Methylation of histone H3 on lysine 79 associates with a group of replication origins and helps limit DNA replication once per cell cycle

Abstract

Mammalian DNA replication starts at distinct chromosomal sites in a tissue-specific pattern coordinated with transcription, but previous studies have not yet identified a chromatin modification that correlates with the initiation of DNA replication at particular genomic locations. Here we report that a distinct fraction of replication initiation sites in the human genome are associated with a high frequency of dimethylation of histone H3 lysine K79 (H3K79Me2). H3K79Me2-containing chromatin exhibited the highest genome-wide enrichment for replication initiation events observed for any chromatin modification examined thus far (23.39% of H3K79Me2 peaks were detected in regions adjacent to replication initiation events). The association of H3K79Me2 with replication initiation sites was independent and not synergistic with other chromatin modifications. H3K79 dimethylation exhibited wider distribution on chromatin during S-phase, but only regions with H3K79 methylation in G1 and G2 were enriched in replication initiation events. H3K79 was dimethylated in a region containing a functional replicator (a DNA sequence capable of initiating DNA replication), but the methylation was not evident in a mutant replicator that could not initiate replication. Depletion of DOT1L, the sole enzyme responsible for H3K79 methylation, triggered limited genomic over-replication although most cells could continue to proliferate and replicate DNA in the absence of methylated H3K79. Thus, prevention of H3K79 methylation might affect regulatory processes that modulate the order and timing of DNA replication. These data are consistent with the hypothesis that dimethylated H3K79 associates with some replication origins and marks replicated chromatin during S-phase to prevent re-replication and preserve genomic stability.

Figure 1. H3K79Me2 containing chromatin is associated preferentially with replication initiation sites genome-wide.

A–E. Screenshots of replication initiation data visualized with the integrated Genome Viewer (http://www.broadinstitute.org/igv/). A chromosome map is shown at the top, and the region-of-interest is delineated by a circle. The analyzed region is shown underneath the ideogram, with map coordinates indicated. The Replication panel shows the distribution of replication initiation events (ratio of reads obtained from a nascent strand preparation, and reads obtained from a corresponding control genomic DNA preparation) of the region-of-interest obtained from our published database, see for details. Regions abundant in H3K79Me2 immunoprecipitated chromatin from H3K79Me2 ChIP-Seq in K562 cells are shown below the initiation profile as reads per kilobase per million aligned reads (RPKM). Ref-Seq genes are aligned above the experimental data. A. The distribution of H3K79Me2 and replication initiation events in the MYC locus. B. The distribution of H3K79Me2 and replication initiation events in the DBF4 locus. C. The distribution of H3K79Me2 and replication initiation events in the LCORL locus. D. The distribution of H3K79Me2 and replication initiation events in the BAG1 locus. E. The distribution of H3K79Me2 and replication initiation events in the UBAP1 locus. F. A box plot comparing the relative enrichment of replication initiation events in chromatin featuring H3K79Me2 obtained by ChIP-Seq as described in methods with various other chromatin features as reported in the UCSC genome browser (for details, see. Boxes indicate distributions of the second and third quartiles and whiskers, 95^th percentiles. The horizontal lines in the boxes represent medians. Values lower than a unit were converted into 1. Methylation of H3K79 exhibits a marked enrichment in replication initiation events that was higher than any other measured histone modification or transcription factor. G. A histogram showing the replication enrichment ratio (calculated as in A, B) for genomic regions as a function of their distance from the closest H3K79Me2 interaction sites. A box plot version of the same histogram is shown as Figure S1.

Figure 2. Preferential enrichment of initiation events in H3K79Me2 containing chromatin at the G1 and G2 phases of the cell cycle.

An asynchronously growing population of K562 cells was fractionated into separate populations of G1, S-phase and G2/M cells using centrifugal elutriation. Chromatin from each separate cell cycle phase population was isolated and analyzed by H3K79Me2 ChIP-Seq as described in the legend to Figure 1. A. The distribution of H3K79Me2 and replication initiation events in the MYC locus during the different phases of the cell cycle. B. The distribution of H3K79Me2 and replication initiation events in the DBF4 locus during the different phases of the cell cycle. C. A box plot as described in Figure 1 legend showed the relative enrichment in H3K79Me2 for replication initiation events during the different phases of the cell cycle. Boxes indicate distributions of the second and third quartiles and whiskers, 95^th percentiles; values lower than a unit were converted into 1. All, replication initiation ratios in regions that were associated with H3K79Me2 in an asynchronous cell sample representing all stages of the cell cycle; G1, S and G2, regions associated with H3K79Me2 in samples from cells at the appropriate cell cycle phase; S-only, regions that were only associated with H3K79Me2 in S-phase but not in G1 and G2. D. The number of H3K79Me2 associated peaks on chromatin during subsequent stages of cell cycle progression. H3K4Me3 was used as a control. Chi-square test shows that the distributions of H3K79Me2 and H3K4Me3 in G1, S and G2 cells are statistically different (p<0.0001). H3K79 dimethylation was more abundant and exhibited a wider distribution during S-phase, but chromatin regions associated with H3K79Me2 solely in S-phase were not further enriched in replication initiation events. The association of H3K79Me2 with replication initiation events was re-established in G2.

Figure 3. H3K79 dimethylation accompanies replicator activity.

A. Two transgenes containing sequences from the human beta-globin Locus Control Region (HS432), the human beta-globin promoter (GloPro) driving enhanced green fluorescent protein (EGFP) and two variants of the Rep-P replicator were inserted into a single location on murine chromosome 15 in murine erythroleukemia (RL4) cells . Murine cells were utilized to facilitate detection of the exogenous sequences from the human beta globin locus; the murine locus control region was used as a control. One transgene variant (Rep-PWT) contains the native unaltered sequence of Rep-P-2 (starting 87 bp 5′ of the beta-globin promoter) that is essential for replication initiation. The other transgene variant (Rep-PAG1) harbors two point mutations at the Rep-P-2 sequence that prevent initiation of DNA replication. B. Chromatin immunoprecipitation with antibodies directed against H3K79Me2 in RL4 cells carrying wild-type (Rep-PWT) or mutant (Rep-PAG1) transgene cassettes. Each column represents the mean enrichment value inH3K79Me2 calcualted based on real-time PCR amplification of chromatin immunoprecipitation using the indicated primer pairs. Error bars indicate standard deviations. C. Nascent strand abundance analysis in RL4 cells carrying wild type (Rep-PWT) or mutant Rep-P (Rep-PAG1) transgene cassettes. Primers and probes used, listed in Table 3, included GloPro (human beta-globin promoter), EGFP (the EGFP gene), bG59.8 (Rep-P 5′ end sequence), bG61.3 (Rep-P 3′ end sequence), AG (the AG region of Rep-P), and mLCR (murine Locus control region). All except mLCR are sequences from the transgene and their location is illustrated in the map shown in panel A. Sequences from transgenes harboring the active replicator were enriched in H3K79Me2 containing chromatin whereas sequences from the mutant transgene that did not initiate replication, were not.

Figure 4. Depletion of H3K79 methyltransferase DOT1L does not change replication elongation and initiation rates.

HCT116 cells were transfected with siRNA directed against DOT1L or scrambled siRNA control twice with a 48 h interval. A. Levels of H3K79Me2 in total cell proteins collected 72 hours after the second transfection. Actin was used as a loading control. B–D. Cells were labeled sequentially with ldU and CIdU as described in the Methods section. Cells were then harvested and DNA extracted. The DNA was stretched on a silanized microscope coverslip, and visualized with antibodies against DNA containing ldU and CldU . B. An example of combed DNA. Replication fork progression rates were calculated from the length of CldU (red replication tracks) and IdU (green replication tracks) signals. Inter-origin distance was measured by identifying replication initiation events (ori1 to ori3). C. A histogram of the distribution of replication fork speeds measured in DNA fibers from cells transfected with scrambled siRNA. D. A histogram of the distribution of replication fork speed measured in DNA fibers from cells transfected with siRNA targeting DOT1L. Depletion of DOT1L reduced the levels of H3K79Me2 but did not affect DNA replication fork velocity and inter-origin distance (Figure S5).

Figure 5. Effects of Dot1L depletion on cell cycle progression.

HCT116 cells transfected with siRNA directed against DOT1L or scrambled siRNA control three times, first with a 48 h interval followed by a 72 hr interval. Cells were collected for cell cycle analysis by FACS 3 days after the third transfection. EdU were added to cells for 45 minutes before harvesting cells. Click-iT EdU kit from Invitrogen was used to detect replicating cells and DAPI was used to determine DNA content. A. The left panel shows a representative cell cycle profile for cells transfected with control siRNA. The middle panel shows a representative cell cycle profile for cells transfected with siRNA directed against DOT1L. The right panel shows a histogram illustrating the fraction of cells in early, middle and late S-phase (ES, MS and LS, representing the number of cells in the P2, P3 and P4 FACS gates, respectively). Two stars represent a statistically significant change with p value lower than 0.01; three stars represent a statistically significant change with p value lower than 0.001. Dot1L depletion increased the fraction of late S-phase cells. B. The left and middle panels show the cell cycle distribution of the same cells as in A, illustrating the FACS gates used to identify the sub-G1, non-replicating S and >G2/M cell populations. The histogram plots the fraction of the gated populations in DOT1L depleted cells divided by the fraction of the same gated populations in cells transfected with a control siRNA. A star represents a statistically significant change with p value lower than 0.05; two stars represent a statistically significant change with p value lower than 0.01. Table S2 shows the fraction of cells at each cell cycle phase. Dot1L depletion caused a limited increase in the fraction of apoptotic cells, cells with S-phase DNA content that did not incorporate EdU and cells with DNA content greater than G2/M.

Figure 6. Over-replication in DOT1L depleted cells.

HCT116 cells were transfected with siRNA directed against DOT1L or scrambled siRNA control twice with a 48 h interval, and collected 3 day after the second transfection. (A–C) Cells were labeled with 10 µm of EdU for 1 hour before harvest and EdU distribution patterns were visualized along with DNA content measurement using DAPI. DNA/DAPI content was quantified in cells exhibiting early and late S phase EdU distribution using the Pathway imaging system (BD). DNA content distribution in early S-phase cells and late S-phase cells determined by EdU patterns in HCT116 cells transfected with control scrambled siRNA (panel A) or DOT1L siRNA (panel B). Cross shows mean and square box shows median of DNA content. C. Example images for diffuse “early” EdU pattern, large punctuate structures of “late” EdU pattern and re-replicated larger cells (> = G2 by DAPI DNA content) with early EdU patterns. DOT1L depleted cells, but not control cells, contained a population of cells with DNA content greater than 4N exhibiting early replication patterns. D. BrdU density gradients measuring DNA re-replication. Top panel: HCT116 cells were transfected with control siRNA or Dot1L siRNA as described above and labeled with Bromodeoxyuridine (BrdU) for 18 hours before harvest. Genomic DNAs were fractionated on CsCl gradients and BrdU substituted DNA was detected using anti-BrdU antibodies on a membrane. BrdU substituted DNA is denser (heavier) than unsubstituted DNA. LL: unsubstituted DNA; HL: semi-substituted DNA; HH: fully substituted. Bottom panel: Serial dilutions of unsubstituted genomic DNA and fully BrdU substituted DNA sample (isolated from cells incubated with BrdU for 48 hours) were used as controls to evaluate the specificity of the anti-BrdU antibody.