Report

MCM2 promotes symmetric inheritance of modified histones during DNA replication

See allHide authors and affiliations

Science  28 Sep 2018:
Vol. 361, Issue 6409, pp. 1389-1392
DOI: 10.1126/science.aau0294

How cells ensure symmetric inheritance

Parental histones with modifications are recycled to newly replicated DNA strands during genome replication, but do the two sister chromatids inherit modified histones equally? Yu et al. and Petryk et al. found in mouse and yeast, respectively, that modified histones are segregated to both DNA daughter strands in a largely symmetric manner (see the Perspective by Ahmad and Henikoff). However, the mechanisms ensuring this symmetric inheritance in yeast and mouse were different. Yeasts use subunits of DNA polymerase to prevent the lagging-strand bias of parental histones, whereas in mouse cells, the replicative helicase MCM2 counters the leading-strand bias.

Science, this issue p. 1386, p. 1389; see also p. 1311

Abstract

During genome replication, parental histones are recycled to newly replicated DNA with their posttranslational modifications (PTMs). Whether sister chromatids inherit modified histones evenly remains unknown. We measured histone PTM partition to sister chromatids in embryonic stem cells. We found that parental histones H3-H4 segregate to both daughter DNA strands with a weak leading-strand bias, skewing partition at topologically associating domain (TAD) borders and enhancers proximal to replication initiation zones. Segregation of parental histones to the leading strand increased markedly in cells with histone-binding mutations in MCM2, part of the replicative helicase, exacerbating histone PTM sister chromatid asymmetry. This work reveals how histones are inherited to sister chromatids and identifies a mechanism by which the replication machinery ensures symmetric cell division.

Histone posttranslational modifications (PTMs) contribute to the establishment and maintenance of epigenetic chromatin states that regulate transcriptional programs during development (1, 2), but the mechanisms that ensure transmission of histone PTM patterns to daughter cells remain unclear. Chromatin is disrupted upon replication fork passage, and nucleosomes are rapidly reassembled on newly synthesized DNA through recycling of evicted parental histones and de novo deposition of new histones (3, 4). The recycling of modified parental histones is a critical step in histone PTM transmission (5), and early studies suggested that parental histones segregate randomly to both daughter DNA strands (6, 7). However, whether histone PTM inheritance is truly symmetric and how parental histones are segregated to the leading and lagging strands of the replication fork remain open questions. Multiple replication origins are used to replicate large metazoan chromosomes, and replication fork directionality (RFD) and leading- and lagging-strand replication therefore alternate along chromosomes (8, 9). Potential biases in segregation of modified parental histones during replication will thus result in a specific pattern of sister chromatid asymmetry. We investigated the distribution of parental and new histones on sister chromatids and linked this distribution to RFD to understand histone segregation.

We developed SCAR-seq (sister chromatids after replication by DNA sequencing) to track histone recycling and de novo deposition genome-wide (Fig. 1A and methods). We differentiated old and new histones H4 by dimethylation at lysine 20 (H4K20me2) (fig. S1A), present exclusively on >80% of old H4 in nascent chromatin (5, 10), and acetylation at lysine 5 (H4K5ac), present on >95% of new H4 (3, 11). Mouse embryonic stem cells were labeled with 5-ethynyl-2′-deoxyuridine (EdU), and nascent mononucleosomes carrying H4K20me2 or H4K5ac were purified sequentially by chromatin immunoprecipitation (ChIP) and streptavidin capture of biotinylated EdU-labeled DNA. The new and parental DNA strands were separated (fig. S1B) and sequenced in a strand-specific manner to score genome-wide sister chromatid histone partition (Fig. 1A and fig. S1C).

Fig. 1 Parental histones segregate to both sister chromatids with a weak bias toward the leading strand.

(A) SCAR-seq technique. Partition of old and new histones is calculated as the proportion of forward (F) (red) and reverse (R) (blue) counts in genomic windows [the range is between −1 (100% reverse strand) and 1 (100% forward strand)]. (B) RFD and partition of H4K20me2 and H4K5ac at a genomic region. Initiation zone centers (lines), active gene orientation (arrowheads), and active enhancers (bars) are shown. Chr3, chromosome 3. RFD is calculated as the proportion between right- and left-moving replication forks [the range is between −1 (100% left moving) and 1 (100% right moving)]. (C) Average RFD (blue) and partition of old (H4K20me2) and new (H4K5ac) histones around initiation zones. (D) Partition at downstream (leading strand) and upstream (lagging strand) edges of initiation zones, with significant partition difference in each replicate (rep) (paired Wilcoxon signed-rank test, P < 5.3 × 10−14).

To determine locally which sister chromatid was replicated preferentially by the leading strand, we measured RFD by Okazaki fragment sequencing (OK-seq) (methods) (8). Replication initiation zones (n = 2,844) (fig. S2A) were comparable to those in humans (8) and Caenorrhabditis elegans (12), ranging in size and efficiency (fig. S2, B and C), and were mostly intergenic (fig. S2D), enriched in enhancer-associated features [H3 acetylation at lysine 27 (H3K27ac), H3 monomethylation at lysine 4 (H3K4me1), p300 occupancy, and deoxyribonuclease I–hypersensitive sites], and flanked by active genes [marked by H3 trimethylation at lysine 36 (H3K36me3) and lysine 4 (H3K4me3)] (fig. S2E). Around initiation zones, the partition of old and new H4 showed a weak reciprocal shift, with H4K20me2 and H4K5ac skewed toward leading- and lagging-strand replication, respectively (Fig. 1, B to D, and fig. S3, A to C). The partition amplitude was considerably lower than RFD, suggesting that old histones segregate to both strands but not entirely symmetrically. Analysis of the parental DNA strands showed the complementary partition shift (fig. S3, D and E), excluding an effect of EdU on partition measurements. The partition skew was most pronounced around highly efficient initiation zones (fig. S3F), indicating that DNA replication drives the observed sister chromatid asymmetries. Histone partition skew also tracked with RFD at higher genomic scales (fig. S4)—for example, across replication units with early-replicating borders and late-replicating centers, termed U-domains (8, 9, 13). Together, these results demonstrate that parental histones segregate to both arms of the replication fork with a slight preference for the leading strand, whereas de novo deposition has a comparable bias toward the lagging strand.

Replication timing is related to chromosome organization in topologically associating domains (TADs) (9, 14, 15). TAD borders (16) are enriched in initiation zones (fig. S5A) (8, 13) and showed a reciprocal histone partition skew (Fig. 2, A and B, and fig. S5B). B-compartment TADs (transcriptionally inactive) displayed stronger RFD and partition shifts than transcriptionally active A-compartment TADs (17) (Fig. 2A), possibly because of increased internal initiation within active TADs (Fisher’s exact test, odds ratio 2.2, P < 2.2 × 10−16) or an effect of transcription. To investigate partition asymmetries over genes, we tracked H3K36me3 (fig. S5C) present on parental histones in gene bodies (5, 18, 19). H3K36me3 partition skewed moderately toward leading-strand replication, consistent with H4K20me2 partition (Fig. 2A and fig. S5, C to E), and was stronger over active genes and codirectional with transcription (Fig. 2C) (8). Further, the correlation of chromatin interaction directionality with histone partition was weaker than the correlation between chromatin interaction directionality and RFD (Fig. 2B), suggesting that although histone partitioning is driven by RFD it can be affected by transcription. Active enhancers often coincided with initiation zone centers, and promoters tended to be flanking (20) (fig. S5F), suggesting that enhancer activity affects partitioning in neighboring regions. Consistently, both RFD and histone partition asymmetries were greater around active enhancers (21) and super-enhancers that control cell type–specific genes (22) (Fig. 2D) (Mann-Whitney U test, P < 1.1 × 10−16).

Fig. 2 TAD borders and genes flanking active enhancers show skewed histone PTM inheritance to sister chromatids.

(A) Average RFD and partition of H4K5ac, H4K20me2, and H3K36me3 across scaled TADs split by compartment class (16, 17). (B) Spearman correlation of Hi-C directionality index (16) with RFD, histone PTM partition [colors as in (A)], and transcriptional directionality measured by precision nuclear run-on (PRO-seq) (29). (C) Density distribution of H3K36me3 partition and RFD in active and inactive forward (red)- and reverse (blue)-strand genes and intergenic regions. (D) Average RFD and histone PTM partition centered at enhancers (21) and super-enhancers (22).

MCM2, part of the replicative helicase, is proposed to recycle parental histone H3-H4 via its N-terminal histone-binding domain (HBD) (11, 2325). Using genome editing, we mutated two critical residues in the HBD [Tyr81→Ala (Y81A) and Tyr90→Ala (Y90A), yielding MCM2-2A] (24, 25) (fig. S6A) that disrupt histone binding (fig. S6B) (25) without affecting cell cycle progression (fig. S6C). Notably, in MCM2-2A mutants, partition of old and new histones was strongly skewed toward leading and lagging strands, respectively, generating partition ratios similar to RFD in amplitude and pattern (Fig. 3, A to C, and figs. S6D and S7). Moreover, partition of new and old histones showed strong anticorrelation in MCM2-2A (Fig. 3D and fig. S7D), indicating segregation to opposite strands. H3K36me3 occupancy was not altered in MCM2-2A cells (fig. S8, A and B), indicating that histone partitioning rather than recycling was perturbed. The association between H3K36me3 partition and transcriptional directionality was reduced in MCM2-2A cells (fig. S8, C to E), further indicating increased replication-driven sister chromatid asymmetry in parental and new histones. Sister chromatid asymmetry was also strongly increased at TAD borders, around enhancers (fig. S8, F and G), and across important developmental loci (e.g., the Hox clusters) (fig. S9) in MCM2-2A cells, which thus provides a model to address histone PTM inheritance in development. The high correlation between H4K20me2 partition in MCM2-2A cells and RFD prompted us to map H4K20me2 partition break points. They showed strong colocalization with initiation zones mapped by OK-seq (Fig. 3E and methods), suggesting H4K20me2 SCAR-seq in MCM2-2A as a method to map replication dynamics.

Fig. 3 MCM2 histone binding is required for parental histone recycling to the lagging strand.

(A) Average RFD and partition of old (H4K20me2 and H3K36me3) and new (H4K5ac) histones in the wild type (WT) (solid lines) and MCM2-2A (dashed lines) around initiation zones. (B) RFD and histone PTM partition at a genomic region in the WT and MCM2-2A. Initiation zone centers (lines), active gene orientation (arrowheads), and active enhancers (bars) are shown. (C) Scatterplots of RFD and histone PTM partition in the WT and MCM2-2A. Spearman’s rank correlation coefficient is shown in the top left corner of each plot. (D) Scatterplot of H4K20me2 versus H4K5ac partition in MCM2-2A. Spearman’s rank correlation coefficient is shown in the top right corner. (E) Fraction of initiation zones with the nearest distance to predicted H4K20me2 partition break points or random H4K20me2 bins in the WT and MCM2-2A. Horizontal dotted lines represent random mean fractions. (F) Model for segregation of parental histones H3-H4 in WT and MCM2-2A cells.

In summary, SCAR-seq revealed that parental histone segregation is almost symmetrical, with a weak inherent preference for the leading strand (Fig. 3F, left), creating modest sister chromatid asymmetries that might be mitotically transmitted as new histones acquire PTMs with slow kinetics (5). Importantly, MCM2 histone chaperone activity promotes balanced segregation of old histones to leading and lagging strands, thereby ensuring inheritance of histone-based information to both sister chromatids. This is consistent with chaperoning of old histones by MCM2 (11, 23) and cryo–electron microscropy data placing the MCM2 HBD in front of the fork (26, 27). We envisage that MCM2 recycles parental histones to the lagging strand (Fig. 3F, right), whereas a separate pathway deposits parental histones on the leading strand. In this vein, it is conceivable that histone segregation can be regulated during development to drive asymmetric cell fates (28).

Supplementary Materials

www.sciencemag.org/content/361/6409/1389/suppl/DC1

Materials and Methods

Figs. S1 to S9

Tables S1 to S3

References (3042)

Data S1

References and Notes

Acknowledgments: We thank P. Lansdorp for initial discussions on SCAR-seq; K. Stewart-Morgan and C. Hammond for comments on the manuscript; A. Fossum for help with fluorescence-activated cell sorting; and the Groth, Andersson, and Brakebusch laboratories for discussions. Funding: Research in the Groth lab was supported by the Independent Research Fund Denmark (4092-00404), the European Research Council (CoG no. 724436), the Novo Nordisk Foundation (NNF14OC0012839), the Lundbeck Foundation (R198-2015-269), and the Danish Cancer Society. Research in the Andersson lab was supported by the Independent Research Fund Denmark (6108-00038B) and the European Research Council (StG no. 638173). Author contributions: N.P. and A.G. conceived of the project and designed experiments. N.P. developed and performed SCAR-seq. M.D. and R.A. developed computational methods. M.D. performed computational analyses with support from N.P. and R.A. C.B.S. and A.S. performed genome editing. A.W. characterized MCM2-2A cells and assisted with SCAR-seq. A.G. and R.A. supervised the project. N.P. and M.D. wrote the manuscript with input from A.G., R.A., and A.W. Competing interests: The authors declare no competing interests. Data and materials availability: Data have been deposited to NCBI GEO under accession number GSE117274.
View Abstract

Navigate This Article