFP–SE, but not YFP, could be detected after respective incubations of the YFP–SE- or YFP-containing supernatants of cell lysates with the GST–HYL1 beads (Figure 5c). Similarly, YFP–HYL1 but not YFP showed positive labelling when the supernatants were incubated with the GST–SE beads (Figure 5d). Note that no YFP, YFP–SE or YFP–HYL1 signals were detected following incubation with the GST beads (Figure 5c,d). All these results provide a strong indication that SE and HYL1 can form a protein complex, which may be required for miRNA accumulation. Since SE and HYL1 are both located in the nucleus, our results also suggest that the SE–HYL1 protein complex may regulate accumulation of miRNA before miRNA is exported from the nucleus to the cytoplasm.

It was reported that HYL1 and DCL1 can form a protein complex, and these two proteins are required for pri-miRNA processing (Kurihara et al., 2006). In addition, the biochemical activity of SE appears to be related to the pri-miRNA processing, as suggested by the analysis of one miRNA, miR166 (Grigg et al., 2005). To investigate the possible activity of SE in processing other miRNAs, we measured several pri-miRNAs in the se-1 mutant according to a method described by Juarez et al. (2004) (Figure 6a), using hyl1-1 and dcl1-7 as positive controls. We also included hen1-2 in the experiments for a possible negative control, because hen1 mutant plants also showed the reduced mature miRNA levels and the HEN1 function is known to be required for miRNA stability but may not relate to pri-miRNA processing (Li et al., 2005b; Yu et al., 2005a). Our results showed that, similar to those in hyl1-1 and dcl1-7, all pri-miRNA levels analysed in the se-1 mutant were elevated compared to those in the wild type (Figure 6b,c), though the pri-miRNA content remained unchanged in the hen1 mutant (Figure 6d). These results suggest that the reduced mature miRNA levels in the se mutant plants may be caused by an inefficient cleavage of pri-miRNAs.

Figure 6.

 SE is critical in pri-miRNA processing.
(a) A schematic diagram showing the primer positions in the RT-PCR experiments. The oligo(dT) primer was used for the first-strand cDNA synthesis. All PCR products in the experiments are about 100 bp in length.
(b–d) Accumulation of pri-miRNAs in se-1 and hyl1-1 (b), dcl1-7 (c) and hen1-2 (d). Polymerase chain reaction thermal cycles for pri-miRNA detection in each mutant are shown to the right of each panel, and ACTIN was used as a positive control. The lower panels are PCR products that serve as controls for monitoring contamination from genomic DNA. Primers for these experiments were designed according to an intergenic sequence between genes At2g19810 and At2g19820. With 31 thermal cycles, the PCR reactions for these controls did not yield products. The contaminated DNA could be detected after 36 thermal cycles.

SE and HYL1 genes are both expressed in many different plant tissues (Figure 7), suggesting that the SE–HYL1 protein complex functions in multiple developmental processes. HYL1 can physically interact with DCL1, and SE, HYL1 and DCL1 are all involved in pri-miRNA processing (Figure 6; Kurihara et al., 2006). These results suggest that these three proteins may act in the same complex in processing pri-miRNAs, and at least the SE–HYL1 and HYL1–DCL1 associations are direct. It was previously reported that several severe dcl1 alleles produce aberrant embryos (Castle et al., 1993; Errampalli et al., 1991; McElver et al., 2001), similar to those in the se-1 hyl1-1 double mutant. Therefore, the embryonic lethality of the se hyl1 double mutant may be caused by the severely disrupted miRNA pathway. All these findings underscore the importance of the SE–HYL1–DCL1 controlled miRNA pathway in fundamental life processes.

Figure 7.

 Reverse transcriptase PCR analyses of HYL1 and SE expression patterns. Both HYL1 and SE transcripts were detected in different tissues of wild-type plants (Col).

In animals, pri-miRNA processing requires Drosha and Pasha/DGCR8, which exist in Microprocessor (Gregory et al., 2004). Although mass spectrometric sequencing revealed that this protein complex may contain about 20 different proteins, Drosha and Pasha/DGCR8 are the most critical components in pri-miRNA processing. It was shown that the addition of both Drosha and DGCR8 recombinant proteins together reconstituted the miRNA processing activity to similar levels seen with native complex, though Drosha or DGCR8 alone showed no processing activity (Gregory et al., 2004; Han et al., 2004a). Components known to be necessary in pri-miRNA processing appear to be varied between plants and animals. In addition to the RNase III proteins (DCLs in plants and Drosha in animals) and the double-stranded RNA-binding proteins (HYL1 in plants and Pasha/DGCR8 in animals), a requirement for the C2H2 zinc finger protein functions in animal pri-miRNA processing has not been reported. It is possible that since the functional domains between Drosha/Pasha (DGCR8) and DCL1/HYL1 proteins are not completely the same, functions of the Drosha/Pasha (DGCR8) pair that are required for pri-miRNA processing may not be fully accomplished by those of the DCL1/HYL1 pair. Therefore, the functions of SE must be additionally included in normal pri-miRNA processing in plants. Alternatively, a protein with similar SE functions in pri-miRNA processing in animals may not yet have been identified. The results from the in vitro reconstitution experiment using recombinant Drosha and DGCR8 cannot rule out an in vivo requirement for the SE-like protein functions in pri-miRNA processing. It is reported that SE-like proteins exist in animals, though the amino acid similarity is relatively low compared with their plant homologues (Prigge and Wagner, 2001). In particular, loss of function in a zebrafish SE-like protein resulted in embryonic lethality (Golling et al., 2002), mimicking the SE function in Arabidopsis embryo development. It would be of interest to determine whether functions of the animal SE-like proteins are related to miRNA biogenesis in the future.

It is known that SE is predominantly expressed in the adaxial domain of a leaf primordium (Prigge and Wagner, 2001), whereas HYL1 is expressed throughout the leaf primordium (Yu et al., 2005b). In addition, SE and HYL1 proteins have their own preferentially concentrated regions in the nucleus, and se and hyl1 have different extents of reduction of some miRNA levels (for example, the miR167 and miR159 levels) in leaves. All these suggest that the two proteins may have some other distinct functions for specific developmental processes that do not involve the SE–HYL1 protein complex.

Experimental procedures

Plant materials and growth conditions

Seeds of hen1-2, sgs3-11, hyl1-1 and the L1 line were kindly provided by X. Chen (UC Riverside, USA), R. S. Poethig (University of Pennsylvania, USA), N. Fedoroff (Penn State University, USA) and H. Vaucheret (Institut National de la Recherche Agronomique, France), respectively. Seeds of se-1 were obtained from the Arabidopsis Biological Resource Center (ABRC). Plants were grown on soil according to our previous conditions (Chen et al., 2000).

RT-PCR, miRNA filter hybridization and GUS staining

Ribonucleic acid extraction and reverse transcription were according to our previous methods (Xu et al., 2003), using whole seedlings or rosette leaves of 20-day-old plants. Real-time RT-PCR was carried out as described previously (Li et al., 2005a), with the following gene-specific primers: 5′-AGTTGTTGTATCCTGGGTGTAGCA-3′, and 5′-CCGTTGGTGGTGGTGGAGAC-3′ for MYB33, 5′-TGGACCACCGCAGAGACAAT-3′ and 5′-CATCATACGCTGGAAGACGACT-3′ for AGO1, 5′-ATCTGTGGTCACAACTCC-3′ and 5′-TAGCGACCTCTCACAAAC-3′ for REV, 5′-CCAACGGGACTGAGAACGAACA-3′ and 5′-CGGTGGAGCGGGAAGGAAT-3′ for CUC1, 5′-TGAGACAAAGCCTACACAGATGGA-3′ and 5′-GATGATGCAACCCGACTTTTTTATG-3′ for SPL10, 5′-GGTGGCCTGGTTCAAAATGGAG-3′ and 5′-CGGAAGAGGGTGATGATGATAC-3′ for ARF3, 5′-AAGGGCTACCGAGAAGAGAACATT-3′ and 5′-AAGCGTGGGATACAGAAGTCAACA-3′ for At5g18040, and 5′-TGGCATCACACTTTCTACAA-3′ and 5′-CCACTGAGCACAATGTT-3′ for ACTIN. The primers for probing ta-siRNAs or miRNAs are as follows: 5′-GGGTCTTACAAGGTCAAGAAAA-3′ for ta-siR2142, 5′-TACGCTATGTTGGACTTAGAA-3′ for ta-siR850, 5′-GTGCTCTCTATCTTCTGTCA-3′ for miR157, 5′-TAGAGCTCCCTTCAATCCAAA-3′ for miR159, 5′-TGCACGTGCCCTGCTTCTCCA-3′ for miR164, 5′-GGGGGATGAAGCCTGGTCCGA-3′ for miR165, 5′-TAGATCATGCTGGCAGCTTCA-3′ for miR167, and 5′-TTCCCGAGCTGCACCAAGCGA-3′ for miR168. Primers for analysing SE and HYL1 expression are as follows: 5′-CTGATTCCGTCGATAACCGTCTCC-3′ and 5′-CAGGCCTCCCACCCATTTCAC-3′ for SE, and 5′-ATGACCTCCACTGATGTTTCCTC-3′ and 5′-CATACTCCTGCAACCGAC-3′ for HYL1. Primers for detecting pri-miRNAs are as follows: 5′-CATAGGTTTGAGAGTGATG-3′ and 5′-CATATTTTATCATCCACATGC-3′ for pri-miR157c, 5′-GATCCCATAAGCCCTAAT-3′ and 5′-GAAAGAAGATGTAGAGCT-3′ for pri-miR159a, 5′-CCATTGACGATTGCATCCTCG-3′ and 5′-TTGATGGAGAAGCAGGGCAC-3′ for pri-miR164c, 5′-GNTCTCGGACCAGGCTTCA-3′ and 5′-NNYCATSATTACACCAATCTG-3′ for pri-miR166 (Grigg et al., 2005), 5′-GATCTGCTACGGTGAAGTC-3′ and 5′-ATCTAATCGAGACTGATCTC-3′ for pri-miR167a, and 5′-GATAGTAGAGTCTCACCATC-3′ and 5′-CGATTCAGTTGATGCAAGGC-3′ for pri-miR168a. The primer pair for monitoring DNA contamination is 5′-ATAAAGATGAAGTATCCCATTC-3′ and 5′-ACCTTCTTGTTTGATTGTATTG-3′. Filter hybridization was performed as previously described (Li et al., 2005a).

To enrich the ta-siRNAs for filter hybridizations, total RNA was treated with a final concentration of 5% polyethylene glycol (PEG) 800 and 0.5 m NaCl for 2 h on ice. After centrifugation at 13 200 g for 10 min, the RNA was ethanol-precipitated, and about 10 μg was loaded on to each lane in gel separation. The GUS activity was detected according to our previous methods (Li et al., 2005a), except that the plant tissues were stained at 37°C for 1 h.

Yeast two-hybrid assay

The cDNA fragments encoding the entire putative HYL1 and SE proteins were PCR-amplified, verified by sequencing and cloned into SmaI–XhoI or SmaI–SalI restriction sites of the MATCHMAKER two-hybrid vector pGADT7 or pGBKT7 (Clontech, Palo Alto, CA, USA) to generate pGADT7-HYL1 and pGBKT7-SE, respectively. Protein–protein interaction was examined according to the manufacturer's protocol (Clontech).

Plant cell transformation and estradiol induction

The SE and HYL1 cDNA fragments (see above) were first inserted into a YFP-containing plasmid resulting in the in-frame YFP–SE and YFP–HYL1 fusions. The YFP–SE and YFP–HYL1 DNA fragments were then subcloned into the XhoI and SpeI sites of an estradiol-inducible plant transformation vector, pER8 (Zuo et al., 2000), respectively. The resulting constructs pER8–YFP–SE and pER8–YFP–HYL were introduced into the tobacco BY2 cells by Agrobacterium-mediated transformation (Shen, 2001). Expression of the YFP–SE or YFP–HYL1 fusion proteins was induced by the addition of estradiol to the media at a final concentration of 4 μm.

GST pull-down assays

The above cDNA fragments were inserted into the vector pGEX-4T1, resulting in pGEX-4T1–SE and pGEX-4T1–HYL1, respectively. Purification of GST-fused proteins was performed according to the manufacturer's recommendation under non-denaturing conditions (Amersham-Pharmacia Biotech, Uppsala, Sweden). The purified GST, GST–SE and GST–HYL1 proteins were fixed to glutathione-Sepharose 4B beads (Amersham-Pharmacia Biotech). Total protein extracts from tobacco BY2 cells, which contain YFP, YFP–HYL1 or YFP–SE, were used in pull-down assays according to a previously described method (Yu et al., 2004). The pull-down fractions were then analysed by Western blotting using polyclonal anti-GFP antibody (Molecular Probe Inc., Leiden, The Netherlands) at a 1:5000 dilution.

Acknowledgements

We thank X. Chen for hen1-2, R. S. Poethig for sgs3-11, H. Vaucheret for L1, N. Fedoroff for hyl1-1 and ABRC for se-1 seeds, J. Liu for technical assistances with ta-siRNA analysis, and J. Sheen and H. Ma for discussion and critical reading of this manuscript. This research was supported by grants from the CNSF (30421001, 30370751 and 90208009) and the SSC (04JC14077) to HH, and a grant from SSC (04JC14017) to AD. This research is also partially supported by a grant from SIBS for the Plant Reproductive Development to HM.

Ancillary