NATURE Vol 447 14 June 2007 ARTICLES the striking difference in histone modification enrichments in TSSs The ENCODE Project provided a unique opportunity to examine residing near versus those more distal to CPG islands(see Fig. 5 and whether individual histone modifications on human chromatin can Supplementary Information section 3.6), including information be correlated with the time of replication and whether such correla This suggests that despite the marked differences in histone modifi- early replication. Our studies also tested whether segments showing cations among these TSS classes, a single predictor can be made, interallelic variation in the time of replication have two different sing the interactions between the different histone modification types of histone modifications consistent with an interallelic vari levels ation in chromatin state In summary, we have integrated many data sets to provide a more DNA replication data set. We mapped replication timing across the mplete view of regulatory information, both around specific sites ENCODE regions by analysing Brd-U-labelled fractions from syn (TSSs and DHSs)and in an unbiased manner From analysing mul- chronized Hela cells(collected at 2 h intervals throughout S phase tiple data sets, we find 4, 491 known and novel TSSs in the ENCODE on tiling arrays (see Supplementary Information section 4.1). regions, almost tenfold more than the number of established genes. Although the HeLa cell line has a considerably altered karyotype, This large number of TSSs might explain the extensive transcription correlation of these data with other cell line data(see below)suggests described above; it also begins to change our perspective about reg- the results are relevant to other cell types. The results are expressed as ulatory information--without such a large TSS catalogue, many of the time at which 50% of any given genomic position is replicated the regulatory clusters would have been classified as residing distal to (TR50), with higher values signifying later replication times. In add promoters. In addition to this revelation about the abundance of tion to the five 'activating histone marks, we also correlated the TR50 promoter-proximal regulatory elements, we also identified a consid- with H3K27me3, a modification associated with polycomb-mediated the basis of the presence of DHSs. Our study of distal regulatory framework, the histone data were smoothed to 100-kb resolution, elements was probably most hindered by the paucity of data gener- and then correlated with the TR50 data by a sliding window correla- ated using distal-element-associated transcription factors; neverthe- tion analysis(see Supplementary Information section 4.2). The less, we clearly detected a set of distal-DHS-associated segments continuous profiles of the activating marks, histone H3K4 mono- bound by CTCF or MYC. Finally, we showed that information about di-, and tri-methylation and histone H3 and H4 acetylation, ar hromatin structure alone could be used to make effective predic- generally anti-correlated with the TR50 signal(Fig. 7a and St tions about both the location and activity of TSSs. low a predominantly positive correlation with late-replicating seg Replication ments(Fig. 7a; see Supplementary Information section 4.3 for addi Overview. DNA replication must be carefully coordinated, both tional analysis across the genome and with respect to development. On a larger scale, Although most genomic regions replicate in a temporally specific early replication in S phase is broadly correlated with gene density window in S phase, other regions demonstrate an atypical pattern of and transcriptional activity-66; however, this relationship is not replication(Pan-S) where replication signals are seen in multiple universal, as some actively transcribed genes replicate late and vice parts of s phase. We have suggested that such a pattern of replication versa. Importantly, the relationship between transcription and stems from interallelic variation in the chromatin structure..If one DNA replication emerges only when the signal of transcription is allele is in active chromatin and the other in repressed chromatin, averaged over a large window(>100 kb), suggesting that larger- both types of modified histones are expected to be enriched in the scale chromosomal architecture may be more important than the Pan-S segments. An ENCODE region was classified as non-specifi ctivity of specific genes (or Pan-S)regions when >60% of the probes in a 10-kb window 1.6Mb 口Eary 3]32m3 00153000153200153400153600163800 Genomic positio Figure 7 Correlation between replication timing and histone odifications. a, Comparison of two histone modifications(H3K4me2 and H3K27me3), plotted and the time for 50% ofthe DNA to replicate(TR50), indicated for ENCODE region ENm006. The colours on the curves reflect the correlation strength in a sliding 250-kb window. b, Differing levels of histone modification for 807 E2007 Nature Publishing Groupthe striking difference in histone modification enrichments in TSSs residing near versus those more distal to CpG islands (see Fig. 5 and Supplementary Information section 3.6), including information about the proximity to CpG islands did not improve the predictors. This suggests that despite the marked differences in histone modifi￾cations among these TSS classes, a single predictor can be made, using the interactions between the different histone modification levels. In summary, we have integrated many data sets to provide a more complete view of regulatory information, both around specific sites (TSSs and DHSs) and in an unbiased manner. From analysing mul￾tiple data sets, we find 4,491 known and novel TSSs in the ENCODE regions, almost tenfold more than the number of established genes. This large number of TSSs might explain the extensive transcription described above; it also begins to change our perspective about reg￾ulatory information—without such a large TSS catalogue, many of the regulatory clusters would have been classified as residing distal to promoters. In addition to this revelation about the abundance of promoter-proximal regulatory elements, we also identified a consid￾erable number of putative distal regulatory elements, particularly on the basis of the presence of DHSs. Our study of distal regulatory elements was probably most hindered by the paucity of data gener￾ated using distal-element-associated transcription factors; neverthe￾less, we clearly detected a set of distal-DHS-associated segments bound by CTCF or MYC. Finally, we showed that information about chromatin structure alone could be used to make effective predic￾tions about both the location and activity of TSSs. Replication Overview. DNA replication must be carefully coordinated, both across the genome and with respect to development. On a larger scale, early replication in S phase is broadly correlated with gene density and transcriptional activity59–66; however, this relationship is not universal, as some actively transcribed genes replicate late and vice versa61,64–68. Importantly, the relationship between transcription and DNA replication emerges only when the signal of transcription is averaged over a large window (.100 kb)63, suggesting that larger￾scale chromosomal architecture may be more important than the activity of specific genes69. The ENCODE Project provided a unique opportunity to examine whether individual histone modifications on human chromatin can be correlated with the time of replication and whether such correla￾tions support the general relationship of active, open chromatin with early replication. Our studies also tested whether segments showing interallelic variation in the time of replication have two different types of histone modifications consistent with an interallelic vari￾ation in chromatin state. DNA replication data set. We mapped replication timing across the ENCODE regions by analysing Brd-U-labelled fractions from syn￾chronized HeLa cells (collected at 2 h intervals throughout S phase) on tiling arrays (see Supplementary Information section 4.1). Although the HeLa cell line has a considerably altered karyotype, correlation of these data with other cell line data (see below) suggests the results are relevant to other cell types. The results are expressed as the time at which 50% of any given genomic position is replicated (TR50), with higher values signifying later replication times. In addi￾tion to the five ‘activating’ histone marks, we also correlated the TR50 with H3K27me3, a modification associated with polycomb-mediated transcriptional repression70–74. To provide a consistent comparison framework, the histone data were smoothed to 100-kb resolution, and then correlated with the TR50 data by a sliding window correla￾tion analysis (see Supplementary Information section 4.2). The continuous profiles of the activating marks, histone H3K4 mono-, di-, and tri-methylation and histone H3 and H4 acetylation, are generally anti-correlated with the TR50 signal (Fig. 7a and Sup￾plementary Information section 4.3). In contrast, H3K27me3 marks show a predominantly positive correlation with late-replicating seg￾ments (Fig. 7a; see Supplementary Information section 4.3 for addi￾tional analysis). Although most genomic regions replicate in a temporally specific window in S phase, other regions demonstrate an atypical pattern of replication (Pan-S) where replication signals are seen in multiple parts of S phase. We have suggested that such a pattern of replication stems from interallelic variation in the chromatin structure59,75. If one allele is in active chromatin and the other in repressed chromatin, both types of modified histones are expected to be enriched in the Pan-S segments. An ENCODE region was classified as non-specific (or Pan-S) regions when .60% of the probes in a 10-kb window H3k27me3 1.6 Mb (ENm006) Enrichment Enrichment TR50 152,800 4 3 2 1 0 2.5 1.5 0.5 4.0 3.5 3.0 153,000 153,200 153,400 153,600 153,800 H3k4me2 Genomic position (kb) a bPer cent enrichment Early Mid Late Pan-S –80 –40 0 40 80 120 H3K27me3.HeLa H3K4me1.HeLa H3K4me2.HeLa H3K4me3.HeLa H3ac.HeLa H4ac.HeLa H3K4me1.GM H3K4me2.GM H3K4me3.GM H3ac.GM H4ac.GM Figure 7 | Correlation between replication timing and histone modifications. a, Comparison of two histone modifications (H3K4me2 and H3K27me3), plotted as enrichment ratio from the Chip-chip experiments and the time for 50% of the DNA to replicate (TR50), indicated for ENCODE region ENm006. The colours on the curves reflect the correlation strength in a sliding 250-kb window. b, Differing levels of histone modification for different TR50 partitions. The amounts of enrichment or depletion of different histone modifications in various cell lines are depicted (indicated along the bottom as ‘histone mark.cell line’; GM 5 GM06990). Asterisks indicate enrichments/depletions that are not significant on the basis of multiple tests. Each set has four partitions on the basis of replication timing: early, mid, late and Pan-S. NATURE| Vol 447|14 June 2007 ARTICLES 807 ©2007 NaturePublishingGroup