Report

Genome-Wide Kinetics of Nucleosome Turnover Determined by Metabolic Labeling of Histones

See allHide authors and affiliations

Science  28 May 2010:
Vol. 328, Issue 5982, pp. 1161-1164
DOI: 10.1126/science.1186777

Abstract

Nucleosome disruption and replacement are crucial activities that maintain epigenomes, but these highly dynamic processes have been difficult to study. Here, we describe a direct method for measuring nucleosome turnover dynamics genome-wide. We found that nucleosome turnover is most rapid over active gene bodies, epigenetic regulatory elements, and replication origins in Drosophila cells. Nucleosomes turn over faster at sites for trithorax-group than polycomb-group protein binding, suggesting that nucleosome turnover differences underlie their opposing activities and challenging models for epigenetic inheritance that rely on stability of histone marks. Our results establish a general strategy for studying nucleosome dynamics and uncover nucleosome turnover differences across the genome that are likely to have functional importance for epigenome maintenance, gene regulation, and control of DNA replication.

Nucleosome disassembly and reassembly, or turnover, is necessary for epigenome maintenance, but the mechanisms that are responsible remain unclear (1). One approach to this problem has been to map enrichment of the universal histone replacement variant, H3.3 (26), which requires complete unwrapping of DNA from around the histone core for its replication-independent deposition to occur. Genome-wide profiling of steady-state amounts of H3.3 from Drosophila melanogaster S2 cells indicated that nucleosome replacement occurs most prominently across transcribed regions of active genes and at promoters and binding sites of trithorax group (trxG) and polycomb group (PcG) proteins (2, 3). Similar results were obtained for HeLa cells (7) and Caenorhabditis elegans embryos (8). A more direct approach, which can measure dynamics but is limited to yeast, is to express constitutive and inducible histone transgenes and to measure the relative incorporation of their encoded tagged histones (911). These studies indicated that turnover rates were high at promoters and chromatin boundary elements but low within transcribed regions. Both approaches are limited by the requirement for transgenes and tags and by the time lag during induction.

We have developed a general method for directly measuring the kinetics of nucleosome turnover that overcomes these limitations (12). Our strategy uses cotranslational incorporation of the methionine (Met) surrogate azidohomoalanine (Aha) into proteins and subsequent ligation of biotin to Aha-containing proteins through the [3+2] cycloaddition reaction between the azide group of Aha and an alkyne linked to biotin (13, 14) (fig. S1A). To measure nucleosome turnover rates, we treated cells briefly with Aha, coupled biotin to nucleosomes containing newly incorporated histones, affinity-purified with streptavidin, washed stringently to remove nonhistone proteins, and analyzed the affinity-purified DNA by using tiling microarrays. We call this strategy CATCH-IT, for covalent attachment of tags to capture histones and identify turnover. In Drosophila S2 cells, Aha was incorporated into total protein, including histones, in a translation-dependent manner (fig. S1, B to D). Incorporation of Aha into newly synthesized histones increased in proportion to the length of Aha treatment for at least 3 hours.

To measure nucleosome turnover across the genome, we starved late log-phase S2 cells for methionine for 30 minutes and treated them with Aha for 3 hours, performed the biotin coupling reaction on isolated nuclei, and digested chromatin with micrococcal nuclease to yield mostly mononucleosomes. Biotin-tagged nucleosomes containing newly synthesized histones were then isolated by using streptavidin beads and washed with a urea-salt solution to remove H2A/H2B dimers and other DNA-binding proteins (fig. S2), leaving only (H3/H4)2 tetramers (15), and the remaining DNA was labeled and hybridized to high-density tiling microarrays along with the corresponding input DNA. For comparison, we also performed streptavidin precipitations using chromatin from S2 cells expressing biotin-tagged H3.3 over a 2-day period as previously described (2). When array data for all genes with annotated ends were divided into quintiles by gene expression and aligned by gene ends and when log2 ratios of precipitated DNA/input DNA were averaged across genes, we found that CATCH-IT and H3.3 profiles were highly similar (Fig. 1, A and C). In addition, CATCH-IT landscapes corresponded overall to steady-state H3.3 landscapes (Fig. 2, blue and green tracks), albeit with better definition of chromatin features. Such correspondence confirms that CATCH-IT depends on nucleosome dynamics, and we attribute the better resolution of CATCH-IT to its capturing turnover kinetics rather than steady-state replacement.

Fig. 1

CATCH-IT marks sites of histone replacement and reveals kinetics. (A) Gene ends analysis of a CATCH-IT experiment from a 3-hour Aha treatment (pulse). All 9820 genes from FlyBase r5.13 with annotated 5′ and 3′ ends were grouped by gene expression quintiles (top 20% to bottom 20% based on GEO GSM333845) and aligned by gene ends versus the log2 ratios of precipitated DNA/input DNA averaged across genes. (B) Same as in (A) but with cells treated with Aha for 3 hours (pulse) followed by a 1.5-hour Met treatment (chase). (C) Same as in (A) but for biotin-tagged H3.3 nucleosomes. (D) Scatter plot showing the correlation between the pulse and the difference between the pulse and chase with use of all 2.1 million probes on the array. (E) Same as in (D) but with use of only genic probes from highest and lowest gene expression quintiles.

Fig. 2

Chromatin landscapes of CATCH-IT, ORC, H3.3, and salt fractions. Chromatin landscapes over a representative euchromatic region of chromosome arm 3R from a CATCH-IT time course of 20-, 40-, and 60-min Aha treatment. Also shown are ORC binding; biotinylated H3.3 nucleosomes; and successive 80 mM, 600 mM, and insoluble chromatin salt fractions (18). Genes are shown above chromatin tracks. Those above the line are on the top strand and those below the line are on the bottom strand. Right section shows a magnification of the indicated region of the left section.

As a control, we also treated S2 cells with Met rather than Aha and obtained a featureless profile, confirming that the signals obtained with Aha represent newly synthesized histones (fig. S3). We also found that independent biological replicates of the CATCH-IT procedure yielded similar results (fig. S4).

We next asked whether CATCH-IT could measure nucleosome turnover kinetics. After treating cells with a 3-hour pulse of Aha, a sample was taken, and then cells were switched back to Met-containing media for a 1.5-hour chase; both samples were processed as before. As expected, the chase resulted in an overall reduction of signal across expressed genes when compared with the pulse signal (Fig. 1, A and B). We also compared the difference between the pulse and chase signals to the pulse signal on scatter plots by using either all 2.1 million probes or only genic probes from genes in each expression quintile. As expected for kinetic measurements, nucleosomes undergoing the highest turnover also tended to show the largest decrease during the chase, and this trend depended on gene expression (Fig. 1, D and E, and fig. S5). We conclude that CATCH-IT captures the dynamics of histone replacement. In addition, these results show that turnover across gene bodies is highly dependent on expression level (Fig. 1, A and B). We suggest that the high turnover seen for gene bodies in Drosophila but low rate seen for yeast (10) reflects biological differences between the two organisms in processes that evict or retain nucleosomes during transcription.

To estimate rates of nucleosome turnover, we removed successive samples after treating with Aha for 20, 40, and 60 min. We observed that the nucleosome turnover landscapes generated at each time point were highly similar (Fig. 2, blue tracks), and this was confirmed by ends analysis (fig. S6). We also examined turnover at sites of epigenetic regulatory elements, as represented by sites of binding of the trxG proteins GAF (GAGA factor or trithorax-like) and Zeste and of the PcG proteins EZ (enhancer of zeste) and PSC (posterior sex combs) (16, 17). Comparison of average turnover profiles across GAF, Zeste, and EZ+PSC sites showed that CATCH-IT identified these sites as regions of high turnover relative to surrounding regions, with higher turnover at GAF than at EZ+PSC sites (Fig. 3, A and B, and fig. S7, A and B). There was also better peak delineation with CATCH-IT than was observed for H3.3 or low salt-soluble chromatin, which represents classical “active” chromatin (18) (Fig. 3, A to D, and fig. S7). By using the 20-min data sets from two independent experiments to calculate turnover rates (fig. S8), we obtained mean lifetimes of ~1 hour for the peak just downstream of the transcriptional start site of active genes, ~1 hour for GAF sites, and ~1.5 hours for EZ+PSC sites (table S1). These estimates are conservative, because any delay in incorporation of Aha would lead us to overestimate mean lifetimes. Therefore, nucleosomes within active genes and at epigenetic regulatory elements turn over multiple times during each ~20-hour cell cycle.

Fig. 3

Kinetics of nucleosome turnover at epigenetic regulatory elements and sites of ORC binding. (A) Average CATCH-IT time course signals over GAF and EZ+PSC binding sites. (B) Same as in (A) but for a pulse-chase experiment with a 3-hour Aha pulse and 1.5-hour Met chase. (C) Average biotinylated H3.3 signals over GAF and EZ+PSC binding sites. (D) Average signals from chromatin salt fractions over GAF and EZ+PSC binding sites. (E to P) Plots of various chromatin signals aligned at peaks of ORC binding and divided into quintiles by ORC peak score (20).

We also compared CATCH-IT turnover landscapes to binding sites for the ORC2 subunit of the origin recognition complex (ORC), which specifies replication initiation (19). We observed a notable correspondence, as exemplified by visual examination of a typical gene-rich region of the Drosophila genome (Fig. 2, magenta tracks). Although H3.3 amounts also showed a correspondence to ORC amounts (Fig. 2, green tracks) as previously observed (20), the resemblance of CATCH-IT profiles to ORC profiles was far more conspicuous. To better evaluate these correspondences, we aligned CATCH-IT, H3.3, and other chromatin profiles around the 5135 ORC peaks (20) and divided them into quintiles on the basis of ORC binding score (Fig. 3, E to G). The nested peaks indicate a quantitative relationship between ORC binding and nucleosome turnover, suggesting that turnover facilitates ORC binding. In contrast, other chromatin features that would be expected for open or dynamic chromatin, including nucleosome density, mononucleosome/oligonucleosome ratio (a measure of micrococcal nuclease accessibility), H2Av (an H2A.Z histone variant enriched in active chromatin), and salt-soluble nucleosomes, show little if any dependence on ORC abundance (Fig. 3, H to P). Our findings support the hypothesis that replication origins are determined by chromatin, not by sequence features (20, 21). The better quantitative correspondence of ORC to CATCH-IT data than to other chromatin measurements implies that the ORC occupies DNA that is made accessible by nucleosome turnover. In support of this interpretation, we note that very similar correspondences are seen when CATCH-IT data are aligned with GAF sites (fig. S9) and that GAF directs nucleosome turnover in vivo (22, 23).

Our direct strategy for measuring the kinetics of nucleosome turnover does not rely on transgenes or antibodies but rather uses native histones and generic reagents. Thus, CATCH-IT provides a general tool for studying activities that influence nucleosome turnover. With use of CATCH-IT, we found direct evidence that epigenetic maintenance involves nucleosome turnover, a process that erases histone modifications (10). The fact that EZ is responsible for di- and trimethylation of H3K27, but the nucleosomes that it modifies turn over faster than a cell cycle, argues against proposals that histone modifications required for cellular memory themselves transmit epigenetic information (24). Rather, by simply increasing or decreasing accessibility of DNA to sequence-specific binding proteins, regulated nucleosome turnover may perpetuate active or silent gene expression states and facilitate initiation of replication.

Supporting Online Material

www.sciencemag.org/cgi/content/full/328/5982/1161/DC1

Materials and Methods

Figs. S1 to S9

Table S1

References

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank T. Furuyama for suggesting this approach, members of our lab for helpful discussions, and the Hutchinson Center Genomics Shared Resource for microarray processing. This work was supported by NIH grant 1R21DA025758 to S.H. and NIH Postdoctoral Fellowship 1F32GM083449 to R.B.D. All data sets can be found in GEO: GSE19788.
View Abstract

Navigate This Article