578 | Nature | Vol 577 | 23 January 2020
Article
H2A.Z nucleosomes (Extended Data Fig. 4f ). Next, we assembled H2A.Z
polynucleosomes with an increasing ratio of an H4KC20me2 octamer.
We found that ORC1 binding was weak when the density of H4KC20me2
was 25%, increasing gradually with increasing density of H4KC20me2
(from 50% to 100%) (Extended Data Fig. 4g). These results suggest that
ORC1 binds to chromatin in an H4K20me2-dosage-dependent manner.
Thus, H2A.Z binds SUV420H1 directly to promote H4K20me2 deposi-
tion on H2A.Z nucleosomes, and the enhanced H4K20me2 deposition
is essential for recruiting ORC1 to H2A.Z nucleosomes.
H2A.Z controls replication origin firing
To investigate how H2A.Z regulates H4K20me2 deposition and ORC1
binding at the genome-wide level, we used chromatin immunoprecipita-
tion with DNA sequencing (ChIP-seq) to map 58,642, 99,574 and 100,917
peaks for H2A.Z, H4K20me2 and ORC1, respectively, in HeLa cells
(Fig. 3a). The H4K20me2 ChIP-seq signal was markedly reduced after
both SUV420H1 and SUV420H2 were knocked down (Extended Data
Fig. 5a). H4K20me1 and H4K20me3 partially overlap with H4K20me2
(Extended Data Fig. 5b), and their ChIP-seq signal increased slightly at
the H4K20me2 peak regions after H2AFZ knockdown (Extended Data
Fig. 5c). These results validate the specificity of our H4K20me2 ChIP-seq
data. The peaks of H2A.Z, H4K20me2 and ORC1 overlap highly with each
other (Fig. 3a), and both H4K20me2 and ORC1 levels correlate positively
with the H2A.Z level (Extended Data Fig. 5d, e). Moreover, after H2AFZ
knockdown, the H4K20me2 and ORC1 levels decrease noticeably at
the H2A.Z peak regions that overlap with both H4K20me2 and ORC1
(Fig. 3b, c). Mass-spectrometry analysis of H4K20me2 after H2AFZ
knockdown validates the global decrease in H4K20me2 (Extended
Data Fig. 5f ). These results support the critical role of H2A.Z in regu-
lating H4K20me2 and ORC1 levels genome wide. Given that ORC1 is
degraded during S phase^17 , we analysed ORC1 binding to chromatin in
G1/S-phase-arrested cells. The results show that the chromatin fraction
of ORC1—and levels of SUV420H1, MCM2 and H4K20me2—decreased
after H2AFZ knockdown (Extended Data Fig. 5g), excluding an effect of
cell-cycle change on ORC1 binding after H2AFZ knockdown.
To investigate whether H2A.Z regulates the firing of replication ori-
gins, we mapped the active replication origins in HeLa cells by nascent-
strand sequencing (NS-seq)^18. We found that treatment with RNase A
markedly reduced the nascent-strand signal (Extended Data Fig. 5h).
Genome-wide, we detected 41,850 nascent-strand peaks (normalized
by the nascent-strand signal following treatment with RNase A), 47.2%
of which co-localized with H2A.Z, H4K20me2 and ORC1 simultaneously
(Fig. 3a). Moreover, this group of nascent-strand peaks had a higher
read density than other nascent-strand peaks (Fig. 3a). The nascent-
strand signal also decreased notably after H2AFZ knockdown (Fig. 3b,
c), suggesting that H2A.Z is essential for origin firing.
To further analyse the regulatory role of H2A.Z in origin firing, we
defined the H2A.Z-regulated H4K20me2 peaks as ‘ZD-K20’ (H2A.Z-
dependent-H4K20me2) (Extended Data Fig. 5i), and the remaining
H4K20me2 peaks as ‘ZI-K20’ (H2A.Z-independent-H4K20me2). We
also defined ‘ZD-ORC1’ (H2A.Z-dependent-ORC1) and ‘ZI-ORC1’ (H2A.Z-
independent-ORC1) (Extended Data Fig. 5j). Remarkably, we found
that 73.3% of ZD-K20 peaks overlap with 45.7% of ZD-ORC1 peaks, and
that 70.2% of the nascent-strand peaks overlap with ZD-K20 and ZD-
ORC1 simultaneously (Fig. 3d). We further defined ZD-ORC1 peaks thatSUV420H1
H4SUV420H1
H4DNAH2A H2A.ZH4H4H2AH2A. ZH2AH2A.Z
H2AH2A.ZWild typeR257AK333ASUV420H1SUV420H1
Input InputInput Flag-IPORC1
H4K20me2
ORC2
MCM2
H4
Flag++++++++++++++++++++++++H2A –
H2A.Z
SUV420H 1
SAM
ORC1
ORC1
H4ORC1
H4InputH4K20me2++++++++++++++++++++++++H2A.Z –
H2A.Z/H4K20A
SUV420H 1
SAM
ORC1
ORC1
H4ORC1
H4H4K20me2ORC1
H4ORC1
H4H4K20me2Histone
methyltransferase
productH2AH2A.ZSUV420H (^1) SAM
H2A.Z
mononucleosome
Biotin pulldownORC1
H4K20me2
e fgh
b d
H4
H4
H4
H4
H2AH2A.Z
SUV420H1:
a
(^3) H
autograph
Commassie
blue
(^3) H
autograph
Commassie
blue
0 × 1 × 2 ×
1 × 2 × 1 × 2 ×
0 × 1 × 2 × H2A.
ZD97N/S98K
H2AH2A.
Z
Mut3H2A.
ZD97N
H2A.
ZS98K
H2A. Z
D97N/S98K
H2A
.ZD97N/S
98K
H2AH2A.
Z
H2A. Z
D97N/
S98K
H2AH2A.
Z
H2A.
ZD97N/S98K
H2AH2A.ZMut3
H2A
.ZD97N
H2A.
ZS98
K
c
CP
M si
gnal
(×^10
3 )
CP
M si
gnal
(×^10
3 )
P < 1 × 10 –4
H2A H2A.Z
P < 1 × 10 –4
0
2
4
8
10
6
P < 1 × 10 –4
0
2
4
8
10
6 P < 1 ×^10 –4
BiotinPulldow
n
BiotinPulldow
n
BiotinPulldow
n
Inpu
t
BiotinPulldow
n
Inpu
t
BiotinPulldow
n
Fig. 2 | H2A.Z binds SUV420H1 to promote H4K20me2 deposition and
thereby recruit ORC1. a, b, Upper panels,^3 H autograph showing
methyltransferase activity of SUV420H1 on wild-type H2A or H2A.Z
mononucleosomes (a) and on mononucleosomes containing H2A.Z point
mutations (b). Bottom, quantitative analysis of the^3 H signal by liquid
scintillation. Mut3 is an H2A.Z mutant whose acid patch is replaced with the
corresponding region of H2A (Extended Data Fig. 3e). CPM, counts per minute.
c, Western blot analysis of SUV420H1 from biotin pulldown samples. d, Western
blot analysis of wild-type and mutant SUV420H1 from mononucleosome biotin
pulldown samples. e, Left, diagram showing the histone methyltransferase
activity of SUV420H1 on nucleosomes (top) and the subsequent binding of
ORC1 to the products of SUV420H1’s methyltransferase activity, detected by
biotin pulldown. SAM, S-adenosyl methionine. Right, western blot analysis of
the binding of ORC1 to SUV420H1 products. f, Western blot analysis of the
binding of ORC1 to SUV420H1’s histone methyltransferase products, using
H2A.Z mononucleosomes containing wild-type H4 or mutant H4K 20A as
substrates. g, Western blot analysis of the binding of ORC1 to SUV420H1
products, using wild-type H2A.Z or H2A.ZD97N/S98K mononucleosomes as
substrates. h, Western blots showing the distribution of ORC1, ORC2, MCM2
and H4K20me2 on H2A, H2A.Z and H2A.ZD97N/S98K mononucleosomes.^3 H
quantification data in a, b are mean ± s.e.m.; n = 3 biological replicates; two-
tailed, unpaired t-test. The^3 H autograph experiments in a, b and western blots
in c–h were independently repeated three times with similar results. H4 was
used as a loading control and sample processing control. For gel source data,
see Supplementary Fig. 1.