Better explicating the strengths and shortcomings of these models will help refine projections and improve transparency in the years ahead.
Image credit: Witsawat.S.
-
Edited by James E. Dahlberg, University of Wisconsin Medical School, Madison, WI (received for review August 2, 2004)
Abstract
In mammalian cells, the nuclear export receptor, Exportin 5 (Exp5), exports pre-microRNAs (pre-miRNAs) as well as tRNAs into the cytoplasm. In this study, we examined the function of HASTY (HST), the Arabidopsis ortholog of Exp5, in the biogenesis of miRNAs and tRNAs. In contrast to mammals, we found that miRNAs exist as single-stranded 20- to 21-nt molecules in the nucleus in Arabidopsis. This observation is consistent with previous studies indicating that proteins involved in miRNA biogenesis are located in the nucleus in Arabidopsis. Although miRNAs exist in the nucleus, a majority accumulate in the cytoplasm. Interestingly, loss-of-function mutations in HST reduced the accumulation of most miRNAs but had no effect on the accumulation of tRNAs and endogenous small interfering RNAs, or on transgene silencing. In contrast, a mutation in PAUSED (PSD), the Arabidopsis ortholog of the tRNA export receptor, Exportin-t, interfered with the processing of tRNA-Tyr but did not affect the accumulation or nuclear export of miRNAs. These results demonstrate that HST and PSD do not share RNA cargos in nuclear export and strongly suggest that there are multiple nuclear export pathways for these small RNAs in Arabidopsis.
Genetic analyses of the juvenile-to-adult transition in Arabidopsis have produced several genes required for the expression of the juvenile phase. The first genes to be identified were HASTY (HST) (1) and PAUSED (PSD) (2). Loss-of-function mutations in these genes have similar phenotypes. In addition to interfering with the expression juvenile traits, these mutations reduce the growth rate of the root and shoot, affect the morphology of the shoot apical meristem, cause aberrant phyllotaxy in the inflorescence, and disrupt floral morphogenesis (3-5). HST and PSD are members of the importin β family of nucleocytoplasmic transport receptors (6, 7). PSD is the Arabidopsis ortholog of the tRNA export receptor, Exportin-t (Exp-t) (4, 5), and likely plays a role in tRNA export because it is capable of rescuing the tRNA export defect of a los1 mutation in Saccharomyces cerevisiae (4). The function of HST is more difficult to predict because its orthologs in S. cerevisiae and mammals have completely different functions. The mammalian ortholog of HST, Exportin 5 (Exp5), exports pre-microRNAs (pre-miRNAs) (8-11), tRNAs (12, 13), a viral hairpin RNA (14), and proteins associated with these and other double-stranded (ds) RNAs (15-17). The yeast ortholog of HST, Msn5p, exports several different types of phosphorylated proteins (18) and imports replication protein A (19), a protein involved in DNA replication and repair.
miRNAs are endogenous 20- to 22-nt RNAs that negatively regulate gene expression at the posttranscriptional level (20, 21). In plants, miRNAs function primarily by directing the cleavage of their targets in the middle of the miRNA recognition element; in animals, they act primarily as translational repressors. In animals, pre-miRNAs are excised from longer transcripts in the nucleus by the RNase III endonuclease, Drosha, and are then exported to the cytoplasm, where they are processed from the imperfectly paired stem of the pre-miRNA by Dicer (22, 23). Arabidopsis does not have an ortholog of Drosha, indicating that miRNAs are probably processed differently in plants. Several proteins required for miRNA biogenesis (including the Dicer-like protein, DCL1) are located in the nucleus in Arabidopsis (24-27), raising the possibility that miRNAs are actually processed to completion in the nucleus. Supporting this conclusion, there is indirect evidence that miR159 is present in the nucleus as a ds 21-nt molecule (24). Whether miRNAs direct the cleavage of mRNAs in the nucleus or cytoplasm in plants is still unknown.
Here, we describe the intracellular distribution of miRNAs and tRNAs in Arabidopsis and the effect of hst and psd mutations on the accumulation of these RNAs. We show that miRNAs are processed into their mature size in the nucleus and are present in both the nucleus and the cytoplasm, predominantly as single-stranded (ss), functional molecules. There is little functional overlap between HST and PSD, because loss-of-function mutations in HST reduce the level of most miRNAs but do not affect the accumulation of tRNAs, whereas a mutation in PSD affects the processing of tRNA-Tyr but has no major effect on miRNA accumulation. The observation that hst and psd do not block the accumulation of miRNAs and tRNAs in the cytoplasm suggests that there are multiple export receptors for these RNAs in Arabidopsis.
Materials and Methods
Plant Material and Growth Conditions. All of the stocks used in this study were in a Columbia genetic background. The hst-1, hst-3, hst-6, psd-13, and hst-6 psd-13 double mutants are described in refs. 1, 3, and 4.
To generate plants overexpressing the putative precursor of miR156a, a 396-bp intergenic region harboring the miR156a sequence was amplified by using a forward primer that introduces a HindIII site (5′-GGCACAAAGCTTAAAAGATTCTCATCG-3′) and a reverse primer that introduces a BamHI site (5′-CTGGATCCGAATGATTAAAGGCTAAAGGT-3′) at the ends of this sequence. This PCR product was cloned downstream of the CaMV 35S promoter in pEZT-CL, and transgenic plants were produced by using the floral dip method (28).
hst-6 was crossed to a line carrying the L1 (35S::GUS) transgene (29) and a line carr ying both the 306-1 (35S::hairpin-GUS) and 6b4 (35S::GUS) transgenes (30). Plants homozygous for hst-6 and L1, and plants homozygous for hst-6 and heterozygous for 306-1 and 6b4 were stained for GUS activity as described in ref. 31. The latter plants were identified in the progeny of a backcross to hst-6, using PCR primers specific for each of these transgenes.
RNA Blot Analysis. Total RNA was extracted by using TRIzol (Invitrogen). RNA from immature rosette leaves was isolated from 3-week-old plants grown in Petri dishes on 1/2× Murashige and Skoog salts, 0.5 g/liter Mes, and 0.8% agar (pH 5.7; continuous fluorescent illumination, 22°C). Mature rosette leaves and floral buds were obtained from plants grown in pots in Metromix 200 (Scotts) under the same growth conditions. For the analysis of the total level of miRNAs and tRNAs, 30 μg of RNA was subjected to electrophoresis on an 8 M urea/15% denaturing polyacrylamide gel. The amount of RNA loaded in the case of fractionated samples is described in the figure legends. RNA was transferred to Hybond N+ membranes (Amersham Pharmacia) and hybridized with [γ-32P]-labeled probes. Small interfering RNA (siRNA) blots were generated by using low-molecular-weight RNA purified from total RNA as described in ref. 32. Ten or 20 μg of this sample was subjected to electrophoresis as described above.
Oligonucleotide probes were labeled with T4 polynucleotide kinase (New England Biolabs). Hybridizations were carried out at 40°C by using ULTRAhyb-Oligo hybridization buffer (Ambion, Austin, TX), and blots were washed as described by the manufacturer. The following sequences were used as oligonucleotide probes: sequences complimentary to miRNAs (www.sanger.ac.uk), In-tRNA-Tyr (TCGAACCAGCGACCTAAAGATTGTCTGCAAACACTACAGTCTTC), tRNA-Tyr (TCGAACCAGCGACCTAAAGAT), tRNA-Met (TCGAACTCTCGACCTCAGGAT), U6 (AGGGGCCATGCTAATCTTCTC), 5S rRNA (GAGGGATGCAACACGAGGACTT), ASRP1511 (AAGTATCATCATTCGCTTGGA) ASRP255 (TACGCTATGTTGGACTTAGTT), ASRP02 (GTTGACCAGTCCGCCAGCCGAT), and ASRP1003 (ATGCCAAGTTTGGCCTCACGGTCT). Cluster2 and AtSN1 probes were prepared as described in ref. 33.
For mRNA analysis, 10 μg of total RNA was subjected to electrophoresis on a 1.2% agarose gel, transferred to Hybond N+ membranes (Amersham Pharmacia), and hybridized with [32P]-labeled probes. Probes were generated from cDNA by PCR and labeled by using the Prime-it II Random Primer Labeling kit (Stratagene) according to the manufacturer's instructions. Hybridization was carried out at 68°C by using PerfectHyb Plus hybridization buffer (Sigma), and blots were washed as described by the manufacturer.
Cellular Fractionation. Tissue from 3-week-old immature rosettes or floral buds was frozen in liquid nitrogen and ground to a powder with a mortar and pestle. Cell wall-disrupting buffer (10 mM potassium phosphate, pH 7.0/100 mM NaCl/10 mM 2-mercaptoethanol/1 M hexylene glycerol) was then added to make a thick slurry. This mixture was filtered through miracloth to remove large chunks of tissue and centrifuged at 1,500 × g for 10 min at 4°C to pellet nuclei and cell debris. After centrifugation, the supernatant was collected and re-centrifuged at 13,000 × g for 15 min at 4°C. The supernatant of this second centrifugation was saved for the cytoplasmic fraction. The first pellet was washed with nuclei preparation buffer (NPB) (10 mM potassium phosphate, pH 7.0/100 mM NaCl/10 mM 2-mercaptoethanol/1 M hexylene glycerol/10 mM MgCl2/0.5% Triton X-100) and centrifuged at 1,500 × g for 10 min at 4°C. After centrifugation, the supernatant was discarded and the pellet was washed with NPB. Washing and centrifugation were repeated four to five times, and the final pellet was saved for the nuclear fraction. RNA was extracted from the cytoplasmic and nuclear fractions as described in ref. 34.
Results
miRNAs Are Processed in the Nucleus. Intermediate stages in miRNA processing are difficult to detect in plants. To solve this problem, several groups have taken advantage of the observation that miRNAs and intermediates in miRNA processing accumulate in virus-infected plants and in plants transformed with viral proteins that interfere with miRNA processing (27, 35-39). As an alternative approach, we produced transgenic Arabidopsis plants overexpressing a 396-nt genomic sequence containing the predicted miR156a pre-miRNA (Fig. 1A ); high-level expression of this sequence was driven by the CaMV 35S promoter. All of the transgenic lines produced with this construct had a latef lowering phenotype that was attributable to the down-regulation of the miR156 target, SPL3 (G.W. and R.S.P., unpublished observations), indicating that this construct produces a functional miR156.
Fig. 1.
The intracellular distribution of miR156 and mir156a *. (A) Diagram of the 35S::miR156a construct. (B) Northern blots of total RNA isolated from nuclear and cytoplasmic fractions of 16-day-old wild-type (Col) and transgenic (TG) plants expressing miR156a under the regulation of the CaMV 35S promoter. Blots were hybridized with oligonucleotide probes complementary to the functional (miR156) or nonfunctional (miR156a *) strands of miR156a. The RNA loaded on these blots represents ≈8% (50 μg) of the cytoplasmic yield and 30% (10 μg) of the nuclear yield. U6 small nuclear RNA and tRNA-Met were used as markers of nuclear and cytoplasmic RNA, respectively.
To determine the site(s) of miRNA processing, RNA was prepared from nuclear and cytoplasmic fractions of transgenic and wild-type Col leaves, and hybridized with probes for miR156 or the nonfunctional strand, miR156a *. Two miR156-related transcripts were observed in fractions of Col plants (Fig. 1B ). The smaller of these transcripts was 20 nt, which is the size of most of the miR156 clones that have been sequenced (40). The larger transcript was 21 nt and may represent an alternative form of miR156 or the closely related miRNA, miR157. Both of these transcripts were more abundant in the cytoplasm than in the nucleus. Because of the relatively small amount of RNA in the nucleus, it was necessary to load a larger proportion the nuclear RNA preparation than the cytoplasmic RNA preparation (see legend for Fig. 1). As a result, the difference in the abundance of miRNA in the nucleus and cytoplasm is actually greater than is illustrated in Fig. 1B . No transcripts corresponding to the miR156a * strand were detected in Col in this experiment, although this strand has been detected in prolonged exposures of unfractionated Col seedling RNA (data not shown). Transgenic plants had a very large amount of a 20-nt miR156 transcript in the cytoplasm and a smaller amount of this transcript in the nucleus (Fig. 1B ). Small amounts of miR156a *-related transcripts were also detected in both the nucleus and the cytoplasm of transgenic plants; three transcripts, between 19 and 22 nt in length, were present in the cytoplasm, whereas the nucleus contained 22- and 23-nt miR156a *-related transcripts. Transcripts corresponding to the predicted 84-nt miR156a hairpin were not observed in either wild-type or transgenic plants. Although several 60- to 90-nt transcripts hybridized to the miR156 probe in the nucleus of transgenic plants, these transcripts did not hybridize to the miR156a * probe and, thus, do not correspond to the miR156a:miR156a * hairpin. Whether these transcripts are actual intermediates in miR156a processing is unknown.
The presence of mature miRNAs in the nucleus cannot be explained by contamination of the nuclear fraction with cytoplasmic RNA because different types of miR156a * transcripts were detected in the cytoplasmic and nuclear fractions (Fig. 1B ). Additional evidence that there was insignificant contamination of the nuclear preparation with cytoplasmic RNA is provided by the distribution of tRNA-Met in these fractions (Fig. 1B ). tRNA-Met is more abundant in the cytoplasm than miR156 and would therefore be expected to be more abundant than miR156 in nuclear RNA preparations if the miR156 in these preparations is cytoplasmic in origin. In contrast, we found that tRNA-Met was present at approximately the same absolute level as miRNAs in our nuclear RNA preparations. Indeed, the nuclear:cytoplasmic ratio of tRNA-Met was 10-fold lower than that of miR156 (0.02 vs. 0.21 for the blot illustrated in Fig. 1B ), indicating that most of the miR156 present in the nuclear fraction does not represent contamination with cytoplasmic RNA.
We conclude that the miR156:miR156a * duplex is excised from the primary miR156a transcript in the nucleus and is then exported to the cytoplasm either as a ds or ss molecule. The vast excess of miR156 relative to mir156a * in both the nucleus and the cytoplasm indicates that miR156 exists primarily as a ss molecule in both locations, presumably in association with proteins that protect it from degradation.
HST, but Not PSD, Plays a Role in miRNA Processing. To determine whether HST and PSD play a role in miRNA processing, we examined miRNA levels in hst-1, hst-6, hst-3, and psd-13 plants. hst-1, hst-6, and psd-13 are putative null alleles, and hst-3 is a 3-nt deletion in the first exon of HST that replaces amino acids D36S37 with alanine (3). All three hst mutations reduced the abundance of miR156, miR159, and miR171 but had no effect on miR172 (Fig. 2A ). The observation that hst-3 reduces miRNA accumulation is particularly significant because this mutation blocks the interaction between HST and AtRAN1 in a yeast two-hybrid assay (3). Exp5 (8, 10, 11) and other exportins in the importin β family (7) only bind their cargoes in association with Ran-GTP. The observation that hst-3 interferes with miRNA accumulation, therefore, suggests that HST interacts with at least some miRNAs in a Ran-dependent manner and supports the hypothesis that HST plays a direct role in miRNA processing and/or export. psd-13 had no obvious effect on the accumulation of these four miRNAs.
Fig. 2.
hst decreases the accumulation of many, but not all, miRNAs. (A) Northern blot of total RNA from Col, hst-3, hst-1, hst-6, and psd-13 rosette leaves hybridized sequentially with probes complementary to the indicated miRNAs. (B) Blots of total RNA from immature rosette leaves, fully expanded rosette leaves, and flower buds hybridized sequentially with probes complementary to the indicated miRNAs. The intensity of the hybridization signal relative to Col is indicated and was calculated after normalization to U6.
To determine the scope of HST function, we examined the effect of hst-1 on the accumulation of a large number of miRNAs in total RNA isolated from immature and mature rosettes and floral buds (Fig. 2B ). Interestingly, the effect of hst-1 varied for different miRNAs and in different tissues. miR159, miR161, miR165, miR171, and miR168 were reduced to approximately the same extent in all tissues. miR156 accumulation was reduced dramatically in immature rosettes and less so in mature rosettes and flowers. The accumulation of miR160, miR162, and miR169 was slightly reduced in both immature and mature rosettes but was essentially normal in flowers. miR172 and miR163 were present at normal, or perhaps slightly elevated, levels in all of these tissues. miR319 was undetectable in vegetative tissue but was present at a normal or slightly elevated level in mutant flowers. Thus, HST appears to be required for the biogenesis or stability of many miRNAs but is required to different extents by different miRNAs and in different tissues. It is intriguing that miR172 is one of the few miRNAs that is completely unaffected by hst. miR172 is unusual among plant miRNAs in that it acts primarily to repress translation rather than to cleave mRNAs (41, 42). Perhaps miR172 associates with its mRNA targets in the nucleus and is exported by mRNA export pathways, as is the case for a variety of translational repressors in animals (43).
To further explore the function of HST and PSD, we examined the accumulation of miRNAs in hst-1, psd-13, and hst-6 psd-13 double mutants (Fig. 3). Only mature-length miRNAs were visible on Northern blots of nuclear and cytoplasmic RNA isolated from leaves (Fig. 3A ) and flowers (Fig. 3B ) of Col and mutant plants. miR156a * and miR165a * were undetectable in both Col and mutant plants (Fig. 3A ). All of the miRNAs we examined were present in both the nucleus and the cytoplasm but were more abundant in the cytoplasm. The difference in the abundance of miRNA in the nucleus and cytoplasm is greater than illustrated here because a larger fraction of the nuclear preparation was loaded on each gel. hst-1 had different effects on different miRNAs. It reduced the level of miR156, miR165, and mir159 by approximately the same amount both the nucleus and cytoplasm but had little or no effect on the accumulation of miR172 and miR163.
Fig. 3.
hst-1, but not psd-13, reduces the accumulation of miRNAs in both the nucleus and cytoplasm. Blots of RNA from nuclear and cytoplasmic fractions of 3-week-old rosettes (A) and flowers (B) were hybridized sequentially with probes complementary to the indicated miRNAs and the siRNA, ASRP255. Fifty micrograms of RNA was loaded was loaded in the case of cytoplasmic samples; this represented ≈10% of the cytoplasmic yield. The amount of nuclear RNA loaded on these blots was as follows: A,3 μg/15%; B Left,10 μg/20%; B Right,20 μg/40%. The intensity of the hybridization signal relative to Col is indicated and was calculated after normalization to 5S rRNA.
The Targets of miRNAs Accumulate in hst-1. To assess the functional significance of the decrease in miRNA abundance in hst-1, we examined the expression of genes targeted by miR171, miR156, and miR162. As expected, the abundance of the full-length mRNA of these targets was elevated in hst-1 (Fig. 4). In the case of SCL6-IV, this was accompanied by a corresponding decrease in the level of the 3′ cleavage fragment resulting from miR171-directed cleavage (44). The effect of hst on the miR162 target, DCL1, was particularly interesting because of its implications for the site of miRNA activity. DCL1 is represented by three transcripts: a 6-kb full-length transcript, and two smaller fragments that result from defective splicing of intron 14 (45). The miR162 miRNA recognition element is absent from the 3′ 2.5-kb fragment resulting from this missplicing event. A probe that hybridizes to this 3′ fragment revealed that both this fragment and the full-length DCL1 transcript are elevated in hst-1 (Fig. 4). Because this 3′ fragment lacks the miR162 miRNA recognition element, this result suggests that the DCL1 is subject to miR162-directed cleavage before it is completely processed. Other mutants that disrupt miRNA biogenesis have a similar effect on DCL1 transcript accumulation (45). Along with the observation that miRNAs are present in the nucleus predominantly as ss molecules (Figs. 1 and 4), this result suggests that there is RNA-induced silencing complex (RISC) activity in the nucleus in Arabidopsis.
Fig. 4.
miRNA targets accumulate in hst-1. Northern blot of total RNA from 3-week-old rosettes hybridized sequentially with probes to the indicated mRNAs. The cognate miRNA is indicated in parentheses. Actin was used as a loading control.
PSD Regulates tRNA Processing and Export. Previously we showed that PSD partially rescues the tRNA export defect of los1 in yeast, indicating that it has the capacity to export tRNAs in yeast (4). PSD cannot be the only tRNA export receptor in Arabidopsis, however, because null alleles of this gene are viable (4, 5). HST is an obvious candidate for a second tRNA export receptor because psd, hst double mutants have a much more severe phenotype than either single mutant (although they are viable) (4), and because Exp5 has been shown to export tRNAs (12, 13). To examine the role of these proteins in tRNA export, we examined the accumulation of tRNA-Tyr and tRNA-Met in the nuclear and cytoplasmic fractions of floral tissue from Col, hst-6, psd-13, and hst-6 psd-13 double mutants. We were particularly interested in the effect of these mutations on tRNA-Tyr because accumulation of unspliced tRNAs is diagnostic of defects in tRNA export in yeast (reviewed in ref. 46). tRNA-Tyr is encoded by 74 genes in Arabidopsis, 69 of which have an intron in the anticodon loop (www.tigr.org/tigr-scripts/e2k1/neuk_rna.spl?db=ath1).
Hybridization with probes spanning the intron in tRNA-Tyr revealed that psd-13 increases the level of unspliced tRNA-Tyr in the nucleus and decreases the level of spliced tRNA-Tyr in the cytoplasm (Fig. 5). psd-13 slightly increased the level of intron-less tRNA-Met in the nucleus but did not affect the cytoplasmic level of this tRNA. hst-6 did not affect the abundance of tRNA-Tyr or tRNA-Met in the nucleus or cytoplasm. Furthermore, hst-6 psd-13 double mutants had essentially the same distribution and level of these tRNAs as psd-13. These results are consistent with a role for PSD in tRNA export but demonstrate that it is not absolutely required for this process. The molecular phenotype of hst, psd double mutants (Figs. 3 and 4) indicates that the severe morphological phenotype of these plants (4) probably arises from simultaneous defects in miRNA and tRNA processing, rather than from an enhanced defect in one or the other process.
Fig. 5.
The effect of hst and psd on tRNA accumulation. Blot of cytoplasmic and nuclear RNA from Col, hst-6, psd-13, and hst-6 psd-13 floral buds hybridized with probes complementary to tRNA-Tyr and tRNA-Met. The In-tRNA-Tyr probe hybridizes both to unspliced and spliced forms of this tRNA. The intensity of the hybridization signal relative to Col is indicated and was calculated after normalization to 5S rRNA. The arrow indicates the band used for quantitation.
HST Is Not Required for the Processing of siRNAs. siRNAs differ from miRNAs in that they arise from long, completely paired dsRNA precursors. Mature siRNAs are similar in size to mature miRNAs but typically accumulate in Arabidopsis as ds molecules, rather than as ss molecules (24, 39, 47). Arabidopsis produces two major size classes (20-21 and 23-24 nt) of siRNAs from endogenous and exogenous (e.g., transgene) transcripts (26, 48). The small class is associated with posttranscriptional gene silencing, whereas the large class is associated with transcriptional silencing via heterochromatin formation (26, 48, 49).
To determine whether HST plays a role in siRNA processing, we examined the effect of hst on the accumulation of several different endogenous siRNAs, and on posttranscriptional gene silencing mediated by transgene-derived siRNAs. Two different transgenic systems were investigated. The L1 line contains a 35S::GUS transgene that is silenced posttranscriptionally by a mechanism that depends on the RNA-dependent RNA-polymerase, RDR6 (Fig. 6B ) (50). For this reason, the transgene in L1 is thought to encode a ss GUS transcript. The 306-1 transgene (35S::ΔGUS-SUG) encodes a transcript that has two identical 558-nt GUS sequences in inverted orientation (51). Posttranscriptional gene silencing initiated by this hairpin transcript is monitored via a second, unlinked 35S::GUS transgene, 6b4 (Fig. 6C ). Silencing in this system does not depend on RDR6 (50).
Fig. 6.
HST is not required for the production/stability of endogenous siRNAs or for transgene silencing. (A) Blots of low-molecular-weight RNA hybridized with probes to the indicated siRNAs. (B) The expression of the L1 transgene in mature rosettes of Col, hst-6, and sgs2-1 (rdr6) plants. Col and hst-6 have similar low levels of GUS expression, whereas GUS expression is elevated in sgs2-1. (C) The expression of 35S::GUS in the absence (Left) and presence (Right) of a 35S::hairpin GUS. hst-6 does not prevent silencing of 35S::GUS.
hst-1 had no obvious effect on the accumulation of endogenous siRNAs in either vegetative or floral tissue (Fig. 6A ). ASRP1511, ASRP255, AtSN1, ASRP02, and Cluster 2 siRNAs were present at the same level in mutant and wild-type plants, whereas ASRP1003 showed inconsistent patterns of accumulation in mutant and wild-type vegetative and floral tissues. ASRP255 was also present at roughly the same level in the nucleus of hst, psd, and hst, psd plants (Fig. 3B ). In addition, hst-6 did not interfere with posttranscriptional gene silencing initiated by L1 or by the 306-1 hairpin (Fig. 6 B and C ). We conclude that HST is not required for the biogenesis or function of most siRNAs.
Discussion
The biogenesis of miRNAs in plants is poorly understood because intermediates in this process are present in very small amounts in wild-type plants. Indeed, these intermediates have only been observed in one case (27). We approached this problem by characterizing the types of miRNAs present in nucleus- and cytoplasm-enriched fractions of wild-type plants and transgenic plants overexpressing miR156a. In addition, we studied the phenotype of mutations in two proteins potentially involved in the nuclear export of miRNAs: HST and PSD.
All of the miRNAs examined in this study were found as mature-length, 20- to 21-nt molecules, in both the nucleus and the cytoplasm. Molecules corresponding in size to the pre-miRNAs observed in mammals were not observed in wild-type plants or in plants engineered to overexpress miR156a. This result is consistent with the results of Papp et al. (2003) and suggests that miRNAs are excised from the primary transcript in the nucleus and are exported from the nucleus in their mature form. Whether the absence of pre-miRNAs in Arabidopsis is due to the speed with which processing occurs, or reflects a unique mechanism of miRNA processing in plants, remains to be determined. The “star” strands of miR156a and miR165 were difficult to detect in wild-type plants. However, low levels of miR156a * were found in both the nucleus and cytoplasm of plants overexpressing this miRNA, implying either that there is a mechanism for exporting ds molecules or that these star strands are exported independently of the functional strand.
In animals, miRNAs must function in the cytoplasm because they act primarily to inhibit translation. The site at which miRNAs act in plant cells is unknown, although there is indirect evidence that they function in the cytoplasm in wheat (52). We found that miRNAs are significantly more abundant in the cytoplasm than in the nucleus and, thus, probably function primarily in this location. It is intriguing, however, that both miR156 and miR165 are present in the nucleus predominantly as ss molecules. Studies employing animal-cell extracts indicate that miRNAs and siRNAs are unwound when they are incorporated into RISC (53-55). This also appears to be true in plants, because viral proteins that interfere with RISC activity result in the accumulation of both miRNA strands (35, 38). The abundance of functional miRNAs in both the nucleus and the cytoplasm, therefore, suggests that miRNAs actually function in both locations.
The mammalian ortholog of HST, Exp5, has at least two functions. In addition to exporting pre-miRNAs, tRNAs, and other types of dsRNAs, Exp5 also plays an important role in stabilizing pre-miRNAs: Cells deficient in Exp5 have reduced amounts of pre-miRNAs, and mutations that block the interaction between a pre-miRNA and Exp5 lead to the degradation of the pre-miRNA (8-10, 12, 13). Previously, we showed that HST is located near the nuclear membrane and interacts with At-RAN1 in a yeast two-hybrid assay, supporting the conclusion that it functions as a nucleocytoplasmic transport receptor (3). The possibility that HST might transport miRNAs, tRNAs, and other small RNAs was suggested both by its similarity to Exp5 and by the genetic interaction between hst and mutations in genes involved in miRNA, siRNA, and tRNA biogenesis or function (4, 31, 33). Our results indicate that HST is required for the biogenesis or stability of some miRNAs. Although we have no convincing evidence that HST is involved miRNA export because hst does not cause miRNAs to accumulate in the nucleus, it is difficult to know how interpret this result because of the general reduction in miRNA levels in mutant plants. The observation that miRNA levels are reduced in hst-3 is significant because this mutation that blocks the interaction between HST and the small GTPase, RAN1. Because exportins in the importin β family only interact effectively with their cargoes in association with Ran-GTP (7), this result suggests that HST interacts directly with mature miRNAs or their precursors and is consistent with a role for this protein in miRNA export. Nevertheless, there is no evidence that HST is absolutely required for miRNA export or that it has a role in the processing/export of tRNAs. In this respect, HST and Exp5 appear to be functionally distinct. This selectivity may indicate that HST is able to discriminate between small RNAs by virtue of their secondary structure or by the proteins with which they associate. Whatever the case, the phenotype of hst must reflect the existence of multiple pathways for the nucleocytoplasmic transport of miRNAs in Arabidopsis.
What is the identity of these HST-independent export pathway(s)? HST has no close relatives in the Arabidopsis genome (3), so alternative export receptors are difficult to predict. PSD is an obvious candidate, but the observation that psd-13 had no effect on miRNA accumulation makes this possibility unlikely. Another possibility is that miRNAs leave the nucleus by diffusion. This seems unlikely, however, because most miRNAs in the nucleus are ss and are probably stabilized by their association with various proteins. If nuclear miRNAs are associated with proteins, they may be directed to protein export pathways by virtue of this association. Perhaps the most interesting possibility is that miRNAs are exported from the nucleus in association with their target mRNAs. This hypothesis predicts that the requirement for HST will vary with the expression level of the mRNA to which a miRNA hybridizes, and it could account for the tissue-specific effect of hst on miRNA accumulation. In this regard, it is intriguing that hst has a lesser effect on miRNA levels in immature flower buds, where mRNA levels are relatively high, than in expanded leaves (Fig. 2B ). The development of an experimental system in which the interaction between HST and its substrates can be studied in detail is an important goal for future research. Genetic and molecular studies of suppressors and enhancers of hst mutations may also reveal the identity of other miRNA export pathways and provide new insights into the mechanism by which miRNAs and other small RNAs are processed in Arabidopsis.
Acknowledgments
We thank Jeongsik Yong and Christine Hunter for helpful comments on the manuscript. Doris Wagner provided much useful advice on nuclear-cytoplasmic fractionation. This work was supported by National Science Foundation Grant IBN 0346050.
Footnotes
-
↵ § To whom correspondence should be addressed. E-mail: spoethig{at}sas.upenn.edu.
-
Author contributions: M.Y.P. and R.S.P. designed research; M.Y.P. and A.G.-S. performed research; G.W. and H.V. contributed new reagents/analytic tools; M.Y.P. and R.S.P. analyzed data; and R.S.P. wrote the paper.
-
This paper was submitted directly (Track II) to the PNAS office.
-
Abbreviations: ds, double-stranded; miRNA, microRNA; siRNA, small interfering RNA; ss, single-stranded.
- Copyright © 2005, The National Academy of Sciences
References
Nuclear processing and export of microRNAs in Arabidopsis
Sign up for Article Alerts
Jump to section
You May Also be Interested in
Did the Caribbean sweep into the western Amazon millions of years ago, shaping the region’s rich biodiversity?
Image credit: Tacio Cordeiro Bicudo (University of São Paulo, São Paulo, Brazil), Victor Sacek (University of São Paulo, São Paulo, Brazil), and Lucy Reading-Ikkanda (artist).
Settlements 4,200 years ago may have suffered from overpopulation before drought and lower temperatures ultimately made them unsustainable.
Image credit: Andrea Ricci.
Marianne Alleyna, Aimy Wissa, and Ophelia Bolmin explain how the click beetle amplifies power to pull off its signature jump.
Some songbird parents might improve their own fitness by manipulating their offspring into leaving the nest early, at the cost of fledgling survival, a study finds.
Image credit: Gil Eckrich (photographer).