pifications. Chromosome breakpoints detected
in T cell cancers frequently reside in chromatin
regions associated with H3K4 methylations.
Our data provide new insights into the function
of histone methylation and chromatin organiza-
tion in genome function.
INTRODUCTION
Eukaryotic DNA is packaged into a chromatin structure
consisting of repeating nucleosomes formed by wrapping
146 base pairs of DNA around an octamer of four core his-
tones (H2A, H2B, H3, and H4). The histones, particularly
their N-terminal tails, are subject to a large number of
posttranslational modifications (Kouzarides, 2007). His-
near the 3
0
end of genes.
Significant progress has also been made in characteriz-
ing global levels of histone modifications in mammals.
Several large-scale studies have revealed interesting in-
sights into the complex relationship between gene ex-
pression and histone modifications. Generally, high levels
of histone acetylation and H3K4 methylation are detected
in promoter regions of active genes (Bernstein et al., 2005;
Kim et al., 2005; Roh et al., 2005, 2006), whereas elevated
levels of H3K27 methylation correlates with gene repres-
sion (Boyer et al., 2006; Lee et al., 2006; Roh et al.,
2006). In addition to the promoter regions, these modifica-
tions are also detected in intergenic regions as both
sharply localized peaks and wide-spread domains. The
H3 acetylation and H3K4me1 signals outside of promoter
regions have been correlated with functional enhancers inResource
High-Resolution Profi
Methylations in the H
Artem Barski,
1,3
Suresh Cuddapah,
1,3
Kairong Cui,
1,3
Ta
Gang Wei,
1,3
Iouri Chepelev,
2
and Keji Zhao
1,
*
1
Laboratory of Molecular Immunology, National Heart, Lung, and
2
Department of Human Genetics, Gonda Neuroscience and Gen
Los Angeles, CA 90095, USA
3
These authors contributed equally to this work and are listed al
*Correspondence: zhaok@nhlbi.nih.gov
DOI 10.1016/j.cell.2007.05.009
SUMMARY
Histone modifications are implicated in influ-
encing gene expression. We have generated
high-resolution maps for the genome-wide
distribution of 20 histone lysine and arginine
methylations as well as histone variant H2A.Z,
RNA polymerase II, and the insulator binding
protein CTCF across the human genome using
the Solexa 1G sequencing technology. Typical
patterns of histone methylations exhibited
at promoters, insulators, enhancers, and tran-
scribed regions are identified. The mono-
methylations of H3K27, H3K9, H4K20, H3K79,
and H2BK5 are all linked to gene activation,
whereas trimethylations of H3K27, H3K9, and
H3K79 are linked to repression. H2A.Z associ-
ates with functional regulatory elements, and
CTCF marks boundaries of histone methylation
domains. Chromosome banding patterns are
correlated with unique patterns of histone mod-tone modifications are implicated in influencing gene ex-
pression and genome function by establishing global
chromatin environments and orchestrating DNA-basedling of Histone
uman Genome
e-Young Roh,
1,3
Dustin E. Schones,
1,3
Zhibin Wang,
1,3
Blood Institute, NIH, Bethesda, MD 20892, USA
etics Research Center, University of California, Los Angeles,
habetically.
biological processes. Among the various modifications,
histone methylations at lysine and arginine residues are
relatively stable and are therefore considered potential
marks for carrying the epigenetic information that is stable
through cell divisions. Indeed, enzymes that catalyze the
methylation reaction have been implicated in playing crit-
ical roles in development and pathological processes.
Remarkable progress has been made during the past
few years in the characterization of histone modifications
on a genome-wide scale. The main driving force has
been the development and improvement of the ‘‘ChIP-
on-chip’’ technique by combining chromatin immunopre-
cipitation (ChIP) and DNA-microarray analysis (chip). With
almost complete coverage of the yeast genome on DNA
microarrays, its histone modification patterns have been
extensively studied. The general picture emerging from
these studies is that promoter regions of active genes
have reduced nucleosome occupancy and elevated his-
tone acetylation (Bernstein et al., 2002, 2004; Lee et al.,
2004; Liu et al., 2005; Pokholok et al., 2005; Sekinger
et al., 2005; Yuan et al., 2005). High levels of H3K4me1,
H3K4me2, and H3K4me3 are detected surrounding tran-
scription start sites (TSSs), whereas H3K36me3 peaksvarious cell types (Heintzman et al., 2007; Roh et al., 2005;
Roh et al., 2007). The apparently opposite modifications,
H3K4me3 and H3K27me3, colocalize in regions termed
Cell 129, 823–837, May 18, 2007 ª2007 Elsevier Inc. 823

‘‘bivalent domains’’ in embryonic stem cells, which have
been suggested to function in the differentiation of these
cells (Bernstein et al., 2006). Similar domains also exist
in differentiated T cells (Roh et al., 2006). Although H3K9
methylation has been implicated in heterochromatin
formation and gene silencing, a large-scale analysis sug-
gested that H3K9me3 is enriched in many active pro-
moters (Squazzo et al., 2006). However, extensive loca-
tion analysis of only a few histone methylations has been
completed at a limited genome coverage and resolution
in human cells (Bernstein et al., 2007). Comprehensive
and high-resolution colocalization analysis of all histone
methylations for the entire human genome is required to
understand the functional correlation of various histone
lysine and arginine methylations in processes such as
transcription and DNA repair.
ChIP-on-chip has a number of limitations, including the
large sets of arrays needed to cover the mammalian ge-
nome and the potential bias introduced by amplification
(Bernstein et al., 2007). We have previously successfully
combined ChIP with serial analysis of gene expression
(SAGE) in a method termed genome-wide mapping tech-
nique (GMAT) (Roh et al., 2004), to map several histone H3
modifications in human T cells, including H3 K9/K14 di-
acetylation (Roh et al., 2005), H3K4 trimethylation, and
H3K27 trimethylation (Roh et al., 2006). However, the
method is limited by a relatively low resolution of 500–
1000 bp and the considerable cost of sequencing in order
to cover the entire genome. These limitations made a high-
resolution map of histone modifications in the human
genome unapproachable. To overcome these problems,
we have employed the Solexa 1G Genome Analyzer,
which performs massive parallel signature sequencing
on unique molecular arrays, to directly sequence ChIP
DNA from mononucleosomes generated by microccocal
nuclease (MNase) digestion of native chromatin. We dem-
onstrate that this is a comprehensive, quantitative, and
cost-effective method to analyze histone modification
and protein target sites in large genomes such as the hu-
man genome. Our data indicate novel functions of histone
methylation and chromatin organization in gene expres-
sion. The data are made available at http://dir.nhlbi.nih.
gov/papers/lmi/epigenomes/hgtcell.html.
RESULTS
Direct Sequencing Analysis of ChIP DNA Samples
Using Solexa 1G Genome Analyzer (ChIP-Seq)
To resolve histone modification signals to individual nucle-
osomes, we used mononucleosome templates generated
by MNase digestion of native chromatin for ChIP. All the
ChIP samples were confirmed using real-time quantitative
PCR analysis of known target sites before further analysis
using the Solexa 1G Genome Analyzer. The sequencingprocedure requires a one-step adaptor ligation and limited
PCR amplification (17 cycles) of ChIP DNA molecules, fol-
lowed by cluster generation and sequencing-by-synthesis
(Figure 1A). The image files generated by the analyzer are
824 Cell 129, 823–837, May 18, 2007 ª2007 Elsevier Inc.processed to produce DNA sequence data using the
Solexa Analysis Pipeline (see Experimental Procedures).
One sequencing run of this procedure can generate
more than 20 million sequence tags of up to 36 bp each.
A large majority of these short tags contain sufficient se-
quence information to be mapped unambiguously to the
human genome. To distinguish between this procedure
and our previous procedure, GMAT, or other similar pro-
cedures that incorporate the SAGE protocol, we have
termed this direct sequencing procedure ‘‘ChIP-Seq.’’
Since the ChIP-Seq method is analogous to direct count-
ing of the molecules in the ChIP DNA samples, it requires
minimal normalization. The number of tags detected for
a particular nucleosome is directly proportional to the
modification level of that nucleosome.
To determine the reproducibility and reliability of ChIP-
Seq, we compared two independent experiments carried
out using the H3K4me3 antibody by scatter analysis. The
high degree of correlation shown in Figure 1B indicates
that ChIP-Seq is indeed a reliable method for analyzing
histone modifications in the human genome. To further
determine whether ChIP-Seq can reproduce our previous
GMAT results (Roh et al., 2006), we compared the
H3K4me3 distribution patterns at a locus on chromosome
19 obtained from both methods. The data obtained using
ChIP-Seq are largely consistent with the GMAT data
(Figure 1C). However, the resolution with ChIP-Seq was
dramatically higher. Furthermore, ChIP-Seq was more
sensitive and generated less false-negative regions. For
example, an H3K4me3 modification peak was found lo-
calized to the enhancer region downstream of the IL-13
gene using ChIP-Seq, while no signal was detected using
GMAT (data not shown). The IL-2 gene, which is rapidly in-
duced by T cell receptor signaling, exhibited a strong peak
of H3K4me3 signals about 1.5 kb upstream of its TSS with
the ChIP-Seq analysis, whereas only a single copy tag
was detected in the same region using GMAT (Figure 1D).
The improved sensitivity and specificity can be attributed
to using native mononucleosomes for ChIP and obtaining
10-fold more tags.
Correlation of Histone Methylations at Promoter
Region with Gene Expression
Gene promoters near transcription start sites contain crit-
ical regulatory elements necessary for transcription. Previ-
ous large-scale and genome-wide analyses have revealed
that several histone modifications including H3K4 methyl-
ation and H3K9 acetylation are enriched in promoter
regions, whereas H3K36me3 is elevated in the transcribed
regions of active genes. To evaluate the contribution of
other histone methylations and histone variants in human
cells, we have now determined the genome-wide distribu-
tion at single-nucleosome resolution of 20 histone lysine
and arginine methylations; one histone variant (H2A.Z);RNA polymerase II (Pol II); and the insulator binding
protein, CTCF. To correlate these modifications with
gene transcription, 12,726 human genes, whose expres-
sion levels in human resting CD4
+
T cells are known