Dot1 binding induces chromatin rearrangements by histone methylation-dependent and -independent mechanisms

Abstract

Background: Methylation of histone H3 lysine 79 (H3K79) by Dot1 is highly conserved among species and has been associated with both gene repression and activation. To eliminate indirect effects and examine the direct consequences of Dot1 binding and H3K79 methylation, we investigated the effects of targeting Dot1 to different positions in the yeast genome.

Results: Targeting Dot1 did not activate transcription at a euchromatic locus. However, chromatin-bound Dot1 derepressed heterochromatin-mediated gene silencing over a considerable distance. Unexpectedly, Dot1-mediated derepression was established by both a H3K79 methylation-dependent and a methylation-independent mechanism; the latter required the histone acetyltransferase Gcn5. By monitoring the localization of a fluorescently tagged telomere in living cells, we found that the targeting of Dot1, but not its methylation activity, led to the release of a telomere from the repressive environment at the nuclear periphery. This probably contributes to the activity-independent derepression effect of Dot1.

Conclusions: Targeting of Dot1 promoted gene expression by antagonizing gene repression through both histone methylation and chromatin relocalization. Our findings show that binding of Dot1 to chromatin can positively affect local gene expression by chromatin rearrangements over a considerable distance.

Figure 1

Dot1 is a derepressor. (A) Targeting of LexA or LexA fusion proteins to LexA operators proximal to a promoterless HIS3 gene or a telomeric URA3 gene. A telomeric URA3 reporter is silenced by the silencing complex (SirCx) that spreads from the telomeric repeats (TEL). LexA-Dot1 was targeted to LexA operators between the telomeric repeats and URA3 in a barrier assay, and to LexA operators distal of the telomeric repeats and URA3 in a desilencing assay. (B) Cells were plated in 10-fold serial dilutions on selective media with or without histidine. Transcriptional activation of HIS3 leads to growth on media lacking histidine (strain L40). LexA alone is indicated with a dash. The transcriptional activator domain of Adr1 was used as a positive control. (C) Barrier and desilencing assays of Dot1 and Rpd3 targeted to telomere VIIL (strains NKI5128 and NKI5376). A strain without LexA operators (NKI5240) and LexA alone were used as controls. Cells were plated in 10-fold serial dilutions on selective media (yeast culture; YC) with or without 5FOA. Cells that silence URA3 can grow on 5FOA media whereas cells that express URA3 cannot. (D) URA3 silencing was not disrupted upon targeting of human Lamin C (pLexA-Lamin), a mutant form of human CyclinE (R130A; pLexA-MCycE) or yeast Ecm5 (pLexA-Ecm5; NKI5128). (E) Barrier and desilencing assay of Dot1 at the HMLα mating-type locus with an inverted I-silencer (HMLi; YQY10, YQY09). A strain without LexA operators was used as a negative control (YXB85-n). (F) Chromatin immunoprecipitation (ChIP) using specific antibodies against Sir2 and Sir3 [24] was followed by quantitative PCR to determine binding to telomeric URA3 and HMLα upon targeting of Dot1 or Dot1^G401R (NKI5128). Average ChIP signals were normalized to input levels and Sir protein binding at URA3 relative to HMLα was plotted (n = 2, +/- SE <). Similar results were obtained with an active (ACT1) reference gene (see Additional File 1). (G) Immunoblot analysis of Sir2 and Sir3 protein levels in a strain expressing LexA, LexA-Dot1 or LexA-Dot1^G401R (NKI5128). Pgk1 was used as loading control.

Figure 2

Dot1 is a methyltransferase-dependent and -independent derepressor. (A) Outline of Dot1 deletion mutants showing the N terminus (white), the methyltransferase domain (black) and the H4 binding domain (grey). The G401R mutation (*) abolishes the catalytic activity of Dot1. All fusion proteins contained LexA and a V5 tag at the N terminus. (B) Protein and H3K79 methylation levels of Dot1 mutants described in (A) were determined in a dot1Δ strain lacking LexA operators (NKI5070). H3K79 methylation levels were determined of a wild-type strain or a dot1Δ strain (NKI5376 and NKI5378) expressing LexA or LexA-Dot1. Protein expression and H3K79 methylation were determined by immunoblot analysis using a V5 antibody and antibodies specific for H3K79 mono-, di- and trimethylation or the histone H3 C terminus. (C) Barrier (NKI5128 and NKI5129; left) and desilencing assay (NKI5376 and NKI5378; right) in the presence and absence of DOT1. Deletion of DOT1 results in reduced silencing that could be bypassed at 37°C [24,25]. Both LexA-Dot1 and LexA-Dot1^G401R were still able to disrupt URA3 silencing in a dot1Δ strain at 37°C, showing that derepressor activity does not require involvement of endogenous Dot1. Note that URA3 silencing is not completely lost in dot1Δ at 30°C (NK5378). This is caused by enhancement of telomeric silencing by the TRP1 gene distal to URA3 (see Additional file 2A). (D) Derepressor activity of Dot1^G401R and Dot1 deletion mutants at telomere VIIL (strains NKI5240, NKI5128 and NKI5376). Serial dilutions as presented before were quantified and plotted as bar graphs. (E) Derepressor activity of Dot1^G401R and Dot1 deletion mutants at HMLα (strains YXB85-n and YQY09). Growth on 5FOA observed for LexA-Dot1^G401R, LexA-Dot1^172-582 and LexA-Dot1^1-237 when targeted to LexA operators was caused by colonies that were uracil auxotrophs, which most likely represent URA3 mutants. (F) Derepressor activity of Dot1 and Dot1 mutants at the native chromosomes XIL, XVR and XVIL as described previously [82] (NKI2229, NKI2230 and NKI2231). (G) Dot1 and Dot1 mutants were expressed in dot1Δ strains with or without LexA operators, expressing wild-type histone H3 or histone H3 with K79 mutated to arginine (H3K79R), to determine whether K79 methylation is required for the methyltransferase-dependent derepressor activity (NKI6045, NKI6047, NKI6049 and NKI6051). Strains were grown at 37°C to enhance URA3 silencing.

Figure 3

Derepression by Dot1 requires the histone acetyltransferase Gcn5. (A) Derepressor activity of Dot1 and Gcn5 in a wild-type strain (NKI5128 and NKI1088), a dot1Δ strain (NKI5129 and NKI6020) or a gcn5Δ strain (NKI5399 and NKI6018). Strains lacking endogenous Dot1 were grown at 37°C to suppress the silencing defect, (see Figure 2C). We found that Gcn5 had a more prominent role in the barrier than in the desilencing assay of Dot1. URA3 silencing in the dot1Δ strain at 37°C in the desilencing assay (NKI1088 background) was limited compared with URA3 silencing in the dot1Δ at 37°C in the barrier assay (NKI1084 background), which is the result of the telomeric context (see Additional file 2A). (B) URA3 expression (n = 2 +/- SEM) was determined by reverse transcriptase-PCR and normalized to SIR3 expression, which is expressed at similar levels as URA3 and at equal levels in wild-type cells and histone modifier mutants [24]. (C) WT, dot1Δ and gcn5Δ strains with a URA3 gene at its endogenous euchromatic location were grown on media with or without uracil (BY4702, NKI3006 and NKI1107). Growth on media lacking uracil requires activation of URA3 by Ppr1, which is not affected by the loss of Dot1 or Gcn5. (D) LexA, LexA-Dot1 and LexA-Dot1^G401R were expressed from a high copy 2 μ plasmid (used for all experiments described here) in a wild-type strain (GCN5; NKI5128) or a gcn5Δ strain (NKI5399), and compared with wild type strains expressing LexA fusion proteins from a single-copy CEN (centromere sequences) plasmid. Each LexA-tagged protein also contained a V5 tag, which was used for immunoblot detection. Pgk1 was used as loading control. (E) Barrier assay of the LexA-tagged proteins expressed from the plasmids described in (D).

Figure 4

Tethered Dot1 disrupts telomere anchoring. (A) Telomere anchoring was measured in strain GA-1459. TEL VIR was visualized by binding of a GFP-LacI fusion protein to the Lac operators (indicated by green boxes). Subnuclear position was scored relative to the nuclear envelope tagged by a GFP-Nup49 fusion in approximately 100 to 300 nuclei. (B) Derepressor activity of targeted Dot1 at TEL VIR was measured in strains NKI1117 and NKI1118. (C) Localization data are represented in bar graphs as the percentage of spots in one of three concentric zones of equal surface. The dashed line at 33% corresponds to a random distribution. Spots observed in zone 1 represent telomeres localized to the nuclear periphery. (D) Two different statistical tests were performed. First, we tested whether telomeres targeted with LexA fusion proteins showed a random distribution over the three zones in the cell. Second, whether telomeres targeted with LexA-fusion proteins had a similar distribution to that of telomeres targeted with LexA alone. A significant difference for each Dot1 protein could be identified in at least one of the two tests. Asterisk indicates statistically significant differences (P < 0.05) from random telomere distribution (P_(random)) or from telomere distribution upon LexA targeting (P_(LexA)). The number of cells analyzed is indicated by n.