- © 2010 American Society of Plant Biologists
Abstract
We previously identified Arabidopsis thaliana mutants defective in sense transgene posttranscriptional gene silencing (S-PTGS) that defined six loci; here, we describe mutants that define nine additional loci, including HYPER RECOMBINATION1 (HPR1), SILENCING DEFECTIVE3 (SDE3), and SDE5. Our analyses extend previous findings by showing that the requirement for the putative RNA helicase SDE3 is inversely proportional to the strength of the PTGS inducer and that the putative RNA trafficking protein SDE5 is an essential component of the trans-acting small interfering RNA (tasiRNA) pathway and is required for S-PTGS but not inverted repeat transgene-mediated PTGS (IR-PTGS). Our screen also identified HPR1 as a PTGS actor. We show that hpr1 mutations negatively impact S-PTGS, IR-PTGS, and tasiRNA pathways, resulting in increased accumulation of siRNA precursors and decreased accumulation of mature siRNA. In animals, HPR1/THO1 is a member of the conserved RNA trafficking THO/TREX complex, which also includes TEX1/THO3. We show that tex1 mutants, like hpr1 mutants, impact TAS precursor and mature tasiRNA levels, suggesting that a THO/TREX complex exists in plants and that this complex is important for the integrity of the tasiRNA pathway. We propose that both HPR1 and TEX1 participate in the trafficking of siRNA precursors to the ARGONAUTE catalytic center.
INTRODUCTION
RNA silencing regulates gene expression through the action of small RNAs and serves as a eukaryotic defense response that thwarts invading RNA deriving from transposons, viruses, and transgenes (Mallory and Vaucheret, 2006; Ding and Voinnet, 2007; Kloc and Martienssen, 2008; Voinnet, 2009). In plants, 24-nucleotide small interfering RNAs (siRNAs) mediate transcriptional gene silencing through DNA methylation and chromatin modifications while 21-nucleotide siRNA and microRNA (miRNA) mediate posttranscriptional gene silencing (PTGS) through RNA cleavage and translational inhibition (Mallory and Vaucheret, 2006; Ding and Voinnet, 2007; Kloc and Martienssen, 2008; Voinnet, 2009). Transgene-based forward genetic screens have identified mutants defective for S-PTGS (sense transgene-mediated PTGS) and IR-PTGS (inverted repeat transgene-mediated PTGS) and have served to decipher endogenous PTGS pathways.
During S-PTGS, transgenes produce double-stranded RNA (dsRNA) that is processed into siRNA duplexes by the type III RNase DICER-LIKE2 (DCL2) (Mlotshwa et al., 2008). siRNAs are methylated at their 3′ end by the methyltransferase HUA ENCHANCER1 (HEN1) (Boutet et al., 2003; Li et al., 2005; Yu et al., 2005) and then bound by ARGONAUTE1 (AGO1), which cleaves complementary target RNA (Morel et al., 2002; Baumberger and Baulcombe, 2005). AGO1-mediated cleavage generates RNA fragments that escape degradation by 5′→3′ (XRN) and 3′→5′ (EXO) exoribonucleases (Gazzani et al., 2004; Souret et al., 2004; Gy et al., 2007), due to the protecting activity of SUPPRESSOR OF GENE SILENCING3 (SGS3), and are transformed into dsRNA by RNA-DEPENDENT RNA POLYMERASE6 (RDR6) (Dalmay et al., 2000; Mourrain et al., 2000; Yoshikawa et al., 2005; Elmayan et al., 2009). These dsRNAs are processed into siRNA duplexes by DCL4 to produce secondary siRNA (Dunoyer et al., 2005; Blevins et al., 2006; Bouche et al., 2006; Deleris et al., 2006; Fusaro et al., 2006) that are bound by AGO1 and guide cleavage of target RNA, establishing an amplification loop that reinforces silencing and contributes to the systemic propagation of S-PTGS from cell to cell (short-distance signaling) and through the vasculature (long-distance signaling) (Palauqui et al., 1997; Voinnet et al., 1998; Brosnan et al., 2007; Dunoyer et al., 2010a; Molnar et al., 2010).
During IR-PTGS, the transcription of inverted repeat transgenes produces self-complementary transcripts that are processed into siRNA duplexes by DCL4 without the need of an RNA-dependent RNA polymerase (Smith et al., 2000). S-PTGS and IR-PTGS pathways are DCL4, HEN1, and AGO1 dependent (Dunoyer et al., 2005, 2007). Forward genetic screens based on an IR trigger expressed specifically in the phloem implicated components of the endogenous 24-nucleotide siRNA pathway (CLASSY1 [CLSY1], NUCLEAR RNA POLYMERASE D1a (NRPD1a), and RDR2) but not components of the endogenous 21-nucleotide siRNA pathway in short-distance signaling of IR-PTGS (Dunoyer et al., 2005, 2007; Smith et al., 2007). By contrast, grafting experiments implicated components of both the 24-nucleotide siRNA pathway (NRPD1a and RDR2) and the endogenous 21-nucleotide siRNA pathway (RDR6) in long-distance signaling of IR-PTGS triggered by a constitutive promoter (Brosnan et al., 2007), suggesting distinct mechanisms for short-distance and long-distance signaling (Dunoyer and Voinnet, 2008).
Forward genetic screens based on the line L1, which carries a posttranscriptionally silent p35S:β-glucuronidase (GUS) sense transgene, and the line 2a3, which carries a p35S:NIA2 sense transgene that triggers cosuppression of the NITRATE REDUCTASE1 (NIA1) and NIA2 endogenous genes, identified a series of S-PTGS–deficient mutants called sgs. Previous analyses defined sgs1 (one allele), sgs2/sde1/rdr6 (26 alleles), sgs3/sde2 (10 alleles), sgs4/ago1 (14 alleles), sgs5/hen1 (one allele), and sgs6/met1 (two alleles), but 27 mutants remained unclassified (Elmayan et al., 1998; Fagard et al., 2000; Morel et al., 2000, 2002; Mourrain et al., 2000; Boutet et al., 2003; Adenot et al., 2006). Here, we define nine additional sgs loci, and we describe sgs7/sde5, sgs9/hpr1, and sgs13/sde3 mutants. SDE3 encodes a putative RNA helicase that was previously identified in an amplicon-based genetic screen together with rdr6/sde1, sgs3/sde2, nrpd1a/sde4, and sde5 (Dalmay et al., 2000; Herr et al., 2005; Hernandez-Pinzon et al., 2007), but SDE3 still awaits an endogenous function. Both SDE5 and HPR1 encode putative RNA trafficking proteins. Similar to rdr6 and sgs3, sde5 mutants impair S-PTGS and the endogenous trans-acting siRNA (tasiRNA) pathway but not IR-PTGS, suggesting that SDE5 acts specifically in the first two pathways. By contrast, hpr1 mutants compromise S-PTGS, IR-PTGS, and the endogenous tasiRNA pathway, indicating that HPR1 acts at a step common to these three pathways. HPR1/THO1 is homologous to a member of the conserved THO/TREX complex (Reed and Cheng, 2005), and we show that a mutation in TEX1/THO3, which is homologous to another member of this complex, also impaired the production of tasiRNA. We propose that a THO/TREX complex also exists in plants and that this complex participates in the trafficking of siRNA precursors.
RESULTS
Forward Genetic Screens Based on the L1 and 2a3 Lines Identify at Least 15 PTGS Loci
The 60 mutants recovered from the L1 screen were classified in three groups based on the degree to which they affected PTGS. The first group consisted of 44 mutants producing 100% PTGS-deficient progeny. This group included one sgs1 allele, 24 sgs2/rdr6 alleles, seven sgs3 alleles, 11 sgs4/ago1 alleles, and one sgs5/hen1 allele (Elmayan et al., 1998, 2009; Fagard et al., 2000; Mourrain et al., 2000; Morel et al., 2002; Boutet et al., 2003; Adenot et al., 2006). The second group consisted of 11 mutants producing a consistent number of both PTGS-deficient and PTGS-efficient progeny at each generation, indicating that the mutations partially impaired S-PTGS. This group included three sgs4/ago1 hypomorphic alleles, two sgs6/met1 alleles, and six unclassified mutants (Morel et al., 2002; Boutet et al., 2003; Mallory et al., 2009). The third group contained five mutants that showed a delay in the onset of S-PTGS but eventually triggered S-PTGS in 100% of the progeny. The 21 mutants recovered from the 2a3 screen also affected PTGS to varying degrees and thus were classified according to the same criteria. The first group consisted of six mutants, including two sgs2/rdr6 alleles and three sgs3 alleles (Adenot et al., 2006; Elmayan et al., 2009). The second and third groups consisted of 12 and three unclassified mutants, respectively.
Complementation analyses were performed on six unclassified mutants derived from the L1 screen and six unclassified mutants derived from the 2a3 screen. The six L1-derived mutants defined six novel complementation groups (sgs7 to sgs12). One 2a3-derived mutant was a hypomorphic rdr6 allele (rdr6-8, which exhibited a T→A nucleotide change that resulted in a Y→N amino acid change at protein position 228), while the five other 2a3-derived mutants defined three complementation groups. To determine the overlap between these two screens, the L1 locus was introduced into 2a3-derived mutants, and the 2a3 locus was introduced into L1-derived mutants. All L1-derived mutants (sgs1 to sgs12) protected 2a3 against S-PTGS generally better than they protected L1 against S-PTGS. By contrast, the three novel 2a3-derived mutants delayed L1 S-PTGS, although their effect on 2a3 was more pronounced. Complementation tests revealed that these three mutants defined three unique loci (hereafter, referred to as sgs13 to sgs15). Altogether, these results show that at least 15 loci control S-PTGS in Arabidopsis thaliana.
SGS13/SDE3 Is Required for S-PTGS Triggered by Weak Silencers Only
The sgs13 mutation was mapped to an 888-bp deletion that caused a 202–amino acid truncation at the C terminus of the SDE3/At1g05460 gene (Figure 1A). SDE3 encodes a putative RNA helicase required for S-PTGS triggered by the GxA amplicon (Dalmay et al., 2001). The sgs13 mutation, hereafter referred to as sde3-6, impaired 2a3 S-PTGS and slightly delayed L1 S-PTGS. At 11 d after germination (DAG), NIA siRNAs were undetectable and NIA mRNA accumulated in 2a3/sde3-6 plants to a level comparable to 2a3/rdr6 plants (Figure 1B). At 40 DAG, none of the plants (n = 100) showed signs of NIA cosuppression, indicating that sde3 prevents 2a3 S-PTGS. By contrast, at 11 DAG, GUS activity in L1/sde3-6 plants was higher than L1 controls, although not as high as in L1/rdr6 plants, and GUS siRNAs were detectable, although the level was consistently lower than in L1 controls (Figure 1B). Confirming that the effect of sde3-6 on L1 was only partial, analyses of GUS activity, GUS mRNA, and GUS siRNA accumulation at 40 DAG indicated that L1/sde3-6 was as silenced as control L1 plants (Figure 1B). To decipher whether the limited effect of sde3-6 on L1 S-PTGS was due to the residual activity of the remaining first 800 amino acids of the protein or if SDE3 was simply a nonessential component of S-PTGS, we crossed L1 to sde3-1, which has a 14-bp deletion in exon 4, causing a frameshift that results in the truncation of the protein at amino acid 491 (Dalmay et al., 2001; Figure 1A). Because the sde3-1 allele originally was recovered in the C24 ecotype, we backcrossed it twice to L1 (in the Columbia-0 [Col-0] ecotype) before comparing its effect on L1 to sde3-6. The sde3-1 mutation, like the sde3-6 mutation, had only a minor effect on L1 PTGS at 11 DAG and was fully silenced at 40 DAG (Figure 1C), confirming that sde3 mutations delay the onset of L1 S-PTGS but do not prevent its establishment.
Analysis of S-PTGS in sde3 Mutants.
(A) Schematic representation of the SDE3 gene with the position and sequence changes introduced by the sde3-1 and sde3-6 mutations indicated. Black boxes represent exons, gray boxes untranslated regions, and thin lines introns and intergenic regions. Dash lines represent the sequences deleted in the sde3 mutants.
(B) Low molecular weight and high molecular weight (LMW and HMW, respectively) RNA gel blots of aerial parts of 11-d-old seedlings and mature rosette leaves (40 DAG) of the indicated plant lines were probed with an RNA GUS or DNA NIA2 probe. 25S rRNA and U6 snRNA hybridizations served as loading controls for HMW and LMW blots, respectively. Normalized values are indicated below each lane. Mean GUS activity is indicated in fluorescence units per min per μg of total protein (FU/min/μg protein).
(C) GUS activity (in FU/min/μg protein) of the indicated genotypes at 11 (black) and 40 DAG (gray).
(D) Percentage of Hc1 silenced plants at 40 DAG in the indicated genotypes. n = number of tested plants.
(E) LMW and HMW RNA gel blots of mature rosette leaves of the indicated genotypes were probed with DNA oligonucleotides complementary to TAS1 [si-480(+)], TAS2 (si-F), TAS3 (tasi-ARF), and DNA complementary to TAS2 precursor, respectively. The expected migration positions of primary TAS RNA precursors (pri) and the 5′ and 3′ cleavage products generated after miR173-guided cleavage are indicated. 25S rRNA and U6 snRNA hybridizations served as loading controls for HMW and LMW blots, respectively. Normalized values with the Col control set to 1 are indicated.
sde3 mutations prevented S-PTGS triggered by two-component systems (GxA and 2a3) but only slightly delayed S-PTGS triggered by a single-component system (L1). To confirm this effect, we tested the effect of sde3 on the single-component system Hc1. The Hc1 line carries the same p35S:GUS transgene as L1 but only triggers S-PTGS in 20% of the plants, whereas L1 triggers S-PTGS in 100% of the plants (Elmayan et al., 1998; Gy et al., 2007). sde3-6 abolished Hc1 S-PTGS (Figure 1D), indicating that SDE3 is not specific to two-component systems. Rather, it suggests that SDE3 is required for weak S-PTGS reporters such as Hc1, 2a3, and GxA but is dispensable for the strongly silenced L1 reporter.
The fact that sde3 has an effect on L1 (although weaker than that on Hc1, 2a3, and GxA) prompted us to further investigate the effect of sde3 mutations on the endogenous tasiRNA pathway, which requires many of the same components as L1 PTGS. Consistent with previous analyses of sde3-4 and sde3-5 (Vazquez et al., 2004; Dunoyer et al., 2010b), mature tasiRNA accumulated at wild-type levels in sde3-6 (Figure 1E). TAS precursors and cleavage products also accumulated at wild-type levels in sde3-6 (Figure 1E), confirming that SDE3 is not required for the tasiRNA pathway.
SGS7/SDE5 Is Required for S-PTGS Triggered by Strong Silencers
The sgs7 mutation was mapped to a G→A nucleotide change at the acceptor site of the second intron of SDE5/At3g15390 (Figure 2A). SDE5 encodes a putative RNA export protein required for GxA S-PTGS (Hernandez-Pinzon et al., 2007). The sgs7 mutation, hereafter referred to as sde5-4, resulted in decreased accumulation of SDE5 mRNA (Figure 2B). sde5-4 impaired both L1 S-PTGS (Figure 2C) and 2a3 S-PTGS (Figure 2D). At 11 DAG, GUS siRNAs were undetectable and GUS mRNA accumulated at high levels in sde5-4, similar to a null rdr6/sgs2-1 mutant (Figure 2C). However, the effect of sde5-4 on L1 silencing was not as strong as that of null rdr6/sgs2-1 or sgs3 mutations, which completely impair L1 silencing because at 40 DAG, 20% (n = 80) of L1/sde5-4 exhibited low GUS activity and accumulated GUS siRNA similar to silenced L1 controls (Figure 2E). These results suggest either that SDE5 is not as essential as RDR6 and SGS3 for L1 S-PTGS or that the sde5-4 mutation could be a partial loss of function. To test this hypothesis, L1 was introduced in the T-DNA insertion mutant sde5-3 (Mallory and Vaucheret, 2009; Figure 2A), which likely is a null allele (Figure 2B). Both sde5-3 and sde5-4 had similar effects on L1 S-PTGS at 11 DAG (i.e., high GUS activity, high GUS mRNA level, and undetectable GUS siRNA; Figure 2C), but, in contrast with the sde5-4 mutant, sde5-3 protected L1 against S-PTGS with 100% efficiency, indicating that sde5-4 is a partial loss of function and that SDE5 is required for S-PTGS.
Analysis of S-PTGS in sde5, rdr6, and sgs3 Mutants.
(A) Schematic representation of the SDE5 gene with the position and sequence changes introduced by the sde5-3 and sde5-4 mutations indicated. Black boxes represent exons (sequence in capital letters), gray boxes untranslated regions, and thin lines introns (sequence in lowercase). Black and gray arrows represent the position of the primers used for RT-PCR of sde5-4 and sde5-3, respectively.
(B) RT-PCR analyses of the sde5-4 and sde5-3 mutants. EF1∝ served as a standard for RT-PCR.
(C) LMW and HMW RNA gel blot analyses of aerial parts of 11-d-old seedlings of the indicated mutant plants were probed with DNA complementary to GUS mRNA. 25S rRNA and U6 snRNA hybridizations served as loading controls for HMW and LMW blots, respectively. Normalized values are indicated. GUS activity is reported in FU/min/μg protein.
(D) Cosuppressed 2a3 line (left) and cosuppression-resistant 2a3/sde5-4 mutant (right). NIA cosuppresion is visualized by chlorosis.
(E) LMW RNA gel blot of mature rosette leaves (40 DAG) of silenced (left) and nonsilenced (right) sde5-4 mutants and L1 control plants were probed with an RNA GUS probe. U6 snRNA hybridizations served as loading controls.
[See online article for color version of this figure.]
SDE5 Is a Core Component of the tasiRNA Pathway
SDE5 was previously implicated as having a limited but nonessential role in the tasiRNA pathway. Indeed, similar to our sde5-4 allele, the previously characterized sde5-2 allele (Hernandez-Pinzon et al., 2007) did not exhibit the accelerated phase change zippy phenotype typical of dcl4-2, rdr6/sgs2-1, and sgs3-1 null mutants impaired in the tasiRNA pathway (Peragine et al., 2004; Yoshikawa et al., 2005; Figure 3A). By contrast, sde5-3 exhibited a zippy phenotype (Figure 3A), suggesting that like sde5-4, sde5-2 could be a partial loss-of-function allele, prompting us to reexamine the contribution of SDE5 to the tasiRNA pathway. Mature tasiRNAs were undetectable in sde5-3, similar to rdr6/sgs2-1 and sgs3-1 null alleles (Figure 3B), whereas they accumulated at 30% of wild-type levels in sde5-2 (Hernandez-Pinzon et al., 2007; Dunoyer et al., 2010b) confirming that sde5-2 is a partial loss-of-function mutant. RNA gel blot analysis of TAS precursors and cleavage products revealed an overaccumulation of TAS cleavage products in the sde5-3 null allele (Figure 3B). This increase in TAS cleavage products was not as strong as that observed in the null rdr6/sgs2-1, but it was stronger than in the partial loss-of-function rdr6-8 (Figures 2C and 3B) and the neomorphic sgs3-3 allele (Figure 3B), which overprotects TAS cleavage products from degradation (Elmayan et al., 2009). Although the exact role of SDE5 in the tasiRNA pathway remains elusive, these data indicate that the contribution of SDE5 to the production of mature tasiRNA is more central than originally reported.
Analysis of the tasiRNA Pathway in sde5, rdr6, and sgs3 Mutants.
(A) Pictures of the indicated mutant and L1 control plants.
(B) LMW and HMW RNA gel blot analyses of mature rosette leaves of the indicated mutant and L1 control plants. LMW RNA gel blots were probed with DNA oligonucleotides complementary to TAS1 [si-480(+)], TAS2 (F), and TAS3 (tasiARF). HMW RNA gel blots were probed with DNA complementary to the TAS1a and TAS2 precursors. The expected migration positions of primary TAS RNA precursors (pri) and the 5′ and 3′ cleavage products generated after miR173-guided cleavage are indicated. 25S rRNA and U6 snRNA hybridizations served as loading controls for HMW and LMW blots, respectively, and normalized values with the L1 control set to 1 are indicated.
[See online article for color version of this figure.]
SGS9 Encodes the Putative Ortholog of HPR1, a Component of the Conserved THO/TREX Complex
The sgs9 mutation was mapped to a G→A nucleotide change that introduces a stop codon in exon14 of At5g09860 (Figure 4A). At5g09860 encodes a protein that shares homology with HPR1/THO1, one of the eight components of the RNA trafficking THO/TREX complex conserved among fungi, invertebrates, and mammals (Reed and Cheng, 2005). The sgs9 mutation, hereafter referred to as hpr1-1, suppresses L1 S-PTGS with 80% efficiency. Analysis of a bulk of plants at 11 DAG revealed that GUS siRNAs were below detectable levels and that GUS mRNA accumulated to high levels (Figure 4B). Consistent with the GUS mRNA levels, GUS activity in hpr1-1 was lower than in the null sgs3-1. At 40 DAG, 80% of the plants exhibited high GUS activity, while 20% exhibited low GUS activity similar to silenced L1 controls. This high GUS activity correlated with increased GUS mRNA levels and undetectable GUS siRNA, whereas low GUS activity correlated with low GUS mRNA levels and detectable GUS siRNA (Figure 4B). Confirmation that hpr1-1 is the cause of L1 reactivation was shown by the analysis of a second allele, hpr1-2, which carries a T-DNA inserted in the eighth intron of HPR1 (Figure 4A) that compromises the production of a functional HPR1 mRNA (Figure 4C). Homozygous hpr1-1, hpr1-2, and F1 hybrids between hpr1-1 and hpr1-2 exhibited identical developmental defects (Figures 4D and 4E). Moreover, F1 hybrids between L1/hpr1-1 and hpr1-2 showed impaired L1 S-PTGS (Figure 4F), confirming that hpr1 mutations compromise S-PTGS.
Analysis of S-PTGS in hpr1 Mutants.
(A) Schematic representation of the HPR1 gene with the position and sequence changes introduced by the hpr1-1 and hpr1-2 mutations indicated. Black boxes represent exons, gray boxes untranslated regions, and thin lines introns. Black arrows represent the position of the primers used for RT-PCR.
(B) LMW and HMW RNA gel blots of aerial parts of 11-d-old seedlings and mature rosette leaves (40 DAG) of silenced (left) and nonsilenced (right) hpr1-1 mutants and L1 control plants were probed with an RNA GUS probe and a DNA GUS probe, respectively. 25S rRNA and U6 snRNA hybridizations served as loading controls for HMW and LMW blots, respectively, and normalized values are indicated. GUS activity is indicated in FU/min/μg protein.
(C) RT-PCR analysis of the hpr1-2 mutant. EF1∝ served as a quantification standard for RT-PCR.
(D) Pictures of the indicated mutant and wild-type control plants.
(E) hpr1-1, hpr1-2, an F1 plant of the cross hpr1-1 × hpr1-2, and an F1 plant of the control cross L1 × hpr1-2.
(F) GUS activity is indicated in FU/min/μg protein in rosette leaves of 40-d-old plants for each indicated genotype. Number of tested plants, n = 10. Standard deviations are indicated.
[See online article for color version of this figure.]
The Putative Orthologs of HPR1 and TEX1, Two Members of the Conserved THO/TREX Complex, Contribute to the tasiRNA Pathway
All S-PTGS mutants identified so far in the L1 screen also impair the endogenous tasiRNA pathway. Indeed, ago1, hen1, rdr6, sde5, and sgs3 lack tasiRNA (Peragine et al., 2004; Vazquez et al., 2004; Adenot et al., 2006; Figure 3B). TAS1, TAS2, and TAS3 tasiRNA levels were reduced in hpr1-1 mutant seedlings (Figure 5A). Similarly, the hpr1-2 allele exhibited a reduction in TAS2 tasiRNA accumulation (Figure 5B), implicating HPR1 in the tasiRNA pathway. Because HPR1/THO1 is part of the conserved, multicomponent THO/TREX complex in fungi, invertebrates, and mammals, we searched for additional mutants impaired in the putative orthologs of other THO/TREX components to determine if they also are necessary for the integrity of the tasiRNA pathway. We could not find homozygous plants in the progeny of tho2/THO2 heterozygous plants, suggesting that THO2 is an essential gene. By contrast, homozygous mutants impaired in the putative ortholog of TEX1/THO3 were viable and exhibited developmental defects similar to hpr1 (Figure 5E), as well as a reduction in tasiRNA accumulation (Figure 5C). These data indicate that the putative orthologs of two THO/TREX components are needed for proper tasiRNA accumulation and suggest that a similar THO/TREX complex exists in plants. To determine at which step the production of tasiRNA is compromised in hpr1 and tex1 mutants, we analyzed the accumulation of TAS2 precursors and cleavage products. Both hpr1 and tex1 mutants overaccumulated TAS2 precursors (Figure 5D). The increase in TAS2 precursor accumulation was much stronger in tex1 than in hpr1, consistent with a stronger reduction of mature tasiRNA accumulation in tex1 (Figure 5C). In addition, the contribution of HPR1 to the tasiRNA pathway depended on the tissue or developmental stage analyzed because the decrease in mature tasiRNA accumulation was stronger in hpr1 seedlings than in hpr1 rosette leaves (cf. Figures 5A and 5C, respectively).
Analysis of the tasiRNA Pathway in hpr1 and tex1 Mutants.
(A) LMW RNA gel blot of seedlings of hpr1-1 mutant and Col-0 control plants was probed with DNA oligonucleotides complementary to miR173, TAS1 [si-480(+)], TAS2 (si-F), and TAS3 (tasi-ARF).
(B) LMW RNA gel blot of mature rosette leaves of hpr1-2 and Col-0 control plants. LMW RNA gel blot was probed with DNA oligonucleotides complementary to TAS2 (si-F), TAS3 (tasi-ARF), and miR173.
(C) LMW RNA gel blot of mature rosette leaves of hpr1 and tex1 mutants and the Col-0 control was probed with DNA oligonucleotides complementary to miR173, TAS1 [si-480(+)], TAS2 (si-F), and TAS3 (tasi-ARF).
(D) HMW RNA gel blot of mature rosette leaves of the two hpr1 alleles and tex1 mutant plants. HMW RNA gel blot was probed with DNA complementary to TAS2 precursor. The expected migration positions of primary TAS RNA precursor (pri) and the 5′ and 3′ cleavage products generated after miR173-guided cleavage are indicated. U6 snRNA hybridizations and 25S rRNA served as loading controls for all LMW and HMW blots, respectively, and normalized values with Col-0 and WS control set to 1 are indicated.
(E) Developmental phenotypes of 16-d-old hpr1 and tex1 mutants compared with Col-0.
[See online article for color version of this figure.]
HPR1 but Not SDE5 Is Involved in the IR-PTGS Pathway
Because SDE5, HPR1, and TEX1 encode putative RNA trafficking proteins involved in S-PTGS, we examined if they also contribute to IR-PTGS triggered by the JAP construct expressing PDS dsRNA under the control of the phloem-specific SUC2 promoter (Smith et al., 2007). Previous analyses indicated that sde5-2 is not impaired in PDS IR-PTGS (Hernandez-Pinzon et al., 2007). However, the elucidation of sde5-2 as a hypomorphic allele called for a reexamination of the role of SDE5 using a null allele. Introduction of the JAP3 IR-PTGS locus in sde5-3 did not impact the PDS IR-PTGS phenotype (Figure 6A). PDS IR precursors remained undetectable, similar to the JAP3 control (Figures 6B), and PDS siRNA accumulation was unchanged (Figures 6C), indicating that SDE5 is dispensable for IR-PTGS, similar to RDR6 and SGS3 (Beclin et al., 2002; Dunoyer et al., 2005). By contrast, introduction of the JAP3 IR-PTGS locus in hpr1-1 resulted in a reduced PDS IR-PTGS phenotype (Figure 6A), increased PDS IR precursor accumulation (Figure 6B), and reduced PDS siRNA accumulation (Figure 6C), indicating that HPR1 is involved in IR-PTGS.
Analysis of IR-PTGS in sde5 and hpr1 Mutants.
(A) Progression of JAP3 silencing resulting from the spreading of PDS siRNAs from the companion cells of the phloem is evidenced by photobleaching in hpr1-1 and sde5-3 mutants compared with the JAP3 control.
(B) RT-PCR of SUC2:PDS transgene-derived RNA in the indicated mutants and control JAP3 line. gDNA indicates genomic DNA controls. EF1∝ served as a quantification standard for RT-PCR.
(C) LMW RNA gel blot of mature rosette leaves of the indicated mutant plants was probed with a DNA probe complementary to PDS. PDS siRNA size is indicated on the left. U6 snRNA hybridization served as a loading control. Normalized values with the JAP3 control set to 1 are indicated.
DISCUSSION
SDE3 Requirement in S-PTGS Is Inversely Proportional to the Silencer Strength
Due to the enormous interest in PTGS in the last 15 years, several forward genetic screens have been performed in plants, including four S-PTGS screens (Elmayan et al., 1998; Dalmay et al., 2000; Adenot et al., 2006; Herr et al., 2006), two IR-PTGS screens (Dunoyer et al., 2005; Smith et al., 2007), two miRNA screens (Brodersen et al., 2008; Yu et al., 2010), and two tasiRNA screens (Peragine et al., 2004; Cuperus et al., 2009). When combined with results from reverse genetic approaches, our knowledge of the PTGS pathways in plants has expanded to include AGO1, CLSY1, DCL2, DCL4, FCA, FPA, FLD, FVE, FY, HEN1, NRPD1a, NRPD2, and RDR2 in IR-PTGS (Dunoyer et al., 2005, 2007; Baurle et al., 2007; Smith et al., 2007; Manzano et al., 2009) as well as AGO1, AGO10, DCL2, DCL4, DDM1, ESP1, ESP3, ESP4, ESP5, FRY1, HEN1, MET1, RDR6, SDE3, SDE5, SGS3, XRN2, XRN3, and XRN4 in S-PTGS (Dalmay et al., 2000, 2001; Fagard et al., 2000; Morel et al., 2000, 2002; Mourrain et al., 2000; Boutet et al., 2003; Gazzani et al., 2004; Herr et al., 2006; Gy et al., 2007; Hernandez-Pinzon et al., 2007; Mlotshwa et al., 2008; Mallory et al., 2009; Mallory and Vaucheret, 2009). Our results clearly indicate that there are a larger number of proteins that play a role in S-PTGS and that these proteins have varying degrees of importance in the different silencing systems. Among three of the S-PTGS systems used to screen for PTGS-deficient mutants, one, L1, is a strong silencer, whereas two, 2a3 and GxA, are more weakly silenced. Indeed, L1 was the line showing the most precocious onset of S-PTGS among the different lines carrying the p35S:GUS transgene (Elmayan et al., 1998). By contrast, because early cosuppression of the p35S:NIA2 transgene and NIA1 and NIA2 endogenous genes is lethal, only transgenic lines such as 2a3, which triggers S-PTGS later in development, could be propagated (Elmayan et al., 1998). Consistent with L1 being a strong silencer, GUS siRNAs appear earlier in development in the L1 line than NIA siRNAs, which appear in the 2a3 line (V. Jauvion and H. Vaucheret, unpublished data). Moreover, all mutants recovered from the L1 screen impaired 2a3 S-PTGS, whereas some mutants recovered from the 2a3 screen have only modest effects on L1 S-PTGS. This last group of mutants includes sde3, which was also recovered from the GxA screen (Dalmay et al., 2000). The GxA amplicon system also can be considered a weak silencer. Indeed, the p35S:GFP transgene (G) by itself is not silenced, and only in the presence of the p35S:PVX-GFP amplicon (A) is GxA PTGS efficient. The sde3 mutations did not alleviate GxA PTGS as efficiently as sde1/sgs2/rdr6, sde2/sgs3, and sde5/sgs7 (Dalmay et al., 2000; Hernandez-Pinzon et al., 2007), suggesting that the strength of the GxA system is between that of L1 and 2a3. We ruled out the possibility that SDE3 could be specific to two-component systems (2a3 and GxA) by showing that sde3 mutations abolish S-PTGS of the Hc1 line, which carries the same transgene as the L1 line but triggers S-PTGS less efficiently than L1 (Elmayan et al., 1998; Gy et al., 2007). Because the tasiRNA pathway requires RDR6, SDE5, and SGS3 but not SDE3, the L1 S-PTGS reporter system serves as a good mimic of the tasiRNA pathway. At present, although SDE3 is involved in Hc1, 2a3, and GxA S-PTGS and possesses a putative RNA helicase domain, it still awaits assignment of an endogenous function.
SDE5 Is a Core Component of S-PTGS and tasiRNA Pathways
Previous characterization of SDE5 was based on hypomorphic sde5 alleles and suggested a limited role of SDE5 in silencing pathways (Hernandez-Pinzon et al., 2007). Our analysis of a null sde5 allele shows that SDE5 plays a more crucial role in the S-PTGS and tasiRNA pathways than previously appreciated. We show that a sde5 null allele is not impaired in IR-PTGS but is deficient for S-PTGS and tasiRNA production, indicating that, like RDR6 and SGS3, SDE5 is a core component of the S-PTGS and tasiRNA pathways but is dispensable for IR-PTGS. It remains unclear if the tasiRNA pathway takes place in the nucleus, the cytoplasm, or both. RDR6 and SGS3 likely localize in the cytoplasm, whereas DCL4 is nuclear (Hiraguri et al., 2005; Glick et al., 2008; Elmayan et al., 2009; Kumakura et al., 2009). However, caution should be taken because all localization experiments were done using constructs expressed under the control of the 35S promoter, which can affect protein localization. Indeed, RDR6 was observed in the nucleus in another report (Luo and Chen, 2007). A plausible scenario would be that TAS cleavage products are exported from the nucleus to the cytoplasm where they are protected against degradation by SGS3 and converted to dsRNA by RDR6 before reentering the nucleus to be processed into siRNA by DCL4. Whether SDE5 participates in the export of these RNAs from the nucleus to the cytoplasm or their import from the cytoplasm to the nucleus remains unknown. TAS cleavage products are detected in sde5 null alleles, indicating that the absence of SDE5 does not prevent TAS precursors from being cleaved. If SDE5 participates in the export of TAS cleavage products from the nucleus to the cytoplasm, in sde5 mutants, nuclear-retained TAS cleavage products would not be transformed into dsRNA by RDR6 but also would not be protected by SGS3. The overaccumulation of TAS cleavage products observed in sde5 is consistent with this hypothesis. However, it also is consistent with the hypothesis that SDE5 imports TAS dsRNA from the cytoplasm to the nucleus because TAS dsRNA that could not be processed into siRNA by DCL4 would also overaccumulate in the cytoplasm.
HPR1 Acts in IR-PTGS, S-PTGS, and tasiRNA Pathways
In contrast with the S-PTGS–specific SDE3, SDE5 is required for both S-PTGS and the tasiRNA pathways, but not the IR-PTGS pathway. On the other hand, HPR1 plays a role in S-PTGS, tasiRNA, and IR-PTGS pathways, indicating that it is involved in a step common to these three pathways. In yeast, Drosophila melanogaster, and human, HPR1/THO1 is a member of the THO/TREX RNA trafficking complex and is required for the regulation of a subset of endogenous mRNA (Reed and Cheng, 2005). Arabidopsis hpr1 mutants exhibit mild developmental defects, which cannot be attributed solely to the partial impairment of the tasiRNA pathway. Rather, these defects are likely explained by a general, but limited, role of HPR1 in the trafficking of a subset of endogenous RNA, similar to the role that it plays in yeast, Drosophila, and human (Reed and Cheng, 2005). Although the exact role of HPR1 in the trafficking of endogenous mRNA remains to be elucidated, the implication of HPR1 in S-PTGS, IR-PTGS, and tasiRNA pathways reinforces the idea that proteins of the endogenous RNA metabolism pathways are shared among the different RNA silencing machineries. Indeed, the Cap binding proteins CBP20 and CBP80 are involved in the miRNA pathway (Gregory et al., 2008; Laubinger et al., 2008), and the putative splicing factor ESP3, the RNA 3′ end formation factors ESP1, ESP4, and ESP5, and the 5′→3′ exoribonucleases XRN2, XRN3, and XRN4 are involved in the S-PTGS pathway (Gazzani et al., 2004; Herr et al., 2006; Gy et al., 2007), while the 3′ RNA processing proteins FCA, FPA, and FY are involved in IR-PTGS (Baurle et al., 2007; Manzano et al., 2009),
HPR1/THO1 is part of a multicomponent THO/TREX complex that is conserved among fungi, invertebrates, and mammals. Three lines of evidence suggest the existence of a plant THO/TREX complex that includes, at least, the putative orthologs of HPR1/THO1 and TEX1/THO3: (1) Arabidopsis HPR1 and TEX1 are homologous to their animal counterparts. Arabidopsis HPR1 is 29 and 28% identical to Drosophila and human HPR1, respectively, and Arabidopsis TEX1 is 50 and 48% to Drosophila and human TEX1, respectively. (2) Arabidopsis hpr1 and tex1 mutants display overlapping developmental defects, and (3) both Arabidopsis hpr1 and tex1 mutants show increased accumulation of uncleaved TAS precursors and decreased accumulation of mature tasiRNA. This overaccumulation of uncleaved TAS precursors is similar to what has been observed in dcl1 mutants, which lack miRNAs that specifically initiate the tasiRNA pathway by cleaving TAS precursors (Allen et al., 2005; Yoshikawa et al., 2005). Because of these similarities, and because neither hpr1 nor tex1 affects miR173 accumulation, we propose that the THO/TREX complex acts in an upstream step of tasiRNA biogenesis that is downstream of miRNA production and, thus, likely acts at the level of miRNA-guided AGO-mediated cleavage of TAS precursors (Figure 7). Because S-PTGS and IR-PTGS involve AGO1-mediated cleavage of siRNA precursors, we propose that the THO/TREX complex generally participates in the trafficking of siRNA precursors to AGO catalytic centers. However, tex1 and hpr1 had stronger effects on TAS1 and TAS2, which depends on AGO1, than on TAS3, which depends on AGO7, suggesting that the implication of the THO/TREX complex toward different RNA may depend on their localization and/or association with other proteins.
Model for the Putative Action Site of the THO/TREX Complex in the tasiRNA Pathway.
miRNA precursors are transcibed by PolII from MIRNA genes. CAP BINDING PROTEIN20 (CBP20) and CBP80 are recruited on the cap of miRNA precursors that are subsequently processed into small RNA duplexes by the type III RNase DCL1, which is aided by the DOUBLE-STRANDED RNA BINDING PROTEIN1 (DRB1/HYL1). SERRATE (SE), a C2H2 zinc finger protein, colocalizes in the D-bodies where DCL1 dicing takes place. Small RNA duplexes are methylated at their 3′ end by HEN1 to protect them from degradation. Finally, one strand of the duplex is loaded by an AGO protein, which mediates either mRNA cleavage (AGO1), translational repression (AGO1 and AGO10/ZLL), or DNA methylation (AGO4). The tasiRNA pathway requires the miRNA pathway. TAS mRNAs produced by PolII are exported/trafficked by the THO/TREX complex and then targeted by AGO1 and AGO7 loaded with a complementary miRNA. Cleavage fragments are protected from degradation and transformed into dsRNA by SGS3, SDE5, and RDR6. This dsRNA is processed by DCL4, assisted by DRB4, to produce tasiRNA duplexes that are methylated by HEN1. tasiRNAs are loaded on AGO1 and guide mRNA cleavage both in trans and cis.
[See online article for color version of this figure.]
The abrogation of HPR1/THO1, THO2, and TEX1/THO3 had different consequences, suggesting that if a plant THO/TREX complex exists, it is a dynamic structure in which each member plays a distinct and maybe tissue-specific role. Indeed, homozygous tho2 mutants could not be obtained, indicating that THO2 is essential. By contrast, homozygous tex1 and hpr1 mutants were viable and exhibited overlapping mild developmental phenotypes, indicating that TEX1 and HPR1 are not as essential as THO2. Finally, tex1 had a stronger effect on the tasiRNA pathway than hpr1, although the contribution of HPR1 appeared to depend on the tissue or developmental stage.
Because S-PTGS and IR-PTGS involve the propagation of short-distance and long-distance silencing signals, it also is tempting to speculate that the RNA trafficking proteins HPR1 and TEX1 play a role in the trafficking of these systemic signals. Supporting this hypothesis, hpr1 mutations did not abolish S-PTGS or IR-PTGS but delayed their establishment and/or impacted their efficiency. Grafting experiments paired with high-throughput sequencing technology using hpr1 and tex1 mutants will provide a platform to ascertain the contribution of these proteins to the systemic aspects of S-PTGS and IR-PTGS.
METHODS
Plant Material
The L1, 2a3, and JAP3 lines and the rdr6/sgs2-1, sde3-1, sde5-3, sgs3-1, and sgs3-3 mutants have been described before (Elmayan et al., 1998, 2009; Mourrain et al., 2000; Dalmay et al., 2001; Smith et al., 2007; Mallory and Vaucheret, 2009). The hpr1-1, rdr6-8, sde3-6, and sde5-4 mutants were isolated in the L1 and 2a3 screens and are described in the results section. The T-DNA insertion mutant hpr1-2 (FLAG_204H05) derives from the Institut National de la Recherche Agronomique (INRA) T-DNA collection (Samson et al., 2002) and was obtained from INRA. The T-DNA insertion mutants tho2-1 (SALK_072011) and tex1-4 (SALK_100012) derive from the SALK T-DNA collection (Alonso et al., 2003) and were obtained from the Nottingham Arabidopsis Stock Centre (NASC). Analyses were performed either on 11-d-old seedlings (aerial parts except for Figure 5A for which entire seedlings were taken) grown on Bouturage media (Duchefa) or rosettes leaves of 40-d-old plants grown on Bouturage media for 14 d and transferred to soil in controlled growth chambers. All plants were grown in standard long-day conditions (16 h of light/8 h of dark) at 21°C.
Molecular Analyses
DNA sequencing was performed as described before (Gy et al., 2007; Mallory and Vaucheret, 2009). RNA extraction, gel blot analyses, and quantification of GUS activity were performed as described before (Gy et al., 2007; Mallory and Vaucheret, 2009). GUS activity was quantified by measuring the quantity of 4-methylumbelliferone product generated from the substrate 4-methylumbelliferyl-β-d-glucuronide (Duchefa) on a fluorometer (Fluoroscan II; Thermo Scientific), and fluorescence values were normalized to total protein extracted, which was quantified using an absorbance microplate reader (Elisa Elx 808; Avantec) and the Bradford protein assay. All RNA gel blot analyses were performed using 10 μg of total RNA, except for PDS siRNA analysis (Figure 6C), which was performed using 20 μg of total RNA. GUS, PDS, TAS, U6, miR173, and 25S probes have been described before (Gy et al., 2007; Smith et al., 2007; Elmayan et al., 2009; Mallory and Vaucheret, 2009). TAS1a and TAS2 probes allowed detection of precursors as well as both 5′ and 3′ cleavage products (Elmayan et al., 2009). Hybridization signals were quantified using a Fuji phosphor imager and normalized to a U6 oligonucleotide probe, and a 25S DNA probe, for LMW and HMW gel blot analyses, respectively. All RNA gel blots were performed twice using biological replicates. For cDNA synthesis, RNAs were extracted with the Plant RNeasy kit (Qiagen), treated with DNaseI (Invitrogen), and 2 μg of DNA-free RNA was reverse transcribed with oligo-dT (Invitrogen). PCR was performed using Taq DNA polymerase (Invitrogen) according to the manufacturer's protocol. Each reaction was performed on 5 μL of 1:60 dilution of the cDNA and synthesized in a 25-μL total reaction. Specific oligonucleotide pairs are listed in Supplemental Table 1 online. The number of PCR cycles for each DNA product was determined using the minimum number of cycles in which a detectable DNA fragment was amplified but not saturated. Ethidium bromide–stained DNA gels were imaged using a Gel Doc XR (Bio-Rad). Forty cycles of amplification were used at a 54°C annealing temperature for sde5-4, 34 cycles of amplification were used at a 57°C annealing temperature for sde5-3, 34 cycles of amplification were used at a 59°C annealing temperature for HPR1, 40 cycles of amplification were used at a 58°C annealing temperature for the JAP3 transgene, and 24 cycles of amplification at a 62°C annealing temperature were used for EF1∝. The results were standardized to the expression level of EF1∝. Protein alignments were performed with MUSCLE (Edgar, 2004), and the similarity between alignments was calculated using the Prodist module of the PHYLIP package (Felsenstein, 1989).
Accession Numbers
Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: AGO1 (At1g48410), RDR6/SGS2 (At3g49500), SDE3 (At1g05460), SDE5 (At3g15390), SGS3 (At5g23570), HPR1/THO1 (At5g09860), TEX1/THO3 (At5g56130), THO2 (At1g24706), MIR173 (At3g23125), NIA1 (At1g77760), NIA2 (At1g37130), TAS1a (At2g27400), TAS2 (At2g39681), TAS3a (At3g17185), and the multigene family EF1∝ (At1g07920, At1g07930, and At1g07940).
Supplemental Data
The following material is available in the online version of this article.
Supplemental Table 1. List of DNA Oligonucleotides Used in This Study.
NOTE ADDED IN PROOF
When this work was in press, Yelina et al. (2010) recovered tex1 mutants in a screen for IR-PTGS deficiency and showed that TEX1 is required for the production of endogenous siRNAs.
Acknowledgments
We thank David Baulcombe for the JAP3 line, INRA for the hpr1-2 mutant, NASC for tho2-1 and tex1-4 mutants, Virginie Gasciolli, Isabelle Gy, and Maud Rivard for technical assistance, Hervé Ferry and Bruno Letarnec for plant care, and Allison Mallory for helpful discussions and for assisting in the writing of the manuscript. This work was supported by a PhD fellowship from the Région Ile-de-France to V.J. and ANR-06-BLAN-0203 and ANR-06-POGM-007 grants to H.V.
Footnotes
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: Hervé Vaucheret (herve.vaucheret{at}versailles.inra.fr).
↵[C] Some figures in this article are displayed in color online but in black and white in the print edition.
↵[W] Online version contains Web-only data.
- Received May 12, 2010.
- Revised July 27, 2010.
- Accepted August 5, 2010.
- Published August 26, 2010.
References
