Nature | Vol 577 | 23 January 2020 | 577Supplementary Table 1). We confirmed the enrichment of ORC1, ORC2
and MCM2 by western blotting (Fig. 1d).
As ORC1 might stabilize the binding of other ORC subunits at origins
during G1 phase^10 , we tested the interaction between H2A.Z and ORC1
using the LacO/LacI targeting system. However, we did not find a direct
interaction (Extended Data Fig. 1g), suggesting that a bridge is required
to recruit ORC1 onto H2A.Z nucleosomes. The bromo adjacent homol-
ogy (BAH) domain of ORC1 has been reported to specifically recognize
the H4K20me2 peptide^5 , suggesting that ORC1 may be recruited onto
H2A.Z nucleosomes through H4K20me2. Indeed, we found that both
H4K20me2 and SUV420H1 were enriched on H2A.Z nucleosomes, in
both unsynchronized (Fig. 1e) and G1-synchronized (Extended Data
Fig. 1h) cells. Mass-spectrometry analysis of H2A and H2A.Z mononu-
cleosomes showed that whereas H4K20me2 is abundant on both H2A
and H2A.Z nucleosomes, it is relatively more enriched on the H2A.Z
variant (Fig. 1f and Extended Data Fig. 2a, b). In addition, knockdown of
SUV420H1 alone or of both SUV420H1 and SUV420H2 (which encode the
two enzymes^11 that catalyse methylation of H4K20me2) abolished the
enrichment of ORC1, ORC2 and MCM2 on H2A.Z nucleosomes (Fig. 1g
and Extended Data Fig. 2c–e). These results suggest that H2A.Z recruits
ORC1 in an H4K20me2-dependent manner. However, knockdown of
SUV420H2 alone had little effect on the binding of ORC1 onto H2A.Z
nucleosomes (Extended Data Fig. 2d, e).H2A.Z recruits SUV420H1 to deposit H4K20me2
Methyltransferase assays showed that, compared with H2A mononu-
cleosomes, H2A.Z mononucleosomes greatly enhanced the histone-
methylation activity of recombinant human SUV420H1 (Fig. 2a and
Extended Data Fig. 3a). Mass-spectrometry analysis of the modifica-
tion products from the histone-methyltransferase reactions revealed
that the main product of SUV420H1 activity on H2A.Z nucleosomes
is H4K20me2, and that this is much more common on H2A.Z than on
H2A nucleosomes (Extended Data Fig. 3b). We validated this result
by western blotting (Extended Data Fig. 3c). It has been reported that
SUV420H1 produces H4K20me2 from H4K20me1 in vivo^12. In line with
this, we found that H2A.Z also promoted the activity of SUV420H1 on
nucleosomes containing H4KC20me1 (with KC being a lysine replaced
by cystine for chemical modification; Extended Data Fig. 3d).
Next, we generated four chimaeric mutants of H2A.Z, containing
regions that had been replaced by the corresponding regions of H2A
(Extended Data Fig. 3e). When the acidic patch of H2A.Z was substi-
tuted with the corresponding domain of H2A, the activity of SUV420H1
was markedly reduced (Extended Data Fig. 3e). The residues D97 and
S98 in the acidic patch have been reported^13 ,^14 to be critical for the
structural and biological functions of H2A.Z. Our results show that
SUV420H1 has very low methylation activity on H2A.ZD97N/S98K mutant
nucleosomes (Fig. 2b). Using mononucleosome pulldown assays, we
found that SUV420H1 binds more strongly to H2A.Z nucleosomes
than to H2A nucleosomes (Fig. 2c). Moreover, mutation of D97 and
S98 to N97 and K98 in H2A.Z impaired the binding of SUV420H1 to
nucleosomes (Extended Data Fig. 3f ). We confirmed the binding of
SUV420H1 to H2A.Z nucleosomes in vivo using the LacO/LacI target-
ing system (Extended Data Fig. 3g, h). Next, we simulated the binding
between SUV420H1 and H2A.Z mononucleosomes using structural
data for SUV420H1 (ref.^15 ) and the H2A.Z nucleosome^16. We found that
the R257 and K333 residues of SUV420H1 are important for the interac-
tion with H2A.Z nucleosomes (Extended Data Fig. 4a, b). Indeed, both
SUV420H1R257A and SUV420H1K333A mutants could not bind H2A.Z nucle-
osomes as efficiently as wild-type SUV420H1 (Fig. 2d). The methylation
activity of these two mutants on H2A.Z nucleosomes was also reduced
(Extended Data Fig. 4c). Together, these data show that residues D97
and S98 of H2A.Z, and R257 and K333 of SUV420H1, are essential for
the binding and enhancing activity of SUV420H1.
A pulldown assay of biotinylated mononucleosomes showed that
binding of ORC1 to H2A.Z nucleosomes is substantially enhanced
by SUV420H1-catalysed H4K20me2 (Fig. 2e). Moreover, ORC1 binds
weakly to the histone-methyltransferase products of H2A.Z/H4K20A
nucleosomes (Fig. 2f) or H2A.ZD97N/S98K nucleosomes (Fig. 2g). We con-
firmed that the enrichment of H4K20me2, ORC1, ORC2 and MCM2,
on H2A.Z nucleosomes is greatly impeded by H2A.ZD97N/S98K mutation
in vivo (Fig. 2h). Using H2A.Z nucleosomes containing H4KC20me2
as substrates (Extended Data Fig. 4d), we found that mutations in the
BAH domain of ORC1, which abolish the interaction between ORC1 and
H4K20me2 (ref.^5 ), greatly impaired the interaction between ORC1 and
H2A.Z nucleosomes containing H4KC20me2 (Extended Data Fig. 4e).
These results support the idea that the binding of ORC1 to H2A.Z
nucleosomes depends on the BAH domain of ORC1 and on SUV420H1-
catalysed H4K20me2.
To test whether the density of H4K20me2 has an effect on ORC1
binding, we first assembled H2A and H2A.Z mononucleosomes con-
taining zero (unmodified H4), one (heterotypic, 50%) or two (homo-
typic, 100%) H4 histones with the H4KC20me2 modification. As the
H4KC20me2 density increased, ORC1 bound more strongly to H2A andd fMCM2Input Flag-IP
ORC1
ORC2H4
Flag
gORC1
ORC2FlagH4MCM2shGFP shSUV420H 1Input
shGFP shSUV420H 1Flag-IP0204060801000246810121416Percentageof cellsRelease from G2/M (h)G1 – IAA
S – IAA
G2 – IAA
S + IAA
G2 + IAAG1 + IAAbcORC1
ORC2MCM 4
MCM 6MCM 2PCNAMCMMCM 53 MCM 7RPAH2A (log 10 (n+0.5)H2A.Z (log(n 10
+0.5))01201(^290) P = 0.0369
80
70
60
50
K20me2/K20 (%
)
H2AH2A.Z
H2A H2A.ZH2A H2A.Z
H2AH2A.ZH2AH2A.Z
H2AH2A.ZH2AH2A.Z H2AH2A.ZH2AH2A.Z
100
e
H4K20me2
H4K20me1
SUV420H 1
H4
Flag
H3K36me3
H4K20me3
Input Flag-IP
a
siNC
siH2A.Z
OD
450
Time (days)
siNCsiH2A.
Z
H4
H2A.Z
0
0.5
1
1.5
2
2.5
01234
Fig. 1 | H2A.Z interacts with the pre-replication complex. a, Analysis of cell
proliferation (left) and western blots (right) for HeLa cells transfected with
negative-control short interfering (si) RNA (siNC) or siRNA that targets H2A.Z
(siH2A.Z). OD 450 , optical density at 450 nm (an indicator of cell density). b, FACS
analysis of the cell-cycle progression of control cells (−IA A; solid lines) and IA A-
treated cells (+IA A; dash lines). The yellow shading shows that the cells were
arrested at G1 phase and delayed proceeding into S phase after IA A-induced
H2A.Z depletion. c, The total number of peptides identified from the
immunoprecipitation of Flag-tagged H2A or H2A.Z nucleosomes from three
independent experiments, plotted as log10(n + 0.5) with jitter. The diagonal
line represents the threshold of twofold enrichment on H2A.Z nucleosomes.
d, e, Western blot analysis of ORC1, ORC2 and MCM2 (d) or histone
modifications and SUV420H1 (e) from immunoprecipitation of Flag-tagged
H2A or H2A.Z mononucleosomes. f, Mass-spectrum analysis of the H4K20me2
modification from the samples in Fig. 1e. The y axis shows the percentage of
H4K20 peptides that are modified as H4K20me2. g, Western blot analysis of
ORC1, ORC2 and MCM2 from immunoprecipitation of Flag-tagged H2A or
H2A.Z mononucleosomes from cells stably expressing a control short hairpin
(sh) RNA (shGFP) or shRNA against SUV420H1 (shSUV420H1). GFP, green
f luorescent protein. Data in a, b are mean ± s.d.: a, n = 6 technical replicates;
b, n = 3 biological replicates. Data in f are mean ± s.e.m.; n = 4 biological
replicates, two-tailed, paired t-test. Western blots in a, d, e, g were
independently repeated three times with similar results, and H4 was used as a
loading control and sample processing control. For gel source data, see
Supplementary Fig. 1. For the FACS gating strategy, see Supplementary Fig. 3.