Chromatin is composed of DNA and histones, which provide a unified platform for regulating DNA-related processes, mostly through their post translational modification. What happens to this epigenetic landscape during DNA replication?
The epigenetic landscape is largely perturbed during DNA replication. First, specific modifications are used to define replication origins, or are deposited during the course of DNA polymerase progression. In addition, histones must be removed from the DNA template to enable replication fork progression and then restored in the wake of the fork.
In addition, during replication, DNA content increases, requiring additional histones for its wrapping (I). Newly synthesized histones are used for wrapping the newly replicated DNA (II). These new histones lack position-specific modifications, but instead are generally modified on multiple residues. Therefore, the chromatin landscape must also be reproduced on the newly replicated DNA.
We studied how epigenetic information is transmitted during the cell cycle by looking on dynamics. The dynamics of retrieval may reveal functional information – For example, if modifications are recovered after a very long delay, it may suggest that other processes (not coupled to DNA replication) are involved in the recovery.
We followed the genome-wide pattern of multiple histone modifications during and after replication. We synchronized budding yeast to the beginning of S-phase, and measured histone modifications using ChIP-Seq. In addition we also measured expression levels (with RNA-Seq), and the replication profile using DNA-Seq. In the right graph, a chromosome is shown. The x axis is location on the chromosome, along time (y-axis). This is the duration of S-phase in this experiment. It is evident that some regions have already replicated, as HU arrests cells after replication has begun. These orange peaks from which replication begins correspond to known origins of replication in yeast.
We asked – when does the modification increase, compared to the time of replication of the same region. We saw that some modifications appeared on replicated DNA quite immediately (e.g. H4K16ac, H3K4me1), whereas all tri-methylation were deposited with a significant delay that extended for over 20 min.
We next examined the effect of transcription-related parameters on this delay. We started by separating genes that replicate at the same time but have different levels of expression (top). We saw that this delayed deposition was largely correlated with gene transcription strength and gene promoter structure (bottom), so that actively transcribing genes were tri-methylated significantly faster than low-expressing ones. Therefore we conclude that replication-independent processes, rather than active memory, account for the recovery of histone tri-methylation pattern following DNA replication.
When examining the delays, we were intrigued by one modification in particular. H3K9ac is a modification extensively studied in the context of transcription. It correlates with transcription levels, and is found mainly in the beginning of genes. However, in our cell-cycle experiment, H3K9ac, was deposited ahead of the replication fork in a distance of 5kb from the fork (therefore – a negative delay), everywhere in the genome.
We next asked which enzyme is responsible for this deposition ahead of the fork. Two enzymes acetylate H3K9: Gcn5 and Rtt109. We deleted either and repeated the same experiment (also for WT) by a different synchronization method (alpha-factor). We observed that when Rtt109 the deposition of H3K9 ahead of the fork disappears. In fact, it followed replication.
We examined the raw H3K9ac pattern along the chromosomes. In the slide, see Chr V. Each box represents the chromosome in a different strain (WT, gcn5-deleted, rtt109-deleted), along time from top to bottom. When looking on the raw H3K9ac in the different strains we observed a nice separation of the pattern in the mutants: When only Rtt109 is present (middle), a replication-like pattern is observed. In contrast, when only Gcn5 is functional, we observe lines that are not changing throughout time, and these are the beginning of genes. It is evident that the pattern of H3K9ac in wild type combines the replication- and transcription-associated patterns. Therefore we have two different enzyme (Gcn5, Rtt109) responsible for one histone mark (K9ac), that act in different context – Gcn5 in transcription, in the beginning of genes, and Rtt109 in replication, everywhere in the genome during replication.

Chromatin dynamics during DNA replication

By Raz Bar-Ziv*, Yoav Voichek & Naama Barkai

Link to Manscript

DOI

Questions: Replication, H3K9ac, H3K56ac, all,

Yes. We continued this study by examining whether H3K9ac is related to supercoiling stresses that accumulate in front of the replication fork. Naturally, these stresses are relieved by DNA topoisomerases. We profiled H3K9ac during replication also in strains depleted of topoisomerases. When topoisomerases are depleted, replication stops.

In the figure, on the left (C), the replication profile is shown on a chromosome. When both top2 and top1 are depleted from cells, replication (in orange), stops. On the right side (D), you can see the H3K9ac signal. Notably, in contrast to the transient H3K9ac wave in wild-type cells, where replicated regions were deacetylated following the passing of the fork, H3K9ac remained stable in topoisomerase-depleted cells. This may indicate continuous acetylation, triggered by supercoiling, or an inability of histone deacetylases to access the supercoiled regions.

We observed that the spread of H3K9ac stopped upon replication arrest, and it did not progress further into unreplicated regions.

In the figure, on the left you see all the regions in the genome that were replicated, ordered by their left. On the right side, the acetylation on K9 on the same regions is plotted. The acetylation does not spread more than the 5kb, suggesting the H3K9ac and the replication process are coupled.

Therefore, the progression of H3K9ac in front of the replication fork depends on active replication or on other processes affected by the torsional stress.

Since we synchronized the cells with HU, cells are synchronized to the beginning of S-phase. At this stage, some of the origins of replication have already intiatied replication. Therefore, already at this time point, some regions show double the DNA amount.

These regions are missing on the chromosome representation since they were excluded due to alignment. If a sequence aligns to more that one region in the genome, it was excluded from the analysis.

We defined the time delay between DNA replication and histone modification using two measures. First, we grouped genomic regions into eight clusters based on their time of replication. Comparing the cluster-averaged increase of DNA content and histone marks, confirmed that some modifications are deposited immediately upon replication while other recovered with a delay. Next, to obtain a measure of this delay, that is independent of cluster assignment, we performed a cross-correlation analysis, defining the time-shift that best aligns the genome-wide increase in DNA content and changes in each modification. The bar graphs in the figure represent the the typical time delay, the time of maximum cross-correlation.

These refer to the organization of the promoter and its architecture. Some genes lack a nucleosome free region proximal to the transcription start site (OPN, Occupied Proximal Nucleosome), while others have a pronounced proximal nucleosome free region (Depleted Proximal Nucleosome, DPN). OPN genes are  more responsive to environmental changes, display a higher cell-to-cell variability (noise) and tend to have a TATA box in their promoters.

The fast recovery associated with promoters that lack a TATA-box or that are depleted of a nucleosome near their TSS, are features that correlate also with slow histone turnover. 

We profiled also H3K56ac, which is deposited on newly synthesized histones prior to DNA incorporation. We observed that H3K9ac is deposited in front of the fork (~5 kb ahead), while H3K56ac increases later (with the replication fork).

To control for possible cross-reactivity of the H3K56ac and H3K9ac antibodies (Drogaris et al. 2012), we repeated our experiment in strains harboring mutated H3, incapable of being either K9- or K56-acetylated (depicted H3K9A, and H3K56A). Indeed, in H3K9A mutant, H3K56ac increased concomitantly with the DNA content, while in the H3K56A mutant the pre-replication induction of H3K9ac was maintained.

Further studies are needed to define the function of H3K9ac during replication, and why it precedes the replication fork. Deletion of Rtt109 or mutating H3K9 did not perturb S-phase progression. This mutant was previously shown to increase genomic instability, although this effect was attributed to the loss of H3K56ac, which also controls expression homeostasis during S-phase. The pre-replication H3K9ac wave may contribute to the smooth progression of the replication machinery, perhaps by recruiting or stabilizing nucleosome remodeling complexes