Erasure of histone acetylation by Arabidopsis HDA6 mediates large-scale gene silencing in nucleolar dominance

Abstract

Nucleolar dominance describes the silencing of one parental set of ribosomal RNA (rRNA) genes in a genetic hybrid, an epigenetic phenomenon that occurs on a scale second only to X-chromosome inactivation in mammals. An RNA interference (RNAi) knockdown screen revealed that the predicted Arabidopsis histone deacetylase, HDA6, is required for rRNA gene silencing in nucleolar dominance. In vivo, derepression of silenced rRNA genes upon knockdown of HDA6 is accompanied by nucleolus organizer region (NOR) decondensation, loss of promoter cytosine methylation, and replacement of histone H3 Lys 9 (H3K9) dimethylation with H3K4 trimethylation, H3K9 acetylation, H3K14 acetylation, and histone H4 tetra-acetylation. Consistent with these in vivo results, purified HDA6 deacetylates lysines modified by histone acetyltransferases whose substrates include H3K14, H4K5, and H4K12. HDA6 localizes, in part, to the nucleolus, supporting a model whereby HDA6 erases histone acetylation as a key step in an epigenetic switch mechanism that silences rRNA genes through concerted histone and DNA modifications.

Figure 1.

HDA6 is required for rRNA gene silencing in nucleolar dominance. (A) Diagrammatic representations of A. thaliana, A. arenosa, and A. suecica chromosome compositions. (B) Organization of transferred DNAs (T-DNAs) containing RNAi-inducing transgenes. The T-DNA, delimited by left and right border sequences (LB and RB), contains a selectable herbicide-resistance gene and an inverted repeat of target gene cDNA fragments (500–700 bp), with a chalcone synthase (CHSA) intron spacer, expressed by the Cauliflower Mosaic Virus (CaMV) 35S promoter and terminated by octopine synthase (OCS) 3′ sequences. (C) rRNA gene organization surrounding internal transcribed spacer 1 (ITS1) with PCR primer (arrows) and HhaI sites indicated. (D) Screening A. suecica RNAi lines targeting predicted Rpd3-like and Sir2-like HDACs. (Lanes 1–5) Controls show HhaI-digested RT–PCR products of A. thaliana (At), A. arenosa (Aa), nontransformed A. suecica (As), or TSA-treated A. suecica. In the remaining panels, RNA from five independent RNAi lines targeting each HDAC was tested. (E) S1 nuclease protection analysis. Lanes 1 and 2 show A. thaliana and A. arenosa RNA controls, demonstrating probe specificity. In the remaining lanes, RNA of wild-type (lanes 3,4), TSA-treated (lane 5), or HDA6-RNAi lines of A. suecica were probed for A. thaliana-like or A. arenosa-like rRNA transcripts. (F) HDA6 mRNA levels are knocked down in RNAi lines. RNA from nontransformed or HDA6-RNAi lines was incubated ± RT. Resulting cDNA was amplified by PCR using primers for HDA6 and a PFK internal control.

Figure 2.

HDA6 localizes to the nucleolus and is required for facultative heterochromatin formation and interphase condensation at underdominant NORs. (A) Immunolocalization of HDA6-Flag using an anti-Flag monoclonal primary antibody (red signal). The tagged protein was engineered by modifying a genomic HDA6 clone under control of its own promoter region. Nuclei were counterstained with DAPI (blue). Controls demonstrate the lack of Flag signal in a wild-type (nontransgenic) nucleus or a HDA6-Flag transgenic nucleus if the primary anti-Flag antibody is omitted and only the secondary antibody is used. (B) Organization of rRNA genes at NORs. Genes encoding 18S, 5.8S, and 25S structural rRNA precursors are separated by intergenic spacers. FISH probe and gene promoter (arrow) locations are indicated. (C, panels A–C) Immunolocalization of H3K9me2 (red signals), FISH localization of A. thaliana rRNA genes (AtNORs; green signals), and the merged image of panels A and B plus a DAPI-stained image (panel C) in meristematic root-tip cell interphase nuclei. (Panels D–F) H3K4me3 (red) and silent AtNORs. (Panels G–I) H3K9me2 (red) and dominant A. arenosa NOR (AaNOR) localization. (Panels J–L) H3K4me3 (red) and AaNORs. (D, panels A–C) H3K9me2 (red) and AtNORs (green) in interphase nuclei of meristematic root tip cells of TSA-treated plants. (Panels D–F) H3K4me3 (red) and AtNORs (green) in TSA-treated plants. (Panels G–I) H3K9me2 (red) and AtNORs (green) in an HDA6-RNAi plant. (Panels J–L) H3K4me3 (red) and AtNORs (green) in an HDA6-RNAi plant.

Figure 3.

Purified HDA6 has HDAC activity. (A) Affinity purification of Flag-HDA6 expressed in transgenic plants. Extracts of wild-type or Flag-HDA6-overexpressing A. thaliana was incubated with anti-Flag resin. A Coomassie blue-stained SDS-PAGE gel of proteins eluted using excess Flag peptide is shown. (B) Coomassie-stained SDS-PAGE gel of His-tagged recombinant Arabidopsis HATs HAG1, HAG2, and HAG5 after purification on nickel-agarose. (C) HAT activity of HAG1, HAG2, and HAG5. Broccoli histones were labeled using ³H-acetyl CoA and the resulting SDS-PAGE gel was Coomassie-stained and subsequently subjected to fluorography. Histone H3 and H4 bands were definitively identified using mass spectrometry. H2A and H2B variants with overlapping migration patterns precluded definitive assignment of H2A and H2B bands. (D) HDA6 deacetylates full-length histones acetylated by HAG1, HAG2, and HAG5. Histones labeled by the HATs were incubated with equal aliquots of Flag-HDA6 or wild-type protein eluted from anti-Flag resin. The fluorogram and Coomassie-stained histone bands are shown. (E) HDA6 deacetylates histone N-peptides. Anti-Flag resin incubated with extracts of wild-type or Flag-HDA6-expressing plants was washed extensively then incubated with HAG5-labeled H4 peptide immobilized on agarose beads. ³H (from ³H-acetyl CoA) released into the reaction buffer was measured by scintillation counting. (F) HDA6 is a TSA-sensitive HDAC. HAG5-labeled broccoli histones were incubated with HDA6 ± TSA and then subjected to SDS-PAGE and fluorography.

Figure 4.

HAG1, HAG2, and HAG5 monoacetylate histone H3 and H4 N-peptides. (A) Mass spectrum of unmodified H4 peptide (M) differentially protonated (H) to generate 2+, 3+, and 4+ charge states. (B,C) Mass spectra of H4 peptide acetylated by HAG2 or HAG5. Unmodified and monoacetylated peptides (M + Ac) were detected in the 2+, 3+, and 4+ charge states. (D) Mass spectrum of unmodified histone H3 peptide. (E) Mass spectrum of the HAG1-acetylated histone H3 peptide. (F) Determination of HAG1, HAG2, and HAG5 specificities. N-peptides bearing pre-existing, nonradioactive acetyl groups on individual lysines were incubated with HAG1, HAG2, or HAG5 and ³H-acetyl CoA, then subjected to SDS-PAGE and fluorography.

Figure 5.

Transcriptional derepression of rRNA genes in A. suecica HDA6-RNAi lines is accompanied by changes in rRNA gene histone methylation and histone acetylation. (A) Diagram of an rRNA gene intergenic spacer highlighting the transcription initiation site (+1; arrow) and S1 and ChIP probe locations. (B) Pre-rRNA transcripts detected using S1 nuclease protection with A. thaliana- or A. arenosa-specific probes. (C) ChIP dot-blot analysis. Duplicate samples of wild-type or HDA6-RNAi plant chromatin, blotted in adjacent rows, were hybridized to A. thaliana- or A. arenosa-specific probes (see diagram). (Columns 1–3) Five percent, 2.5%, or 1.25% of the input chromatin in the ChIP reactions. (Column 4) Protein A beads in the absence of antibodies. (Columns 5–11) Chromatin immunoprecipitated with the indicated antibodies.

Figure 6.

HDA6 is required for cytosine hypermethylation at silenced promoters. ChIP–chop PCR was used to evaluate cytosine methylation density within A. thaliana- and A. arenosa-derived promoters in wild-type and HDA6-RNAi lines. Ten percent of the immunoprecipitated chromatin dot-blotted in Figure 5 was incubated with (lanes 5,7,9,11) or without (lanes 4,6,8,10) McrBC, then PCR was used to amplify A. thaliana or A. arenosa rRNA gene promoter regions. Hypermethylated DNA digested by McrBC is not amplified. (Lanes 1–3) Input controls used 0.1%, 0.05%, or 0.025% of the chromatin subjected to ChIP and show that the assay is semiquantitative.

Figure 7.

A model for the epigenetic control of rRNA gene on/off states. ChIP–chop PCR data indicate that active and silenced rRNA genes are marked by distinctive DNA and histone modifications that are mutually reinforcing. The model predicts de novo cytosine methylation and histone deacetylation as key events in switching from the transcriptionally permissive to the repressive state. Likewise, loss of promoter cytosine methylation and histone hyperacetylation are likely key events in switching from the silent to the active state.