Histone octamer dissociation is not required for in vitro replication of simian virus 40 minichromosomes.

Replication of chromosomal templates requires the passage of the replication machinery through nucleosomally organized DNA. To gain further insights into these processes we have used chromatin that was reconstituted with dimethyl suberimidate-cross-linked histone octamers as template in the SV40 in vitro replication system. By supercoiling analysis we found that cross-linked histone octamers were reconstituted with the same kinetic and efficiency as control octamers. Minichromosomes with cross-linked nucleosomes were completely replicated, although the efficiency of replication was lower compared with control chromatin. Analysis of the chromatin structure of the replicated DNA revealed that the cross-linked octamer is transferred to the daughter strands. Thus, our data imply that histone octamer dissociation is not a prerequisite for the passage of the replication machinery and the transfer of the parental nucleosomes.

Replication of eukaryotic genomes involves both the duplication of DNA and the assembly into chromatin. Formation of chromatin during replication consists of two distinct reactions: the transfer of parental nucleosomes from the unreplicated DNA to the daughter strands and the assembly of new nucleosomes from newly synthesized histones (reviewed in Refs. [1][2][3]. The assembly of new nucleosomes during DNA replication is mediated by chromatin assembly factor-1, a heterotrimeric protein originally purified from human nuclear cell extracts (4). Chromatin assembly factor-1 mediates the first step of nucleosome assembly, namely the deposition of newly synthesized, posttranslationally modified H3/H4 tetramers onto the replicating DNA. This reaction is dependent on ongoing DNA replication (5) and as shown recently this process is mediated by the proliferating cell nuclear antigen, a DNA polymerase clamp. Proliferating cell nuclear antigen binds directly to p150, the largest subunit of chromatin assembly factor-1, and the two proteins colocalize at sites of DNA replication in cells (6). In a second step, which appears to be independent of ongoing DNA replication, two histone H2A/H2B dimers are deposited onto the H3/H4 tetramer precursor particle to complete the full nucleosome core (5).
The mechanism of the transfer of parental nucleosomes to the replicated DNA has been the subject of intense research. Experiments with reconstituted chromatin (7) or with SV40 minichromosomes (8,9) led to the conclusion that the parental nucleosomes remain associated with the DNA during the passage of the replication machinery. However, this association must be weak because a large excess of competitor DNA can cause the dissociation of histones during replication, indicating that histones/nucleosomes remain only loosely associated with the DNA during passage of the replication fork (10). In fact, electron microscopic examination of replicating SV40 minichromosomes revealed that replication forks move up to the next prefork nucleosome and that new nucleosomes appear on the replicated daughter strands at average distances of about 250 nucleotides behind the fork (11). Based on these observations, Randall and Kelly (9) suggested that advancing replication forks release positively charged amino acid side chains in histones of the prefork nucleosome. Released positively charged amino acids then immediately gain contact with newly synthesized DNA in daughter strands. Removal of the amino-terminal histone domains by trypsin treatment of SV40 minichromosomes revealed, however, that tailless nucleosomes are efficiently transferred to the daughter strands excluding a participation of the basic histone domains in nucleosome transfer (12).
By using SV40 minichromosomes photoreacted with psoralen under moderately destabilizing conditions, it has been shown that on the average two nucleosomes are destabilized in front of the replication fork (13). This seems to be caused by a dissociation of histone H1 from one to two nucleosomes immediately in front of the fork. In addition the size of the last prefork nucleosome indicates a dissociation of one or two H2A/ H2B dimers (13). This is consistent with in vivo data, showing that nucleosomes dissociate into H2A/H2B dimers and H3/H4 tetramers (14,15). Thus it was shown that H3/H4 tetramers selectively deposit on newly replicated DNA, whereas new H2A/H2B dimers associate with either new or old H3/H4 tetramers (14,15). These conclusions were supported by in vitro replication studies (5,10,16,17) and biochemical and electron microscopic analysis of SV40 minichromosomes replicated under cell-free conditions. These studies revealed that a subnucleosomal particle, most probably an H3/H4 tetramer, is the transferred unit (10), which is complemented in a second step by the association of H2A/H2B dimers.
To investigate whether the disruption of histone octamers is necessary for the passage of the replication machinery we compared chromatin, reconstituted with either control or with cross-linked histone octamers, as substrates in the SV40 in vitro replication system. We found that histone octamer crosslinking does not impede the passage of the replication machinery and the transfer of the parental nucleosomes to the daughter strands.
Chromatin Reconstitution-Nucleosomal cores were reconstituted onto supercoiled SV40 DNA (20) by using a modification of the salt dilution method of Germond et al. (44). In a standard reconstitution reaction control and cross-linked histone octamers were mixed in an initial volume of 20 l with 6 g of supercoiled SV40 DNA in 2 M NaCl, 10 mM Tris-HCl, pH 7.4, 0.5 mM EDTA, pH 8.0. Samples were diluted at successive intervals of 60 min to contain 1.12, 0.8, 0.6, 0.4, and finally 0.12 M NaCl in the same buffer at room temperature. Samples were centrifuged at 12,000 ϫ g for 10 min to remove aggregated material and stored on ice until use.
Analysis of DNA Topology-200 ng of mock assembled DNA and chromatin reconstituted with increasing amounts of control or crosslinked histones (0.8, 1.2, 1.8:1 histone to DNA weight ratios) were incubated in a 50-l reaction with 5 units of DNA topoisomerase I (Promega) in 50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, pH 8.0, 1 mM dithiothreitol, 20% glycerol, and 50 mM NaCl for 1 h at 37°C. To stop the reaction, SDS (0.2% (w/v)) and EDTA (15 mM) were added followed by ethanol precipitation.
In Vitro Replication of Reconstituted Chromatin-SV40 T-Ag was prepared from insect cells (Sf9) infected with a recombinant baculovirus (21) by immunoaffinity chromatography (22). Cytosolic HeLa S100 extracts were prepared exactly as described (23). In standard experiments, 300 ng of mock-assembled DNA or chromatin reconstituted with increasing amounts of control or cross-linked histones (0.8, 1.2, 1.8:1 histone to DNA weight ratios) were incubated with 2.5 g of T-Ag and 350 g of cytosolic S100 extract proteins in a 190-l reaction for 2 h at 37°C exactly as described (24). For determination of the incorporated nucleotides, 1/10 of the replication assay was precipitated with 10% trichloroacetic acid.
Pulse-Chase Analysis-For pulse-chase analysis, SV40 in vitro replication assays were performed in a two step procedure as described by Fotedar et al. (30) with the following modifications. In the first step, 2.5 g of SV40 T-Ag were incubated with 250 ng of reconstituted control or cross-linked chromatin (1.2:1 histone to DNA weight ratio) or 250 ng of mock reconstituted SV40 DNA and 650 g of dialyzed S100 replication extract in a 190-l assay (31). This mixture was preincubated in the presence of 3 mM Mg 2ϩ /ATP, 30 mM Hepes-KOH, pH 7.8, 0.5 mM dithiothreitol, 40 mM creatine phosphate, and 0.24 g/ml creatine kinase at 26°C. Elongation was initiated by the addition of the remaining three ribonucleosides (80 M each of CTP, GTP, and UTP), 100 M dGTP and dCTP, and 3 Ci of [␣-32 P]dATP to label the newly synthesized DNA. After 30 s, the pulse was terminated by the addition of an 1000-fold excess of cold dATP and 100 M dTTP. After the indicated time points (30 s to 4 min) aliquots of the different reactions were removed and stopped by the addition of 0.6 volume stop solution (2.0% SDS, 60 mM EDTA). After proteinase K digestion, phenol extraction, and ethanol precipitation, purified products were resuspended in 50 mM NaOH, 1 mM EDTA and separated by 1.2% alkaline-agarose gel electrophoresis (1 V/cm, 16 h) in 30 mM NaOH, 1 mM EDTA and visualized by autoradiography.
Micrococcal Nuclease Footprint Analysis-For micrococcal nuclease digestion, 300 ng of mock-assembled SV40 DNA and reconstituted control and cross-linked chromatin (1.2:1 histone to DNA weight ratio) were preincubated in a 50-l reaction with cytosolic S100 replication extract in the presence or absence of 2.5 g of T-Ag as described for pulse-chase analysis. After a 30-min incubation at 26°C, preinitiated complexes were digested with 150 units micrococcal nuclease for 0.5, 5, and 20 min at 26°C in the presence of 3 mM CaCl 2 . DNA purification and Southern blot analysis were done as described (32) with a 32 Plabeled oligonucleotide complementary to the SV40 origin sequence (nucleotides 5217-5242, lower strand of the SV40 genome) (24).

Chromatin Reconstitution with Cross-linked Histone
Octamers-Histone octamers were purified from HeLa nuclei; one half served as the untreated control, whereas the second half was cross-linked with dimethyl suberimidate. As demonstrated before by circular dichroism spectra, DNase I digestion, electron microscopy, and DNA supercoiling assays cross-linked histone octamers have properties that are similar to those of native control octamers (33). To check the efficiency of dimethyl suberimidate cross-linking, control and cross-linked octamers were analyzed by SDS-PAGE (Fig. 1, input). The cross-linked histone octamer migrates with an apparent molecular mass of ϳ100,000 dalton, with no detectable contamination of lower molecular mass components. By using salt gradient dialysis we then reconstituted SV40 DNA into chromatin with control and cross-linked histone octamers. Following reconstitution the integrity of the cross-linked histone octamer was confirmed by SDS-PAGE (Fig. 1, chromatin), where we detected only one band with ϳ100,000 dalton and no dissociated free histones.
The effect of histone octamer cross-linking on the efficiency of chromatin assembly was then examined by DNA supercoiling analysis (Fig. 2). As one nucleosome introduces one constrained negative supercoil into closed circular DNA templates (34), the change in DNA linking number can be taken as a measure of nucleosome cores assembled on DNA templates from control and cross-linked octamers. SV40 DNA was reconstituted by salt gradient dialysis with increasing amounts of control and cross-linked histone octamers. Nonconstrained supercoils were relaxed by topoisomerase I, the DNA was purified and investigated by agarose gel electrophoresis (Fig. 2). We found that relaxed form II DNA is converted into a topoisomer ladder and subsequently into form I DNA by the addition of increasing amounts of histone octamers. No difference in the efficiency and kinetic of nucleosome assembly could be detected between control and cross-linked octamers, indicating that both templates carry the same number of nucleosomes. We have also investigated the reconstitution products by micrococcal nuclease digestion and obtained very similar digestion products with the familiar ladder of mono-, di-, and trimeric nucleosomal DNA size (data not shown, see Refs. 9, 24, and 31) Histone Octamer Dissociation Is Not Required for Passage of the Replication Machinery-To see whether the dissociation of histone octamers is necessary for a passage of the replication machinery through nucleosomally organized DNA we used chromatin, reconstituted with control and cross-linked histones, as templates in the SV40 in vitro replication system (Fig.  3). Chromosomal DNA was incubated for 2 h in the presence of the SV40 T-Ag and [␣-32 P]dATP. Replication products were purified and analyzed by agarose gel electrophoresis and auto-radiography (Fig. 3A). The incorporation of radioactive nucleotides was determined by trichloroacetic acid precipitation (Fig. 3B). Protein-free DNA replicates very efficiently in this system. However, the replication efficiency of chromatin was reduced, and the extent of reduction depended on the amounts of nucleosomes. As demonstrated before, the decreased template activity of reconstituted chromatin is because of the packaging of the origin sequences into nucleosomes, which renders the DNA inaccessible to the SV40 T-Ag and thus inactive for replication (10,24,35). Comparison of the replication efficiencies of templates with control and cross-linked histone octamers revealed that the efficiency is more drastically reduced with templates containing cross-linked histones. This could be seen both from the autoradiographic signal (Fig. 3A) and the incorporated nucleotides (Fig. 3B). Importantly however, we detected completely replicated form I DNA with molecules containing cross-linked histone octamers, indicating that histone dissociation is not necessary for the passage of the replication machinery.
To further investigate the reasons for the reduced replication efficiency of templates containing cross-linked octamers and to discriminate between an effect on the initiation or elongation of replication we performed pulse-chase experiments. To this end initiation complexes were formed in the presence of the SV40 T-Ag, dialyzed cytosolic S100 replication extract and ATP/ Mg 2ϩ using protein-free DNA and chromatin reconstituted with control and cross-linked octamers (histone:DNA weight ratio 1:1.2) as template. During this preincubation step no T-Ag-dependent DNA synthesis was measured (data not shown). DNA elongation was initiated by the addition of ribonucleoside triphosphates, dGTP and dCTP, to preincubated reactions and pulse labeled with [␣-32 P]dATP. After 30 s, dTTP was added, and the elongation reactions were chased with an excess of unlabeled dATP (Fig. 4). The products were then analyzed under denaturing conditions by alkaline gel electrophoresis (Fig. 4A), and the incorporated nucleotides were determined by trichloroacetic acid precipitation (Fig. 4B). The average lengths of the replication products were evaluated by densitometric scanning of the autoradiograms (Fig. 4C). This kind of pulse-chase analysis allows two different measure-

FIG. 1. Analysis of control and cross-linked histones octamers.
To investigate the purity and integrity of the individual histones and the extent of histone cross-linking before (input) and after chromatin reconstitution (chromatin), 5 g of control and cross-linked histones either before or after chromatin reconstitution were precipitated with trichloroacetic acid and analyzed on a 6 -18% step SDS-PAGE. The gel was stained with Coomassie Brilliant Blue R. HM, histone marker; HMW, high molecular weight marker. The molecular masses are indicated in kDa.  ments: (i) the intensities of the autoradiographic signal and the efficiencies of nucleotide incorporation correspond to the fraction of minichromosomes participating in replication and (ii) comparison of the lengths of the replication products gives the rate of DNA chain elongation. We found an ϳ3-fold difference in the initiation rate between control and cross-linked octamers as visualized by the autoradiographic signal (Fig. 4A) and the efficiency of nucleotide incorporation (Fig. 4B). However, after 4 min of chase we measured only a 1.3-fold difference in the product lengths between control and cross-linked chromatin (Fig. 4C). This supports the conclusion that the reduced replication efficiency of templates containing cross-linked octamers is mainly because of a reduced initiation rate.
We have shown recently that topoisomerase I and II have a stimulatory effect on the initiation of chromatin replication (31). Higher amounts of topoisomerases seem to be necessary for the removal of topological links in chromatin compared with protein-free DNA. In addition nucleosomal organization is a critical determinant for the amount of topoisomerases needed for efficient replication (31,36). Therefore cross-linked histone octamers might represent a stronger barrier for supercoil diffusion compared with control octamers, requiring a higher amount of topoisomerases for efficient replication. To test this assumption the preinitiation complex was formed in the absence or presence of additional topoisomerases (31). We found that both templates were stimulated by the addition of topoisomerases, the difference in the number of initiation events between control and cross-linked templates, however, remained constant (data not shown).
Another possibility for the reduced replication efficiency of templates containing cross-linked octamers could be a lower accessibility for the SV40 T-Ag to the origin sequences. To investigate the efficiency of T-Ag binding we performed digestion experiments with micrococcal nuclease (MNase) (Fig. 5).
To release and quantitate all DNA fragments bound by T-Ag the chromatin was extensively digested with MNase. The MNase protection pattern at specific sites was revealed by Southern blotting and hybridization with oligonucleotides com-plementary to the origin sequences. Upon extensive digestion with MNase a subnucleosomal fragment of around 80 bp became evident. This fragment consisted of origin DNA protected from MNase digestion by bound T-Ag. As recently shown (31), this fragment is protected by T-Ag against MNase digestion on protein-free DNA. Importantly, the same amount of DNA was protected by T-Ag in templates containing control and cross-

FIG. 4. Influence of histone cross-linking on initiation and elongation of chromatin replication.
To investigate the influence of the cross-linked histone octamers on initiation and elongation of replication mock-assembled SV40, DNA and chromatin reconstituted with either control or cross-linked histone octamers (1.2: 1 histone to DNA weight ratio) were investigated by pulse-chase analysis. A, at the indicated time points (30 s to 4 min) aliquots were removed and the isolated DNA was analyzed by 1.2% alkaline-agarose gel electrophoresis and autoradiography. As molecular weight marker 32 P-labeled SV40 DNA fragments were used. The sizes are indicated in bases (b). B, DNA synthesis was quantitated by trichloroacetic acid precipitation. C, the average lengths of the newly synthesized DNA fragments were determined by densitometric scanning of equally exposed autoradiograms (A), and the values were blotted against the elongation time. Open squares, DNA; closed circles, control; closed triangles, cross-linked octamers.

FIG. 5. Binding of the SV40 T-Ag to the origin is not significantly altered in cross-linked chromatin.
After initiation complex formation initiated SV40 DNA, control and cross-linked chromatin (1.2:1 histone to DNA weight ratio) were digested under replication conditions with micrococcal nuclease. Digestion was stopped at the indicated time points; the DNA was purified and analyzed by Southern blotting and hybridization with an oligonucleotide located within the SV40 origin sequence (nucleotides 5217-5242, lower strand of the SV40 genome). As a molecular weight marker a 32 P-labeled 123-bp ladder was used (M). The size of the marker is given in bp. The positions of mono-, di-, and trimer and the T-Ag footprint are indicated on the right. linked histone octamers, indicating that cross-linked histone octamers do not restrict the access of T-Ag to their binding sites.
Transfer of Parental Nucleosomes to the Replicated Daughter Molecules Is Not Impeded by Octamer Cross-linking-The precise mechanism of parental nucleosome transfer is still not completely understood. To determine whether histone octamer dissociation is necessary for the transfer of the parental nucleosomes, we investigated the chromatin structure of the daughter strands of templates containing control or crosslinked histone octamers by micrococcal nuclease digestion of the replicated daughter strands. Because of the absence of chromatin assembly factor-1 chromatin assembly from newly synthesized histones should not occur in this system (5). However, some free histone H2A/H2B dimers do exist in the extract and may confound the analysis of nucleosomes on replicated DNA. Therefore the extracts were depleted using SV40 DNA coupled to paramagnetic beads (29). To enable an efficient removal of H2A/H2B dimers, the coupled DNA was preassembled with purified H3/H4 tetramers (27,28). Depletion of S100 replication extracts was achieved by two consecutive incubations with the coupled beads. Bound proteins were eluted from the beads, precipitated with trichloroacetic acid, and analyzed by SDS-PAGE and silver staining (Fig. 6A). We found that H2A/H2B dimers associate with the H3/H4 tetramers from the input template only during the first incubation. This demonstrates that free H2A/H2B dimers could be efficiently removed at one incubation with H3/H4 bearing DNA. The depleted extract was used for in vitro replication of templates containing control and cross-linked histone octamers. Replicated chromatin was digested with micrococcal nuclease, and the purified products were analyzed by agarose gel electrophoresis and autoradiography (Fig. 6B). Because an assembly of new nucleosomes does not occur in this system, replicated daughter strands contain on an average only half of the num-ber of nucleosomes of the parental template. Therefore, micrococcal nuclease digestion of the daughter strands mainly results in mono-and dinucleosomes (8,10). We found with both templates a prominent monomeric band, demonstrating that the cross-linked histone octamer has been transferred to the daughter strands and that histone octamer dissociation is not necessary for the transfer of the parental nucleosomes. DISCUSSION It may be obvious that histone-DNA and histone-histone interactions within the nucleosome must be significantly altered to allow the passage of the replication machinery, consisting of DNA polymerases, helicase (s), ligases, proliferating nuclear antigen, replication protein A, and additional proteins (37). However, the studies reported here demonstrate that a dissociation of the histone octamer is not an absolute requirement for the passage of the replication fork. In fact, chromatin templates carrying dimethyl suberimidate-cross-linked histone octamers are completely replicated under the in vitro conditions (Fig. 3). Furthermore, cross-linked parental histone octamers are transferred to replicated DNA as shown under tightly controlled conditions (Fig. 6). These data are consistent with transcriptional studies that have shown that histone octamer dissociation is not required for transcription elongation through arrays of nucleosome cores in vitro (19), indicating that RNA and DNA polymerases are able to deal with the intact histone octamer.
Whether this mechanism functions in vivo is not clear. It is possible that histone octamers are displaced as intact units by the action of DNA polymerases and only subsequently undergo partial or complete dissociation, given the instability of histone octamers under physiological ionic conditions (38). Indeed, early experiments with density-labeled amino acids indicated that old and new histones do not intermix and that old nucleosomes are transferred as intact units to the daughter strands (39,40). Biochemical studies with purified phage replication proteins and a template containing a few artificially assembled nucleosomes also suggest that the whole histone octamer is transferred to the daughter strands without displacement (7). However, other in vitro studies showed that old nucleosomes dissociate during replication to give histone H3/H4 tetramers and histone H2A/H2B dimers. Old histone H3/H4 tetramers remain on DNA and combine on replicated DNA with either old or new histone H2A/H2B dimers to form postreplicative nucleosomes (15). These data are in agreement with electron microscopic analysis of replicating SV40 minichromosomes, suggesting that the H3/H4 tetramer is the transferred unit, which in a second step, binds histone H2A/H2B dimers to complement the full nucleosome core (10). In addition by using replicating SV40 minichromosomes isolated from lytically infected cells it has been shown that two nucleosomes in front of the replication fork are destabilized, probably because of the dissociation of histone H1 and one or two H2A/H2B dimers (13). From these data it was concluded that the advancing replication fork may induce a disassembly of the octamer in front of the fork.
By using pulse-chase analysis we have found that the number of active templates and thus the initiation rate of chromatin templates containing cross-linked histone octamers is reduced compared with control chromatin (Fig. 4). One possibility is that cross-linked octamers restrict the access of the SV40 T-Ag to the origin sequences and thereby inhibit efficient initiation of replication. We excluded this possibility and demonstrated that the same amount of T-Ag is bound to chromatin containing control or cross-linked histone octamers (Fig. 5). A second possibility is suggested by recent studies showing that nucleosomal organization determines the accessibility for topoisomerases (31,36) and thus the efficiency of initiation of FIG. 6. Transfer of the parental nucleosomes is not inhibited by the cross-linking of the histone octamers. To investigate the ability of cross-linked histone octamers to be transferred to the newly synthesized daughter strands 300 ng of SV40 DNA and reconstituted control or cross-linked chromatin (1.2:1 histone to DNA weight ratio) were replicated in a cytosolic S100 extract depleted of endogenous histones H2A and H2B. A, to investigate the efficiency of depletion of the S100 extract, bound proteins were eluted (2 M NaCl, 1ϫ Laemmliloading buffer), precipitated with trichloroacetic acid, and separated on an 18% SDS-PAGE. Input, 1-g beads preassembled with H3/H4 tetramers. M, histone marker. B, the replicated material was digested with micrococcal nuclease and stopped after 1, 5, and 20 min. Purified DNA was analyzed on a 1.3% Tris-glycine-agarose gel and autoradiography. As marker a 32 P-labeled 123-bp ladder was used. The sizes are indicated in bp.
replication. In fact cross-linked histone octamers could represent a more stringent barrier for supercoil diffusion than untreated control octamers. However, additional topoisomerases stimulated the replication of cross-linked and control minichromosomes to the same extent. A third possibility is that crosslinking of histone octamers might inhibit interactions with other as yet unidentified proteins necessary for efficient initiation of replication. There could be chaperone-like proteins that facilitate transient detachment of histones from the DNA template during replication and re-assembly onto DNA daughter strands after replication fork passage. A candidate protein could be a factor termed FACT recently purified from human cell nuclei that facilitates efficient transcript elongation through nucleosomally organized DNA (41,42). Indeed, the finding that the yeast counterpart of FACT (Spt16/pob3) interacts specifically with DNA polymerase ␣ (43) suggests a role for FACT in chromatin replication and histone displacement.