Folding of Chromatin in the Presence of Heterogeneous Histone H1 Binding to Nucleosomes*

We have reconstituted oligonucleosome complexes containing histone H1 starting from a synthetic DNA template, consisting of 12 tandemly arranged 208-base pair fragments of the 5 S rRNA gene, purified HeLa histone octamers, and histone H1. A ratio of histone H1 per histone octamer used in the reconstitution (0.8–0.9 mol of histone H1/mol of histone octamer) similar to that observed in vivo was used. The reconstituted chromatin complexes exhibit a salt-dependent folding, which is almost indistinguishable from that exhibited by chromatin fragments obtained from nuclease digestion of native chromatin. The folding of this reconstituted chromatin complex seems to be rather independent of the symmetrical or asymmetrical position occupied by H1 in the individual nucleosomes. Binding of histone H1 to the oligonucleosome complexes, under the stoichiometric binding conditions used, had no inhibitory effect on the transcriptional potential of these complexes.

We have reconstituted oligonucleosome complexes containing histone H1 starting from a synthetic DNA template, consisting of 12 tandemly arranged 208-base pair fragments of the 5 S rRNA gene, purified HeLa histone octamers, and histone H1. A ratio of histone H1 per histone octamer used in the reconstitution (0.8 -0.9 mol of histone H1/mol of histone octamer) similar to that observed in vivo was used. The reconstituted chromatin complexes exhibit a salt-dependent folding, which is almost indistinguishable from that exhibited by chromatin fragments obtained from nuclease digestion of native chromatin. The folding of this reconstituted chromatin complex seems to be rather independent of the symmetrical or asymmetrical position occupied by H1 in the individual nucleosomes. Binding of histone H1 to the oligonucleosome complexes, under the stoichiometric binding conditions used, had no inhibitory effect on the transcriptional potential of these complexes.
For almost a decade an important research effort has been made in several laboratories, including ours, trying to reconstitute chromatin fibers upon addition of purified H1 to homogenous oligonucleosome fibers (1). The reconstitution of homogenous oligonucleosomal fibers onto a synthetic DNA template, consisting of a defined number of tandemly arranged fragments of the 5 S rRNA gene from the sea urchin Lytechinus variegatus (2) was put forward as a promising system with high expectations for the structural characterization of the chromatin fiber. Although the system has proven to be beneficial for the structural characterization of the chromatin fiber in the absence of histone H1 linker histone (3,4), attempts to reconstitute this latter histone into native-like chromatin structures have been only partially successful (1). In fact, no evidence has been provided to show that addition of H1 to the oligonucleosome complexes, reconstituted onto the tandemly arranged 5 S rRNA gene fragments, results in a folding of a fiber resembling that of native chromatin structures. Indeed, the major structural role of the linker histones in chromatin is to increase the extent of compaction (folding) of the polynucleosomal fiber upon what appears to be a rather labile binding (5-7) (see also van Holde (8)).
Recent linker-histone knockout experiments carried out in Tetrahymena have provided in vivo support for the structural role of histone H1 (9) (see also Ausió (10)). The binding of linker histones to DNA in the chromatin fiber complex is weak compared with that of core histones. Linker histones can be removed from chromatin at lower ionic strengths than any of the core histones (11) and they can easily move to occupy different positions even at relatively low salt concentrations (12).
The presence of more than one binding position for the linker histones to nucleosomal DNA had been long proposed (13,14). In recent years, characterization of the interaction of linker histones with a nucleosome reconstituted onto a single fragment of the Xenopus borealis 5 S rRNA gene has provided evidence for asymmetrical binding (off-axis location) (15) of these histones in this reconstituted nucleosome particle (15,16). Such binding is in contrast to earlier symmetrical (on-axis) models in which the folded domain of histone H1 is positioned on the dyad axis of the nucleosome, symmetrically binding to the flanking exiting and entering duplexes of DNA (17,18). In both instances, binding of H1 confers an extra 20-bp 1 DNA protection from digestion by micrococcal nuclease although the distribution of the protection sites is symmetrical in one case (17,18) and asymmetrical in the other (16). The experimental support in favor of these different binding models, including an alternative one which has also been recently proposed (19), has stirred controversy on the issue of linker histone binding in chromatin (20).
We initially thought that the asymmetry of binding of histone H1 to the nucleosomes reconstituted onto the X. borealis 5 S rRNA gene could perhaps account for the apparent inability of this histone to fold the oligonucleosome fibers that had been reconstituted from tandemly repeated 5 S rRNA fragments of the sea urchin gene. In this report we show that histone H1 from HeLa cells can induce the folding of an oligonucleosome complex reconstituted from HeLa core histones and a DNA template consisting of 12 tandemly arranged 208 bp fragments of the 5 S rRNA gene of the sea urchin L. variegatus.

DNA and Histones
DNA templates (208-12) consisting of 12 tandemly arranged 208-bp fragments of the 5 S rRNA gene of the sea urchin L. variegatus (2) were prepared as described elsewhere (4). Chromatin from HeLa cells was prepared according to Ausió and van Holde (21), and histone octamers were then obtained as previously described (22).

Oligonucleosome Reconstitution
Oligonucleosome complexes were reconstituted onto the 208-12 DNA template by salt-gradient dialysis (40) as described by Garcia-Ramirez et al. (4) with the modifications that will be described next.

Chromatin Reconstitution by Addition of Histone H1 to Oligonucleosome Templates
Two different methods of reconstitution were used. Salt-gradient dialysis and direct mixing.
Salt Gradient Dialysis-During the course of the oligonucleosome reconstitution by salt gradient dialysis (see previous section) the dialysis was stopped at 0.5 M NaCl. At this point the dialysis bag was opened and its contents were concentrated 5-fold (from ϳ50 to ϳ250 g/ml) using a Centricon 100 concentrator device (Amicon Inc., Beverly, MA). The concentrate thus obtained was diluted with 3 volumes of 0.5 M NaCl, 10 mM Tris-HCl, 0.5 mM EDTA, pH 7.5, and was concentrated (3-fold) again using a Centricon 100 unit. This step was repeated once more, and the concentrate solution thus obtained was diluted back (5-fold) to its original volume using the same 0.5 M NaCl buffer. The purpose of these 0.5 M NaCl washes is to remove any extra core histones that are not part of the nucleosome complexes but which might be weakly associated with them (23). At this point the sample was equally divided in two aliquots. Histone H1 prepared from HeLa cells following the method described by Garcia-Ramirez et al. (24) was added to one of them at a ratio of approximately 0.9 -1.0 mol of histone H1/mol of 208-bp DNA fragment. Histone H1 dissolved in 0.5 M NaCl, 10 mM Tris-HCl, 0.1 mM EDTA, pH 7.5, at a concentration of 1 mg/ml was used. An extinction coefficient of A 260 0.1% ϭ 2.0 was used to determine the concentration of histone H1 (25). An equal volume of the 0.5 M NaCl buffer (but without H1) was added to the second aliquot, and the two samples thus obtained were dialyzed next against 0.3 M NaCl, 10 mM Tris-HCl, 0.5 mM EDTA, pH 7.5, for 3 h at 4°C. The samples were finally dialyzed overnight at 4°C against 10 mM NaCl, 10 mM Tris-HCl, 0.5 mM EDTA, pH 7.5.
Direct Mixing-Two batches of oligonucleosome (208-12) complexes were prepared by gradient dialysis (4) to a final NaCl concentration of either 20 mM NaCl or 80 mM NaCl in 10 mM Tris-HCl, 0.5 mM EDTA, pH 7.5, buffer. Histone H1 in the corresponding buffer was then added by direct mixing (while vortexing) to each of these samples. The ratio of histone H1 per nucleosome was kept the same as in the preceding section. The complexes thus obtained were then loaded onto (5-20%) sucrose gradients (26) in the corresponding buffers. The fractions containing the reconstituted chromatin fractions were then pooled together and dialyzed overnight against 10 mM NaCl, 10 mM Tris-HCl, 0.5 mM EDTA, pH 7.5.

Analytical Ultracentrifuge Analysis
Sedimentation velocity experiments were carried out in a Beckman XL-A analytical ultracentrifuge using an aluminum An-55 rotor and double sector aluminum-filled Epon centerpieces. All runs were routinely performed at 26,000 rpm and 20°C. The boundaries were analyzed according to the method of van Holde and Weischet (27) using an XL-A Ultra Scan-Origin Version 2.93 sedimentation data analysis software (B. Demeler, Missoula, MT).
Sedimentation equilibrium experiments were performed also in a Beckman XL-A ultracentrifuge using a titanium An-60 Ti rotor and double sector aluminum-filled Epon centerpieces. Equilibrium runs were carried out at 2400 rpm, 20°C. Data were evaluated using a nonlinear least-squares curve-fitting algorithm (28) contained in the XL-A data analysis software using an ideal one-component model.
The partial specific volumes of the different histone-DNA complexes analyzed in this work were calculated as described elsewhere (29). The partial specific volume values used for DNA, the histone octamer, and histone H1 in the calculations were 0.53 5 , 0.75 3 , and 0.77 7 cm 3 /g, respectively (30 -32). The molecular weight of the histone octamer (M r 108,768) (33) and the molecular weight average of histone H1 (M r 22,000). The partial specific volumes determined in this way were estimated to be 0.63 1 cm 3 /g and 0.64 0 cm 3 /g for the 208-12 reconstituted oligonucleosome complex in the absence or in the presence of ϳ1.0 mol of histone H1 per mol of nucleosome, respectively.

Magnesium Titration
The solubility of the reconstituted oligonucleosome complexes in the presence of MgCl 2 was determined as described elsewhere (26,34).

Determination of Nucleosome Core Particle/Chromatosome
Positioning on Reconstituted 208-12 DNA Nucleosome core particles and chromatosomes reconstituted on the 208-12 fragment were adjusted to 1 mM CaCl 2 and digested with micrococcal nuclease. The time of digestion and the amount of micrococcal nuclease was established from the previous time course of digestion analysis carried out under the same conditions. Digestion was stopped and the DNA deproteinized by adjusting the solution to 5 mM EDTA, 0.25% SDS, and phenol/chloroform extracting. The DNA was precipitated and end-labeled with [␥-32 P]ATP and polynucleotide kinase (35). The approximately 145-bp (nucleosome core particle) and 165-bp (chromatosome) micrococcal nuclease-resistant DNA fragments were isolated by gel electrophoresis on a 4% acrylamide gel (36). After extraction from the acrylamide matrix, the DNA was precipitated and cut with 5 units of restriction enzyme in 20 l of appropriate restriction buffer for 4 h at 37°C. Following this, the digested fragments were phenol/chloroform-extracted, ethanol-precipitated, and resuspended in 95% formamide, 0.25% bromphenol blue, 0.25% xylene cyanol, and 1 mM EDTA. The fragments were denatured at 90°C for 2 min and resolved on a 8% acrylamide gel (19:1 acrylamide:bisacrylamide) containing 8.3 M urea and 1ϫ TBE.

In Vitro Transcription of Reconstituted 208-12 Oligonucleosome Complexes
The HeLa nuclear transcription extracts were purchased from Promega, and the transcriptions were performed per the manufacturer's instructions using 470 ng of 208-12 template (either reconstituted into nucleosome core particles/chromatosomes or as uncomplexed DNA). The buffer composition of the extracts in which the transcription reactions were carried out was 44 mM KCl, 9.1% glycerol, 0.088 mM EDTA, and 0.22 mM dithiothreitol, 2 mM MgCl 2 , 0.4 mM NTPs, 7.4 mM HEPES, 2.9 mM Tris-HCl, pH 7.8. The extracts were supplemented with 150 nM recombinant Xenopus TFIIIA. For each transcription reaction, 500 ng of Xenopus laevis oocyte 5 S rRNA gene DNA was included as an internal control.

RESULTS
Sedimentation Analysis of Chromatin Folding-As it has been mentioned in the introduction, one of the most critical criterion to assess the proper binding of histone H1 to chromatin is its ability to compact the polynucleosomal fiber upon binding. The analytical ultracentrifuge provides one of the most sensitive methods to monitor compaction of the chromatin fiber. Fig. 1A shows the salt-dependent hydrodynamic behavior of 208-12 oligonucleosome complexes in the presence or absence of histone H1, as monitored by analytical ultracentrifugation. As it can be seen there, in vitro reconstitution of nativelike stoichiometric amounts of histone H1 by salt gradient dialysis produces chromatin fibers which follow the salt dependent pattern of folding (as envisaged from the increase in sedimentation coefficient) expected for native chromatin fibers of an equivalent DNA length.
The stoichiometry of histone H1 per nucleosome was determined by sedimentation equilibrium (4) using the molecular weight of the reconstituted complex containing H1 and of the H1 lacking complex which were reconstituted under identical conditions. A ratio of approximately 1-1.1 histone H1 molecules per nucleosome was determined in this way. As seen in Fig. 2A this corresponds to the amounts which are present in native chromatin as assessed by SDS-polyacrylamide gel electrophoresis. The folding of the chromatin fibers reconstituted by salt gradient dialysis is in contrast with the low folding inability (lack of salt-dependent sedimentation coefficient increase) exhibited by the fibers reconstituted by direct mixing in either low (20 mM NaCl) or high (80 mM NaCl) salt (see Fig. 1A). Fig. 1B shows the Mg 2ϩ solubility of the salt gradient dialysis reconstituted chromatin complexes lacking or containing H1. The midpoint of the solubility transitions (point at which 50% of chromatin remains still soluble) was found to be 3.4 mM for the H1-containing complexes and 4.6 for the H1-lacking complexes. The decrease in Mg 2ϩ solubility upon binding of H1 is comparable to that observed for native chicken erythrocyte chromatin (26). It reflects the interactions of histone H1 with the linker regions connecting adjacent nucleosomes (26). The interactions of Mg 2ϩ with oligonucleosomes leading to aggregation and precipitation of the complexes have been characterized thoroughly (37,38).

Determination of Nucleosome Core Particle/Chromatosome
Positioning on Reconstituted 208-12-When digested with micrococcal nuclease, nucleosome core particles protect approximately 145 bp of DNA, while the incorporation of histone H1 increases this protection by 20 bp (39). In this study, the positions of nucleosome core particles and chromatosomes on the 208-12 DNA fragment were mapped by restriction digestion of these micrococcal nuclease resistant fragments. In Fig. 2B, lanes 1 and 2 show the 145-bp (nucleosome core particle) and 165-bp (chromatosome) micrococcal nuclease-resistant fragments, respectively, intentionally overloaded to demonstrate that there were no internal nicks within these fragments. Lane 4 shows the products of the DraI restriction digestion of the 145-bp core particle fragment. This lane shows two bands of 57 and 88 bp, indicating that the nucleosome core particle adopts primarily one position on the individual 208 subunits. Cleavage of these two fragments with MspI (lane 5) had no effect on the 57-bp fragment, but the 88-bp band was cut to yield a 28-bp band, indicating that this fragment originated from of the 3Ј end of the micrococcal nuclease resistant fragment. Fig. 3 schematically represents the position of the nucleosome core particles on the 208 subunits determined from this analysis. It was determined that the position of the core particles on the 208-12 DNA fragment was from Ϫ84 to ϩ61 bp with respect to the transcriptional start site (ϩ6 to ϩ151 with respect to the 208 sequence). When the 165-bp fragment underwent a similar treatment, three "pairs" of bands were observed. The most predominant pair of 88 and 77 bp (representing approximately 60%) is indicative of the "asymmetrical protection" of DNA (see Fig. 3B) seen by Hayes and Wolffe (16). A second pair of bands with sizes 96 and 69 bp suggest an almost symmetrical protection by histone H1 (Fig. 3C). Finally, DraI restriction fragments of approximately 125 bp, which were cleaved by MspI, suggest a third chromatosome position that does not overlap the original core position (Fig. 3D). This implies that the incorporation of histone H1 caused in some instances, the histone octamer to adopt a new position on the 208 fragment. DNase I footprinting analysis (data not shown) did not reveal any major differences between the individual nucleosomes of the H1 lacking or the H1 containing 208-12 oligonucleosome complexes. This result is consistent with a predominantly "asymmetric protection" of nucleosome DNA in the H1 containing complexes (16).
In Vitro Transcription of Reconstituted 208-12 DNA-The same 208-12 DNA oligonucleosomes containing or lacking H1 that had been used for the ultracentrifuge analysis, were transcribed in HeLa cells extracts to determine the possible effects of histone H1 on transcription. No free DNA template was present in these samples. It is important to note in this context that during the construction of the 208-12 fragment, the termination site of the 5 S rRNA gene was removed. Due to this, transcription of 208-12 yields a "ladder" of transcripts resulting from run-on transcription initiating at the many different transcriptional start sites. This can be seen in Fig. 4, lanes 3 and 7, which show the products generated by transcription of naked 208-12 DNA. Lanes 2, 4, and 5 contain a 120-nucleotide band due to the inclusion of the X. laevis oocyte 5 S rRNA gene as an internal control (shown by itself in lane 2). The products generated by transcription of the 208-12 DNA after reconstitution into nucleosome core particles are shown in lane 4. As can be seen, the presence of nucleosomes on the 208-12 DNA significantly reduces the levels of transcription, but does not prevent transcription initiation. Furthermore, the size of the transcripts produced demonstrates RNA polymerase III is capable of transcription elongation through the nucleosome core particle arrays. Finally, lane 5 shows that the addition of histone H1 to the nucleosome arrays does not effect the amount or lengths of transcripts produced as the overall extent of transcription upon addition of H1 is almost indistinguishable from that of H1 depleted oligonucleosomes (Fig. 4, compare lanes 4 and 5). DISCUSSION We have been able to reconstitute H1 onto synthetic oligonucleosome templates to produce chromatin-like fibers which exhibit a very similar salt dependent folding to that of chromatin fragments obtained by micrococcal nuclease treatment of native chromatin (Fig. 1A) (41)(42)(43)(44). In the reconstitution method used, histone H1 is added to the oligonucleosome templates at 0.5 M NaCl, and the mixture is dialyzed down to lower salt concentrations. Presumably the reason why reconstitution works under these conditions is because nucleosomes in 0.5-0.6 M NaCl are in an equilibrium situation in which the histone octamer can freely and reversibly disassociate and associate with the DNA template (23,36,45) and thus it can slide to accommodate for the proper binding of histone H1. These ionic strength conditions may thus represent a thermodynamically favorable situation similar to that which can be achieved under more physiological conditions using ATP dependent chaperone molecules (46 -50). Direct mixing of histone H1 at much lower ionic strengths under conditions in which histone H1 is highly mobile (80 mM NaCl) or not (20 mM NaCl) do not produce a proper reconstitution (see Fig. 1A). Positioning analysis of the nucleosomes upon H1 reconstitution (Fig. 2B) indicate that indeed, at least a fraction of nucleosomes move to accommodate the incorporation of H1. The same analysis shows that the oligonucleosome folding or compaction induced by H1 occurs in the presence of a highly heterogeneous binding of this histone to the individual nucleosomes. Thus our results suggest that folding of the chromatin fiber is not critically dependent on histone H1 occupying a symmetrical position in the chromatosome. In fact most of the positioning obtained by our reconstitution method represents an asymmetrical off-axis binding of histone H1 to the nucleosome. We initially thought this to be responsible for the unsuccessful attempts in reconstituting H1 onto the 5 S rRNA gene oligonucleosome templates. Although H1 binding under our experimental conditions generates a 165 particle (Fig. 2B), DNase I footprinting (results not shown) does not show any major distinctive protection in agreement with previous reconstitutions carried out on single copy templates of the 5 S rRNA gene (16).
It is possible that, as hinted by Crane-Robinson (20), the binding of the folded domain of histone H1 to the nucleosome results in a conformational transition of the core particle which creates two binding sites for the linker DNA. This would explain why the folding of the chromatin fiber does not seem to be much affected by either a "symmetrical" or an off-axis location of H1. After all, this may not be at all surprising. It may simply be a particular reflection of the intrinsic ability of histone H1 and histone H1-related proteins to condense DNA into 30 -40-nm fibers regardless of the presence or not of nucleosomes (51).
We also wanted to take advantage of this reconstituted system to see if it could shed some insight onto the still unresolved enigma of histone H1 and transcription (52) (see Fig. 4). Special attention was paid to the MgCl 2 conditions used in the reaction mixtures in order to prevent any precipitation of H1 containing oligonucleosome templates (44,53). Under the experimental conditions used by us, the presence or absence of H1 in the reconstituted oligonucleosome complexes (Fig. 4, lanes 4 and 5) did not seem to have any major effect on the modulation of RNA polymerase III transcription. This is in contrast to the presence of nucleosomes (Fig. 4, compare lanes 3 and 4) which very significantly decreases the extent of transcription on the 208-12 DNA template as it had already been reported previously (54). Our results are also in agreement with the results obtained with a polymerase II system which observed transcriptional repression by nucleosomes but not H1 in reconstituted Drosophila chromatin (55). However, this contradicts earlier results (56). Further analysis in this direction, designed to clarify this seemingly contradictory results is currently carried out in our laboratory.