Abstract

Transcriptional gene silencing (TGS) of transgenes by promoter-related RNAs has been known for more than a decade. However, the effectiveness and efficiency of silencing of endogenes by single-stranded and inverted repeat (IR) RNA/silencers remain unclear. Here, we demonstrated that a single-stranded antisense (AS) silencer targeting the promoter region can efficiently silence four Arabidopsis endogenes, with comparable efficiency to an IR silencer. In the case of Too Many Mouths (TMM), single-stranded silencers generated mainly 24 nt small RNAs (smRNAs), whereas IR silencers produced a higher proportion of 21–23 nt smRNAs. Heavy CG, CHG and CHH methylations were detected on the TMM promoter in silenced plant lines. We also demonstrated that the silencing and DNA methylation of the TMM promoter was dependent on the presence of the silencer. Chromatin immunoprecipitation (ChIP) assays showed that DNA methylation was accompanied by formation of repressive chromatin structures. Our results suggest that single-stranded silencer transcripts are converted to double-stranded RNA to enter the RdRM (RNA-directed DNA methylation) pathway for TGS of endogenes.

Introduction

More than a decade ago, Matzke and colleagues reported the occurrence of transcriptional gene silencing (TGS) of a transgene NOS promoter in plants by introducing a second transgene (Matzke et al. 1989). They found that TGS of the NOS promoter was associated with promoter DNA methylation (Matzke et al. 1989). Since this first report, mechanisms underlying TGS in plants have been extensively studied for both transgenes and endogenes in both monocots and dicots (for reviews, see Matzke and Matzke 2004, Eamens et al. 2008, Matzke et al. 2009). The current view of TGS suggests that 21–24 nt small RNAs (smRNAs) generated by double-stranded RNA processing machinery (AGO4, DCL3) are targeted to genomic regions with sequence homology. These smRNAs guide DNA methytransferases (DRM1/2, MET1 and CMT3) to the target locus to methylate DNA. Presumably, this is then followed by histone modifications involving histone methytransferases and deacetylases so as to establish a repressive chromatin state at the methylated locus.

Early studies using transgenes as a model system uncovered several basic features related to TGS in plants (Mette et al. 2000, Jones et al. 2001, Sijen et al. 2001). These studies demonstrated that long inverted repeat (IR) RNAs were required to generate smRNAs targeted to transgene promoters. DNA analysis of the silenced transgenes showed increases in both symmetrical and asymmetrical cytosine methylations and the involvement of MET1 in the maintenance of the silenced locus. Similarly, long IRs have also been shown to mediate efficient TGS of a 35S promoter in rice (Okano et al. 2008). Notably, single-stranded silencers in the sense (S) or antisense (AS) orientation were inefficient in producing smRNAs and generating TGS in tobacco and Arabidopsis (Mette et al. 2000).

Although the sequence of events surrounding TGS of transgenes is relatively well defined, it is not known whether similar steps also apply to TGS of endogenous loci (hereafter referred to as endogenes). Surprisingly, only a few cases of TGS of endogenes have been reported to date, all using IR RNAs. Cigan et al. (2005) reported successful and strong silencing of two maize endogenes, and moderate silencing was seen for only a few endogenes in petunia, potato and rice (Sijen et al. 2001, Heilersig et al. 2006, Okano et al. 2008). The inefficiency of IR RNAs to trigger TGS of endogenes was well illustrated in a comprehensive study using the model monocot rice. Six out of seven rice endogenes appeared to be recalcitrant to silencing by smRNAs generated from IR RNAs (Okano et al. 2008). In contrast, in the same experiment, a 35S promoter can be efficiently silenced by IR RNA targeted to this promoter.

These results highlight differences between endogenes and transgenes. Other unsuccessful trials [phytoene desaturase (PDS) and chalcone synthase (CHS)] have also been reported in Arabidopsis (Eamens et al. 2008). Together, these results suggest that endogene promoters may possess some intrinsic properties that can prevent unexpected TGS and therefore there is a need to explore more efficient silencing strategies for these genes. In addition to the observed low success rate of gene silencing of endogene promoters, there is also a discrepancy between TGS and DNA methylation as well as between DNA methylation and histone modifications for these endogenous loci (Okano et al. 2008). For most of the cases in rice, IR RNAs targeting promoter sequences could trigger DNA methylation of homologous sequences, but they failed to induce chromatin modifications and TGS (Okano et al. 2008).

Here, we compared different strategies to bring about TGS on Arabidopsis endogenes. In addition to using the conventional IR RNAs as a silencer, we also explored the use of single-stranded RNAs. Success of TGS of endogenes was correlated with DNA methylation and histone modifications. In contrast to previous studies, we found that single-stranded silencers alone could promote the production of smRNAs and bring about TGS on several Arabidopsis endogenes. DNA methylation profiles and histone modification patterns of the targeted endogenes were consistent with a repressed chromatin state.

Results

Single-stranded silencers can induce TGS of TMM

Mette et al. (2000) previously reported that an IR RNA silencer was able to cause TGS of a transgene but single-stranded silencers in either the S or AS orientation (expressed either separately or together) were ineffective. We compared the effectiveness of these three types of silencer using the Arabidopsis Too Many Mouths (TMM) endogene as a model target gene. We chose TMM because loss-of-function tmm mutants show a clustered stomata phenotype on cotyledons, which is convenient to score. We constructed silencers expressing the TMM promoter sequence (pTMM −9 to −483) in S, AS (with respect to the coding sequence) and IR configuration (Fig. 1a). The first series of silencers contained the TRV2 (Tobacco rattle virus RNA 2) self-processing sequences (Fig. 1a1) whereas the second series contained a NOS poly(A) addition sequences (Fig. 1a2).

Fig. 1

Both single-stranded silencers and the IR silencer can induce TGS at the TMM promoter. (a) Schematic representations of silencer constructs. A 475 bp (−483 to −9) fragment of the TMM promoter region was used as a silencer. (a1) Silencers without a poly(A) addition sequence. A double 35S was used to transcribe the silencer in the sense (S, sense silencer) or antisense (AS, antisense silencer) directions, or to transcribe an inverted repeat (IR) silencer with IR sequences. All 3′ ends of constructs carried a TRV2 self-processing sequence. (a2) Silencers with a NOS poly(A) addition sequence. Silencers (S, AS and IR) were transcribed from a 35S promoter. All 3′ ends of constructs carried a NOS poly(A) addition sequence. (b) Typical clustering of stomata seen in the sense silencer line, S21, the antisense silencer line, AS10, and the IR silencer line, IR8. WT, wild type Col-0. Scale bar = 10 µm. (c and d) Relative TMM transcript levels in WT and transgenic lines expressing the three different types of silencers. (c) Relative TMM transcript levels in T1 transgenic plants. For each silencer construct, a pool of 50–100 T1 transgenic plants was analyzed. (d) Relative TMM transcript levels in representative S silencer lines (S21 and S26), AS silencer lines (AS4 and AS10) and IR silencer lines (IR8 and IR44). AS10NI and IR8NI were progeny plant lines which did not carry any transgene from a T2 segregating population of AS10 and IR8, respectively. Vect., vector control. Data shown are means of three technical repeats ± SD. Similar results were obtained in another independent experiment (Supplementary Fig. S2).

Fig. 1

Both single-stranded silencers and the IR silencer can induce TGS at the TMM promoter. (a) Schematic representations of silencer constructs. A 475 bp (−483 to −9) fragment of the TMM promoter region was used as a silencer. (a1) Silencers without a poly(A) addition sequence. A double 35S was used to transcribe the silencer in the sense (S, sense silencer) or antisense (AS, antisense silencer) directions, or to transcribe an inverted repeat (IR) silencer with IR sequences. All 3′ ends of constructs carried a TRV2 self-processing sequence. (a2) Silencers with a NOS poly(A) addition sequence. Silencers (S, AS and IR) were transcribed from a 35S promoter. All 3′ ends of constructs carried a NOS poly(A) addition sequence. (b) Typical clustering of stomata seen in the sense silencer line, S21, the antisense silencer line, AS10, and the IR silencer line, IR8. WT, wild type Col-0. Scale bar = 10 µm. (c and d) Relative TMM transcript levels in WT and transgenic lines expressing the three different types of silencers. (c) Relative TMM transcript levels in T1 transgenic plants. For each silencer construct, a pool of 50–100 T1 transgenic plants was analyzed. (d) Relative TMM transcript levels in representative S silencer lines (S21 and S26), AS silencer lines (AS4 and AS10) and IR silencer lines (IR8 and IR44). AS10NI and IR8NI were progeny plant lines which did not carry any transgene from a T2 segregating population of AS10 and IR8, respectively. Vect., vector control. Data shown are means of three technical repeats ± SD. Similar results were obtained in another independent experiment (Supplementary Fig. S2).

Fig. 1 shows the results obtained with the first series of constructs (Fig. 1a1). Single-stranded silencers alone as well as a double-stranded IR silencer were able to induce typical clustering of stomata on cotyledons phenocopying tmm mutants (Fig. 1b). Penetration rates of the tmm mutant phenotype determined from a population of random T1 transformants suggested that the silencing efficiency of the single-stranded AS silencer was comparable with that of the double-stranded IR silencer, whereas the single-stranded S silencer was much less effective (Table 1). Quantitative transcript analysis by quantitative reverse transcription–PCR (qRT–PCR) showed that TMM transcript levels were reduced significantly in transgenic plants expressing AS and IR silencers (Fig. 1c, d; Supplementary Fig. S2). TMM transcript levels recovered to wild-type (WT) levels when the transgenic silencers were segregated by genetic crosses (AS10NI and IR8NI in Fig. 1d), indicating that the TGS was silencer dependent.

Table 1

Penetration rate of tmm mutant phenotype in T1 transformants

Type of silencer Phenotype (%)
 
Weak Intermediate Strong None 
S silencer 12.5 1.25 86.25 
AS silencer 15 27.5 20 37.5 
IR silencer 10 22.5 40 27.5 
Type of silencer Phenotype (%)
 
Weak Intermediate Strong None 
S silencer 12.5 1.25 86.25 
AS silencer 15 27.5 20 37.5 
IR silencer 10 22.5 40 27.5 

Constructs carrying S, AS or IR silencers described in Fig. 1a1 were used. The relative proportions of plants showing a weak, intermediate and strong tmm phenotype were scored as described in the Materials and Methods.

None, plants showing the WT phenotype. n = 80.

To test whether the TRV2 self-processing sequence was necessary for the observed TGS, we used a 35S promoter to express the same pTMM sequence but with a NOS 3′ poly(A) addition sequence in the pBA002 binary expression vector (Kost et al. 1998). Similar tmm mutant phenotype were observed in transgenic plants expressing the S, AS and IR series of silencers, although with a slightly lower penetration rate (Supplementary Table S1). Together, these results show that the effectiveness of single-stranded S and AS silencers in inducing TGS of the TMM endogene is not dependent on transcript 3′ end structures.

Genomic insertion positions of single-stranded silencers

For each construct, we selected for further phenotypic investigations two independent lines with a nearly 3 : 1 segregation ratio in the T2 generation. There was a remote possibility that two copies of the transgene expressing the single-stranded silencer might have been inserted into the genome in an inverted orientation. Leaky transcriptional read-through could generate RNA complementary to the pTMM sequences, thus forming double-stranded IR RNAs. Becuse it is difficult to use Southern blot to analyze complex T-DNA insertions or closely juxtapositioned insertions, we mapped the insertion positions of several transgenic lines by thermal asymmetric interlaced PCR (TAIL-PCR) (Liu et al. 1995). We found that all the tested lines carried single T-DNA insertions (Fig. 2). The T-DNA of S21 (sense series, line 21) was located near the end of At2g46940, which encodes a protein with unknown function. The T-DNA of S26 was located at the 5′-untranslated region (UTR) of At1g11860, which encodes a glycine cleavage T-protein family protein. The T-DNA insertion of AS4 was found in the intergenic region between At5g09720, which encodes a magnesium transporter CorA-like family protein, and At5g09730, which encodes a protein similar to a β-xylosidase. The T-DNA of AS10 was inserted at the 5′ UTR of At3g44190 encoding an FAD/NAD(P)-binding oxidoreductase family protein. So far, we have found no evidence that mutation of these genes had any effect on either TMM expression or the tmm mutant phenotype. We further sequenced approximately 13 kb genomic regions surrounding the insertion sites of AS4 and AS10. Only one intact T-DNA was recovered in each transgenic line (Supplementary Fig. S3).

Fig. 2

Genomic insertion locations of T-DNA in Arabidopsis transgenic lines carrying single-stranded silencers. The schematic diagram shows disrupted Arabidopsis genes. In each line, the T-DNA insertion position is indicated by an arrow. White box, UTR; black box, exon coding region; black line, intron region; dotted line, intergenic region. Note that the intergenic and genic regions are not on the same scale in (c). S21 and S26 are sense silencer lines and AS4 and AS10 are antisense silencer lines.

Fig. 2

Genomic insertion locations of T-DNA in Arabidopsis transgenic lines carrying single-stranded silencers. The schematic diagram shows disrupted Arabidopsis genes. In each line, the T-DNA insertion position is indicated by an arrow. White box, UTR; black box, exon coding region; black line, intron region; dotted line, intergenic region. Note that the intergenic and genic regions are not on the same scale in (c). S21 and S26 are sense silencer lines and AS4 and AS10 are antisense silencer lines.

Single-stranded silencers can generate mostly 24 nt smRNAs different from IR-related smRNAs

There is good evidence that TGS can be induced by RNA-dependent DNA methylation (RdDM) with smRNAs, mostly 24–26 nt, derived from exo/endogeneous double-stranded IR RNAs (Wassenegger et al. 1994, Mette et al. 2000, Hamilton et al. 2002). On the other hand, shorter smRNA species (21–22 nt) mainly mediate degradation of target RNAs with sequence homology, resulting in post-transcriptional gene silencing (PTGS) (Hamilton et al. 2002, Vaucheret 2006).

To investigate possible smRNAs linked to TMM silencing by single-stranded and double-stranded silencers, we determined the sequences of smRNAs by the Illumina high-throughput sequencing platform. Except for sample S21, for which we obtained 868,571 reads, we generated >1 million reads for each of the other samples. The read numbers of AS10 and IR8, which we started with purified smRNAs (see the Materials and Methods), were one order of magnitude more than those of unfractionated samples (Supplementary Table S2). After removing irrelevant sequences, approximately 80% of the reads could be mapped to the Arabidopsis genome (TAIR 9) (Supplementary Table S2). For all samples, 24 nt smRNAs represented the dominant species amongst the mapped smRNA population (Supplementary Fig. S4). None of the samples produced any smRNAs corresponding to the TMM coding region. In the untransformed WT sample, we did not recover any smRNAs corresponding to the TMM promoter region. In contrast, we recovered an abundance of smRNAs with perfect sequence homology to the TMM promoter sequence in silenced transgenic plants (Table 2; Fig. 3a). The silencer was required for smRNA production; when the silencer transgene was segregated by genetic crosses, the smRNAs disappeared (Table 2, AS10NI and IR8NI). The majority of smRNAs were derived from the minus strand in all three categories of silencers (Table 2). Among the smRNAs that can be mapped to the TMM promoter region, almost all were 21–24 nt in length. For transgenic plants harboring single-strand silencers, >70% of smRNAs were 24 nt in length. This was different from smRNAs derived from the double-stranded IR silencer in which a significant proportion of 21–23 nt smRNAs can be detected, accounting for approximately 45% of all smRNAs (Fig. 3b). The results suggested a different smRNAs biogenesis mechanism generated by single-stranded silencers and by the double-stranded IR silencer.

Fig. 3

Clone number and size distribution of smRNAs mapped to the TMM promoter region. (a) Normalized clone numbers of smRNAs mapped to the TMM promoter. The TMM promoter-related smRNAs were normalized to the total number of 21–24 nt smRNAs that can be mapped to the Arabidopsis genome. (+) and (−) refer to the sense strand and antisense strand, respectively; rpm denotes read per million reads. (b) Size distribution of smRNAs mapped to the TMM promoter. S21 and S26, sense silencer lines; AS4 and AS10, antisense silencer lines; IR8, IR silencer line; AS10NI and IR8NI were progeny plant lines which did not carrying any transgene from a T2 segregating population of AS10 and IR8, respectively.

Fig. 3

Clone number and size distribution of smRNAs mapped to the TMM promoter region. (a) Normalized clone numbers of smRNAs mapped to the TMM promoter. The TMM promoter-related smRNAs were normalized to the total number of 21–24 nt smRNAs that can be mapped to the Arabidopsis genome. (+) and (−) refer to the sense strand and antisense strand, respectively; rpm denotes read per million reads. (b) Size distribution of smRNAs mapped to the TMM promoter. S21 and S26, sense silencer lines; AS4 and AS10, antisense silencer lines; IR8, IR silencer line; AS10NI and IR8NI were progeny plant lines which did not carrying any transgene from a T2 segregating population of AS10 and IR8, respectively.

Table 2

Summary of smRNA analysis

Sample Total mapped 21–24 nt mapped Mapped to TMM promoter
 
Plus Minus 
WT 4,381,086 924,984 
S21 606,301 190,799 12 86 
S26 818,493 150,757 
AS4 2,073,267 468,597 37 153 
AS10 19,637,148 14,646,135 1,951 3,473 
AS10NI 2,749,728 591,549 
IR8 19,740,551 15,397,037 2,260 4,357 
IR8NI 1,615,622 325,687 
Sample Total mapped 21–24 nt mapped Mapped to TMM promoter
 
Plus Minus 
WT 4,381,086 924,984 
S21 606,301 190,799 12 86 
S26 818,493 150,757 
AS4 2,073,267 468,597 37 153 
AS10 19,637,148 14,646,135 1,951 3,473 
AS10NI 2,749,728 591,549 
IR8 19,740,551 15,397,037 2,260 4,357 
IR8NI 1,615,622 325,687 

smRNA sequences were determined in eight samples.

Total mapped, the total number of clones (19–28 nt) that can be mapped (with perfect sequence identity) to the Arabidopsis genome; 21–24 nt mapped, the total number of 21–24 nt smRNAs that can be mapped to the Arabidopsis genome; mapped to TMM promoter, 21–24 nt smRNAs that can be mapped to the TMM promoter; Plus, plus strand; Minus, minus strand.

Silencer-derived smRNAs induce de novo DNA methylation

Increased DNA methylation at the small interfering RNA-targeted promoter region is generally considered as the hallmark for TGS (Aufsatz et al. 2002a) although it is not always linked to successful gene silencing (Okano et al. 2008). Mette et al. (2000) reported that a single-stranded antisense trigger/silencer can induce low levels of DNA methylation on the silencer transgene but not on the targeted promoter (Mette et al. 2000). To determine whether the observed smRNAs related to the TMM promoter can actually induce DNA methylation at the target promoter, we performed sequencing of bisulfite-treated DNAs to examine the methylation status in detail. Significant increases in methylation of all three kinds of cytosine, CG, CHG (where H denotes A, C or T) and CHH, were observed at the endogenous TMM promoter in S-silencer lines (S21 and S26), AS-silencer lines (AS4 and AS10) and a double-stranded IR RNA silencer line (IR8) compared with WT (Col-0) control (Fig. 4; Supplementary Fig. S5).

Fig. 4

DNA methylation profile at the endogenous TMM promoter region. (a) The region of endogenous TMM promoter analyzed by sequencing of bisulfite-treated DNA. (b–d) DNA methylation levels at CG, CHG (H = A, C or T) and CHH types of cytosine, respectively. S21 and S26, sense silencer lines; AS4 and AS10, antisense silencer lines; IR8, IR silencer line.

Fig. 4

DNA methylation profile at the endogenous TMM promoter region. (a) The region of endogenous TMM promoter analyzed by sequencing of bisulfite-treated DNA. (b–d) DNA methylation levels at CG, CHG (H = A, C or T) and CHH types of cytosine, respectively. S21 and S26, sense silencer lines; AS4 and AS10, antisense silencer lines; IR8, IR silencer line.

To determine whether the DNA methylation at the TMM promoter region was dependent on silencer-generated smRNAs, we used McrBC digestion analysis as a rapid assay. There was no significant DNA methylation in the AS10NI line, which was segregated from the heterozygous AS10 parent and carried no T-DNA insert (Fig. 5). The TMM promoter of transgenic line AS27, which did not produce any smRNAs and did not show any tmm mutant phenotype (data not shown), was also unmethylated (Fig. 5). Therefore, our data indicate that DNA methylation at the TMM promoter region requires the presence of promoter-related smRNAs arising from the silencer RNAs.

Fig. 5

DNA methylation at the TMM promoter region is silencer dependent. Endonuclease McrBC cleaves DNA containing (G/A)mC(N40–3,000) (G/A)mC. (a) 5′ and 3′ regions of the TMM promoter interrogated for DNA methylation. (b) qPCR results with or without digestion with McrBC. WT, wild type Col-0. AS4, AS10 and AS27, antisense silencer lines. AS10NI plants were derived from selfing of heterozygous AS10 and contained no transgenic silencer. U, without McrBC treatment; D, with McrBC treatment.

Fig. 5

DNA methylation at the TMM promoter region is silencer dependent. Endonuclease McrBC cleaves DNA containing (G/A)mC(N40–3,000) (G/A)mC. (a) 5′ and 3′ regions of the TMM promoter interrogated for DNA methylation. (b) qPCR results with or without digestion with McrBC. WT, wild type Col-0. AS4, AS10 and AS27, antisense silencer lines. AS10NI plants were derived from selfing of heterozygous AS10 and contained no transgenic silencer. U, without McrBC treatment; D, with McrBC treatment.

Histone modifications of the TMM promoter region

DNA methylation of the targeted endogenous rice promoter was for the most part not linked to TGS or formation of repressive chromatin (Okano et al. 2008). To determine whether the changes in DNA methylation observed at the Arabidopsis TMM promoter region were accompanied by changes in histone modifications, we used a chromatin immunoprecipitation (ChIP) assay to interrogate H3 acetylation and H3K4 trimethylation for active markers, and H3K9me3 and H3K27me3 for repressive markers. Three primer pairs were designed to examine both promoter (5′ and 3′) and coding regions of the endogenous TMM locus (Fig. 6a). Significant reduction of active markers, H3K9/14Ac and H3K4me3, was observed in AS-silencer and IR-silencer transgenic plants at the TMM promoter regions, compared with WT plants (Fig. 6b, c). Higher levels of H3K9me3 and H3K27me3 were seen at the 3′ region of the TMM promoter, whereas enrichment of these repressive markers was less significant at the 5′ promoter and coding regions (Fig. 6b–d). Overall, both AS-silencer and IR-silencer plants possessed similar patterns of histone modifications which were correlated with DNA methylation at the TMM promoter region mediated by smRNAs.

Fig. 6

Histone H3 modification patterns on the TMM promoter and coding region in plants silenced by the AS or IR silencer. (a) qPCR primers designed to amplify DNA fragments (around 100 bp) corresponding to three different regions of the TMM promoter and coding sequences as indicated. (b–d) Histone H3 modification patterns at the TMM promoter 5′ region, 3′ region and the TMM coding region. Ace, H3K9/14 acetylation; K4me3, H3K4 trimethylated form; K9me3, H3K9 trimethylated form; K27me3, histone H3K27 trimethylated form. Histograms show average values ± SD (n = 3 biological replicates). Significant differences in histone modifications between transgenic lines and Col-0 WT are denoted with asterisks: *P < 0.05; **P < 0.01.

Fig. 6

Histone H3 modification patterns on the TMM promoter and coding region in plants silenced by the AS or IR silencer. (a) qPCR primers designed to amplify DNA fragments (around 100 bp) corresponding to three different regions of the TMM promoter and coding sequences as indicated. (b–d) Histone H3 modification patterns at the TMM promoter 5′ region, 3′ region and the TMM coding region. Ace, H3K9/14 acetylation; K4me3, H3K4 trimethylated form; K9me3, H3K9 trimethylated form; K27me3, histone H3K27 trimethylated form. Histograms show average values ± SD (n = 3 biological replicates). Significant differences in histone modifications between transgenic lines and Col-0 WT are denoted with asterisks: *P < 0.05; **P < 0.01.

Single-stranded silencers can induce TGS of FHY1, HFR1 and PHYB promoters

FHY1 (far-red elongated hypocotyl 1) and HFR1 (long hypocotyl in far red 1) encode proteins involved in positive regulation of the phytochrome A signaling pathway. Under far-red light conditions, elongated hypocotyls were observed in seedlings of fhy1 and hfr1 mutants (Whitelam et al. 1993, Fairchild et al. 2000, Soh et al. 2000). PHYB is one of the five phytochromes in Arabidopsis, and phyB null mutants display longer hypocotyls compared with the WT under red light (Reed et al. 1993). To determine the general applicability of TGS using single-stranded silencers, we constructed S- and AS-silencer vectors targeting promoters of FHY1, HFR1 and PHYB, and generated transgenic plants for phenotypic observations. For HFR1, an IR silencer was also tested for its efficiency to induce TGS.

Fig. 7a and b and Supplementary Fig. S6a show hypocotyl lengths of transgenic lines carrying either an S silencer or an AS silencer targeting the FHY1 or HFR1 promoter. The increase in hypocotyl length was significant although the phenotype was moderate compared with null mutants of fhy1-3 and hfr1-201. qRT–PCR results confirmed the reduction of FHY1 and HFR1 transcript levels in the transgenic lines (Fig. 7c; Supplementary Fig. S6b). Sequencing of bisulfite-treated DNA revealed a significant increase of DNA methylation at the endogenous FHY1 promoter in both the S-silencer (S3 and S4) and the AS-silencer (AS8 and AS11) transgenic plants (Fig. 7d; Supplementary Fig. S7). Elongated hypocotyls of transgenic plants carrying an S silencer and AS silencer targeting the PHYB promoter under red light confirmed that this strategy for TGS also worked for the PHYB locus (Supplementary Fig. S8).

Fig. 7

The single-stranded FHY1 silencer targeting the FHY1 promoter can induce the fhy1 mutant phenotype. (a) Hypocotyl lengths of S silencer lines (S3 and S4) and antisense silencer lines (AS8 and AS11) were greater than those of the wild type (WT, Col-0) under 5 µmol m−2 s−1 far red light (FR) treatment for 4 d. fhy1-3, an FHY1 loss-of-function mutant. Scale bar = 5 mm. (b) Hypocotyl lengths ± SD, n = 20. (c) Relative FHY1 transcript levels detected by qRT–PCR using gene-specific primers. Histograms show average values ± SD (n = 3 technical replicates).

Fig. 7

The single-stranded FHY1 silencer targeting the FHY1 promoter can induce the fhy1 mutant phenotype. (a) Hypocotyl lengths of S silencer lines (S3 and S4) and antisense silencer lines (AS8 and AS11) were greater than those of the wild type (WT, Col-0) under 5 µmol m−2 s−1 far red light (FR) treatment for 4 d. fhy1-3, an FHY1 loss-of-function mutant. Scale bar = 5 mm. (b) Hypocotyl lengths ± SD, n = 20. (c) Relative FHY1 transcript levels detected by qRT–PCR using gene-specific primers. Histograms show average values ± SD (n = 3 technical replicates).

Discussion

Using a conventional double-stranded IR silencer targeting promoter sequences, two Arabidopsis endogenes, TMM and HFR1, can be efficiently silenced (Fig. 1; Supplementary Fig. S6). In the case of rice, only one out of seven endogenes could be moderately silenced by an IR silencer (Okano et al. 2008). Furthermore, using single-stranded silencers targeting promoter sequences, we successfully silenced four Arabidopsis endogenes: TMM, FHY1, HFR1 and PHYB. Our results indicate that the single-stranded AS silencer has comparable efficiency with the double-stranded IR silencer, whereas the single-stranded S silencer has a much weaker effect (Fig. 1; Table 1). These differences in efficiency between AS and S silencer lines were not due to multiple T-DNA insertions or a rearrangement of T-DNA resulting in a complex structure of transgene silencers. Mapping of T-DNA in AS4, AS10, S21 and S26 showed that each of these lines carried a single T-DNA copy (Fig. 2). We also sequenced an approximately 13 kb genomic region surrounding the T-DNA insertion site to confirm that these single-stranded silencers retain their original gene structure in the transgenic lines (Supplementary Fig. S3). Therefore, in the case of AS4 and AS10, our results ruled out any complex T-DNA rearrangement that might generate two copies of the single-stranded silencer in an inverted configuration to produce double-stranded RNAs.

smRNAs is correlated with TGS of the TMM promoter

Mette et al. (2000) previously noted a good correlation between smRNAs and DNA methylation in TGS of transgenes. The double-stranded IR silencer can efficiently produce approximately 24 nt smRNAs and trigger DNA methylation and TGS of homologous sequence in trans. However, neither DNA methylation nor TGS was observed in transgenic plants carrying a single-stranded S or AS silencer alone or even when S and AS silencers were transcribed together (Mette et al. 2000). Other than tobacco and Arabidopsis, TGS initiated by smRNAs targeting promoter sequences was also observed in petunia (Sijen et al. 2001) and maize (Cigan et al. 2005). However, promoter-related smRNAs did not trigger TGS of endogenes in the monocot rice even though promoter DNA methylation was observed (Okano et al. 2008). This result indicates that smRNA-mediated DNA methylation alone is not sufficient to induce TGS, at least in some cases.

Here, we showed that smRNAs targeted to the TMM promoter is necessary for DNA methylation and gene silencing. Several lines of evidence support this claim: (i) in the WT, no smRNAs associated with the TMM locus were recovered by smRNA sequencing (Fig. 3a; Table 2). Consistent with this observation, no promoter DNA methylation was detected (Fig. 4) and the promoter remains active. (ii) All tested silenced transgenic lines produced promoter-related smRNAs and displayed promoter DNA methylation (Figs. 3a, 4). We observed that the majority of smRNAs were derived from the minus strand in all three categories of silencers. The exact reason for this observation is unknown. One possibility is that the smRNAs from the minus strand are more stable, which may be caused by its higher efficiency in cooperating with AGO proteins. (iii) Transgenic lines such as AS27 without promoter-related smRNAs also did not show any DNA methylation of the TMM promoter (Fig. 5). (iv) In silenced lines, the tmm mutant phenotype reverted to WT when the transgenic silencer was segregated by genetic crosses, e.g. AS10NI and IR8NI (Fig. 1), and no TMM promoter-related smRNAs were detected in progeny plants not carrying the silencer transgene (Fig. 3a; Table 2). These results are consistent with a previous study with a double transgene system, in which the silenced promoter was demethylated and restored when the other homologous sequence was removed by self-fertilization or backcrossing (Matzke et al. 1989). On the other hand, there appears to be no strict correlation between smRNA levels, DNA methylation and TGS efficiencies. It is quite possible that there is a threshold smRNA level that is needed to initiate DNA methylation and trigger TGS.

Formation of repressive chromatin is associated with TGS of endogenes

Modifications at specific amino acid positions of histones on nucleosomes play an important role in shaping the chromatin structure, thereby regulating transcription activity in plants (Loidl 2004). Two groups of histone modifications have been identified: (i) active or euchromatin markers associated with regions of actively expressed genes; and (ii) repressive or heterochromatin markers associated with inactive regions in euchromotin or heterochromotin (Loidl 2004, Bernatavichute et al. 2008, Liu et al. 2010). There is increasing evidence for cross-talk between DNA methylation and histone modifications. A reinforcement loop between DNA methylation and H3K9 methylation has been demonstrated in Arabidopsis (Bernatavichute et al. 2008, Law and Jacobsen 2010). However, DNA methylation is not always linked to repressive chromatin formation, which is probably necessary for TGS of endogenes. This is clearly the case in the model monocot rice. In rice, IR silencers targeting seven endogenes produced abundant smRNAs and associated strong DNA methylation. Nevertheless, only one endogene, Se5, was partially silenced, and only in this case was the locus formatted by repressive chromatin states (reduced H3/H4 acetylation and H3K4me2) (Okano et al. 2008). In the case of Arabidopsis, histone deacetylase 6 (HDA6) was shown to be necessary for TGS of the transgene (Aufsatz et al. 2002b). These results with Arabidopsis and rice indicate that gene silencing is correlated with chromatin repression.

In our system, the TGS of TMM by RdDM also led to altered histone modifications on the TMM locus. Active histone marks H3K4me3 and H3K9/14ace decreased while the repressive marks H3K9me3 and H3K27me3 increased in silenced plants compared with the WT. All three genomic regions covering the TMM locus showed a similar pattern, although the region corresponding to the TMM transcription start site displayed the most significant changes (Fig. 5). Comparing histone marks associated with the silenced TMM locus in AS10 and IR8, the single-stranded silencer line AS10 deposited heavier repressive marks H3K9me3 and H3K27me3, and the double-stranded silencer line IR8 abolished more positive marks H3K9/14ace and H3K4me3 (Fig. 5). We do not know whether this difference represents a difference in chromatin silencing induced by a single-stranded silencer and double-stranded silencer, in which the relative proportions of smRNA classes were different (Fig. 3b), or if it was just a variation between different samples. This should be clarified by analysis of more independent transgenic lines in future studies.

Here, we showed that single-stranded silencers could produce smRNAs that enter into the endogenous RdDM pathway to mediate TGS of endogenes. The efficiency of S and AS silencers to recruit RNA-dependent RNA polymerase(s) may differ, leading to the different TGS efficiency between S and AS silencers. For most active promoters, there is increasing evidence supporting the existence of promoter-associated transcripts (PROMPTs, both sense and antisense) (Chekanova et al. 2007, Seila et al. 2008, Neil et al. 2009, Preker et al. 2011). One possible scenario is that such PROMPTs are also transcribed from the upstream region of the TMM promoter. The single-stranded silencer transcripts and these putative PROMPTs may form double-stranded RNAs (dsRNAs) which are then directed to the RdDM pathway. The non-uniform distribution and relative levels between sense and antisense PROMPTs (Chekanova et al. 2007, Neil et al. 2009) may determine the level of dsRNAs leading to different TGS efficiencies between S and AS silencers. The detection of PROMPTs is technically challenging as these transcripts are largely nuclear with low expression levels; moreover, they rapidly turn over owing to the action of RNA exosomes. These low level transcripts can be detected only in exosome mutants and/or by special methods (Seila et al. 2008, Neil et al. 2009, Preker et al. 2011). The possibility of TMM-associated PROMPTs remains to be addressed in future studies.

Materials and Methods

Construct

Using standard PCR and cloning techniques, a 475 bp genomic DNA fragment from the promoter region (−483 to −9, in reference to the transcription start site as +1) of AT1G80080 (TMM) was cloned into a binary plasmid pBAV3. The latter plasmid was derived from pBA002 (Kost et al. 1998) by replacing the NOS poly(A) addition sequence near the right border of T-DNA with a fragment from TRV2 self-processing sequences and replacing the 35S promoter with a double 35S promoter from pBCO-DC-CFP (Wu et al. 2010). The TMM promoter fragment was inserted downstream of the double 35S promoter either in the sense (S silencer) or the antisense (AS silencer) orientation with respect to the TMM coding sequence. To obtain an IR structure, the TMM promoter DNA fragment was cloned first into the vector pSRS, which is derived from pSK-int (Guo et al. 2003) by replacing the intron with a soybean promoter sequence (Mette et al. 2000). The IR structure (IR silencer) was then cloned into pBAV3 under the control of a double 35S promoter. The S, AS and IR silencers targeting the TMM promoter were also cloned into the conventional pBA002 vector (Kost et al. 1998) to generate another series of silencers with a NOS poly(A) addition sequence. By a similar strategy, the silencer constructs for FHY1 (−449 to −1), HFR1 (−1,025 to −1) and PhyB (−945 to +112) were generated.

Plant materials and transformation

We used Arabidopsis thaliana Columbia (Col-0) ecotype as the WT control in this study. All A. thaliana mutants have been described: phyB-9 (Reed et al. 1993), hfr1-201 (Soh et al. 2000) and fhy1-3 (Zeidler et al. 2004). All mutants are in the Columbia (Col-0) ecotype background.

Plants were grown in a growth chamber under long-day (16 h of light/8 h of dark) conditions illuminated by cool-white fluorescent light (100 µmol m−2 s−1) until flowering. Standard floral dip transformation was performed (Zhang et al. 2006).

tmm phenotype observation

After harvest, T1 seedlings were screened on BASTA agar plates. Morphological phenotype was scored under a microscope. For each independent transgenic line, the penetration rate was calculated from 40–80 randomly picked 3-week-old T1 seedlings. We classified the tmm mutant phenotype roughly into three categories: weak, intermediate and strong according to the frequency and number of clustered stomata (Supplementary Fig. S1). When plants from a transformation event showed predominantly two clustered stomata and occasionally three clustered stomata, we scored this as a weak phenotype. Transgenic plants displayed an intermediate tmm mutant phenotype if three clustered stomata were frequently observed. Transgenic plants with a strong tmm mutant phenotype displayed >4 clustered stomata and usually single stomata could not be observed (Supplementary Fig. S1).

DNA methylation analysis

Arabidopsis genomic DNA was extracted from 2-week-old seedlings using a DNeasy Plant Mini Kit (Qiagen). Bisulfite DNA conversion was performed by using 1 µg of genomic DNA and an EpiTech Bisulfite Kit (Qiagen) following the manufacturer’s protocol. PCR was performed using primers located outside of the targeted region and designed for single strand methylation detection. PCR products were then cloned into pCR2.1 using a TA cloning kit (Invitrogen). For each genotype, at least 20 independent clones were sequenced using the M13R primer, and the data were analyzed by Cymate (Hetzl et al. 2007). For McrBC digestion, 250 ng of genomic DNA was digested in a total volume of 20 µl by McrBC (New England Biolabs) following the manufacturer’s instruction. Treated and untreated control samples were then used as templates in real-time PCRs.

RNA isolation and RT–PCR

RNA was extracted from 2-week-old seedlings using an RNeasy Plant Mini Kit (Qiagen) following the manufacturer’s instructions. cDNA synthesis was performed by using the Superscript III First strand synthesis system (Invitrogen) following the manufacturer’s instructions. Real-time PCR was performed using SYBR Premix Ex Taq (TAKARA) in a Biorad CFX96 real-time PCR system. ACTIN2 was used as an internal control. The primers used are listed in Supplementary Table S3. Real-time quantitative PCR was repeated with 2–4 biological replicates, and each sample was assayed in triplicate by PCR.

TAIL-PCR

TAIL-PCR was conducted as described by Liu et al. (1995) with slight modification. In brief, 50 ng of total DNA from each genotype was used as template. After three rounds of PCR amplification with three nested primers located in the pBAV3 vector combined with additional degenerate (AD) primers, the specific bands were purified and sequenced.

Chromatin immunoprecipitation (ChIP) and ChIP-qPCR

A 3 g aliquot of 2-week-old seedlings was used in immunoprecipitation experiments as described by Gendrel et al. (2005) with minor modifications. Cross-linked chromatin pellets resuspended in nuclei lysis buffer were sonicated in a Bioruptor (Bioruptor UCD 200, Diagenode) for 10 min at the maximum level. Samples were sonicated for periods of 30 s. with a 30 s interval in between treatments. Histone H3 trimethyl Lys4 (K4me3) antibody was from Active Motif (Cat. No. 39159) and histone H3 Acetylation, histone H3 trimethyl Lys9 (K9me3) and histone H3 trimethyl Lys27 (K27me3) antibodies were from Milipore (Cat. No. 06-599, 07-442 and 07-449). The purified DNA fragments were analyzed by quantitative PCR as before. For K4me3 and H3Ace, ACTIN2 was used as an internal control. Ta3 and AGAMOUS, which were highly associated with H3K9 methylation and H3K27me3, were selected as a control for K9me3 and K27me3, respectively. Student’s t-test was used to determine statistical significance. The primers used are listed in Supplementary Table S3.

Small RNA analysis

Total RNA was extracted from 2-week-old seedlings by Trizol reagent (Invitrogen) following the manufacturer’s instructions. SmRNA libraries were constructed using a TruSeq Small RNA Sample Prep Kit (Illumina) following the manufacturer’s instructions. Briefly, 1 µg of total RNA or purified smRNA was ligated with 3′ and 5′ adaptors and used as a template for RT–PCR. After PCR amplification, 6 µl of each sample were pooled and separated on a 6% polyacrylamide gel. Gel slices corresponding to approximately 20–30 nt smRNAs were recovered and sequences were determined by Illunina HiSeq in the Genomic Center of the Rockefeller University. The smRNAs sequence data sets are available at the Gene Expression Omnibus (GSE39398). Adaptor sequences were trimmed by local Perl script and only sequences >15 nt were used in further analysis. All retained sequences were mapped to the Arabidopsis genome (TAIR 9 version) by the C program allowing no mismatch.

Hypocotyl phenotype observation

Hypocotyl phenotypes were assayed as described before (Jang et al. 2007). Briefly, sterilized seeds were sown on Murashige and Skoog (MS) plates and stratified at 4°C in the dark for 4 d then exposed to white light for 1 h before incubating under far-red or red light for 4 d at 22°C. Lengths of hypocotyls were recorded.

Supplementary data

Supplementary data are available at PCP online.

Funding

This work was supported by E.I. du Pont de Nemours and Company.

Acknowledgments

We thank members of the Chua lab and Drs, Enno Krebbers, Barbara Mazur, Bobby Williams, Carl Falco and Brian McGonigle for useful discussions.

Disclosures

The authors have no conflicts of interest to declare.

Abbreviations

    Abbreviations
  • AS

    antisense

  • ChIP

    chromatin immunoprecipitation

  • FHY1

    far-red elongated hypocotyl 1

  • HFR1

    long hypocotyl in far red 1

  • IR

    inverted repeat

  • PROMPT

    promoter-associated transcript

  • qRT–PCR

    quantitative reverse transcription–PCR

  • RdDM

    RNA-directed DNA methylation

  • phyB

    phytochrome B

  • S

    sense

  • smRNA

    small RNA

  • TAIL-PCR

    thermal asymmetric interlaced PCR

  • TGS

    transcriptional gene silencing

  • TMM

    Too Many Mouths

  • TRV2

    Tobacco rattle virus RNA 2

  • UTR

    untranslated region

  • WT

    wild type

References

Aufsatz
W
Mette
MF
Van Der Winden
J
Matzke
AJM
Matzke
M
RNA-directed DNA methylation in Arabidopsis
Proc. Natl Acad. Sci. USA
 , 
2002a
, vol. 
99
 (pg. 
16499
-
16506
)
Aufsatz
W
Mette
MF
Van Der Winden
J
Matzke
M
Matzke
AJM
HDA6, a putative histone deacetylase needed to enhance DNA methylation induced by double-stranded RNA
EMBO J.
 , 
2002b
, vol. 
21
 (pg. 
6832
-
6841
)
Bernatavichute
YV
Zhang
X
Cokus
S
Pellegrini
M
Jacobsen
SE
Genome-wide association of histone H3 lysine nine methylation with CHG DNA methylation in Arabidopsis thaliana
PLoS One
 , 
2008
, vol. 
3
 pg. 
e3156
 
Chekanova
JA
Gregory
BD
Reverdatto
SV
Chen
H
Kumar
R
Hooker
T
, et al.  . 
Genome-wide high-resolution mapping of exosome substrates reveals hidden features in the Arabidopsis transcriptome
Cell
 , 
2007
, vol. 
131
 (pg. 
1340
-
1353
)
Cigan
AM
Unger-Wallace
E
Haug-Collet
K
Transcriptional gene silencing as a tool for uncovering gene function in maize
Plant J.
 , 
2005
, vol. 
43
 (pg. 
929
-
940
)
Eamens
A
Wang
MB
Smith
NA
Waterhouse
PM
RNA silencing in plants: yesterday, today, and tomorrow
Plant Physiol.
 , 
2008
, vol. 
147
 (pg. 
456
-
468
)
Fairchild
CD
Schumaker
MA
Quail
PH
HFR1 encodes an atypical bHLH protein that acts in phytochrome A signal transduction
Genes Dev.
 , 
2000
, vol. 
14
 (pg. 
2377
-
2391
)
Gendrel
AV
Lippman
Z
Martienssen
R
Colot
V
Profiling histone modification patterns in plants using genomic tiling microarrays
Nat. Methods
 , 
2005
, vol. 
2
 (pg. 
213
-
218
)
Guo
HS
Fei
JF
Xie
Q
Chua
NH
A chemical-regulated inducible RNAi system in plants
Plant J.
 , 
2003
, vol. 
34
 (pg. 
383
-
392
)
Hamilton
A
Voinnet
O
Chappell
L
Baulcombe
D
Two classes of short interfering RNA in RNA silencing
EMBO J.
 , 
2002
, vol. 
21
 (pg. 
4671
-
4679
)
Heilersig
BHJB
Loonen
AEHM
Janssen
EM
Wolters
AMA
Visser
RGF
Efficiency of transcriptional gene silencing of GBSSI in potato depends on the promoter region that is used in an inverted repeat
Mol. Genet. Genomics
 , 
2006
, vol. 
275
 (pg. 
437
-
449
)
Hetzl
J
Foerster
AM
Raidl
G
Scheid
OM
CyMATE: a new tool for methylation analysis of plant genomic DNA after bisulphite sequencing
Plant J.
 , 
2007
, vol. 
51
 (pg. 
526
-
536
)
Jones
L
Ratcliff
F
Baulcombe
DC
RNA-directed transcriptional gene silencing in plants can be inherited independently of the RNA trigger and requires Met1 for maintenance
Curr. Biol.
 , 
2001
, vol. 
11
 (pg. 
747
-
757
)
Kost
B
Spielhofer
P
Chua
NH
A GFP–mouse talin fusion protein labels plant actin filaments in vivo and visualizes the actin cytoskeleton in growing pollen tubes
Plant J.
 , 
1998
, vol. 
16
 (pg. 
393
-
401
)
Law
JA
Jacobsen
SE
Establishing, maintaining and modifying DNA methylation patterns in plants and animals
Nat. Rev. Genet.
 , 
2010
, vol. 
11
 (pg. 
204
-
220
)
Liu
C
Lu
F
Cui
X
Cao
X
Histone methylation in higher plants
Annu. Rev. Plant Biol.
 , 
2010
, vol. 
61
 (pg. 
395
-
420
)
Liu
YG
Mitsukawa
N
Oosumi
T
Whittier
RF
Efficient isolation and mapping of Arabidopsis thaliana T-DNA insert junctions by thermal asymmetric interlaced PCR
Plant J.
 , 
1995
, vol. 
8
 (pg. 
457
-
463
)
Loidl
P
A plant dialect of the histone language
Trends Plant Sci.
 , 
2004
, vol. 
9
 (pg. 
84
-
90
)
Matzke
M
Kanno
T
Daxinger
L
Huettel
B
Matzke
AJM
RNA-mediated chromatin-based silencing in plants
Curr. Opin. Cell Biol.
 , 
2009
, vol. 
21
 (pg. 
367
-
376
)
Matzke
M
Primig
M
Trnovsky
J
Matzke
A
Reversible methylation and inactivation of marker genes in sequentially transformed tobacco plants
EMBO J.
 , 
1989
, vol. 
8
 (pg. 
643
-
649
)
Matzke
MA
Matzke
AJM
Planting the seeds of a new paradigm
PLoS Biol.
 , 
2004
, vol. 
2
 pg. 
e133
 
Mette
M
Aufsatz
W
Van der Winden
J
Matzke
M
Matzke
A
Transcriptional silencing and promoter methylation triggered by double-stranded RNA
EMBO J.
 , 
2000
, vol. 
19
 (pg. 
5194
-
5201
)
Neil
H
Malabat
C
d‘Aubenton-Carafa
Y
Xu
Z
Steinmetz
LM
Jacquier
A
Widespread bidirectional promoters are the major source of cryptic transcripts in yeast
Nature
 , 
2009
, vol. 
457
 (pg. 
1038
-
1042
)
Okano
Y
Miki
D
Shimamoto
K
Small interfering RNA (siRNA) targeting of endogenous promoters induces DNA methylation, but not necessarily gene silencing, in rice
Plant J.
 , 
2008
, vol. 
53
 (pg. 
65
-
77
)
Preker
P
Almvig
K
Christensen
MS
Valen
E
Mapendano
CK
Sandelin
A
, et al.  . 
PROMoter uPstream Transcripts share characteristics with mRNAs and are produced upstream of all three major types of mammalian promoters
Nucleic Acids Res.
 , 
2011
, vol. 
39
 (pg. 
7179
-
7193
)
Reed
JW
Nagpal
P
Poole
DS
Furuya
M
Chory
J
Mutations in the gene for the red/far-red light receptor phytochrome B alter cell elongation and physiological responses throughout Arabidopsis development
Plant Cell
 , 
1993
, vol. 
5
 (pg. 
147
-
157
)
Seila
AC
Calabrese
JM
Levine
SS
Yeo
GW
Rahl
PB
Flynn
RA
, et al.  . 
Divergent transcription from active promoters
Science
 , 
2008
, vol. 
322
 (pg. 
1849
-
1851
)
Sijen
T
Vijn
I
Rebocho
A
van Blokland
R
Roelofs
D
Mol
JNM
, et al.  . 
Transcriptional and posttranscriptional gene silencing are mechanistically related
Curr. Biol.
 , 
2001
, vol. 
11
 (pg. 
436
-
440
)
Soh
MS
Kim
YM
Han
SJ
Song
PS
REP1, a basic helix–loop–helix protein, is required for a branch pathway of phytochrome A signaling in Arabidopsis
Plant Cell
 , 
2000
, vol. 
12
 (pg. 
2061
-
2074
)
Vaucheret
H
Post-transcriptional small RNA pathways in plants: mechanisms and regulations
Genes Dev.
 , 
2006
, vol. 
20
 (pg. 
759
-
771
)
Wassenegger
M
Heimes
S
Riedel
L
Sänger
HL
RNA-directed de novo methylation of genomic sequences in plants
Cell
 , 
1994
, vol. 
76
 (pg. 
567
-
576
)
Whitelam
GC
Johnson
E
Peng
J
Carol
P
Anderson
ML
Cowl
JS
, et al.  . 
Phytochrome A null mutants of Arabidopsis display a wild-type phenotype in white light
Plant Cell
 , 
1993
, vol. 
5
 (pg. 
757
-
768
)
Wu
HW
Lin
SS
Chen
KC
Yeh
SD
Chua
NH
Discriminating mutations of HC-Pro of Zucchini yellow mosaic virus with differential effects on small RNA pathways involved in viral pathogenicity and symptom development
Mol. Plant Microbe Interact.
 , 
2010
, vol. 
23
 (pg. 
17
-
28
)
Zeidler
M
Zhou
Q
Sarda
X
Yau
CP
Chua
NH
The nuclear localization signal and the C-terminal region of FHY1 are required for transmission of phytochrome A signals
Plant J.
 , 
2004
, vol. 
40
 (pg. 
355
-
365
)
Zhang
X
Henriques
R
Lin
SS
Niu
QW
Chua
NH
Agrobacterium-mediated transformation of Arabidopsis thaliana using the floral dip method
Nat. Protoc.
 , 
2006
, vol. 
1
 (pg. 
641
-
646
)