The EMBO Journal vol.10 no.4 pp.971 -980, 1991
Stepwise assembly of chromatin during DNA replication
Susan Smith' and Bruce Stillman
Cold Spring Harbor Laboratory, PO Box 100, Cold Spring Harbor,
NY 11724, USA
'Present address: Laboratory of Cell Biology, Howard Hughes Medical
Institute, The Rockefeller University, NY 10021, USA
Communicated by J.D.Watson
A cell free system that supports replication-dependent
chromatin assembly has been used to determine the
mechanism of histone deposition during DNA replication.
CAF-I, a human cell nuclear factor, promotes chromatin
assembly on replicating SV40 DNA in the presence
of a crude cytosol replication extract. Biochemical
fractionation of the cytosol extract has allowed separation
ofthe chromatin assembly reaction into two steps. During
the first step, CAF-I targets the deposition of newly
synthesized histones H3 and H4 to the replicating DNA.
This reaction is dependent upon and coupled with DNA
replication, and utilizes the newly synthesized forms of
histones H3 and H4, which unlike bulk histone found in
chromatin, do not bind to DNA by themselves. The
H3/H4-replicated DNA complex is a stable intermediate
which exhibits a micrococcal nuclease resistant structure
and can be isolated by sucrose gradient sedimentation.
In the second step, this replicated precursor is converted
to mature chromatin by the addition ofhistones H2A and
H2B in a reaction that can occur after DNA replication.
The requirement for CAF-I in at least the first step of
the reaction suggests a level of cellular control for this
Key words: chromatin/DNA replication/histones/simian virus
The replication of eukaryotic chromosomes requires not only
the duplication of the DNA, but also accurate reproduction
of the associated chromatin
indicate that specific chromatin structures can influence the
transcriptional capability of the cell (reviewed by van Holde,
1988) and in certain cases, these chromatin structures are
propagated during DNA replication (Weintraub,
Groudine and Weintraub, 1982). Since DNA replication
disrupts the structure ofchromatin, it provides an opportunity
to either change or maintain specific structures and hence,
influence the transcriptional state of the cell. Very little
is known, however, about the molecular mechanisms
involved in propagating and changing chromatin structures.
Understanding how chromosomal proteins are deposited
during DNA replication will provide insights into this
chromatin, assembled on newly replicated DNA? The
©) Oxford University Press
nucleosome core particle, the basic subunit of chromatin,
consists of two copies of each of the four core histones,
H2A, H2B, H3 and H4, organized into an octameric
structure. Electron microscopic analyses of replicating
cellular (McKnight and Miller, 1977) and viral (Cremisi et
al., 1978; Seidman et al., 1978) chromosomes indicate that
nucleosomal structures are assembled on replicating DNA
immediately after passage of the replication fork. It appears,
however, that the structure of newly synthesized chromatin
is somehow different from that ofmature chromatin. Nascent
chromatin is more sensitive than mature chromatin to
micrococcal nuclease digestion, but over time, normal
Hildebrand and Walters, 1976; Levy and Jakob, 1978;
Klempnauer et al., 1980; Cusick et al., 1981, 1983). The
biochemical events that accompany this maturation process,
however, are not well understood.
To study the mechanism ofhistone deposition during DNA
replication, investigators have relied on in vivo studies which
follow the fate of newly synthesized histones that are
incorporated into chromatin during DNA replication. These
studies have yielded an incomplete and somewhat confusing
picture, and to date a single mechanism for histone deposition
is still not generally accepted (Svaren and Chalkley, 1990).
Early studies, which monitored the incorporation of pulse-
labeled histones, indicated a sequential mechanism for
histone deposition where newly synthesized histones H3 and
H4 deposited first on newly replicated DNA, followed by
deposition of new histones H2A and H2B (Worcel et al.,
1978; Cremisi and Yaniv,
performed nearly a decade later, demonstrated that nascent
chromatin consisted of nucleosomes containing newly
synthesized histones H3 and H4 and mostly old histones
H2A and H2B (Jackson, 1987, 1988). These more recent
observations have compromised the interpretation of the
earlier studies because the presence of old (and therefore
unlabeled in pulse-chase experiments) histones H2A and
H2B in newly replicated chromatin would have gone
undetected. In addition, although most studies agree that
newly synthesized histones H3 and H4 deposit exclusively
on newly replicated DNA (Cremisi et al., 1977; Worcel
et al., 1978; Senshu et al., 1978; Cremisi and Yaniv, 1980;
Jackson and Chalkley, 1981a,b, 1985), the mechanism that
targets histones H3 and H4 to the replicating DNA remains
To address the mechanism of chromatin assembly during
DNA replication, we have developed an in vitro system
which closely mimics the in vivo process. Cytosol extracts
derived from human cells support complete and authentic
replication of SV40 origin-containing plasmid DNA in the
presence ofpurified SV40 T antigen (for review see Stiliman,
1989; Challberg and Kelly, 1989). Addition of chromatin
assembly factor I (CAF-I), a multisubunit histone binding
protein isolated from nuclei of human cells, promotes
chromatin assembly during DNA replication (Stillman, 1986;
is recovered (Seale,
1980). Subsequent studies,
S.Smith and B.Stillman
Smith and Stillman, 1989). This in vitro reaction resembles
the in vivo process in at least two important ways. First,
the cytosol histones used in the assembly reaction correspond
to the newly synthesized histone pool and hence are the direct
for chromatin assembly
chromatin assembly in this reaction is dependent upon and
coupled with DNA replication and therefore is likely to
reflect the events that occur during DNA replication in vivo,
when de novo chromatin assembly occurs.
Biochemical fractionation has allowed the separation of
the chromatin assembly reaction into two steps. During the
first step of the reaction, CAF-I deposits newly synthesized
histones H3 and H4 on the DNA in a reaction that
discriminates between replicated and unreplicated DNA. This
replicated precursor then serves as the template for the
deposition of either old or new histones H2A and H2B. We
present a model for the stepwise assembly of chromatin and
discuss possible mechanisms for targeting histone deposition
to replicating DNA.
Biochemical fractionation of the histones in the
In the SV40 in vitro replication system, the cytosol
for DNA replication as well as the four core histones
needed for chromatin assembly. To identify other chromatin
assembly factors and to investigate the mechanism ofhistone
deposition during DNA replication, we attempted to separate
the histones from the replication components through
biochemical fractionation (Figure IA). The cytosol extract
chromatography; a flow-through fraction at 0.6 M NaCl
('0.6M') and a bound fraction, eluted with 2 M NaCl ('2M').
Based upon previous studies we predicted that under these
conditions the replication components would flow through
the column (Tsurimoto and Stillman, 1989), whereas the
all of the cellular proteins required
fractions by phosphocellulose
histones, relatively tight DNA-binding proteins, should
remain bound. This fractionated system was tested for its
ability to support replication-dependent chromatin assembly.
The SV40 replication reaction consists of the cytosol
extract, T antigen, SV40 origin-containing plasmid DNA and
Topoisomerases I and II, found in the cytosol in limiting
amounts, were also added to the reaction. When the DNA
products of this reaction were isolated and analyzed by
agarose gel electrophoresis, autoradiography ofthe replicated
DNA revealed predominantly relaxed monomer circle DNA
(Figure 1B, lane 1). When purified CAF-I was included
in the reaction, the replicated DNA was assembled into
chromatin. This was visualized, after deproteinization, as
supercoiled DNA (Figure iB, lane 2). A comparison of the
replicated DNA (autoradiograph, Figure iB, left panel) with
the total input DNA (ethidium bromide stained gel, Figure
lB, right panel), demonstrated that while all ofthe replicated
DNA was assembled into chromatin, the input, unreplicated
DNA remained relaxed, indicating the preferential assembly
of chromatin on replicating DNA. Similar to the cytosol
extract, the 0.6M flow-through supported DNA replication
(Figure IC, lane 1). In contrast, however, addition ofCAF-I
to the 0.6M flow-through did not result in chromatin
assembly (Figure IC, lane 2), demonstrating that while the
0.6M extract contained all the components required for DNA
replication, it had been depleted of one or more factors
needed for chromatin assembly. These factors were in the
2M step fraction, since addition of this extract restored
chromatin assembly activity (Figure IC, lane 4).
We considered it likely that, because of their ability to
bind tightly to DNA, the histones might be retained on the
phosphocellulose column at 0.6 M NaCl. To determine the
fate of the cytosolic histones during the fractionation scheme
described in Figure 1, the proteins in the extracts were
subjected to analysis by two-dimensional gel electrophoresis.
To detect the very small amounts of histones contained in
the cytosol, fractionated extracts were prepared from cells
Fig. 1. Biochemical fractionation and reconstitution of the chromatin assembly reaction. (A) Fractionation scheme. Human 293 cells were fractionated
into a cytosol and nuclear extract. CAF-I was purified from the nuclear extract (see Materials and methods). The cytosol replication extract was
separated into two components, the 0.6M flow-through and the 2M step fraction. (B) and (C) DNA replication reactions containingpSVOI1, SV40 T
antigen, topoisomerases I and II, nucleoside triphosphates including [a-32P]dATP, the cytosol S100 extract (B) or the 0.6M flow-through fraction (C)
and the indicated components were incubated at 37°C for 60 min. The DNA was then isolated and subjected to electrophoresis in
(B) and (C) Left panel shows the autoradiographs and right panel shows the ethidium bromide-stained gels. The positions of form I and form II
marker DNAs are indicated at right.
I % agarose gels.
Chromatin assembly and DNA replication
that were metabolically pulse-labeled with [14C]lysine and
[14C]arginine. As shown in Figure 2B, the unfractionated
cytosol extract (S100) contained the four core histones: H2A,
H2B, H3 (H3. 1, H3.2, H3.3) and H4 (modified form). Note
that the specific activity of '4C-label for histones H2A and
H2B is significantly lower than for histones H3 and H4
(Smith and Stillman, 1989). Analysis of the fractionated
extracts demonstrated that the 0.6M flow-through contained
histones H3 and H4 (Figure 2C), but was depleted ofhistones
H2A and H2B, which were in the 2M step fraction (Figure
2D). The relatively low affinity of histones H3 and H4
for phosphocellulose was unexpected, considering the
strong DNA-binding properties ascribed to these histones.
However, under identical conditions, histones H3 and H4
Fig. 2. Two-dimensional gel analysis of the histones contained in 14C-labeled fractionated cell extracts. Fractionated extracts were prepared from cells
that had been labeled for 60 min with [14C]lysine and [14C]arginine and were analyzed by two-dimensional gel electrophoresis as described in
Materials and methods. TAU-polyacrylamide gels were used for the first dimension (horizontal direction) and SDS-polyacrylamide gels were used
for the second dimension (vertical direction). (A) Unlabeled marker histones isolated from 293 cell chromatin. The proteins were visualized by
staining with Coomassie brilliant blue. (B), (C) and (D) 14C-labeled histones contained in the fractionated cell extracts. The proteins were visualized
by autoradiography. The migration positions of unlabeled bulk histones isolated from chromatin are indicated by filled in arrows and identified in
(A). The histone species contained only in the fractionated cell extracts but not in bulk chromatin, are shown by open arrows; H3.3 migrates more
rapidly in the first dimension than H3.1 and H3.2 and modified H4 migrates more slowly in the first dimension than the unmodified form of H4.
Fig. 3. Histones H2A and H2B purified from chromatin can substitute for the 2M step and can be added after DNA replication. (A) Chromatin was
isolated from 293 cell nuclei (lane 1); histones H2A and H2B (lane 3) and histones H3 and H4 (lane 2) were eluted by sequential salt extraction.
The proteins were subjected to electrophoresis in a TAU-polyacrylamide gel and visualized by staining with Coomassie brilliant blue.
(B) Substitution with histones H2A and H2B. DNA replication reactions containing the 0.6M flow-through fraction and the indicated components
were incubated at 37°C for 60 min. (C) Time of addition of histones H2A and H2B. DNA replication reactions containing the 0.6M flow-through
fraction and CAF-I were carried out in the absence of histones H2A and H2B (lane 1) or with H2A and H2B added at the start of the reaction (lane
2) or at 45 min (lane 3). Reactions were incubated at 37°C for a total time of 60 min. (B) and (C) The DNA products were isolated and subjected
to electrophoresis in 1% agarose gels. Left panel shows the autoradiographs and right panel the ethidium bromide stained gels.
S.Smith and B.Stillman
purified from chromatin (shown in Figure 3, lane 2), did
bind to phosphocellulose (data not shown), indicating a
difference between the newly synthesized histones H3 and
H4 in the cytosol extract and the chromatin-bound forms
of these histones.
The 2M extract contains histones H2A and H2B but does
not provide any essential replication components. We
therefore determined whether a purified fraction ofhistones
H2A and H2B could substitute for the 2M extract. A purified
fraction of histones H2A and H2B was isolated from 293
cell chromatin by column chromatography and sequential
salt extraction (Figure 3A, lane 3). As shown in Figure 3B,
lane 4, purified histones H2A and H2B substituted for the
2M extract. These results demonstrate that the only essential
chromatin assembly components in the 2M extract are the
histones H2A and H2B. Furthermore, either cytosol or
chromatin bound histones H2A and H2B will function in the
replication-dependent chromatin assembly reaction.
Separation of replication-dependent chromatin
assembly into two steps
Since the chromatin assembly components histones H2A and
H2B were separated from the essential replication factors,
it was possible to determine whether these histones were
required to be present during DNA replication. To address
this question, the time of addition of histones H2A and H2B
to the replication reaction was varied. As shown in Figure
3C, addition of histones H2A and H2B at the start of the
reaction (lane 2) or 45 min after the start ofDNA synthesis
(lane 3) resulted in the same level of supercoiling when
reactions were terminated at 60 min. Note that under the
conditions described here, the vast majority of DNA
synthesis occurs before 45 min(data not shown) and this
completely replicatedDNA is stillsupercoiled by subsequent
addition of histones H2A and H2B. These results demon-
strate that theH2A/H2B-dependent chromatinassembly of
thereplicated DNA can occur after DNAreplication.
Theexperiments described above allowed separation of
the chromatinassembly reaction into two steps. During the
firststepofthereaction, DNA wasreplicated in thepresence
of histones H3 and H4 and CAF-I. The secondstep of the
reactiondepended only uponthe addition ofpurified histones
H2A and H2B and occurred after DNA replication. Although
thisH2A/H2B-dependent chromatinassembly occurred after
DNAreplication, the reaction still maintained the ability to
discriminate between replicated and unreplicated DNA
(Figure 3C, lane 3; compare the replicated DNA, left
panel, with the total input DNA, right panel). These data
suggest that during the first step of this two step reaction,
the replicated DNA is somehow marked for subsequent
CAF-Igenerates a micrococcal nuclease-resistant
structure onreplicating DNA
The first step of the two step reaction contained CAF-I, an
essentialcomponent of the chromatin assembly reaction. To
tryto ascertain whether there is a difference between DNA
replicated in the absence orpresence ofCAF-I, theproducts
of thereplication reactions were analyzed. We havealready
demonstrated that the DNA products of the two reactions
-4 r -
Fig. 4. Micrococcal nuclease digestion products of the replicated
DNA. DNA replication reactions containing the 0.6M flow-through
fraction were incubated at 37°C for a total time of 60 min in the
absence of CAF-I (lanes 1, 4 and 7), the presence of CAF-I (lanes 2,
5 and 8), or in the presence of CAF-I with histones H2A and H2B
included in the last 15 min of the reaction (lanes 3, 6 and 9). The
reactions were then adjusted to 3 mM CaCl2 and digested with
micrococcal nuclease for the indicated times. The DNA was isolated,
subjected to electrophoresis in 2% agarose, and autoradiographed. The
length in base pairs of double-stranded DNA markers are indicated at
left. The positions of the mono-, di- and tri-nucleosomes are indicated
Fig. 5. The chromatin precursor is generated during DNA replication.
DNA replication reactions containing the 0.6M flow-through fraction
were incubated in the absence (A, lane 1) and (B, lanes 1, 3, 5, 7, 9
and I1) or presence (A, lane 2) and (B, lanes 2, 4, 6, 8, 10, 12) of
CAF-I at 37°C for 15 min. (A) The DNA was isolated, subjected to
electrophoresis in 1% agarose and autoradiographed. (B) The reactions
were adjusted to 3 mM CaCI2 and digested with micrococcal nuclease
for the indicated times. The DNA was isolated, subjected to
electrophoresis in 2% agarose, and autoradiographed.
Chromatin assembly and DNA replication
are similar in that they are relaxed monomer circle DNA,
although a slight change in the distribution of topoisomers
in the DNA replicated in the presence of CAF-I can
sometimes be observed (see Figure IC, compare lanes 1 and
2). To probe more finely the structure of the chromatin
precursor, the products of the replication reactions were
analyzed by micrococcal nuclease digestion.
For these experiments, three types of reactions were
considered: DNA replicated in the absence of CAF-I,
DNA replicated in the presence ofCAF-I, or DNA replicated
in the presence of CAF-I followed by addition of histones
H2A and H2B. Histones H3 and H4 fractionate with the
0.6M replication extract (see Figure 2C) and, therefore,
are contained in all three reactions. The DNA-protein
products of these three replication reactions were digested
with micrococcal nuclease for increasing times and the DNA
products isolated and analyzed by agarose gel electro-
phoresis. The results of this analysis, presented in Figure
4, demonstrate that DNA replicated in the presence ofCAF-I
was more resistant to micrococcal nuclease than DNA
replicated in the absence ofCAF-I (Figure 4, compare lanes
1 and 2). Moreover, nuclease digestion of DNA replicated
in the presence of CAF-I yielded a smear of DNA which
migrated in the same region of the gel as the DNA isolated
from mono- and di-nucleosomes (Figure 4, compare lanes
5 and 6). Therefore, DNA replicated in the presence of
CAF-I, but in the absence of histones H2A and H2B,
displayed a relatively nuclease-resistant structure which
resembled mature chromatin but appeared more sensitive to
nuclease digestion. Addition of histones H2A and H2B
converted this precursor to mature chromatin containing
correctly spaced nucleosomes (Figure 4, lane 9).
CAF-I assembles the chromatin precursor during DNA
To determine whether the nuclease-resistant structure was
generated during DNA replication, in association with
passage of the replication fork, DNA replication products
from a very early time point in the replication reaction were
digested with micrococcal nuclease. Although the standard
time for the replication reaction was 60 min, for these
experiments DNA was replicated in the absence or presence
of CAF-I for only 15 min. As shown in Figure 5A, the
replicated DNA products at this early time point did not
appear as completed monomer circles, but rather, migrated
as a smear between forms I and II which correspond to
replicative intermediates (Tsurimoto
this experiment, DNA replication was greatly inhibited in
the presence of CAF-I. While we have previously noted
inhibition ofDNA replication by CAF-I (Smith and Stiliman,
1989), this inhibition appears to be exaggerated at this early
time point. Nevertheless, the DNA-protein products from
these replicative intermediates were digested with micro-
coccal nuclease and the DNA products isolated and analyzed
by agarose gel electrophoresis. The results presented in
Figure SB demonstrate that even at this early time point,
DNA replicated in the presence of CAF-I is more resistant
to micrococcal nuclease digestion than DNA replicated in
the absence of CAF-I. These data suggest that CAF-I can
function during DNA replication to generate the chromatin
Sucrose gradient sedimentation of the chromatin
We have shown above that DNA replicated in the presence
of CAF-I and histones H3 and H4 displays a micrococcal
nuclease-resistant structure that is more sensitive to nuclease
digestion than mature chromatin. This observation, coupled
with the presence of histones H3 and H4 in the 0.6M cytosol
chromatin precursor contained histones H3 and H4. Velocity
sedimentation behavior has previously been used as a means
of analyzing the protein-DNA products of the DNA
replication reactions; replicated DNA assembled into a
mature chromatin structure containing all four core histones
sediments much more rapidly through sucrose gradients than
,' 4-'_ ;
Fig. 6. Sucrose gradient sedimentation of the chromatin precursor. DNA replication reactions containing the 0.6M flow-through fraction were
incubated at 37°C for a total time of 60 min in the absence of CAF-I (A), the presence of CAF-I (B), or the presence of CAF-I with H2A and H2B
included in the last 15 min of the reaction (C). The reaction mixtures were then sedimented through 15-30% sucrose gradients and fractions
collected. The DNA products contained in the even numbered fractions were isolated and subjected to electrophoresis in 1% agarose gels. Top panel
shows the autoradiographs and the bottom panel the ethidium bromide stained gels. The direction of sedimentation was from right to left.
S.Smith and B.Stillman
Fig. 7. Two-dimensional gel analysis of the histones contained in the
chromatin precursor. DNA replication reactions containing the
3H-labeled 0.6M flow-through fraction were incubated at 37°C for a
total time of 60 min in the absence of CAF-I (A), the presence of
CAF-I (B), or the presence of CAF-I with the 3H-labeled 2M step
fraction included in the last 15 min of the reaction (C). The reaction
mixtures were then sedimented through 15-30% sucrose and fractions
collected and analyzed exactly as described in Figure 6. The fractions
from each sucrose gradient that contained the replicated DNA were
then subjected to further analysis. Sucrose gradient fractions containing
DNA replicated in the absence of CAF-I (A), DNA replicated in the
presence of CAF-I (B), or DNA replicated in the presence of CAF-I
followed by addition of the 3H-labeled 2M extract (C), were each
pooled separately. The DNA was isolated and subjected to
autoradiographs and lane 2 the ethidium bromide stained gels. The
proteins were subjected to two-dimensional gel electrophoresis exactly
as described in the legend to Figure 2 (right panels). The proteins
were visualized by autoradiography.
I % agarose gels (left panels). Lane
1 shows the
unassembled, replicated DNA (Stillman, 1986; Smith and
Stillman, 1989). We therefore analyzed the sedimentation
behavior of the intermediate described above.
chromatin precursor contains histones H3 and H4, then it
should sediment more rapidly than DNA replicated in
the absence of CAF-I. DNA replication reactions were
performed in the absence of CAF-I, the presence ofCAF-I,
or in the presence ofCAF-I followed by addition ofhistones
H2A and H2B. The DNA-protein products from these
reactions were then sedimented through sucrose gradients
and the DNA products present in each fraction isolated and
subjected to agarose gel electrophoresis (Figure 6).
The results presented in Figure 6 demonstrate that DNA
replicated in the presence ofCAF-I sedimented more rapidly
through sucrose than DNA replicated in the absence of
CAF-I (compare A and B, top panels). The sedimentation
behavior of the chromatin precursor was similar to that of
Fig.8.Proposed model of CAF-I mediated chromatinassembly during
DNAreplication. CAF-I binds thenewly synthesized histones H3 and
H4 anddeposits thecomplex onnewly replicated DNA to generate the
chromatinprecursor. CAF-I is thenexchanged for histonesH2A and
H2B (either thenewly synthesized or chromatin-bound forms) to
generate mature chromatin. Thedeposition of histones H2A and H2B
constitutespart of the maturation process which may also include
additionalsteps such aspost-translational histone modification. The
solid lines indicate theparental DNA strands and the dashed lines the
newly synthesized DNA. The fate ofparental histones is not addressed
mature minichromosome(compareB andC, top panels)and
thusmay reflectassembly of thereplicated DNA into a
protein-DNA complex. Moreover, in the presence of
CAF-I,while all ofthereplicated DNA exhibited thisrapid
displayeda much slower sedimentationvelocity (compare
topand bottompanels, Figure 6B). Therefore, thepresence
of CAF-I in thereplication reactionproduced arapidly
sedimenting structureonlyon thereplicated DNA.
the input unreplicated DNA
The chromatinprecursor iscomplexed with histones
H3 and H4
Thefindingthat the chromatinprecursor formed in the
presenceofCAF-I sedimentedrapidly throughsucrose was
consistent with itsassemblyinto aprotein-DNAcomplex
containing histones H3 and H4. To test thishypothesis
directly,thereplicatedDNA-protein complexwas isolated
andanalyzedfor its histone content. To allow detection of
thevery small amounts of histones assembled in these
reactions, DNAreplicationreactions wereperformedusing
fractionated extractsprepared from cells that had been
metabolically labeled with[3H]lysine and[3H]arginine.
DNA wasreplicatedwith the 3H-labeled 0.6Mreplication
extract in the absence ofCAF-I, in thepresenceofCAF-I,
or in thepresence ofCAF-I, followedby addition of the
3H-labeled 2Mstepfraction. Note that the 2Mstepis used
here toprovide 3H-labeled histones H2A and H2B. The
DNA-protein products were sedimentedthrough sucrose
gradients exactlyas described inFigure6. Toanalyze the
Chromatin assembly and DNA replication
proteins complexed to the replicated DNA, the sucrose
replicated DNA (see Figure 6A, fraction 14), the rapidly
sedimenting chromatin precursor (see Figure 6B, fractions
4-6), or the rapidly sedimenting mature chromatin (see
fractions 2-4) were
dimensional protein gel analysis. The results presented in
Figure 7 demonstrate that DNA replicated in the presence
of CAF-I (the chromatin precursor) contained histones H3
and H4 (Figure 7B), whereas DNA replicated in the absence
of CAF-I did not (Figure 7A). Addition of the 2M step
fraction to the chromatin precursor resulted in the deposition
of histones H2A and H2B (Figure 7C). Note that in the
absence of CAF-I, the bulk of the histones in the cytosol
extract did not sediment rapidly through sucrose gradients
(data not shown; Smith and Stillman, 1989).
From these data we can draw two important conclusions.
First, in the absence of CAF-I, histones H3 and H4 do not
associate stably with DNA (replicating or non-replicating
DNA) and second, in the presence of CAF-I, histones H3
and H4 form a stable complex with DNA in a reaction that
discriminates between replicating and non-replicating DNA.
Thus, CAF-I is required for the stable deposition ofhistones
H3 and H4 on replicating DNA.
Finally, we note that in addition to the four core histones,
the mature minichromosome contains two proteins which
migrate in the second dimension with histone HI, but migrate
in the first dimension with histone H3 (Figure 7C and Smith
and Stillman, 1989). Further experiments will be required
to determine whether these proteins are H1 related and if
they are deposited specifically during the second step of the
chromatin assembly reaction in association with or following
Model for sequential deposition of histones during
Based upon our findings, we propose a model for stepwise
assembly of chromatin during DNA replication (Figure 8).
In this model, histones H3 and H4, synthesized in the
cytoplasm during S phase, are transported to the nucleus
whereupon CAF-I, a nuclear protein (S.Smith and B.
Stillman, submitted for publication), binds the (H3/H4)2
tetramer and targets this complex to replicating DNA. We
have demonstrated that the deposition of newly synthesized
histones H3 and H4 occurs exclusively on replicating DNA
in close association with the passing replication fork.
Moreover, the newly synthesized forms of histones H3 and
H4 may be essential components of this reaction (discussed
below). The finding that newly synthesized histones H3 and
H4 deposit exclusively on replicating DNA is consistent with
a large body of in vivo evidence (Cremisi et al., 1977;
Worcel et al., 1978; Senshu et al., 1978; Cremisi and Yaniv,
1980; Jackson and Chalkley, 1981a,b, 1985), demonstrating
that in this regard the cell free system closely mimics the
in vivo process, but adds a new level of complexity by
introducing the requirement for the cellular factor CAF-I.
The chromatin precursor generated during the first step
of this two step reaction can be isolated as a stable
intermediate containing histones H3 and H4. We have
demonstrated that this replicated precursor, like nascent
chromatin found in vivo,
is hypersensitive to nuclease
digestion; this is consistent with the suggestion that newly
synthesized chromatin contains only histones H3 and H4
(Worcel et al., 1978; Cremisi and Yaniv, 1980). Moreover,
our preliminary data indicate that in addition to histones H3
and H4, CAF-I is associated with the chromatin precursor;
immunoblot analysis using monoclonal antibodies directed
against CAF-I (Smith,
1990; S.Smith and B.Stillman,
submitted for publication) indicate that CAF-I is contained
in the sucrose gradient isolated chromatin precursor (data
not shown). The model, therefore, depicts CAF-I as
depositing along with histones H3 and H4 on the newly
Next, as part of the chromatin maturation process, CAF-I
exchanges with H2A/H2B dimers to generate nucleosomes
and mature chromatin, perhaps through direct recognition
of CAF-I by H2A and H2B. We have demonstrated that
either the newly synthesized or chromatin-bound forms of
histones H2A and H2B will deposit on the replicated
precursor, indicating that unlike the first step, this exchange
reaction does not require newly synthesized histones H2A
and H2B. This is consistent with in vivo studies indicating
that nascent nucleosomes contain newly synthesized histones
H3 and H4 mixed with either old or new histones H2A and
H2B (Jackson, 1987, 1988).
Newly synthesized histones H3 and H4 differ from
the chromatin-bound forms
We have demonstrated that histones H3 and H4 present in
the 0.6M extract do not stably associate with DNA in the
absence of CAF-I (Figure 7A), a surprising result con-
sidering the strong DNA binding properties that have been
described for these histones. However, most of what is
known about histones H3 and H4 comes from studies using
the chromatin bound forms of these proteins. The newly
synthesized histones H3 and H4 contained in the cytosol
replication extract (which are the precursors for de novo
chromatin assembly in vivo) could differ from the chromatin-
bound forms in several ways. First, the cytosolic histones
may contain specific modifications not found in bulk
chromatin. We have shown by two-dimensional gel analysis
that the H4 in the cytosol extract migrates as the diacetylated
form of this histone (Figure 2B; Smith and Stillman, 1989),
a modification specifically associated with newly synthesized
histone H4 found in vivo (Ruiz-Carrillo et al., 1975; Jackson
et al., 1976; Bonner et al., 1988). Second, newly synthesized
histones H3 and H4 may be associated with other poly-
peptides not found in bulk chromatin, as is the case for the
histones contained in the chromatin assembly extract derived
from Xenopus eggs (Kleinschmidt and Franke,
Dilworth et al., 1987). In support of a difference between
the newly synthesized and chromatin-bound forms ofhistones
H3 and H4, we do find a change in their chromatographic
behavior; under conditions where the cytosolic forms of
histones H3 and H4 flow through phosphocellulose, histones
H3 and H4 purified from chromatin bind quantitatively (data
not shown). Thus, newly synthesized histones H3 and H4
have a reduced affinity for phosphocellulose, which could
be due to histone modification and/or association with other
An important question is whether the newly synthesized
forms of histones H3 and H4 are required for the replication
dependent chromatin assembly reaction described here.
Our preliminary experiments suggest that this may be the
S.Smith and B.StilIman
case. Since the cytosol extract has not yet been depleted of
histones H3 and H4, we are unable to perform a substitution
experiment. We do find, however, that the addition of
even very small amounts of histones H3 and H4 (purified
from chromatin) to the 0.6M replication extract promotes
supercoiling of both replicating and non-replicating DNA
in the presence or absence of CAF-I (data not shown),
suggesting that newly synthesized histones H3 and H4 are
essential for the CAF-I-dependent discrimination between
replicated and unreplicated DNA.
In agreement with this suggestion is the recent demon-
stration by Gruss et al. (1990) that the addition of a crude
nuclear extract (containing large amounts of histones) to the
cytosol replication extract promotes supercoiling of both
replicated and unreplicated DNA. These authors, however,
interpret their results to indicate that nucleosome assembly
is not coupled to DNA replication. We, on the other
hand, suggest that their crude system which contains two
populations of histones (the newly synthesized histones H3
and H4 and the chromatin derived histones) displays two
disparate supercoiling reactions and that true chromatin
assembly in their experiments only occurs during DNA
replication. Their own data (shown in Figure 10 of Gruss
et al., 1990) demonstrate that the replication coupled
supercoiling reaction produces DNA-protein complexes
similar to nucleosome particles found in SV40 and host cell
chromosomes in vivo and confirm our data reported here
and elsewhere (Stillman, 1986; Smith and Stillman, 1989).
In contrast, the DNA-protein complexes produced in the
absence of DNA replication did not resemble the known
structure of nucleosomes based upon the length of DNA
associated with them. We therefore suggest that these
complexes do not reflect the assembly ofchromatin. It may
well be that these DNA-protein complexes are similar to
the DNA -protein complexes that we observe when we add
chromatin-derived histones H3 and H4 to the reactions. It
will be necessary to determine the exact nature of these
replication-independent complexes before their significance
is appreciated. Thus the conclusions by Gruss et al. (1990)
that CAF-I and DNA replication do not play a decisive role
in chromatin assembly do not seem to be warranted at this
This raises the question as to the role ofDNA replication
in nucleosome assembly. In vivo, the bulk of histone
deposition into chromatin occurs during S phase of the cell
cycle and is coupled to DNA replication. Our data, which
reproduce this in vitro, indicate that the CAF-I-dependent
coupling ofchromatin assembly to DNA replication requires
the exclusive use of the newly synthesized forms of histones
H3 and H4 and the availability of these histones may be
tightly controlled. It should also be noted that our results
do not imply that chromatin assembly occurs exclusively
during DNA replication in vivo. Clearly, cell free systems
for the assembly of chromatin in the absence of DNA
replication have been developed using Xenopus egg extracts
(Dilworth and Dingwell, 1988) and it is easy to imagine
many situations in which chromatin needs to be formed in
non-replicating cells. In fact, recent studies using the Xenopus
cell free system have demonstrated a sequential mechanism
for histone deposition in the absence of DNA replication
(Kleinschmidt et al., 1990; Zucker and Worcel, 1990; Sapp
and Worcel, 1990). A role for CAF-I in this in vitro
independent assembly of chromatin in vivo remains to be
determined. One interesting possibility is that theexchange
of histones H2A and H2B observed in non-replicating cells
(Jackson, 1987, 1988) requires CAF-I as an exchange factor.
Finally, it should be noted that the role ofCAF-I and DNA
replication in the redistribution ofold, chromosome-bound
histones H3 and H4 after passage ofthe replication fork has
not been addressed in these cell free systems.
CAF-I targets deposition of histones H3 and H4 to
The demonstration that in the absence of CAF-I, histones
H3 and H4 have no affinity for DNA, points to a key role
for the cellular factor (CAF-I) in targeting newly synthesiz-
ed histones H3 and H4 to the replicating DNA. What is the
mechanism which targets deposition of histones H3 and H4
to replicating DNA? Recently, Fotedar and Roberts (1989)
have used a similar, but unfractionated cell free system of
chromatin assembly to propose a multi-step pathway for
replication coupled chromatin assembly. In this study the
authors describe a DNA-protein particle which they
suggest, although did not show, contains histones H3 and
H4. Unlike the chromatin precursor described in our report,
however, this particle was found on both replicated and
unreplicated DNA and therefore does not reflect the
replication-preferential nature ofchromatin assembly. Since
our results indicate that in the absence of CAF-I, histones
H3 and H4 do not associate with the DNA (replicating or
non-replicating DNA), we favor a model in which CAF-I
binds to histones H3 and H4 and targets the complex to
replicating DNA. CAF-I could modify histones H3 and H4
directly to somehow promote their deposition onto replicating
DNA or a CAF-I-H3-H4 complex could deposit on
replicating DNA. This latter possibility would be consistent
with our preliminary findings (described above) that CAF-I
is contained in the chromatin precursor. IfCAF-I is respon-
sible for localizing the complex to replicating DNA, it must
do so through protein-protein interaction, since purified
CAF-I does not bind DNA directly (Smith and Stillman,
1989). Such interactions could occur with one or more of
the proteins associated with the replication fork. Since most
of the proteins required for DNA replication in this cell free
system have been identified and purified (see Stillman, 1989;
Tsurimoto et al., 1990), we can begin to test this hypothesis
directly. Finally, we note that a two step nucleosome
assembly mechanism was recently proposed, based upon
studies using replicating single-stranded DNA and Xenopus
egg extracts (Almouzni et al., 1990).
Regulation of chromatin assembly in vivo
To investigate the mechanism of histone deposition during
DNA replication we have used a cell free system which,
by many criteria (Stillman, 1986; Smith and Stillman, 1989;
this report), reproduces events that occur in vivo. We present
evidence for a sequential mechanism of chromatin assembly
during DNA replication; histones H3 and H4 deposit first
on replicating DNA, followed by deposition of histones H2A
and H2B. The demonstration that this reaction is catalyzed
cellular protein (CAF-I)
mechanism for the cellular control of this fundamental
process. Interestingly, we have recently shown that CAF-I,
a multisubunit protein, is phosphorylated on at least two
subunits in vivo (Smith, 1990; S.Smith and B.Stillman,
Chromatin assembly and DNA replication
submitted for publication), a modification which could affect
the ability of CAF-I to bind histones. It will be important
to determine whether this phosphorylation can affect the
histone binding and assembly properties of CAF-I both
in vitro and in vivo. Finally,
could regulate access of non-histone proteins to the newly
replicated DNA, thereby influencing the inheritance of
it is possible that CAF-I
Materials and methods
Preparation of cytosol extracts and T antigen
The cytosol replication extracts were prepared from suspension cultures of
human 293 cells as described previously (Stillman and Gluzman, 1985;
Stillman, 1986). The radiolabeled cytosol replication extracts were prepared
in the same way except that prior to harvesting, the cells were incubated
in arginine-free and lysine-free media (MEM Joklik-Modified) supplemented
with 5% dialyzed calf serum for 30 min at 37°C, followed by incubation
in the same media containing
[14C]arginine or 500ACi/ml [3H]lysine and 500 uCi [3H]arginine for
60 min at 37°C. SV40 T antigen was obtained from recombinant baculovirus
vector (941T)-infected Spodotera frugiperda insect cells as described
(Lanford, 1988) and was purified by immunoaffinity chromatography
according to Simanis and Lane (1985) and Stillman and Gluzman (1985).
100ACi/ml[14C]lysine and 100 sCi/ml
Preparation of fractionated cytosol extracts
The cytosol replication extract derived from 16 1 of 293 cells (8 x 109
cells) was adjusted to 0.6 M NaCl and loaded onto a 20 ml phosphocellulose
column (2 x 6 cm) equilibrated in buffer A [25 mM Tris-HCI (pH 7.5),
1 mM EDTA, 0.01% NP-40, 10% glycerol,
phenylmethylsulfonyl fluoride] containing 0.6 M NaCl. The protein that
flowed through the column was dialyzed against buffer A containing 50 mM
NaCl and 20% sucrose (w/v), aliquoted, and stored at -70°C (0.6M
flow-through fraction). The column was then washed with 40 ml of buffer
A containing 0.6 M NaCl and the bound material eluted with 40 ml of buffer
A containing 2 M NaCI. The eluted protein was concentrated in volume
- 8-fold by ultrafiltration (Centricon), aliquoted, and stored at -70°C (2M
1 mM dithiothreitol, 0.1 mM
Preparation of nuclear extracts, CAF-I and histones
Nuclei were isolated from 293 cells and extracted with 0.4 M NaCl as
described previously (Smith and Stillman, 1989). The residual nuclear
material was collected by centrifugation and used for the preparation
of the histones (see below). CAF-I was purified from the supernatant
chromatographic steps (Smith and Stillman, 1989). The highly purified Mono
Q fraction was aliquoted and stored at -700C (CAF-I). For the preparation
of the histones, a chromatin extract was prepared from the residual nuclear
material. The nuclear pellet was washed twice with 60 ml of buffer containing
25 mM NaCl and 8 mM EDTA (pH 8.0) and collected by centrifugation.
The chromatin was prepared and fractionated by hydroxylapatite according
to the procedure of Simon and Felsenfeld (1979). The fractions containing
the purified histone pairs H2A/H2B and H3/H4 were dialyzed againstbuffer
A containing 100 mM NaCl and 20% sucrose (w/v), aliquoted, and stored
at -70°C. Protein concentrations were determined by the method of
Bradford (1976), using bovine serum albumin as a standard.
DNA replication reactions
SV40 DNA replicationin vitro was assayedunder standard conditions similar
to those described previously (Stillman, 1986). Each reaction contained
4pig/mlof plasmid DNA pSVOI1 [whichcontains the SV40 origin from
HindIlI (nucleotide no. 5171) to SphI (nucleotide no. 128) in pUC18 and
is 2.9 kb in size], 22 ug/ml purified SV40 T antigen, 3.2 mg/ml 293 cell
cytosol extract or 6.4 mg/mil 0.6M flow-through fraction, 4Ag/ml purified
topoisomerase I and 1.8Ag/ml purified topoisomerase II. Topoisomerases
I and II were purified from calfthymusnuclear extract by slightmodifications
of published procedures (Liu and Miller, 1982: Schomburg and Grosse,
1986). The replication
components as indicated: 21Ag/ml purified CAF-I, 80 Ag/ml 2M step
fraction or 32 ug/ml H2A/H2B. Reactions were incubated at 370C for
60 min unless otherwise indicated. The 8:1 ratio of H2A/H2B to DNA
(weight:weight, i.e. a 4:1 ratio of each histone to DNA) was requiredto
achieve complete supercoilingof the replicatedDNA. It should be noted,
however, that supercoilingcould be observed at lower H2A/H2B ratios
(i.e. 4:1 and 2:1, data not shown).
Replication product analysis
Reaction mixtures were digested first with ribonuclease A (20 Ag/ml) for
15 min at 37°C in the presence of10 mM EDTA and 0.5% SDS and then
with1mg/mi of pronase for1 h at 37°C. The samples were then extracted
once with phenol:chloroform:isoamyl alcohol (25:24:1) and subjected to
agarose gel electrophoresis in Tris -borate-EDTA buffer (Maniatis et al.,
1982) at -2 V/cm. The gels were stained with ethidium bromide (I/Ag/ml),
then dried and autoradiographed.
Micrococcal nuclease digestions
After the DNA replication reaction, the mixture was adjusted to 3 mM
to the reaction for various times at 30°C. All reactions were terminated
by addition ofEDTA to 10 mM. The DNA was isolated as described above
and separated from unincorporated nucleoside triphosphates by spin dialysis
(Maniatis et al., 1982) with Sephadex G-50 in a 1 ml syringe. The DNA
was subjected to electrophoresis through 2% agarose as described above.
In order to load similar amounts of replicated DNA on the gel, in each
case only 50% of the samples of DNA replicated in the absence of CAF-I
was subjected to agarose gel electrophoresis.
1A1 (15 U) of micrococcal nuclease (Worthington) was added
Sucrose gradient sedimentation
Replication reaction mixtures were layered onto preformed 15-30% sucrose
gradients in 10 mM Tris-HCI (pH 7.5),
0. 15 M NaCI in Beckman SW41 centrifuge tubes. Samples were subjected
to centrifugation at 30 000 r.p.m. for 16 h and fractions were collected from
the bottom. The position of the minichromosome was determined by
subjecting a sample of the sucrose gradient fractions to electrophoresis in
1 mM EDTA, 0.25% NP-40,
Analysis of the proteins contained in the minichromosome was carried out
as described previously (Smith and Stillman, 1989) according to the method
of Shimamura et al. (1988). Sucrose gradient fractions were combined and
layered over a1 ml cushion of 30% sucrose (for the rapidly sedimenting
products) or 15% sucrose (for the slower sedimenting products) and pelleted
in a Beckman SW60 rotor at 30 000 r.p.m. for 16 h. The pellet was
resuspended in H20 and mixed with an equal volume of a solution
containing 0.4 M HCI, 0.5 mg/mi protamine sulfate, 8 M urea and 0.02%
pyronine Y. Proteins were subjected to two-dimensional gel electrophoresis.
For the first dimension, TAU tube gels were run essentially as described
(Alfageme et al., 1974) using a 15% acrylamide gel containing 6 M urea
and 0.37% Triton X-100. After electrophoresis the gels were equilibrated
in 60 mM Tris (pH 6.8), 0.5% ,B-mercaptoethanol and placed over an 18%
SDS-polyacrylamide gel and subjected to electrophoresis as described by
Laemmli (1970). The positions of the marker histones were determined by
co-electrophoresis of core histones isolated from 293 cell chromatin. After
electrophoresis proteins were visualized by staining with Coomassie brilliant
blue followed by fluorography where appropriate; gels were treated
with H20 for 20 min,
radiographed. The histone pairs isolated from 293 cell chromatin were
analyzed by electrophoresis through a TAU-15% polyacrylamide gel
containing 6 M urea and 0.37% Triton X-100. Proteins were visualized
by staining with Coomassie brilliant blue.
1M sodium salicylate for 30 min, dried and auto-
We would like to thank John Diffley for valuable discussion throughout
the course of this work. We gratefully acknowledge Winship Herr, Mike
Mathews and John Diffley for critically reading the manuscript. We also
thank Jim Duffy and Phil Renna for the photography and preparation of
the figures. This work was supported by a grant from the National Institutes
of Health (CA 13106).
Alfageme,C.R., Zweidler,A., McDonald,A. and Cohen,L.H. (1974)J. Biol.
Chem., 249, 3729-3736.
Almouzni,G., Clark,D.J., Mechali,M. and Wolfe,A.P. (1990)Nucleic Acids
Res., 18, 5767-5774.
Bonner,W.M., Wu,R.S., Panusz,H.T. and Muneses,C. (1988) Biochemistry,
Bradford,M.M. (1976)Anal. Biochem., 72, 248-254.
Chalberg,M.D.andKelly,T.J. (1989)Annu. Rev. Biochem., 58, 671-717.
Cremisi,C. and Yaniv,M. (1980)Biochem. Biophys.Res. Commun., 92,
S.Smith and B.Stillman
Cremisi,C., Chestier,A. and Yaniv,M. (1977) Cell, 12, 947-951.
Cremisi,C., Chestier,A. and Yaniv,M. (1978) Cold Spring Harbor. Symp.
Quant. Biol., 42, 409-416.
Cusick,M.E., Herman,T.M., DePamphilis,M.L. and Wasserman,P.M.
(1981) Biochemistry, 20, 6648-6658.
Cusick,M.E., Lee,K.-S., DePamphilis,M.L. and Wasserman,P.M. (1983)
Biochemistry, 22, 3873-3884.
Dilworth,S.M. and Dingwall,C. (1988) BioEssays, 9, 44-49.
Dilworth,S.M., Black,S.J. and Laskey,R.A. (1987) Cell, 51, 1009-1018.
Fotedar,R. and Roberts,J.M. (1989) Proc. Natl. Acad. Sci. USA, 86,
Groudine,M. and Weintraub,H. (1982) Cell, 30, 131-139.
Gruss,C., Gutierrez,C., Burhans,W.C., DePamphilis,M.L., Koller,T. and
Sogo,J.M. (1990) EMBO J., 9, 2911-2922.
Commun., 73, 157-163.
Jackson,V. (1987) Biochemistry, 26, 2315-2325.
Jackson,V. (1988) Biochemistry, 27, 2109-2120.
Jackson,V. and Chalkley,R. (1981a) Cell, 23, 121-134.
Jackson,V. and Chalkley,R. (1981b) J. Biol. Chem., 256, 5095-5103.
Jackson,V. and Chalkley,R. (1985) Biochemistry, 24, 6921-6930.
Jackson,V., Shires,A., Tanphaichitr,N. and Chalkley,R. (1976) J. Mol.
Biol., 104, 471-483.
Kleinschmidt,J.A. and Franke,W.W. (1982) Cell, 29, 799-809.
Seiter,A. and Zentgraf,H. (1990) EMBO J.,
Klempnauer,K.H., Fanning,E., Otto,B. and Knippers,R. (1980) J. Mol.
Biol., 136, 359-374.
Laemmli,U.K. (1970) Nature, 227, 680-686.
Lanford,R.E. (1988) Virology, 167, 72-81.
Levy,A. and Jakob,K.M. (1978) Cell, 14, 259-267.
Liu,L.F. and Miller,K.G. (1981) Proc. NatI. Acad.
Maniatis,T., Fritsch,E.F. and Sambrook,J. (1982) Molecular Cloning: A
Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring
McKnight,S.L. and Miller,O.L. (1977) Cell, 12, 795-804.
Ruiz-Carrillo,A., Wangh,L.J. and Alfrey,V.J.
Sapp,M. and Worcel,A. (1990) J. Biol. Chem., 265, 9357-9365.
Schomburg,V. and Grosse,F. (1986) Eur. J. Biochem., 160, 451-457.
Seale,R.L. (1975) Nature, 255, 247-249.
Seale,R.L. (1978) Proc. Natl. Acad. Sci. USA, 75, 2717-2721.
Seidman,M.M., Garon,C.F. and Salzman,N.P. (1978) Nucleic Acids Res.,
Senshu,T., Fukuda,M. and Ohashi,M. (1978) J. Biochem., 84, 985 -988.
Shimamura,A., Tremethick,D. and Worcel,A. (1988) Mol. Cell. Biol., 8,
Simanis,V. and Lane,D.P. (1985) Virology, 144, 88-100.
Simon,R.H. and Felsenfeld,G. (1979) Nucleic Acids Res., 6, 689-696.
Smith,S. (1990) Replication dependent chromatin assembly in vitro. Ph.D.
Thesis. State University of New York at Stony Brook.
Smith,S. and Stillman,B. (1989) Cell, 58, 15-25.
Stillman,B. (1986) Cell, 45, 555-565.
Stillman,B. (1989) Annu. Rev. Cell Biol., 5, 197-245.
Stillman,B. and Gluzman,Y. (1985) Mol. Cell. Biol., 5, 2051-2060.
Svaren,J. and Chalkley,R. (1990) Trends Genet., 6, 52-56.
Tsurimoto,T. and Stillman,B. (1989) Mol. Cell. Biol., 9, 609-619.
Tsurimoto,T., Melendy,T. and Stillman,B. (1990) Nature, 346, 534-539.
van Holde,K.E. (1988) Chromatin. Springer-Verlag, New York.
Weintraub,H. (1979) Nucleic Acids Res., 7, 781-792.
Worcel,A., Han,S. and Wong,M.L. (1978) Cell, 15, 969-977.
Zucker,K. and Worcel,A. (1990) J. Biol. Chem., 265, 14487-14496.
Sci. USA, 78,
Received on October 22, 1990; revised on January 11, 1991