Subnuclear partitioning of rRNA genes between the nucleolus and nucleoplasm reflects alternative epiallelic states

Abstract

Eukaryotes can have thousands of 45S ribosomal RNA (rRNA) genes, many of which are silenced during development. Using fluorescence-activated sorting techniques, we show that active rRNA genes in Arabidopsis thaliana are present within sorted nucleoli, whereas silenced rRNA genes are excluded. DNA methyltransferase (met1), histone deacetylase (hda6), or chromatin assembly (caf1) mutants that disrupt silencing abrogate this nucleoplasmic-nucleolar partitioning. Bisulfite sequencing data indicate that active nucleolar rRNA genes are nearly completely demethylated at promoter CGs, whereas silenced genes are nearly fully methylated. Collectively, the data reveal that rRNA genes occupy distinct but changeable nuclear territories according to their epigenetic state.

Keywords: DNA methylation; RNA polymerase I; chromatin assembly; gene silencing; histone deacetylation; ribosomal RNA gene; transcription.

Figure 1.

Partitioning of active and silent rRNA genes between the nucleolus and nucleoplasm. (A) Relationships between the nucleolus, NORs, and 45S rRNA gene repeats. The drawing at the top depicts a metaphase chromosome with the remnants of a nucleolus associated with the secondary constriction, the location of active rRNA genes in the preceding interphase. The black oval represents a chromomere in which rRNA genes are assembled into dense heterochromatin. In A. thaliana, insertions/deletions in the 3′ external transcribed region define rRNA gene variant types. (B) Localization of rRNA genes (rDNA) and Pol I. DNA-FISH using an rRNA gene probe (red signals) and immunolocalization of Flag-tagged Pol I using an anti-Flag antibody (green signals) were performed in A. thaliana interphase nuclei. DNA was counterstained with DAPI (gray signals). (C) Subnuclear localization of rRNA genes, pre-rRNA transcripts, and Flag-tagged HDA6. rRNA gene or transcript FISH signals are shown in green, immunolocalized HDA6 is in red, and DAPI-stained DNA is in blue. Merged signals are shown in the right column. (D) DNA-FISH detection of rRNA genes in wild-type (Col-0) and hda6 nuclei. rRNA gene FISH signals are shown in red and are merged with the DAPI (blue) image in the right column. (E) Detection of rRNA gene variant types and their transcripts by PCR using genomic DNA or reverse-transcribed (RT) cDNA of wild-type (Col-0) or hda6 plants. The amplified region is shown in A. (F) Leaf cell homogenate of a FIB2:YFP plant stained with DAPI and subjected to fluorescence microscopy. Chloroplasts fluoresce red, DAPI-stained DNA is blue, and nucleolar FIB2:YFP is green. (G) Purified nuclei obtained by FANS. (H) Purified nucleoli obtained by FANoS. (I) PCR detection of rRNA gene variant types in DNA of purified nuclei (N) or nucleoli (No) of wild-type (Col-0) or hda6 plants. The PCR amplicon is shown in A.

Figure 2.

MET1-dependent CG methylation is required for variant 1-type rRNA gene silencing. (A) rRNA gene variant expression in wild type (Col-0) or drm2-2, drm1 drm2, cmt3-11, drm1 drm2 cmt3, met1-1, or met1 cmt3 mutants. RT–PCR using primers that discriminate variant 1- from variant 2- and 3-type rRNA genes was conducted (see the diagram for primer locations). RT–PCR of ACTIN2 (ACT2) mRNA serves as a positive control. Reactions lacking reverse transcriptase (−RT) serve as negative controls. (B) Frequencies at which individual cytosines are methylated between −316 and +243 relative to the transcription start site (+1), determined by bisulfite sequencing. Wild-type Col-0, drm1 drm2 cmt3 triple mutants, and met1-7 mutants are compared. Approximately 40 independent promoter clones were sequenced for each genotype. Cytosine-depleted regions are compressed on the X-axis. (C,D) Cytosine methylation in the downstream promoter region of rRNA genes in purified nuclei or nucleoli from wild-type or hda6 leaves, determined by bisulfite sequencing. Positions of methylated (filled circles) or unmethylated (open circles) cytosines in CG motifs of 43 independent promoter clone sequences are shown. Methylation haplotypes are grouped according to methylation density. Histograms show frequencies of hypomethylated haplotypes (white), haplotypes with intermediate methylation (gray), or heavily methylated haplotypes (red).

Figure 3.

Reducing rRNA gene number in fas mutants disrupts variant 1 silencing, nucleolus–nucleoplasm partitioning, and CG methylation. (A) rRNA gene localization by DNA-FISH in nuclei of wild type or fas1 or fas2 mutants at G2 and G9. Nuclei were counterstained with DAPI. (B) qPCR analysis of relative rRNA gene numbers in wild type (WT) versus fas1-4 or fas2-4 at G2, G5, G7, or G9. (C) Semiquantitative PCR detection of rRNA gene variant types in genomic DNA of fas1 or fas2 mutants or lines derived from their wild-type siblings at G2, G5, or G9. Amplification products of rRNA gene variants after 20 or 25 cycles of PCR or of ACTIN2 after 30 cycles of PCR resolved by agarose gel electrophoresis. (D) RT–PCR amplification of rRNA gene variant transcripts or an ACTIN2 mRNA control. The lane order is the same as in C. (E) PCR detection of rRNA gene variant types in sorted nuclei or nucleoli of fas2-4 FIB2:YFP plants. (F) Frequencies of haplotypes showing CG hypomethylation, intermediate methylation, or heavy methylation in wild-type or fas2-4 nuclei (N) or nucleoli (No) within the rRNA gene promoter downstream region.

Figure 4.

Model for the organization of rRNA genes in interphase nuclei. (A) The blue circle represents a nucleus visualized by DAPI staining, with the black hole representing the nucleolus. Results of FANS or FANoS experiments indicate that condensed rRNA gene DNA-FISH signals in the nucleoplasm correspond to silent rRNA genes that are heavily methylated at promoter CG motifs. In contrast, active rRNA genes are decondensed, localized within the nucleolus, and CG-demethylated. (B) A single NOR can be composed of condensed, silent rRNA genes external to the nucleolus as well as decondensed, active rRNA genes dispersed within the nucleolus. Changing the number of rRNA genes that spool out from, or are reeled into, the reservoir of rRNA genes at the external periphery of the nucleolus can account for changes in the number of active versus silenced genes during development.