Heterochromatic siRNAs and DDM1 independently silence aberrant 5S rDNA transcripts in Arabidopsis

Abstract

5S ribosomal RNA gene repeats are arranged in heterochromatic arrays (5S rDNA) situated near the centromeres of Arabidopsis chromosomes. The chromatin remodeling factor DDM1 is known to maintain 5S rDNA methylation patterns while silencing transcription through 5S rDNA intergenic spacers (IGS). We mapped small-interfering RNAs (siRNA) to a composite 5S rDNA repeat, revealing a high density of siRNAs matching silenced IGS transcripts. IGS transcript repression requires proteins of the heterochromatic siRNA pathway, including RNA polymerase IV (Pol IV), RNA-DEPENDENT RNA POLYMERASE 2 (RDR2) and DICER-LIKE 3 (DCL3). Using molecular and cytogenetic approaches, we show that the DDM1 and siRNA-dependent silencing effects are genetically independent. DDM1 suppresses production of the siRNAs, however, thereby limiting RNA-directed DNA methylation at 5S rDNA repeats. We conclude that DDM1 and siRNA-dependent silencing are overlapping processes that both repress aberrant 5S rDNA transcription and contribute to the heterochromatic state of 5S rDNA arrays.

Figure 1. Long 5S rDNA-derived transcripts are subject to RNA silencing.

A) Map of Arabidopsis small RNAs matching the 5S rDNA unit repeat, based on analysis of leaf datasets from Rajagopalan et al. (2006) and Kasschau et al. (2007). Small RNA 5′-end positions are indicated on the x-axis, with sequencing reads tallied on the y-axis. Upward bars are matches to the forward strand; downward bars represent reverse strand matches. Read tallies are stacked in 10-bp bins, with size-class indicated by color. The diagram at bottom shows a 5S rRNA gene (thick black arrow), the surrounding intergenic spacers (IGS, gray), and regions probed by RNA blot hybridization (red lines). RT-PCR products obtained with F and R primers are indicated (black lines). B) Searching small RNA datasets for the siR1003 core sequence (yellow box) yielded a family of 21 to 24-nt siRNAs, which all match 5S LT1 transcripts identified by Vaillant et al. (2006) . C) Analysis of 5S LT1 transcript accumulation: RNA samples from inflorescences of wild type (WT), ddm1 and siRNA biogenesis mutants were analyzed by one-step RT-PCR. Reverse transcription was performed using R primer, and PCR performed using F and R primers (panel A, diagram). Control reactions were performed using ACT2 primers. RT enzyme was omitted from duplicate 5S LT1 and ACT2 reactions (no RT). D) Blot analysis of small RNA isolated from material described in panel C. The membrane was sequentially hybridized with a DNA oligonucleotide (oligo) that detects siR1003, an RNA probe for the 3′-flanking IGS region (panel A, diagram), a DNA oligo that detects miR160, and DNA oligos that detect U6 snRNA. Migration of 21-nt and 24-nt RNA oligo size standards is indicated at left.

Figure 2. DDM1 limits IGS siRNA accumulation and asymmetric methylation.

A) Detection of IGS siRNAs (siR1003) in dicer-like (dcl) mutant combinations. Blot analysis of small RNA isolated from leaves of wild type (WT); single mutants dcl2, dcl3, dcl4; double mutants dcl2 dcl3 (dcl2/3), dcl2 dcl4 (dcl2/4), dcl3 dcl4 (dcl3/4); and the triple mutant dcl2 dcl3 dcl4 (dcl2/3/4). B) Localization of IGS-related RNA in interphase nuclei by fluorescent in situ hybridization with an RNA probe for siR1003 (red). DNA was stained with DAPI (white). The size bar corresponds to 5 µm. C) Overaccumulation of IGS siRNAs in the ddm1 background. Blot analysis of small RNA from dcl3 and double mutant dcl3 ddm1 using probes for siR1003 and the larger 3′flanking region (Figure 1A, diagram). WT and ddm1 signals from the same membrane are provided for comparison. D) IGS siRNA overaccumulation persists in out-crossed ddm1. Blot analysis of small RNA isolated from WT, nrpd1 (−/−), ddm1 (−/−) and F1 heterozygotes (+/−) of each mutant crossed to the WT. E) Asymmetric cytosines in the IGS are hypermethylated in met1 and ddm1. Southern blot analysis was performed on Alu I-digested genomic DNA isolated from WT, nrpd1, rdr2, two alleles of dcl3, met1 and ddm1. The probe corresponds to 5S LT1, shown aligned to a representative 5S rDNA repeat unit from Chromosome 5. Alu I and Hae III sites in the IGS region are indicated.

Figure 3. Effect of combined DDM1 and siRNA deficiency on 5S rDNA methylation and aberrant transcripts.

A) 5S rDNA hypermethylation in ddm1 is siRNA-dependent: Southern blot comparison of AluI and HaeIII-digested genomic DNA isolated from inflorescences of wild type (WT), ddm1, and double mutant lines nrpd1 ddm1, rdr2 ddm1, and dcl3 ddm1 (top panel). The probe is the same as in Figure 2E. Dilutions of the above digests were also assayed by PCR using 5S LT1 primers (bottom panel). Samples to which no restriction enzyme was added are controls (no digest). B) 5S LT1 is silenced by two overlapping processes: RNA samples from inflorescences of WT, ddm1 and the double mutant panel were analyzed by one-step RT-PCR, performed as described in Figure 1C. Control reactions were performed with ACT2 primers; reverse transcriptase was omitted from duplicate 5S LT1 and ACT2 reactions (no RT).

Figure 4. Both DDM1 and siRNAs are required for proper 5S rDNA condensation.

A) 5S rDNA localization in interphase nuclei: Fluorescent in situ hybridization (red) was performed with a probe for 5S rDNA. DNA was stained with DAPI (white). The size bar corresponds to 5 µm. The white arrow in the wild type (WT) images indicates 5S rDNA colocalized with a DAPI-stained chromocenter; this colocalization was observed in a majority of WT nuclei (see panel B). In contrast, a majority of nuclei from ddm1 and the double mutants nrpd1 ddm1, rdr2 ddm1 and dcl3 ddm1 showed 5S rDNA localization outside chromocenters (arrows in mutant panels). B) Nuclei from each genotype were scored for 5S rDNA colocalization with chromocenters (white bars), as compared to 5S rDNA not colocalized with chromocenters (black bars). Differences observed between 5S rDNA localization outside chromocenters in ddm1 nuclei, compared to in nrpd1 ddm1 or rdr2 ddm1 nuclei are statistically significant (*). Numbers of nuclei scored: WT (n = 134), ddm1 (n = 158), nrpd1 ddm1 (n = 186), rdr2 ddm1 (n = 215), dcl3 ddm1 (n = 190).

Figure 5. Simultaneous deficiency for DDM1 and siRNAs impairs plant growth.

A) Rosette stage plants (21 days-post-germination) of lines deficient for heterochromatic siRNA biogenesis (nrpd1, rdr2 and dcl3) compared to corresponding double mutant combinations with ddm1. Four representative plants in each population (n = 18) were photographed. B) Fresh weight of aerial portion of plants from each population: plants were individually weighed and mean values plotted with standard errors (bars).

Figure 6. Model for 5S rDNA methylation and aberrant transcript silencing.

A canonical RNA polymerase (e.g., Pol III) is assumed to initiate at the 5S rRNA gene promoter, transcribe through the 5S rRNA gene (120 bp, gray), and continue into the intergenic spacer (IGS, white). Concurrently, dsRNA corresponding to the IGS is generated in a Pol IV and RDR2-dependent manner. DCL3 processes dsRNA into 24-nt siRNAs, with DCL2 and DCL4 being alternate enzymes at this step. RNA-directed DNA methylation (RdDM) resulting from this process is thought to require DRM2. DDM1/MET1 maintenance of silent chromatin and CG methylation represses aberrant IGS transcription, limiting production of siRNA precursors and, by consequence, attenuating RdDM. IGS-derived siRNAs may also guide cleavage of nascent IGS transcripts.