The Stability of Nucleosomes at the Replication Fork

J. Mol. Biol. (1996) 258, 224–239

The Stability of Nucleosomes at the Replication Fork Regula Gasser, Theodor Koller and Jose´ M. Sogo* Institute of Cell Biology ETH-Ho¨nggerberg, CH-8093 Zu¨rich, Switzerland

Purified simian virus (SV40) minichromosomes were photoreacted with psoralen under various conditions that moderately destabilize nucleosomes. This assay allows indirect distinction between stable nucleosomes, partially unravelled nucleosomes and nucleosomes containing (or lacking) histone H1. In replicating molecules the passage of the replication machinery destabilizes the nucleosomal organization of the chromatin fiber over a distance of 650 to 1100 bp. In front of the fork, an average of two nucleosomes are destabilized presumably by the dissociation of histone H1 and the advancing replication machinery. On daughter strands, the first nucleosome is detected at a distance of about 260 nucleotides from the elongation point. This nucleosome is interpreted to contain no histone H1, while no stepwise association of (H3-H4)2 tetramers with H2A/H2B dimers on nascent DNA can be detected in vivo. The second nucleosome after the replication fork appears to contain histone H1. The prolonged nuclease sensitivity of newly replicated chromatin described in the literature therefore may not be due to a slow reassociation of histone H1. 7 1996 Academic Press Limited

*Corresponding author

Keywords: replication; chromatin; nucleosomes; SV40; psoralen

Introduction Replication of eukaryotic chromosomes involves both the duplication of DNA and its assembly into chromatin. Electron microscopy has revealed putative nucleosomes in the vicinity of replication forks on prereplicative as well as on newly replicated sections (McKnight & Miller, 1977; Tsanev & Semionov, 1985). However, the replication machinery moving along the DNA transiently disrupts the chromatin fiber and a random segregation of parental nucleosomes to the daughter strands occurs (Cusick et al., 1984; Sogo et al., 1986; Burhans et al., 1991; reviewed by Gruss & Sogo, 1992). Concerning the transfer mechanism of the parental nucleosomes to the daughter strands it has been postulated that they remain associated with the DNA during the passage of the replication fork (Bonne-Andrea et al., 1990; Krude & Knippers, 1991, 1993; Randall & Kelly, 1992). In contrast, in vitro experiments using a large excess of competitor DNA have shown that histones can be trapped during replication, indicating that they remain only loosely associated with DNA during the passage of the replication fork (Gruss et al., 1993). Moreover, by following the fate of labelled histones in vivo, it has been postulated that the nucleosomes dissociate as Abbreviations used: ss-bubbles, single-stranded bubbles. 0022–2836/96/170224–16 $18.00/0

H2A/H2B dimers and (H3.H4)2 tetramers (Jackson, 1987, 1990). In vivo studies on the mechanism of nucleosome assembly on nascent DNA revealed that newly synthesized histone (H3.H4)2 tetramers were selectively deposited on newly replicated DNA, whereas new H2A/H2B dimers associate with either new or old (H3.H4)2 tetramers (Jackson, 1987, 1990). In vitro studies provide growing evidence for this two-step mechanism of nucleosome assembly (reviewed by Sogo & Laskey, 1994). Whatever the mechanisms and factors involved in nucleosome assembly, a general property of newly replicated chromatin is its increased nuclease sensitivity as compared with that of bulk chromatin (Klempnauer et al., 1980; De Pamphilis & Wassarman, 1982). Therefore, the region of newly assembled nucleosomes has been designated as immature chromatin (Cusick et al., 1981). The processes involved in the maturation of chromatin are still unknown. They may involve changes in histone composition and/or histone modification. Since histone H1 protects linker DNA from nuclease attack and prevents nucleosome sliding, it has been suggested that maturation of newly replicated chromatin is mediated by the addition of histone H1 (Klempnauer et al., 1980; Jackson & Chalkley, 1981a,b; Galili et al., 1981; Schla¨ger, 1982; Cusick et al., 1983). When the appearance of histone H1 in nascent chromatin was followed, a wide time-range of the deposition of 7 1996 Academic Press Limited

Chromatin Structure at Replicating Forks

histone H1 has been reported, ranging from 30 seconds to 10 to 20 minutes (Worcel et al., 1978; Annunziato et al., 1981; Stefanovsky et al., 1990; Bavykin et al., 1993). In order to gain further insights into the process of chromatin replication, we analyzed the stability of nucleosomes in the vicinity of the replication fork. By using psoralen as a probe to detect freely accessible DNA in chromatin, we focused on the nucleosomes immediately in front and behind the replication fork. Simian virus (SV40) chromatin was selected as the model system because its nucleosomal organization is indistinguishable from that of the host cell (Shelton et al., 1980). The experimental approach was to isolate SV40 minichromosomes that subsequently were photoreacted with psoralen under various conditions that moderately destabilize nucleosomes. The comparison of these in vitro destabilized SV40 minichromosomes with replicating molecules cross-linked in vivo revealed that on the average four (21) nucleosomes are destabilized due to the passage of the replication machinery.

Results Psoralen cross-linking of SV40 bulk chromatin under various conditions Psoralen was used as a probe to detect nucleosomal structures. Psoralen intercalates into freely accessible DNA but not into DNA of nucleosomes, which is protected by histones (Hanson et al., 1976; Sogo et al., 1984). Upon photoactivation of psoralen, the DNA in chromatin becomes cross-linked in the linker DNA only. After deproteinizing the DNA and spreading it under denaturing conditions for electron microscopy, DNA that originates from a nucleosome appears as a single-stranded DNA bubble (ss-bubble), and the cross-linked linker DNA as a short duplex region. Therefore, bulk chromatin prepared in this way appears as rows of ss-bubbles (Sogo et al., 1984; Conconi et al., 1984). When soluble H1-depleted chromatin is exposed to very low ionic strength and high pH (pH 10), histone-DNA interactions are altered (Libertini & Small, 1982, 1984; see also Van Holde, 1988). Under such conditions histone-complexed DNA is accessible to psoralen and the DNA appears fairly continuously cross-linked, in contrast to histone H1-containing chromatin, which still leads to rows of ss-bubbles (Conconi et al., 1984). To determine the fate of chromatin structure during DNA replication, SV40 minichromosomes were exposed to moderate nucleosome destabilizing conditions in order to detect possible changes in histone-DNA interactions, which can occur in nucleosomes immediately ahead of and behind the replication fork. Therefore, the treatment of SV40 minichromosomes with psoralen under various destabilizing conditions should allow one to distinguish between loose and tight histone-DNA interactions. Loosened histone-DNA interactions

225 occurring in vivo will result in an increase of psoralen-accessible DNA sites, whereas tight histone-DNA contacts will remain insensitive to psoralen. The effects of the various conditions on the extent of psoralen cross-linking were first analyzed in bulk chromatin, i.e. in non-replicating SV40 minichromosomes. Experimentally, SV40 minichromosomes were isolated, partially purified on a sucrose gradient and extensively dialyzed against the corresponding buffers. Since the buffers used were always of low ionic strength, they will be referred to only according to their pH (pH 7 or pH 10). After dialysis, the SV40 minichromosomes were photoreacted with psoralen, and an aliquot of the samples was analyzed by gel electrophoresis to test for the integrity of the histone proteins (Figure 1a). SV40 minichromosomes isolated in this way are referred to as native minichromosomes. In order to test the effects of a lack of histone H1 on the chromatin structure at either pH 7 or pH 10, part of the SV40 minichromosomes were depleted of histone H1 by centrifugation over a sucrose gradient containing 500 mM NaCl. The further treatment and analysis was exactly as for native SV40 minichromosomes. To minimize the salt-induced rearrangement of nucleosomes (Watkins & Smerdon, 1985), all manipulations of both intact and H1-depleted minichromosomes were done at 4°C. Histone H1-depleted SV40 minichromosomes are referred to as pH 7 −H1 and pH 10 −H1, respectively. Once it was confirmed that no degradation of the histone proteins was detectable (Figure 1b), the DNA cross-linked in chromatin from the parallel aliquots of samples was analyzed by electron microscopy. The criteria used for quantitative analysis of chromatin structure are both the average size of the ss-bubbles and the r-value, which gives the proportion of nucleosomal DNA on a given section of chromatin, i.e. its nucleosomal density (for details see Table 1). As a reference for all the different conditions of psoralen cross-linking, SV40 minichromosomes cross-linked in vivo, i.e. in intact cells, were used. In good agreement with previous results (Sogo et al., 1986) the data of these in vivo minichromosomes reveal that the average size of the ss-bubbles and the r-value correspond to regularly spaced nucleosomes (Figure 2a, Table 1). At pH 7, the DNA of both isolated native and the corresponding histone H1 -depleted non-replicating SV40 minichromosomes show regular rows of ss-bubbles (Figure 1c and d). Their analysis revealed very similar values. The distribution of the ss-bubbles (Figure 2b and c) showed two peaks corresponding to ss-bubbles with a size of roughly 150 nt and 90 nt, respectively (Table 1). The average sizes of the ss-bubbles of the main peaks are similar to those of in vivo cross-linked SV40 minichromosomes. The r-values for pH 7 are slightly reduced as compared with in vivo cross-linked SV40 minichromosomes, but are still in the range of normal chromatin (Table 1). The slight reduction of the r-values reflects the presence of small ss-bubbles, and probably a

226

Figure 1. Characterization of isolated SV40 minichromosomes. At 36 to 40 hours after infection of exponentially growing cells, SV40 minichromosomes were extracted, purified on a sucrose gradient, extensively dialyzed against pH 7 or pH 10 buffers of low ionic strength and photoreacted with psoralen. Protein analysis of native (a) and histone H1-depleted (b) SV40 minichromosomes photoreacted with psoralen after dialysis against pH 7 (lanes 1) and pH 10 (lanes 2) respectively. M, histone marker, in which histones of SV40 host cells were used. DNA from psoralen cross-linked native (c and e) and H1-depleted (d and f) SV40 minichromosomes spread under denaturing conditions. Minichromosomes were photoreacted at pH 7 (c and d) and pH 10 (e and f). The bar represents 500 nucleotides.

Chromatin Structure at Replicating Forks

more efficient cross-linking in in vitro samples compared with in vivo samples, as deduced by the smaller amount of ss-bubbles with a size corresponding to dinucleosomes. Since the average number of ss-bubbles per isolated non-replicating molecule is similar to that in in vivo cross-linked SV40 minichromosomes (Table 1), the most likely interpretation of the small ss-bubbles observed in low ionic strength and pH 7 is that they correspond to partially unravelled nucleosomes. According to earlier observations (Gruss et al., 1993) showing that in vitro assembled subnucleosomal particles containing H3.H4-N1 complexes protect 180 bp of DNA against cross-linking, we believe that in these partially unravelled or swollen nucleosomes only the DNA wrapped around the (H3.H4)2 tetramer remains protected from being cross-linked by psoralen, giving rise to ss-bubbles of about 90 nt, even though the histones H2A and H2B are present. In the following sections the small ss-bubbles are considered to represent partially unravelled nucleosomes (for details, see Discussion). Isolated (non-replicating) native SV40 minichromosomes photoreacted with psoralen at pH 10 (Figure 1e) are similar to pH 7 minichromosomes (compare Figure 1c and e, and Figure 2b and d; although the number of small ss-bubbles per molecule is increased (Table 1)). The increase of the proportion of small ss-bubbles can be attributed to the conditions of low ionic strength and pH 10, which cause a further destabilization of the nucleosomes (see Discussion). Therefore, the observed partitioning of the ss-bubbles into a small and a nucleosomal-size class indicates a progressive destabilization of the nucleosomes. Analysis of the corresponding histone H1depleted SV40 minichromosomes cross-linked at pH 10 showed that the chromatin structure of these molecules was highly disturbed. The DNA molecules were cross-linked to a large extent, and few ss-bubbles, mostly small were visualized (Figures 1f and 2e). The r-value and the number of bubbles per molecule were drastically reduced (Table 1). These results can best be explained by a partial or complete unfolding of the nucleosomes (see Discussion). The data described in this section suggest that ss-bubbles of about 150 nt originate from stable nucleosomes, ss-bubbles of about 90 nt originate from swollen or partially unfolded nucleosomes, and the ‘‘disappearance’’ of ss-bubbles at pH 10, i.e. unravelling of nucleosomes, indicates nucleosomes lacking histone H1. Analysis of replicating SV40 molecules psoralen cross-linked under destabilizing conditions The DNA of cross-linked replicating minichromosomes, collected from the same samples described in the previous section, was analyzed as well. In order to determine the stability of the nucleosomes before and after the passage of the replication fork,

227

Chromatin Structure at Replicating Forks

Table 1. Analysis of bulk (non-replicating) SV40 minichromosomes No. of molecules analyzed

Size of ss-bubbles (nt)

In vivo

12

pH 7 native

10

pH 7 −H1

10

pH 10 native

10

pH 10 −H1

40

— 153 2 47 87 2 15 149 2 29 89 2 12 146 2 29 85 2 11 146 2 28 86 2 16 —

Experiment

% small ss-bubbles

r-value

ss-bubbles per molecule

6

0.81 2 0.06

25 2 1.5

21

0.69 2 0.07

24 2 2.0

22

0.67 2 0.04

24 2 1.9

38

0.62 2 0.05

24 2 1.3

78

0.28 2 0.08

13 2 3.0

The bubble size is given as the mean value and the standard deviation of the corresponding normal distribution shown in Figure 2. The percentage of small bubbles represents the proportion of ss-bubbles constituting the peak of subnucleosomal size. The r-value was calculated as the ratio of the sum of the lengths of single-stranded DNA of one molecule over its total DNA contour length. The r-value is a measure of the nucleosome density on an SV40 minichromosome or on a given section of chromatin (Sogo et al., 1986). As a control of the spreadings, linear DNA of psoralen cross-linked rat liver chromatin was always coprepared and showed a bubble size in the range from 154 to 162(226) nucleotides and an r-value from 0.78 to 0.82(20.10).

parental and daughter strands, respectively, were independently considered. In vivo cross-linked replicating SV40 molecules were again used as a reference. A characteristic late replicative intermediate is shown in Figure 3a. The size and distribution of the ss-bubbles of both the parental daughter strands (Figure 4a and b; Table 2) are similar to those described for non-replicating molecules (compare with Figure 2a and Table 1). The r-value of the parental strand (Table 3) is very close to the r-value of nonreplicating molecules (compare with Table 1). However, the r-value of the daughter strands is reduced (Table 3) due to the prenucleosomal DNA (distance between the branch point of the fork and the first nucleosome on nascent DNA; see Figure 5a). After subtraction of the lengths of these prenucleosomal stretches of continuously crosslinked DNA, the r-value is raised in the same range as the r-value of the parental strand and of non-replicating molecules (Table 3, compare also with Table 1). In replicating molecules of isolated native SV40 minichromosomes, psoralen cross-linked at pH 7 (Figure 3b), the size and distribution of ss-bubbles gave values similar to those for non-replicating molecules (Figure 4c and d). However, the r-values of the parental and daughter strands decrease as compared with those of in vivo cross-linked replicating molecules (Table 3) or to isolated native bulk minichromosomes (Table 1). Especially, the r-value of daughter strands remains slightly reduced even after the substraction of the prenucleosomal stretches (Table 3). Together with the reduced number of ss-bubbles per molecule as compared to in vivo cross-linked minichromosomes (referring to one unit length of SV40; Table 2), this suggests that all these rather small but persistent differences probably indicate destabilization of a few nucleosomes in the vicinity of the replicating fork (see below).

The parental DNA strands of replicating molecules of isolated native SV40 minichromosomes psoralen cross-linked at pH 10 are organized in a rather regular row of ss-bubbles, whereas on the daughter strands the bubble pattern is less regular under these conditions (Figure 3c). In contrast to replicating molecules cross-linked in vivo or at pH 7, the proportion of small ss-bubbles clearly increased (Figure 4e and f; Table 2) and the stretches of cross-linked duplex DNA around the replication fork in both nascent and parental strands appear to be significantly longer (Figure 3c). The r-value of the parental strand is below the range of a normal nucleosomal density. After subtracting the distances of cross-linked duplex prefork DNA (distance between the branch-point of the fork and the last prefork nucleosome; see Figure 5a), the r-value increases (Table 3) close to normal nucleosomal density. The r-value of the daughter strands is drastically reduced corresponding to about half the r-value of the parental strand or of non-replicating molecules. After subtraction of the cross-linked duplex DNA stretches behind the replication fork, the r-value of the daughter strands rises but is still reduced as compared with the parental strands or non-replicating molecules (Table 3 see also Table 1). All these r-values ((+) and (−) prenucleosomal and prefork DNA) indicate a destabilization or disturbance of nucleosomes mainly in close vicinity around the replication fork. Moreover, these data show that the extent of DNA involved in the chromatin disruption process during the passage of the fork is less than that involved in the chromatin reassembly of the newly replicated chromatin (see below). The stability of nucleosomes in front of the replication fork The suggestion that nucleosomes are destabilized at the replication fork prompted us to further

228 examine in detail the nucleosomal organization in this particular region. In replicating molecules cross-linked in vivo the chromatin just in front of the fork appeared either as ss-bubbles (60%; Figure 5b and c) or as cross-linked duplex DNA (40%; Figure 5d) in about the same proportions as previously reported (Sogo et al., 1986). The length of prefork duplex DNA was measured and corresponds on the average to 79 bp (Table 4). The spread of the individual values shows no Gaussian distribution, and therefore the whole range of the values observed in front of the replication forks are given in Table 4. The large spread indicates a heterogeneous organization of this chromatin section, i.e. reflects the dynamics of the replication process. In isolated native molecules cross-linked at pH 7 the proportion of preforks appearing as ss-bubbles was decreased to 36%. The length of cross-linked duplex appearing DNA in front of the fork is, on the average, 131 bp (subnucleosomal length). Both

Chromatin Structure at Replicating Forks

observations appear to indicate that nucleosomes immediately preceding the fork are somehow affected by the advancing replication machinery. In isolated native replicating molecules crosslinked at pH 10 (Figure 5e to g) the parental DNA in front of the fork is organized similar to that in molecules cross-linked at pH 7, i.e. in 30% ss-bubbles and in 70% duplex DNA was found. However, the average length of the prefork duplex appearing DNA was increased to 180 bp on the average (Table 4). Since the length measurements show that more than half of the prefork duplex DNA is longer than 150 bp, we conclude that not only nucleosomes directly involved in the replication-elongation process are affected, but in at least half of the cases additionally one contiguous nucleosome is destabilized. In order to further dissect the events occurring in front of the advancing replication machinery, the ss-bubbles immediately preceding the replication fork were analyzed. It is interpreted to correspond

Figure 2. Length measurements of the ss-bubbles of bulk (nonreplicating) SV40 molecules. Purified DNA from psoralen crosslinked minichromosomes was spread under denaturing conditions for analysis by electron microscopy. The histograms of the length measurements of the ss-bubbles are shown together with superpositions of fitted normal distributions. The cross-linking conditions are indicated in the upper right corner of the panels. −H1 stands for chromosomes depleted of histone H1. Detailed data are summarized in Table 1.

Chromatin Structure at Replicating Forks

229

Figure 3. DNA from psoralen cross-linked replicating SV40 minichromosomes. In a minichromosomes were photoreacted in vivo. In b and c native sucrose gradient-purified minichromosomes were photoreacted after dialysis against buffers of low ionic strength at pH 7 (b) or pH 10 (c). Both unreplicated (between arrows) and daughter sections show the typical array of denatured ss-bubbles. In c the bubble pattern on the daughter strands is less regular than on the parental strands. Arrows designate fork movement. The bar represents 500 nucleotides.

to a nucleosome involved in replication. The definition of the corresponding ss-bubbles is shown in Figure 5a. The size of ss-bubbles involved in replication did not follow a normal distribution but rather showed a plateau (between 60 and 280 nt), under all cross-linking conditions investigated (not shown). This is in complete agreement with the findings of Sogo et al. (1986), and since has been interpreted as the replication fork moving into the nucleosomes. The last prefork ss-bubble (for definition see Figure 5a) of in vivo cross-linked molecules showed a size distribution as that found for ss-bubbles of non-replicating minichromosomes (Table 4; compare with Table 1). In isolated molecules crosslinked at pH 7, this last prefork ss-bubble showed two size distributions similar to the ss-bubbles of bulk chromatin (Table 4, compare with Table 1). One class represents ss-bubbles of nucleosomal length. The other class (about 40%) corresponds to the ss-bubbles in the range of 90 nt. These data indicate a tendency in which, in addition to the nucleosome involved in replication, already the last entire prefork nucleosome (as defined in Figure 5a) appears to be affected by weakened core histone-

DNA interactions. The second last prefork ss-bubble (Table 4) showed a size distribution similar to those of the rest of the parental ss-bubbles or the ss-bubbles of non-replicating molecules (Table 4, compare with Table 1). Since in isolated native molecules cross-linked at pH 10 the lengths of prefork duplex DNA are considerable, among the last ss-bubbles seen on the parental strands only those detected within 50 to 300 bp from the fork correspond to the last prefork ss-bubbles. The last ss-bubbles detected more than 300 bp away from the fork correspond to second last prefork ss-bubbles. The measurements of ss-bubbles according to this definition are summarized in Table 4. The size distribution of the last prefork ss-bubble (90 2 16 nt) shows mainly one class (65% of the ss-bubbles) corresponding to the size distribution observed for bulk histone H1-depleted molecules cross-linked at pH 10 (Figure 2e and Table 1). Moreover, in these native molecules cross-linked at pH 10, in more than half of the cases this last prefork ss-bubble cannot be detected, i.e. is destabilized. Therefore, the observed last prefork ss-bubbles detected at pH 10 are interpreted to represent residual ss-bubbles as seen in the bulk

230

Chromatin Structure at Replicating Forks

Figure 4. Length measurements of the ss-bubbles of DNA from replicating SV40 molecules. Molecules were either cross-linked in vivo or after dialysis to pH 7 or pH 10 as indicated in the upper right corner. Histograms and superpositions of fitted normal distributions are presented. The corresponding data are presented in Tables 2 and 3.

histone H1-depleted minichromosomes, and therefore might correspond to histone H1-depleted nucleosomes. In contrast, but in support of the above interpretation, the second last prefork ss-bubble again showed the same size distribution as that found in native non-replicating molecules at pH 10 (Table 4, compare with Figure 2d and Table 1). In summary, one can conclude that, besides the nucleosome immediately ahead of the replication apparatus, the ‘‘last entire’’ nucleosome detected in vivo in front of the fork has features similar to histone H1-depleted nucleosomes. In favor of this interpretation is: (1) in isolated native molecules cross-linked at pH 10, the prefork DNA is,

on the average, 180 bp instead of 79 bp in vivo; (2) for pH 10, the size distribution of the last prefork ss-bubble in native molecules is similar to those of histone H1-depleted bulk SV40 minichromosomes. The stability of nucleosomes on nascent DNA In light of the proposed two-step process of chromatin maturation, namely (1) conversion of prenucleosomal DNA into immature nucleosome oligomers possibly lacking histone H1, and (2) maturation of newly assembled chromatin into a structure with increased nuclease resistance (Cusick et al., 1983), one could expect that immature

Table 2. Analysis of the ss-bubbles of replicating SV40 molecules Size of ss-bubbles (nt) Experiment

No. of molecules analyzed

In vivo

49

pH 7 native

56

pH 10 native

50

% small ss-bubbles

Parental

Daughter

Parental

Daughter

ss-bubbles per molecule

— 171 2 38 85 2 17 150 2 36 87 2 15 143 2 34

— 163 2 45 81 2 17 143 2 37 85 2 13 148 2 26

5

7

26 2 2.8

16

22

21 2 1.8

42

50

19 2 2.4

The number of molecules analyzed in each experimet is indicated, giving rise to about 500 ss-bubbles on the parental strand and about 1000 ss-bubbles on each daughter strand. The bubble size is given as the mean value and the standard deviation of the corresponding normal distribution shown in Figure 4. The percentage of the small bubbles represents the proportion of ss-bubbles constituting the peak of subnucleosomal size. Coprepared DNA of cross-linked rat liver chromatin showed ss-bubble sizes from 154 to 162 (226) nucleotides.

231

Chromatin Structure at Replicating Forks

Table 3. The r-values of replicating SV40 molecules r-value Experiment In vivo pH 7 native pH 10 native

r-value (−) fork

Parental

Daughter

Parental

Daughter

0.79 2 0.10 0.66 2 0.12 0.56 2 0.11

0.59 2 0.13 0.43 2 0.14 0.31 2 0.10

0.81 2 0.10 0.73 2 0.10 0.64 2 0.11

0.75 2 0.10 0.64 2 0.14 0.51 2 0.13

The r-values were calculated as the ratio of the sum of the lengths of single-stranded DNA of one molecule over its total DNA contour length for the parental and daughter strand, respectively, and are shown to the left with the corresponding standard deviation (2). The r-values (−) fork were calculated as above but after subtracting the lengths of either the duplex DNA just in front of the fork from the total parental strand length or the duplex DNA behind the fork from the total daughter strand length.

nucleosomes might readily be destabilized under the low salt and pH conditions used during cross-linking. If so, this would result, in an increase of the distance between the branch-point of the replication fork and the first ss-bubble on the daughter strands. The lengths of these distances on

the daughter strands were determined in replicating molecules cross-linked under the conditions described. The leading and lagging strands of nascent DNA (see Sogo et al., 1986) were not measured separately, since this distinction was not always possible.

Figure 5. Representative forks of SV40 replicating molecules. a, Definition of ss-bubbles and of duplex DNA stretches at the forks. b, c and d, Minichromosomes photoreacted in vivo. In e, f and g, replication forks from sucrose gradient-purified minichromosomes photoreacted at pH 10 and low ionic strength are shown. The arrowheads indicate the positions of the first ss-bubble on daughter strands, and the arrows point to the last prefork ss-bubble. In e, f and g the distance from the growing point to either the last ss-bubble on the parental strands or to the first ss-bubble on the nascent strands increases significantly. The bar represents 500 nucleotides.

232

Chromatin Structure at Replicating Forks

Table 4. Analysis of the length measurements of duplex DNA and of ss-bubbles in front of the replication fork

Experiment In vivo pH 7 native pH 10 native

Distance from the fork to the last ss-bubble on parental strandsa (bp) 79 min. 0 max. 386 131 min. 0 max. 378 180 min. 0 max. 534

Size of last prefork ss-bubble (nt)

Size of second last prefork ss-bubble (nt)

[82]

— 151 2 42

(5) [108]

— 162 2 28

(7) [99]

[92]

100 2 13 151 2 25

(38) [84]

91 2 16 155 2 17

(18) [90]

[90]

90 2 16 —

(65) [83]

92 2 22 159 2 27

(45) [90]

The length of duplex DNA immediately in front of the replicaton fork was analyzed and the arithmetical averages and standard deviations of the measurements are indicated. The size distribution of the ss-bubbles immediately preceding the replication fork was analyzed. The bubble size (in nucleotides) is given as the mean value and the standard deviation of the corresponding normal distributions. The numbers in parentheses correspond to the percentage of small ss-bubbles constituting the peak of subnucleosomal size. The ss-bubbles measured constitute a single normal distribution, of which the mean value is given. The number of analyzed forks is given in square brackets. a The values in this column do not show a Gaussian distribution. Therefore besides the mean values the spread of the values is indicated by the smallest (min.) and the largest (max.) values.

The average length of prenucleosomal DNA in isolated native replicating molecules cross-linked at pH 7 is similar to that determined for control in vivo cross-linked intermediates (Figure 5b, c and d; Table 5), indicating that immature nucleosomes at physiological pH are not preferentially destabilized even at low ionic strength. However, in isolated native replicating molecules cross-linked at pH 10 (Figure 5e, f and g) the distances from the branch-point of the fork to the first ss-bubble on the daughter strands were significantly increased (Table 5). In comparison with molecules crosslinked in vivo, this average length is shifted by about 180 bp, corresponding to the length of one nucleosome. Therefore, it can be concluded that at pH 10 and low ionic strength, on the average, one (21) immature nucleosome after the replication fork is destabilized in native SV40 molecules. Since our assay allows us to distinguish between stable nucleosomes and partially unravelled nu-

cleosomes of about 90 nt in length, corresponding to (H3.H4)2 tetramers (Gruss et al., 1993), and in order to gain more information about the assembly of nucleosomes on nascent DNA, the first ss-bubbles on the daughter strands behind the replication fork were individually analyzed. Under the assumption of the proposed model of nucleosome assembly (first deposition of (H3.H4)2 tetramers followed by the addition of H2A/H2B dimers; see. Introduction), one might be able to detect these sequential steps under our destabilizing conditions The first ss-bubble behind the replication fork of molecules cross-linked in vivo showed a size distribution (Table 5) similar to that of ss-bubbles of isolated bulk chromatin (Table 1). This indicates that in vivo nucleosome cores are rapidly assembled. In the analysis of isolated native molecules crosslinked at pH 7 the corresponding ss-bubbles showed distributions similar to that of the bulk of ss-bubbles cross-linked under these conditions

Table 5. Analysis of the length measurements of duplex DNA and of ss-bubbles behind the replication fork Experiment

Distance from the fork to the first ss-bubble on daughter strands (bp)

In vivo

270 2 149

[177]

pH 7 native

260 2 187

[224]

pH 10 native

443 2 208

[203]

Size of first ss-bubble on daughter strand (nt) — 144 2 47 93 2 20 133 2 48 119 2 27 —

(10) [176] (26) [228] (75) [198]

In the column to the left, the continuously cross-linked duplex DNA behind the replication fork was measured, which corresponds to the distance between the branch-point of the replication fork and the first ss-bubble on the daughter strand. The average length of prenucleosomal DNA is given by the mean value and the standard deviation of the corresponding normal distributions. The size distribution of the first ss-bubble just behind the replication fork was analyzed exactly as described for Table 4. The percentage of small ss-bubbles is given in parentheses. The number of analyzed forks is given in square brackets.

233

Chromatin Structure at Replicating Forks

(compare values given in Table 5 with those in Table 1). These data are interpreted to indicate that most or all of the first nucleosomes behind the fork are probably rapidly assembled with a full set of core histones. Since the length of prenucleosomal duplex appearing DNA is increased at pH 10 compared with pH 7, the ‘‘first’’ ss-bubbles detected behind the fork in molecules cross-linked at pH 10 probably represent later stages in chromatin assembly. Indeed, in native molecules cross-linked at pH 10, these ss-bubbles show an intermediate size (119(227) nt; Table 5) compatible with the mean value of the two classes of ss-bubbles determined for isolated native bulk chromatin at pH 10 (Table 1). Therefore, these first ss-bubbles correspond presumably to mature nucleosomes. In summary, the described observations confirm that the assembly of nucleosomes on nascent DNA is a very fast process. We showed that the histone H1-dependent maturation of chromatin is completed within a distance up to 650 bp (443(2208) bp; Table 5) behind the fork. However, it was not possible to follow the proposed steps in nucleosome core assembly. Analysis of histone H1-depleted replicating SV40 molecules Histone H1-depleted replicating SV40 minichromosomes cross-linked at pH 7 are almost identical with replicating native molecules cross-linked under the same conditions (Figure 6a; for more details, see Gasser, 1993). The fact that in these two kinds of minichromosomes the nucleosomal organization at the replication forks is indistinguishable (not shown), suggests that the salt treatment (H1 depletion) alone at 4°C does not significantly affect the chromatin structure at nascent or parental strands. In contrast, histone H1-depleted replicating SV40 minichromosomes cross-linked at pH 10 show a highly disturbed chromatin structure (Figure 6b). The ss-bubbles show properties and size distributions similar to those of non-replicating molecules (data not shown, but see Figure 2e and Table 1). The r-values of both the parental and the daughter strands are dramatically decreased (between 0.2 and 0.4), even after subtraction of the continuously cross-linked duplex DNA around the fork, which in these molecules show considerable lengths (386(2273) bp and 693(2466) bp for parental and daughter strands, respectively). The number of ss-bubbles per replicating molecule, corresponding to one unit length of SV40, is reduced to about ten. This reduction of the number of ss-bubbles and of their size is most probably due to the unravelling of the nucleosome cores under the destabilizing conditions used. Since a disturbance of the chromatin structure similar to that observed in histone H1-depleted chromatin at pH 10 was seen in the domains near the replication fork of native SV40 minichromosomes cross-linked under pH 10 conditions but not

Figure 6. DNA of H1-depleted replicating SV40 minichromosomes. Sucrose gradient-purified H1-depleted minichromosomes were photoreacted under nucleosome destabilizing conditions at low ionic strength and pH 7 (a) or pH 10 (b). In b the arrangement and size of ss-bubbles is highly disturbed. Besides predominantly small ss-bubbles, many stretches of continuously crosslinked duplex DNA are detected. Between arrows, the course of the parental strands. Arrows designate fork movement. The bar represents 500 nucleotides.

in the molecules cross-linked at pH 7, we conclude that this disturbance could be due to the absence of histone H1.

Discussion Detection of altered histone-DNA interactions in nucleosomes

Cross-linking of native SV40 chromatin at pH 7 and pH 10 and low ionic strength In order to study the histone-DNA interactions in nucleosomes in the vicinity of the replication fork, isolated replicating SV40 minichromosomes were

234 photoreacted with psoralen under moderate nucleosome destabilizing conditions. As a reference, SV40 minichromosomes psoralen cross-linked in vivo were analyzed, which show ss-bubbles of nucleosomal size and a nucleosome density as observed by Sogo et al. (1986). By analyzing isolated native non-replicating molecules, besides ss-bubbles of nucleosomal size, we find a certain amount of smaller ss-bubbles (about 90 nt) in molecules cross-linked at physiological pH and low ionic strength. The proportion of these small ss-bubbles increases further in molecules crosslinked at pH 10 and low ionic strength, but the total number of ss-bubbles per molecule remains the same as in molecules cross-linked in vivo (Sogo et al., 1986). The small ss-bubbles seen under cross-linking conditions of low ionic strength are probably related to the low-salt transition of nucleosomes, which has been observed to occur below 10 mM ionic, strength (Gordon et al., 1978; Wu et al., 1979; Libertini & Small, 1980). The low-salt transition was suggested from investigations of mononucleosomes depleted of histone H1 and described as a gradual swelling of the nucleosome if the ionic strength was decreased below 26 mM (Wu et al., 1979). By further decreasing the ionic strength, an abrupt conformational change of the nucleosome was observed between 3 and 0.7 mM ionic strength (Gordon et al., 1978; Wu et al., 1979; Martinson et al., 1979; Dietrich et al., 1980). Whereas the gradual swelling has been observed to occur both in histone H1-containing and H1-depleted nucleosomes, the abrupt transition or unfolding of the nucleosome below 3 mM ionic strength was observed only in histone H1-depleted chromatin (Thoma et al., 1979; Buch & Martinson, 1980). During the gradual swelling, disruption of histone H2B-H4 contacts was detected (Martinson et al., 1979), and the unfolding has been explained by electrostatic repulsion forces (Wu et al., 1979; Labhart et al., 1981). A model of the mechanism of the low-salt transition has been presented, which suggests the separation of H2A/H2B dimers from the (H3.H4)2 tetramer (Libertini & Small, 1982; Brown et al., 1991). Since our destabilizing conditions are in the range of 7 mM ionic strength, the swelling of nucleosomes under low-salt conditions as described in the literature has to be taken into account. Considering the model of the low-salt transition, the loosened contacts between H2A/H2B dimers and (H3.H4)2 tetramers probably allow an increased access of psoralen to the nucleosomal DNA. Thereby, mainly DNA interacting with the (H3.H4)2 tetramer remains protected from psoralen cross-linking, resulting in the small ss-bubbles observed. This is in agreement with findings from in vitro reconstitution experiments, in which H3/H4-N1 complexes upon micrococcus nuclease digestion gave a DNA repeat of 65 bp and multiples thereof (Zucker & Worcel, 1990), and where about 80 bp remained protected against psoralen cross-linking (Gruss et al., 1993).

Chromatin Structure at Replicating Forks

This psoralen protection increases to about 165 bp upon addition of histones H2A/H2B in the presence of nucleoplasmin (Gruss et al., 1993). Therefore, the small ss-bubbles of about 90 nt most probably correspond to swollen or partially unravelled nucleosomes, in which mainly the DNA wrapped around the histone (H3.H4)2 tetramer remains protected from being cross-linked by psoralen (in isolated native minichromosomes about 20% at pH 7 and about 40% at pH 10). However, we cannot formally exclude the possibility that other than histone proteins that stick to our minichromosomes might contribute to the observations described here. The higher proportion of small ss-bubbles observed in isolated native SV40 minichromosomes at pH 10 and low ionic strength as compared with pH 7 is consistent with the finding that at high pH the low-salt transition of nucleosomes takes place at about 7 mM salt, whereas at physiological pH the low-salt transition occurs only at 3 mM and below (Libertini & Small, 1982). Investigations of the effects of pH on nucleosome cores (ie. −H1), have shown little change in core particle size and shape, and were interpreted to mean that pH effects induce a loosening rather than an opening of the nucleosomal structure (Libertini & Small, 1984). Furthermore, electron microscopic studies of chromatin fibers showed that native chromatin at high pH and very low ionic strength (1 mM) appeared as ‘‘beads on a string’’ (Labhart et al., 1981). The beads on a string type of fiber has been interpreted to represent partially unravelled nucleosomes (Thoma & Koller, 1981). In contrast, the usual structure of native chromatin at pH 7 corresponded to a loose, zigzag-shaped arrangement of nucleosome beads at very low ionic strength, and to a closed zigzag, or fiber with about two nucleosomes per diameter, in low ionic strength (10 mM; Thoma et al., 1979; Labhart et al., 1981). Thus, the increase in the proportion of swollen nucleosomes, i.e. small ss-bubbles, that we observe is in agreement with the above studies. Our findings are consistent also with the observation that the gradual swelling of nucleosomes takes place in the presence and the absence of histone H1 (Burch & Martinson, 1980).

Cross-linking of histone H1-depleted SV40 chromatin at pH 7 and pH 10, and low ionic strength The analysis of histone H1-depleted SV40 minichromosomes psoralen cross-linked at pH 7 and low ionic strength showed results similar to those obtained with isolated native chromatin cross-linked under these conditions. However, histone H1-depleted minichromosomes photoreacted at pH 10 and low ionic strength revealed a heavily disturbed chromatin structure. The ssbubbles detected were clearly reduced in number compared with pH 7 conditions or native minichro-

Chromatin Structure at Replicating Forks

mosomes (compare r-values and numbers of ss-bubbles per molecule in Table 1) and were predominantly small (Figures 1 and 2). The large amount of psoralen cross-linking seen as the large stretches of duplex appearing DNA and small ss-bubbles suggests strongly an unravelling of the nucleosomes (total or partial, respectively). Since the sequence of SV40 (Fiers et al., 1978) does not consist of stretches of only purine bases of a length greater than 30 to 50 nt, i.e. the resolution of our electron microscopic analysis, the remaining ssbubbles do not originate from a sequence-dependent inefficient cross-linking of purine bases (Cimino et al., 1985). These findings with histone H1-depleted SV40 minichromosomes are compatible with those of Conconi et al. (1984) and Widmer et al. (1988) with histone H1-depleted rat liver chromatin. Since psoralen cross-linking at pH 10 and low salt leads to unravelling of nucleosomes to a large extent, if H1 is absent, but to only a small extent, if H1 is present, cross-linking at pH 10 and low salt allows discernment between H1-containing and H1-depleted chromatin. Disassembly and assembly of chromatin during replication

The fate of prefork chromatin By focusing on the chromatin section immediately in front of the replication fork, we find that upon psoralen cross-linking the DNA appears either as an ss-bubble or as continuously crosslinked duplex DNA. If these particular ss-bubbles were due to replication-related proteins, they were expected to be detectable in front of each fork and to show always the same size. The fact that we find both ss-bubbles of heterogeneous size as well as cross-linked duplex DNA stretches, indicates that these ss-bubbles immediately in front of the fork are probably due to histone-DNA interactions, as concluded by Sogo et al. (1986). The heterogeneous, non-Gaussian size distribution of these particular ss-bubbles has been interpreted as being due to replication forks moving to, as well as into, the nucleosomes. Upon cross-linking under the pH 10 destabilizing conditions described here, the frequency of ss-bubbles immediately in front of the fork decreases by a factor of about 2, and the lengths of the prefork duplex DNA increases also by a factor of 2 (Table 4). This indicates that the nucleosomes closest to the replication forks are destabilized. The large spread of the data might reflect a fluctuating dynamic structure. Since in native molecules at pH 10 mostly cross-linked duplex DNA is seen in the prefork area, whereas H1-containing nucleosomes at pH 10 give rise to a nucleosomal or subnucleosomal ss-bubble (Figures 2 and 4), we conclude that the nucleosomes in putative contact with the replication machinery might be destabilized by the lack of histone H1. The fact that at

235

Figure 7. A model to explain the in vivo dynamics of nucleosomes at the replication fork. Through the passage of the replication machinery, two nucleosomes in the parental strand are disorganized. Histones H1 and H2A.H2B dimers are progressively displaced. In the nucleosome involved in replication, only (H3.H4)2 tetramers remain, the other immediately contiguous ones lack H1. On nascent strands, the first formed nucleosome contains the whole set of core histones, indicating that the deposition of (H3.H4)2 tetramers followed by the addition of H2A.H2B dimers is a very fast process (undetected in our system). Nucleosomes are completed by the addition of H1. Due to the passage of the replication machinery, nucleosomes become destabilized over a distance of 650 to 1100 bp.

pH 7 a bubble in this area shows a non-Gaussian distribution (Sogo et al., 1986; Gasser, 1993) suggests in addition that other factors contribute to the destabilization, presumably the advancing replication apparatus. In native chromatin, cross-linked at pH 10 and low ionic strength, the average length of continuously cross-linked prefork DNA is 180 bp (Table 4) suggesting that, besides the nucleosome involved in replication, also the last prefork nucleosome (for definition, see Figure 5a) lacks histone H1 (see Figure 7). This is consistent with the findings about the last prefork ss-bubbles, which show a bigger proportion of small ss-bubbles than bubbles of bulk SV40 minchromosomes in molecules cross-linked at pH 7 (compare Table 4 with Table 1). Since H2A/H2B dimers were found to dissociate in a non-cooperative manner (Burton et al., 1978), this might indicate that the last prefork nucleosomes possibly become destabilized by a successive disruption of one and then the second H2A/H2B dimer (see Figure 7). Evidence for the dissociation of parental nucleosomes at replication forks has been described for in vivo experiments (Jackson & Chalkley, 1981; Jackson, 1987, 1990). Furthermore, when upon replication initiation the elongation was efficiently slowed, a loss of histone H1 has been observed (D’Anna & Prentice, 1983). Based on these investigations it has been proposed that histone H1 dissociates from initiated replicons (D’Anna & Prentice, 1983; D’Anna & Tobey, 1984). Taking these in vivo data into account, our more precise observations strongly suggest the dissociation of H1 from one to two nucleosomes immediately in front of the fork. The second last prefork nucleosome, however, which shows a size distribution similar to that of the bulk of the ss-bubbles, seems not yet affected by the advancing replication machinery.

236 Chromatin assembly after replication In replicating molecules psoralen cross-linked in vivo or under destabilizing conditions at pH 7, the average length of prenucleosomal DNA (1260 bp; Table 5) agrees well with the values previously determined for the leading and lagging strands by Sogo et al. (1986). However, in isolated native molecules cross-linked at pH 10, the length of this continuously cross-linked DNA behind the fork (443(2208) bp) is increased by about 180 bp as compared with in vivo cross-linked molecules, a length that corresponds to the DNA of one nucleosomal repeat length. Since pH 10 conditions are indicative for the presence or absence of histone H1, increased length of the cross-linked prenucleosomal DNA suggests that, on average, the first nucleosome behind the replication fork lacks histone H1. This implies that within a distance of 450 to 650 bp, the nascent DNA is needed to package a complete nucleosome, including histone H1 (Figure 7). For the first time we clearly show in situ the transition in newly replicated chromatin from nucleosome cores to nucleosomes containing histone H1, and thus are able to demonstrate the histone H1-dependent maturation of chromatin, as has been suggested (Cusick et al., 1983; D’Anna & Prentice, 1983). Such a rapid assembly of histone H1 as we observe has been indicated by Annunziato et al. (1981) and suggested by Bavykin et al. (1993). Histone H1 also rapidly associates with newly reformed nucleosomes in repaired nucleosome regions of chromatin (Smerdon et al., 1982; for a review, see Smerdon, 1989). Since histone H1 prevents sliding of nucleosomes, our data are in agreement with findings that show that the phase relationship between DNA sequence and chromatin structure is established within 400 bp of replication forks (Tack et al., 1981). Mature chromatin has been described to show the same nuclease resistance as bulk chromatin (reviewed by DePamphilis & Wassarman, 1980, 1982). Like newly replicated cellular chromatin, newly replicated SV40 minichromosomes reveal a hypersensitivity to non-specific endonucleases, a characteristic of immature chromatin (Klempnauer et al., 1980; Cusick et al., 1981, 1983). Whereas cellular host chromatin has been shown matured after 15 minutes (Klempnauer et al., 1980), SV40 minichromosomes required 25 to 30 minutes chase periods to acquire the same nuclease sensitivity as non-replicating molecules (Klempnauer et al., 1980; Cusick et al., 1981). Since we show that histone H1 is assembled 450 to 650 bp behind the replication fork, it is likely that the extent of nuclease resistance as seen in bulk chromatin becomes established at a later stage. Since SV40 minichromosomes are replicated within 15 to 25 minutes, including replication initiation, termination as well as the separation of the sibling molecules, we conclude that the full maturation of newly replicated SV40 chromatin does not take place immediately during the course of replication, but occurs rather

Chromatin Structure at Replicating Forks

postreplicative. Our findings indicate therefore that the association of histone H1 is not the only process required for the maturation of chromatin. Prolongated nuclease sensitivity has been observed also in newly repaired chromatin regions containing H1 (for a review, see Smerdon, 1989). We conclude that the suggestion of Cusick et al. (1983) and D’Anna & Prentice (1983) considering the association of histone H1 as the essential step for chromatin maturation, is not sufficient to explain maturation of chromatin. Additional processes are necessary, which might involve modification of the histones (Perry & Annunziato, 1989, 1991; Smith & Stillman, 1991). Besides the binding of histone H1, many experimental data are available concerning the assembly of nucleosome cores on nascent DNA, both from in vivo as well as in vitro studies. From biochemical in vivo studies (Jackson, 1990), it has been shown that newly synthesized histones (H3.H4)2 deposit as tetramer on newly replicated DNA and that the nucleosomes are completed by the assembly of either old or new H2A/H2B dimers and H1. These conclusions drawn from in vivo experiments are further supported by in vitro replication studies (Fotedar & Roberts, 1989; Almouzni et al., 1990; Smith & Stillman, 1991; Gruss et al., 1993). However, by direct analysis of the psoralen imprinting nucleosomal structures at the replication fork, and as summarized in Figure 7, this two-step assembly of nucleosome cores as deduced from in vitro replication studies is not detectable in vivo. We find that already the first ss-bubble behind the replication fork shows nucleosomal size in molecules cross-linked in vivo. Also under destabilizing conditions at pH 7, the size distribution of the first ss-bubble behind the fork on nascent DNA indicates that most of these ss-bubbles show nucleosomal size (Table 5). The slightly increased proportion of small ss-bubbles generally detected on daughter strands (Table 2) might be an indication for the proposed two-step mechanism of nucleosome assembly. However, this finding might equally well indicate weaker binding of newly assembled histones (Smith et al., 1984; Bavykin et al., 1993). We conclude that in vivo the first nucleosome behind the replication fork is already constituted by a complete set of core histones. Thus, our data neither confirm nor disprove whether in vivo the assembly of nucleosome cores occurs in two steps. However, our findings imply a very quick and efficient assembly of nucleosome cores on newly replicated DNA, which occurs too fast to be further dissected in vivo so far.

Materials and Methods Preparation of SV40-infected cells and isolation of SV40 nucleoprotein complexes TC7 or CVI monkey cells were infected with SV40 strain Rh911, (DePamphilis et al., 1975) as described by

237

Chromatin Structure at Replicating Forks

Stahl & Knippers (1983). At 36 to 40 hours after infection the cells were either directly used for psoralen cross-linking (in in vivo experiments) or, after cell lysis, SV40 minichromosomes were extracted in hypotonic buffer (Su & DePamphilis, 1978). SV40 nucleoprotein complexes were purified at 4°C in sucrose gradients containing 10% to 30% sucrose (w/v), 5 mM triethanolamine-HCl (pH 7), 0.2 mM EDTA, 1 mM aminoacetonitrile bisulfate (for native minichromosomes) or in addition 500 mM NaCl (for H1-depleted minichromosomes). After centrifugation, fractions were collected and dialyzed at 4°C against various destabilizing conditions. Dialysis of purified SV40 minichromosomes Isolated SV40 nucleoprotein complexes and purified by sucrose gradient centrifugation, were brought to destabilizing conditions, i.e. low ionic strength and high pH, by dialyzing first against low-salt buffer pH 7 (5 mM triethanolamine chloride (pH 7.0), 0.2 mM EDTA, 1 mM aminoacetonitrile bisulfate) for at least four hours with two buffer changes. Part of the material was further dialyzed against this buffer, the other part was dialyzed against low-salt buffer (pH 10 (5 mM diethanolamine chloride (pH 10), 0.2 mM EDTA, 1 mM aminoacetonitrile bisulfate) for 16 to 24 hours by changing the buffer twice. The pH in the samples was checked after dialysis. Protein analysis suggested that no significant protease activity was present in the samples. Psoralen cross-linking For the in vivo experiments, SV40-infected cells in culture dishes of 60 mm were placed on ice. The medium was removed, 1.5 ml of fresh cold medium and 0.05 volume of 4,5',8-trimethylpsoralen stock solution in ethanol (200 mg/ml) was added. After incubation for five to ten minutes in the dark, the cells were irradiated for five minutes with UV light of 366 nm (model B-100 A; Ultra Violet Products Inc., San Gabriel, CA). The procedure was repeated three times with a total irradiation time of 20 minutes. The light source was mounted at a distance of 6 to 7 cm above the culture dish. For soluble chromatin experiments, purified SV40 minichromosomes dialyzed to the appropriate buffers, were placed in Eppendorf or polypropylene tubes and photoreacted with psoralen as described above. DNA extraction and spreading for electron microscopy The DNA extraction procedures used after photocrosslinking of purified minichromosomes are described by Sogo et al. (1986). The SV40 DNA cross-linked in vivo was extracted from the photoreacted cells by a modification of the Hirt procedure (DeBernardin et al., 1986). Purified SV40 DNA was nicked with DNase I (DeBernardin et al., 1986) and spread for electron microscopy under denaturing conditions (Sogo et al., 1986). As spreading control, DNA of rat liver chromatin (cross-linked at pH 7 and 100 mM salt) was coprepared. Quantitative analysis The size of ss-bubbles and duplex DNA stretches was calculated from the contour length of the SV40 DNA (5243 bp; Fiers et al., 1978). The contour length

measurements were made with a Hewlett-Packard digitizer on enlarged photographic prints. A total magnification of 180,000× was used (corresponding to 150 bases/cm). In all histograms the data were fit with linear superposition of a Gaussian distribution (Koller et al., 1978).

Polyacrylamide gel electrophoresis Proteins were analyzed by SDS-PAGE in 8% to 15% (w/v) polyacrylamide gels according to Laemmli (1970). For the qualitative analysis of histone proteins, polyacrylamide gels were run according to modified protocols described by Thoma et al. (1979).

Acknowledgements We thank Drs M. J. Smerdon and F. Thoma for helpful discussions and for critical reading of the manuscript. We thank Mrs H. Mayer-Rosa for her excellent technical help. The project was supported by the ‘‘Schweizerische Krebsliga’’ and the ‘‘Schweizerischer Nationalfonds zur Fo¨rderung der wissenschaftlichen Forschung’’ (31-41 827.94).

References Almouzni, G., Clark, D. J., Me´chali, M. & Wolffe, A. P. (1990). Chromatin assembly on replicating DNA in vitro. Nucl. Acids Res. 18, 5767–5774. Annunziato, A. T., Schindler, R. K., Thomas, C. A., Jr & Seale, R. L. (1981). Dual nature of newly replicated chromatin. J. Biol. Chem. 256, 11880–11886. Bavykin, S., Srebreva, L., Banchev, T., Tsanev, R., Zlatanova, J. & Mirzabekov, A. (1993). Histone H1 deposition and histone-DNA interactions in replicating chromatin. Proc. Natl Acad. Sci. USA, 90, 3918–3922. Bonne-Andrea, C., Wong, M. L. & Alberts, B. M. (1990). In vitro replication through nucleosomes without histone displacement. Nature, 343, 719–726. Brown, D. W., Libertini, L. J. & Small, E. W. (1991). Fluorescence anisotropy decay of ethidium bound to nucleosome core particles. 1. Rotational diffusion indicates an extended structure at low ionic strength. Biochemistry, 90, 5293–5303. Burch, J. B. E. & Martinson, H. G. (1980). The roles of H1, the histone core and DNA length in the unfolding of nucleosomes at low ionic strength. Nucl. Acids Res. 8, 4969–4987. Burhans, W. C., Vassilev, L. T., Wu, J. R., Sogo, J. M., Nallaseth, F. S. & DePamphilis, M. L. (1991). Emetine allows identification of origins of mammalian DNA replication by imbalanced DNA synthesis, not through conservative nucleosome segregation. EMBO J. 10, 4351–4360. Burton, D. R., Butler, M. J., Hyde J. E., Phillips, D., Skidmore, C. J. & Walker, I. O. (1978). The interaction of core histones with DNA equilibrium binding studies. Nucl. Acids Res. 5, 3643–3663. Cimino, G. R., Gamper, H. B., Isaacs, S. T. & Hearst, J. E. (1985). Psoralens as photoactive probes of nucleic acid structure and function: organic chemistry, photochemistry and biochemistry. Annu. Rev. Biochem. 54, 1151–1193.

238 Conconi, A., Losa, R., Koller, Th. & Sogo, J. M. (1984). Psoralen-cross-linking of soluble and of H1-depleted soluble rat liver chromatin. J. Mol. Biol. 178, 920–928. Cusick, M. E., Herman, T. M., DePamphilis, M. L. & Wassarman, P. M. (1981). Structure of chromatin at DNA replication forks: prenucleosomal DNA is rapidly excised from replicating SV40 chromosomes by micrococcal nuclease. Biochemistry, 20, 6648–6658. Cusick, M. E., Lee, K. S., DePamphilis, M. L. & Wassarman, P. M. (1983). Structure of chromatin at DNA replication forks: nuclease hypersensitivity results from both prenucleosomal DNA and an immature chromatin structure. Biochemistry, 22, 3873–3884. Cusick, M. E., DePamphilis, M. L. & Wassarman, P. M. (1984). Dispersive segregation of nucleosomes during replication of SV40 chromosomes. J. Mol. Biol. 178, 249–271. D’Anna, J. A. & Prentice, D. A. (1983). Chromatin structural changes in synchronized cells blocked in early S phase by sequential use of isoleucine deprivation and hydroxyurea blockade. Biochemistry, 22, 5631–5640. D’Anna, J. A. & Tobey, R. A. (1984). Changes in histone H1 content and chromatin structure of cells blocked in early S phase by 5-fluoro-deoxyuridine and aphidicolin. Biochemistry, 23, 5024–5029. DeBernardin, W., Koller, Th. & Sogo, J. M. (1986). Structure of in vivo transcribing chromatin as studied in SV40 minichromosomes. J. Mol. Biol. 191, 469–482. DePamphilis, M. L. & Wassarman, P. M. (1980). Replication of eukaryotic chromosomes: a close-up of the replication fork. Annu. Rev. Biochem. 49, 627–666. DePamphilis, M. L. & Wassarman, P. M. (1982). Organization and replication of papovavirus DNA. In Organization and Replication of Viral DNA (Kaplan, A. S., ed.), pp. 37–114, CRC Press, Boca Raton, FL. DePamphilis, M. L., Beard, P. & Berg, P. (1975). Synthesis of superhelical SV40 deoxyribonucleic acid in cell lysates. J. Biol. Chem. 250, 4340–4347. Dietrich, A. E., Eshaghpour, H., Crothers, D. M. & Cantor, C. R. (1980). Effect of DNA length on the nucleosome low salt transition. Nucl. Acids Res. 8, 2475–2487. Fiers, W., Contreras, R., Haegeman, G., Rogiers, R., van de Voorde, A., van Heuverswyn, H., van Herreweghe, J., Volckaert, G. & Ysebaert, M. (1978). Complete nucleotide sequence of SV40 DNA. Nature, 273, 113–120. Fotedar, R. & Roberts, J. M. (1989). Multistep pathway for replication-dependent nucleosome assembly. Proc. Natl Acad. Sci. USA, 86, 6459–6463. Galili, G., Levy, A. & Jakob, K. M. (1981). Changes in chromatin structure at the replication fork. II. The DNPs containing nascent DNA and a transient chromatin modification detected by DNase I. Nucl. Acids Res. 9, 3991–4005. Gasser, R. (1993). I. Regulation of transcription of ribosomal RNA genes. II. The stability of the nucleosomes at the replication fork. PhD thesis, Swiss Federal Institute of Technology, Zu¨rich. Gordon, V. C., Knobler, C. M., Olins, D. E. & Schumaker, V. N. (1978). Conformational changes of the chromatin subunit. Proc. Natl Acad. Sci. USA, 75, 660–663. Gruss, C. & Sogo, J. M. (1992). Chromatin replication. BioEssays, 14, 1–8. Gruss, C., Wu, J., Koller, Th. & Sogo, J. M. (1993). Disruption of the nucleosomes at the replication fork. EMBO J. 12, 4533–4545.

Chromatin Structure at Replicating Forks

Hanson, C. V., Shen, C. J. & Hearst, J. E. (1976). Cross-linking of DNA in situ as a probe for chromatin structure. Science, 193, 62–64. Jackson, V. (1987). Deposition of newly synthesized histones: new histones H2A and H2B do not deposit in the same nucleosome with new histones H3 and H4. Biochemistry, 26, 2315–2325. Jackson, V. (1990) . In vivo studies on the dynamics of histone–DNA interaction: evidence for nucleosome dissolution during replication and transcription and a low level of dissolution independent of both. Biochemistry, 29, 719–731. Jackson, V. & Chalkley, R. (1981a). A new method for the isolation of replicative chromatin: selective deposition of histone on both new and old DNA. Cell, 23, 121–134. Jackson, V. & Chalkley, R. (1981b). A reevaluation of new histone deposition on replicating chromatin. J. Biol. Chem. 256, 5095–5103. Klempnauer, K. H., Fanning, E., Otto, B. & Knippers, R. (1980). Maturation of newly replicated chromatin of SV40 and its host cell. J. Mol. Biol. 136, 359–374. Koller, Th., Ku¨bler, O., Portmann, R. & Sogo, J. M. (1978). High resolution physical mapping of specific binding sites of E. coli RNA polymerase on the DNA of bacteriophage T7. J. Mol. Biol. 120, 121–131. Krude, T. & Knippers, R. (1991). Transfer of nucleosomes from parental to replicated chromatin. Mol. Cell. Biol. 11, 6257–6267. Krude, T. & Knippers, R. (1993). A nucleosome assembly factor is a constituent of SV40 minichromosomes. Mol. Cell. Biol. 13, 1059–1068. Labhart, P., Thoma, F. & Koller, Th. (1981). Structural changes of soluble rat liver chromatin induced by the shift in pH from 7 to 9. Eur. J. Cell Biol. 25, 19–27. Laemmli, U. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, 227, 680–685. Libertini, L. J. & Small, E. W. (1980). Salt induced transitions of chromatin core particles studied by tyrosine fluorescence anisotropy. Nucl. Acids Res. 8, 3517–3534. Libertini, L. J. & Small, E. W. (1982). Effects of pH on low-salt transition of chromatin core particles. Biochemistry, 21, 3327–3334. Libertini, L. J. & Small, E. W. (1984). Effects of pH on the stability of chromatin core particles. Nucl. Acids Res. 12, 4351–4359. Martinson, H. G., True, R. J. & Burch, J. B. E. (1979). Specific histone-histone contacts are ruptured when nucleosomes unfold at low ionic strength. Biochemistry, 18, 1082–1089. McKnight, S. L. & Miller, O. L., Jr (1977). Electron microscopic analysis of chromatin replication in the cellular blastoderm Drosophila melanogaster embryo. Cell, 12, 795–804. Perry, C. A. & Annunziato, A. T. (1989). Influence of histone acetylation on the solubility, H1 content and DNase I sensitivity of newly assembled chromatin. Nucl. Acids Res. 17, 4275–4291. Perry, C. A. & Annunziato, A. T. (1991). Histone acetylation reduces H1-mediated ncleosome interactions during chromatin assembly. Exp. Cell Res. 196, 337–345. Randall, S. K. & Kelly, T. J. (1992). The fate of parental nucleosomes during SV40 DNA replication. J. Biol. Chem. 267, 14259–14265. Schla¨ger, E. J. (1982). Replicative conformation of parental

Chromatin Structure at Replicating Forks

nucleosomes: salt sensitivity of DNA-histone interaction and alteration of histone H1 binding. Biochemistry, 21, 3167–3174. Shelton, E. R., Wassarman, P. M. & DePamphilis, M. L. (1980). Structure, spacing and phasing of nucleosomes on isolated forms of mature simian virus 40 chromosomes. J. Biol. Chem. 255, 771– 782. Smerdon, M. J. (1989). DNA excision repair at the nucleosome level of chromatin. In DNA Repair Mechanisms and their Biological Implications in Mammalian Cells (Lambert, M. W., ed.), pp. 271–294, Plenum Publishing Corp., New York. Smerdon, M. J., Watkins, J. F. & Lieberman, M. W. (1982). Effect of histone H1 removal on the distribution of ultraviolet-induced deoxyribonucleic acid repair synthesis in chromatin. Biochemistry, 21, 3879–3885. Smith, P. A., Jackson, V. & Chalkley, R. (1984). Two-stage maturation process for newly replicated chromatin. Biochemistry, 23, 1576–1581. Smith, S. & Stillman, B. (1991). Stepwise assembly of chromatin during DNA replication in vitro. EMBO J. 10, 971–980. Sogo, J. M. & Laskey, R. A. (1995). Chromatin replication and assembly. In Chromatin Structure and Gene Expression (Elgin, S., ed.), pp. 49–70, IRL-Press, Oxford. Sogo, J. M. Ness, P. J., Widmer, R. M., Parish, R. W. & Koller, Th. (1984). Psoralen-cross-linking of DNA as a probe for the structure of active nucleolar chromatin. J. Mol. Biol. 178, 897–928. Sogo, J. M., Stahl, H., Koller, Th. & Knippers, R. (1986). Structure of replicating Simian virus 40 minichromosomes. J. Mol. Biol. 189, 189–204. Stahl, H. & Knippers, R. (1983). SV40 large tumor antigen on replicating viral chromatin: tight binding and localization on the viral genome. J. Virol. 47, 65–76. Stefanovsky, V., Dimitrov, S., Russanova, V. & Pashev, I.

239 (1990). Histone H1 and H4 are present near the replication fork. Mol. Biol. Rep. 14, 231–235. Su, R. T. & DePamphilis, M. L. (1978). SV40 DNA replication in isolated replicating viral chromosomes. J. Virol. 28, 53–65. Tack, L. C., Wassarman, P. M. & DePamphilis, M. L. (1981). Chromatin assembly: relationship of chromatin structure to DNA sequence during SV40 replication. J. Biol. Chem. 256, 8821–8828. Thoma, F. & Koller, Th. (1981). Unraveled nucleosomes, nucleosome beads and higher order structures of chromatin: influence of non-histone components and histone H1. J. Mol. Biol. 149, 709–733. Thoma, F., Koller, Th. & Klug, A. (1979). Involvement of histone H1 in the organization of the nucleosome and of the salt-dependent superstructures of chromatin. J. Cell Biol. 83, 403–427. Tsanev, R. G. & Semionov, E. P. (1985). Ultrastructural features of replicating chromatin in the Drosophila embryo. Eur. J. Cell Biol. 39, 469–474. Van Holde, K. E. (1988). Chromatin, Springer-Verlag, New York. Watkins, J. F. & Smerdon, M. J. (1985). Nucleosome rearrangement in vitro. 1. Two phases of salt-induced nucleosome migration in nuclei. Biochemistry, 24, 7279–7287. Widmer, R. M., Koller, Th. & Sogo, J. M. (1988). Analysis of the psoralen-cross-linking pattern in chromatin DNA by exonuclease digestion. Nucl. Acids Res. 16, 7013–7024. Worcel, A., Han, S. & Wong, M. L. (1978). Assembly of newly replicated chromatin. Cell, 15, 969–977. Wu, H. M., Dattagupta, N., Hogan, M. & Crothers, D. M. (1979). Structural changes of nucleosomes in low-salt concentrations. Biochemistry, 18, 3960–3965. Zucker, K. & Worcel, A. (1990). The histone H3/H4-N1 complex supplemented with histone H2A-H2B dimers and DNA topoisomerase I forms nucleosomes on circular DNA under physiological conditions. J. Biol. Chem. 265, 14487–14496.

Edited by J. Karn (Received 14 November 1995; received in revised form 8 February 1996; accepted 13 February 1996)