In Vitro Binding of H1 Histone Subtypes to Nucleosomal Organized Mouse Mammary Tumor Virus Long Terminal Repeat Promotor*

The binding of all known linker histones, named H1a through H1e, including H10 and H1t, to a model chromatin complex based on a DNA fragment containing the mouse mammary tumor virus long terminal repeat promotor was systematically studied. As for the histone subtype H1b, we found a dissociation constant of 8–16 nm to a single mononucleosome (210 base pairs), whereas the binding constant of all other subtypes varied between 2 and 4 nm. Most of the H1 histones, namely H1a, H1c, H1d/e, and H10, completely aggregate polynucleosomes (1.3 kilobase pairs, 6 nucleosomes) at 270–360 nm, corresponding to a molar ratio of six to eight H1 molecules per reconstituted nucleosome. To form aggregates with the histones H1t and H1b, however, greater amounts of protein were required. Furthermore, our results show that specific types of in vivo phosphorylation of the linker histone tails influence both the binding to mononucleosomes and the aggregation of polynucleosomes. S phase-specific phosphorylation with one to three phosphate groups at specific sites in the C terminus influences neither the binding to a mononucleosome nor the aggregation of polynucleosomes. In contrast, highly phosphorylated H1 histones with four to five phosphate groups in the C and N termini reveal a very high binding affinity to a mononucleosome but a low chromatin aggregation capability. These findings suggest that specific S phase or mitotic phosphorylation sites act independently and have distinct functional roles.

The binding of all known linker histones, named H1a through H1e, including H1 0 and H1t, to a model chromatin complex based on a DNA fragment containing the mouse mammary tumor virus long terminal repeat promotor was systematically studied. As for the histone subtype H1b, we found a dissociation constant of 8 -16 nM to a single mononucleosome (210 base pairs), whereas the binding constant of all other subtypes varied between 2 and 4 nM. Most of the H1 histones, namely H1a, H1c, H1d/e, and H1 0 , completely aggregate polynucleosomes (1.3 kilobase pairs, 6 nucleosomes) at 270 -360 nM, corresponding to a molar ratio of six to eight H1 molecules per reconstituted nucleosome. To form aggregates with the histones H1t and H1b, however, greater amounts of protein were required. Furthermore, our results show that specific types of in vivo phosphorylation of the linker histone tails influence both the binding to mononucleosomes and the aggregation of polynucleosomes. S phase-specific phosphorylation with one to three phosphate groups at specific sites in the C terminus influences neither the binding to a mononucleosome nor the aggregation of polynucleosomes. In contrast, highly phosphorylated H1 histones with four to five phosphate groups in the C and N termini reveal a very high binding affinity to a mononucleosome but a low chromatin aggregation capability. These findings suggest that specific S phase or mitotic phosphorylation sites act independently and have distinct functional roles.
H1 histones are a heterogeneous group of at least five subtypes with closely related but nonetheless different primary structures (1,2). Two further H1 subtypes are known: the histone H1 0 , which is found in nonreplicative tissues (3,4) and in rapidly proliferating cells (5), and the testis-specific histone variant H1t (6). The various linker histones containing a globular central region flanked by highly basic and hydrophilic C-and N-terminal tails (7,8) bind to the nucleosome and promote the organization of nucleosomes to a higher order structure (9,10).
There is evidence that histone H1 may interact differently with transcriptionally active and inactive regions of chromatin (11). Linker histones are also thought to modulate nucleosome position (12,13) and to influence replication efficiency in vitro (14).
The presence of this large number of various H1 histone subtypes and their possible posttranslational modifications, such as phosphorylation (15), make it very probable that H1 histones play numerous structural and functional roles in chromatin. Until now, no specific role for the various variants has been established although Kaludov et al. (16) showed that the mouse histone H1b binds preferentially to a regulatory sequence within a mouse H3.2 replication-dependent histone gene. Previous analysis of the structural role of H1 histones demonstrated that three subfractions of H1 histones differ in their effectiveness in condensing DNA fibers into ordered aggregates (17) and that histone subtype H1t, compared with other subtypes, differs in its ability to condense chromatin (18,19). Furthermore, differences in the binding of H1 variants to DNA or phosphorylated H1 histones to DNA have been shown (20,21).
The present work was undertaken to extend these earlier binding studies using an in vitro assay developed by Wolffe and Hayes (22), enabling us to systematically study the binding of all known linker histones to a model chromatin complex. The complex is based on a DNA fragment containing the mouse mammary tumor virus long terminal repeat (MMTV LTR) 1 promotor, the chromatin structure of which has been well characterized (23). The MMTV LTR reproducibly acquires a series of six positioned nucleosomes (24). We have extended our systematic study of the histone-chromatin complexes formed by all known various H1 subtypes, H1a to H1e, H1 0 , and H1t, to differently phosphorylated H1 histones according to the various phases of the cell cycle. The in vitro results presented here suggest that (i) differences in binding to a single mononucleosome (210 bp) and (ii) differences in aggregation of polynucleosomes (1.3 kbp, 6 nucleosomes) exist among the various linker histones. Furthermore, we found that (iii) specific types of in vivo phosphorylation of the linker histone tails do not influence binding to mononucleosomes, whereas (iv) aggregation of polynucleosomes was decreased in the case of highly phosphorylated mitotic H1 histones.

EXPERIMENTAL PROCEDURES
Culture and Synchronization of Cells-NIH 3T3 fibroblasts were grown in monolayer cultures and cultivated in DMEM (Life Technologies, Inc.) supplemented with 10% fetal calf serum, penicillin (60 g/ ml), and streptomycin (100 g/ml) in the presence of 5% CO 2 . To obtain cells from the G 1 phase, S phase, and mitosis, we used a cell synchronization method described by Talasz et al. (15). Routine microscopic examination and FACS analysis showed that more than 90% of the cells were in the various cell cycle phases.
Isolation of H1 Histones-For isolation of H1 histones, nuclei from * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. mouse liver or NIH fibroblasts were prepared as described by Zubay and Doty (25) and Talasz et al. (15). H1 histones were isolated from the resulting nuclear preparations by extracting with 5% perchloric acid at 4°C for 1 h. The mixture was centrifuged at 10,000 ϫ g for 20 min. The supernatant was mixed with trichloroacetic acid to the final concentration of 20% and allowed to stand for 1 h at 4°C. The precipitate was washed once with acetone-HCl and thereafter with acetone, resuspended in water containing 10 mM 2-mercaptoethanol, and freeze-dried.
Reversed-phase HPLC-The equipment used consisted of two Beckman model 114M pumps, a 421A system controller, and a model 165 variable-wavelength UV-visible detector. The effluent was monitored at 210 nm and the peaks were recorded using Beckman System Gold software. Protein separation was performed on a Nucleosil 300 -5 C4 column (12.5 or 25 cm ϫ 0.8 cm, 5-m beads, 300 Å). The freeze-dried proteins were dissolved in 0.04 M 2-mercaptoethanol/water containing 0.1% trifluoroacetic acid, and samples of 100 -300 g of histones were loaded onto the column. At a constant flow rate of 1 and 1.5 ml/min, the H1 histones were eluted using a 45-min linear gradient from 34 to 54% solvent B (solvent A, water containing 15% ethylene glycolmonomethylether and 0.1% trifluoroacetic acid; solvent B, ethylene glycolmonomethylether (15%)/70% acetonitrile (85%) with 0.1% trifluoroacetic acid) (26 -28).
Capillary Electrophoresis-High performance capillary electrophoresis of histone H1 proteins was performed with slight modifications according to Lindner et al. (29,30). In short, an untreated capillary (57/50 cm) was used, and separation was performed in a 10 mM triethylamine/H 3 PO 4 buffer system (pH 2.0) containing 90 mM perchloric acid and 0.02% hydroxypropylmethyl cellulose.
Plasmids and DNA Fragments-The plasmid pMMTV-CAT was prepared from Escherichia coli cells harboring this plasmid using the alkali lysis method and the PC-500 Nucleobond kit (Macherey-Nagel). Plasmid DNA was digested either with restriction endonucleases SacI and BamHI (Boehringer Mannheim) at 37°C (2 units/g of DNA overnight), which deliberated a 210-bp insert containing part of the MMTV LTR promotor region, or with restriction endonucleases BamHI and HindIII (Boehringer Mannheim) at 37°C (2 units/g of DNA overnight), which liberated a 1.3-kbp insert containing the whole MMTV LTR promotor region. The inserts were purified from the digest on 1% agarose gels in 1ϫ TBE buffer (90 mM Tris, 90 mM boric acid, 2.5 mM Na 2 EDTA, pH 8.3) using the GenElute Agarose Spin Columns (Supelco, Bellefonte). The 210-bp and 1.3-kbp DNA fragments were 5Ј-end labeled with T4 polynucleotide kinase and used for nucleosome reconstitution experiments. The naked nonspecific DNA (146 bp) was recovered from mouse liver mononucleosome particles with phenol-chloroform extraction after proteinase K (50 g/1 mg of DNA) digestion in the presence of 1% SDS at 37°C for 2 h. Mononucleosomes were recovered from mouse liver nuclei as described in Ref. 31.
Nucleosome Reconstitution-Nuclei from mouse liver were prepared (25), and the crude nuclear extract was centrifuged in a Beckman SW28 rotor through a 2.2 M sucrose (50 mM Tris-HCl, pH 7.5, 1 mM MgCl 2 , 0.1 mM phenylmethanesulfonyl fluoride, 0.1 mM benzamidine) cushion at 25,000 rpm for 1 h at 4°C to remove residual cytoplasmic components. Soluble chromatin was obtained by brief micrococcal nuclease (Sigma) digestion of the nuclei (31). Digestion of nuclei with 10 units of micrococcal nuclease per 1 mg of DNA for 30 s at 37°C in 5 mM Tris-HCl, pH 7.5, 0.3 M sucrose, 1 mM CaCl 2 (Buffer 1) produces chromatin lengths of 10 -30 nucleosomes. To remove linker histone H1 this soluble chromatin was treated with Dowex 50 W-X2 (Sigma) in 0.4 M NaCl, 0.05 M sodium phosphate buffer (pH 7.4). Soluble chromatin without H1 histones was transferred to a dialysis bag (with a molecular size limit of 3 kDa) and dialyzed against Buffer 1 for 4 h. Thereafter, chromatin was redigested with 60 units of micrococcal nuclease per 1 mg of DNA for 10 min. Core particles were separated on 5-20% (w/w) sucrose gradients containing 5 mM Tris-HCl, pH 7.5, 1 mM Na 2 EDTA by centrifuging in a Beckman SW28 rotor at 25,000 rpm for 17 h at 4°C. Nucleosomes were reconstituted onto radiolabeled DNA fragments by exchange with mouse liver core particles (donor chromatin) according to Drew and Travers (32). In the case of reconstitution onto a 1.3-kbp DNA fragment, the original 6 l of exchange reaction containing ϳ3200 ng of donor chromatin, 520 ng of naked nonspecific DNA (146 bp), and 200 ng of labeled 1.3-kbp fragment of the MMTV promotor was incubated in 1 M NaCl for 1 h (all incubations at room temperature). By adding 2.75 l of TE 1 M NaCl was then diluted to 0.8 M NaCl followed by addition of 4.55 l of TE to 0.6 M NaCl, for 30 min each (TE consisted of 10 mM Tris-HCl, pH 8.0, 1 mM EDTA). Thereafter, it was diluted to 0.4 M NaCl by addition of 9.1 l of TE and to 0.1 M NaCl by addition of 82 l TE, for 30 min each. The salt concentration was finally diluted to 0.05 M NaCl with 109 l of TE. In this condition, 100% of the labeled 1.3-kbp MMTV promotor fragment was assembled to a saturated oligonucleosome chain. In the case of reconstitution onto 210-bp DNA fragments, the original 6 l of exchange reaction containing ϳ1300 ng of donor chromatin, ϳ260 ng of naked nonspecific DNA (146 bp), and 80 -100 ng of labeled 210-bp fragment of the MMTV promotor was incubated for 1 h (all incubations at room temperature) in 1 M NaCl. This was then diluted from 1 M NaCl to 0.05 M NaCl with TE as described for the exchange reaction onto 1.3-kbp. About 50 -60% of the labeled MMTV LTR promotor 210-bp fragment was assembled to mononucleosome cores as monitored with electrophoresis.
Linker Histone Binding Experiments-To ensure accurate protein concentration determination BCA protein assay (Pierce), spectrophotometry at 230 nm and reversed-phase HPLC with peak integration were used for concentration measurement of H1 histones. Additionally, we used the dotMETRIC protein assay (Geno Technology, Inc.), because the results obtained with this assay are not dependent on the amino acid composition of the protein, making the assay independent of protein-to-protein variation. Nucleosomes reconstituted with approximately 7 ng of radioactively end-labeled 210-bp MMTV LTR DNA in the presence of 110 ng of unlabeled nonspecific DNA were incubated with various amounts of histone H1 subtypes (see legend to Samples from reconstitution of nucleosomes or DNA with linker histones were incubated at room temperature for 15-20 min and loaded directly onto running 0.7% agarose gels in 0.1ϫ TB (1ϫ TB is 90 mM Tris, 90 mM boric acid, pH 8.3) or in 0.5ϫ TBE. Running buffer was circulated throughout the experiment at a rate sufficient to prevent formation of either pH or ion gradients. After electrophoresis, the gels were dried and autoradiographed. The quality and quantity of the dried gels were determined by PhosphorImager analysis (Molecular Dynamics Inc.).
Amplification and Polymerase Chain Reaction Radiolabeling-For amplification and internal labeling of a 210-bp DNA fragment corresponding to the first A nucleosome of the MMTV LTR promotor (24) Micrococcal Nuclease Digestion-Mononucleosomes (100 ng of specific internally labeled 210-bp DNA and 1600 ng of nonspecific 146-bp DNA) in the presence or absence of 200 ng of histone H1 (w/w ratio of histone to DNA was 0.12) were digested with 0.15, 0.3, or 0.6 units of micrococcal nuclease (Sigma) for 5 min at 37°C. Incubation with histone H1 was performed as described above. Ca 2ϩ was adjusted to 2 mM concomitantly with the addition of micrococcal nuclease. Micrococcal nuclease digestions were terminated by adding of EDTA (5 mM), SDS (0.25%, w/v), and proteinase K (0.5 g/ml). Digestion with proteinase K lasted 2 h at 37°C. DNA was recovered by phenol extraction. The labeled DNA fragments were separated by electrophoresis in nondenaturing 9% polyacrylamide gels with 1ϫ TBE as running buffer. After electrophoresis, the gels were dried and autoradiographed. Alkaline Phosphatase Digestion of Mitotic H1 Histones-H1 histones from mitotic cells were digested with alkaline phosphatase basically as described (33,34). About 50 g of H1 histones in 140 l of 10 mM Tris-HCl, pH 8.0, and 1 mM phenylmethanesulfonyl fluoride was mixed with 2 units of E. coli alkaline phosphatase (Sigma), corresponding to 40 units/mg histone. Incubation at 22°C was terminated after 16 h by adjusting the incubation mixture to 5% perchloric acid. The histone H1 was resolved as described above.

RESULTS
To test mononucleosome cores for their ability to bind the various linker histones we first fractionated mouse liver H1 histones using reversed-phase HPLC. The resulting subtypes and subfractions obtained were designated according to the nomenclature of Lennox et al. (35). In mouse liver we found H1a, H1b, H1c, and H1 0 as pure subtypes, whereas one subfraction was a mixture containing the variants H1d and H1e (Fig. 1A) (28). Histone subtype H1t was obtained by means of reversed-phase HPLC fractionation of mouse testis (Fig. 1B) (36). Mononucleosome cores were prepared as described under "Experimental Procedures" and reconstituted with a DNA fragment of the MMTV LTR promotor. We used a 210-bp fragment corresponding to the first A nucleosome described by Archer et al. (24). We assembled about 50 -60% of the labeled MMTV LTR DNA with a single octamer, whereas the rest remained unbound and served as a naked DNA control (Fig. 2A, lane 2). Fig. 2 shows the binding pattern of the mouse H1 histones H1a to H1e, H1 0 , H1t, and total H1 histones from mouse liver and testis. Similar to reconstitution experiments performed by other groups (37), we found that the reconstitutes were prefer-entially bound by the various linker histone subtypes over the naked DNA ( Fig. 2A, lanes 3-8). After the octamer DNA complex shifted completely to the H1 octamer DNA complex, the H1 histone started to bind to naked DNA (disappearance of free DNA) and an H1 DNA complex appeared ( Fig. 2A, lanes 7 and  8). This phenomenon is visible at titration with H1a, H1c, H1d/e, H1 0 , and total liver H1, but not clearly pronounced with H1b, H1t, and total testis H1 histones.
Nucleoprotein gels, such as those shown in Fig. 2A, were quantitatively analyzed to determine the relative nucleosome binding affinity of the various linker histone subtypes (Fig. 2B). This analysis revealed that none of the subtypes has significantly different affinity, with the exception of histone H1b. For this H1b subtype we found a dissociation constant of 8 -16 nM, whereas the dissociation constant of all other subtypes varied between 1 and 3 nM.
In the cases of H1b and H1t, the binding to free DNA appears to be less strong than in the case of H1a. To test whether the various H1 subtypes have different affinities for free DNA we directly measured the binding of H1 subtypes to not nucleosomal organized DNA. These results (Fig. 3) further confirmed our finding that H1b and H1t have less stronger affinities not only to nucleosomal organized DNA but also to free DNA.
As shown by Ura et al. (38), micrococcal nuclease digestion of H1 histone-bound chromatin leads to an accumulation of a kinetic intermediate containing all the histones and 166 -168 bp of DNA, known as the chromatosome (39). We used this chromatosomal stop (40) as a diagnostic tool for correct incorporation of the various linker histone subtypes into chromatin. Chromatosome-length DNA was observed regardless of the histone subtype used to reconstitute the MMTV LTR promotor mononucleosomal cores. Fig. 4 compares total H1 histones from mouse liver and H1b subtype (other subtypes not shown). With an H1/DNA ratio (w/w) of 0.12, corresponding to a calculated molar ratio of about one MMTV LTR promotor DNA H1 histone molecule per reconstituted nucleosome core, both total H1 from To test the influence of phosphorylation of the histone tails on binding to the MMTV LTR promotor DNA octamer complex, we prepared linker histones in various phosphorylation states (see under "Experimental Procedures"). Total H1 histones from mouse NIH 3T3 fibroblasts were prepared from synchronized cells passing through G 1 phase, S phase, or mitosis. Capillary electrophoresis was used to measure the phosphorylation state of total H1 histones (Fig. 5). In growth-arrested cells at G 1 phase, linker histones are unphosphorylated, whereas in late S phase, the H1 variants exist as a combination of molecules containing one to three phosphate groups, according to the particular subtype (15). The fastest migrating components in capillary electrophoresis are the unphosphorylated H1 histones (29,36) (Fig. 5A, peaks between 18 and 20.5 min). Histones from S phase cells (Fig. 5B) and, to a great extent, highly phosphorylated H1 histone subtypes from cells in mitosis containing four to five phosphate groups migrate at lower speed. In the case of the most highly phosphorylated H1 histones, the main peaks are visible at 20.5-21.5 min (Fig. 5C). When we digested the mitotic H1 histones with alkaline phosphatase, only peaks between 18 and 20.5 min were visible, corresponding to unphosphorylated H1 histones (Fig. 5D). Fig. 6A shows the association of cell cycle-specific linker histones with the MMTV LTR nucleosome core, whereas Fig.  6B presents the quantitative analysis of these nucleoprotein gels. Both the unphosphorylated linker histones prepared from G 1 cells and the medium phosphorylated H1 histones from S phase seem to have a binding constant of about 3-5 nM, with a total shift to the chromatosome at 32 nM. Interestingly, the very highly phosphorylated H1 histones from mitotic cells also have a binding constant of about 3-5 nM and shifted completely to the chromatosome with 32 nM (Fig. 6A, mitosis). Khadake and Rao (19) showed that a mixture of histone subtypes H1b, H1c, H1d, and H1e retarded the mobility of a polynucleosome in a concentration-dependent manner, whereas H1t had only little effect. We extended our study to compare all H1 subtypes. To check the properties of chromatin aggregation (condensation) of the various linker histones, we used an MMTV LTR polynucleosome core template of a defined length of 1.3 kbp, which is organized in six phased nucleosomes (24) after reconstitution with liver core histones, as described under "Experimental Procedures." Bartolomé et al. (41) showed that about six nucleosomes are sufficient to produce compact structures, such as chromatin fragments, of higher molecular weight, and many other laboratories using different techniques reported results concerning the association of oligonucleosomes to form higher order structures (42,43). Therefore, our model FIG. 5. High performance capillary electrophoresis of total H1 histones obtained from synchronized NIH cells. Capillary electrophoresis was used to measure the phosphorylation state of H1 histones from growth-arrested cells in G 1 (A), late S phase (B), and mitosis (C). Unphosphorylated H1 histones are the fastest-migrating component, appearing between 18 and 20.5 min. S phase H1 histones (zero to three phosphates per molecule) show a shift to lower migrating peaks. This shift is clearly pronounced in highly phosphorylated H1 histones from mitotic cells (four or five phosphates per molecule), resulting in the main peaks at 20.5-21.5 min. After digestion of mitotic H1 histones with alkaline phosphatase (D), the high performance capillary electrophoresis picture is comparable to that of G 1 phase H1 histones (A).
FIG. 6. Binding activity of various phosphorylated H1 histones to mononucleosomes. A, nucleoprotein gel shift assay of the nucleosome binding behavior of unphosphorylated H1 histones from G 1 phase, medium phosphorylated H1 histones from S phase, and highly phosphorylated from mitosis to the MMTV LTR nucleosome. The positions of free DNA, nucleosome (Nuc), and H1-nucleosome complexes (Nuc-H1) are indicated. Nucleosomes reconstituted with radioactively end-labeled 210-bp MMTV LTR DNA were incubated with 0, 0.375, 0.75, 1.5, 3, 6, and 12 ng of H1 subtypes, corresponding to 0, 1, 2, 4, 8, 16, and 32 nM (lanes 2-8, respectively). Lane 1, free 210-bp DNA. Complexes were separated on 0.7% agarose gels, and the gels were autoradiographed. B, binding titration of various phosphorylated linker histones obtained from G 1 , S, and mitosis. The fraction of H1 histone-bound nucleosomes observed on autoradiographs like those shown in A was determined and plotted against free H1 histone concentration. The mean and S.D. of three experiments are shown.
consisting of chromatin fragments containing six nucleosomes is suitable for studying the influence of linker histones on the formation of the polynucleosome complex. The MMTV LTR polynucleosomes were mixed with increasing quantities of the various H1 histone subtypes and the formed complexes analyzed on agarose gels (Fig. 7A). Most of the H1 histones, namely H1a, H1c, H1d/e, and H1 0 , aggregate completely at 270 -360 nM, corresponding to a molar ratio of six to eight H1 molecules per reconstituted nucleosome. In the case of both histone subtypes H1t and H1b, higher amounts of these proteins are needed to aggregate (Fig. 7A, H1b, H1t, and H1b, higher concentration). We tested the influence of H1t on aggregation using total testis H1 histones, because more than 30% of mouse testis linker histones belong to histone subtype H1t and, in-deed, the result is similar to that of H1t alone (Fig. 7B). Fig. 7B summarizes the quantitative analysis of the MMTV LTR polynucleosome H1 aggregation experiments.
Finally, we compared the complex formation of unphosphorylated (G 1 phase), medium phosphorylated (S phase), and highly phosphorylated (mitosis) H1 histones with the 1.3-kbp MMTV LTR oligonucleosome. It is evident from Fig. 8 that the mobility of the H1 histones prepared from G 1 cells progressively decreased with increasing protein/DNA ratio (Fig. 8A, G 1  phase, lanes 3-7). At an H1 histone concentration of 90 -270 nM (w/w ratio, 0.2-0.6) aggregates with increasing molecular weight were formed, with completely immobile aggregates occurring at 270 -360 nM (Fig. 8A, G 1 phase, lanes 7 and 8). S phase H1 histones show nearly the same pattern of aggregation formation as compared with that of G 1 phase H1 histones. A slightly higher concentration of histones, however, was required to create totally immobile aggregates (Fig. 8A, S phase). The highly phosphorylated H1 histones prepared from mitotic cells, on the other hand, had a lesser effect on the mobility of the polynucleosome as compared with G 1 or S phase histones (Fig. 8A, compare lane 3 to lane 8, G 1 , S phase, and mitosis). A progressive decrease in mobility was not visible between a 0.2 and a 0.6 w/w ratio, and higher amounts of linker histones were needed for the formation of aggregates with reduced mobility or completely immobile aggregates (w/w ratio of between 0.8 and 1.0). As a control, we digested the mitotic H1 histones with alkaline phosphatase (see Fig. 5D) to test the behavior of unphosphorylated H1 from mitotic cells. These histones are comparable to G 1 and S phase histones in their ability to bind oligonucleosomes and in the formation of aggregates. Quantitative analyses of gels like those shown in Fig. 8A are given in Fig. 8B. DISCUSSION Most of the H1 subtypes reveal high similarity in binding affinity to mononucleosomes despite their differences in amino acid sequences. For example, H1 0 , which showed only 30 -38% sequence similarity as compared with the other subtypes (44), binds with the same affinity (45). Because the globular domain of the linker histones can be regarded as the nucleosomebinding domain (46), a sequence difference in the globular domain of H1b as compared with all other H1 histones could be responsible for our finding that H1b has the lowest binding affinity. Interestingly, only three amino acids in the globular domain at positions 54, 59, and 68 were found to be different (glycine instead of serine or alanine, proline instead of alanine or glycine, and glycine instead of alanine or hydroxyproline) (44). It is not clear, however, whether these differences explain the decreased binding to the MMTV LTR promotor. A recently published paper (16) showed that H1b interacts preferentially with highly specific DNA sequences but poorly with nonspecific DNA.
Comparison of the aggregation behavior of the various linker histone subtypes using gel retardation revealed that higher amounts of H1b, H1t, or total H1 from testis are necessary to reach full aggregation. Furthermore, a difference in quality of shifting could be seen among the various H1 histones. The total histones from liver and H1a, H1c, and H1d/e showed a steady increase in polynucleosome complex formation, whereas titration with increasing amounts of H1b and H1t did not cause a continuous shift but an abrupt occurrence of immobile aggregates. This behavior can be interpreted as a diminished cooperative binding capability of these H1 subtypes.
There are several indications that somatic variants of H1 histones differ in their ability to induce DNA condensation (18,47). The results obtained with different methods, however, were variable (17,48). Khadake and Rao (19) compared a mixture of H1 histones (H1b, c, d, and e) and H1t for their ability to decrease the mobility of 5S DNA chromatin using an agarose gel retardation assay and explained the diminished binding of H1t with the lack of S/TPKK motifs in the C terminus. Our finding that the binding behavior to chromatin of histone H1b is similar to that of H1t cannot be explained by the lack of T/SPKK motifs in the C terminus, because H1b has four motifs in the C terminus (44,49).
The data available on the influence of the cell cycle-dependent phosphorylation of linker histones on chromatin structure are conflicting (50). Because we used gel retardation in our experiments, the results are not directly comparable with circular dichroic or sedimentation studies (14, 21, 51, 52) but can support some of the earlier findings. An unexpected and inter- esting result of our studies is the binding ability of medium and highly phosphorylated H1 histones from S phase or mitotic cells to the mononucleosome, which is comparable to that of unphosphorylated H1 histones. Neither the S phase-specific phosphorylation of H1 histones, which occurs only in the C terminus (53), nor the highly phosphorylated mitotic H1 histones, in which both N and C termini are phosphorylated, show reduced binding capability, indicating that neither type of phosphorylation influences the binding of the globular domain to the mononucleosome.
We used the H1 histones with various phosphorylation patterns to titrate them to oligonucleosomal chromatin. These experiments revealed that H1 histones from G 1 and S phase show nearly the same pattern of aggregation formation, whereas H1 histones from mitosis have a lesser effect on the mobility of the polynucleosome as compared with G 1 or S phase histones. The chromatin binding and aggregation experiments reported here suggest that S phase-specific phosphorylation with one to three phosphate groups at specific sites in the C terminus influences neither the binding to a mononucleosome nor the aggregation of polynucleosomes. In contrast, highly phosphorylated histones show a very high binding affinity to a mononucleosome but low chromatin aggregation capability. These findings suggest that phosphorylation at both tails (N and C termini) is necessary to influence aggregation. In our in vitro system consisting of only DNA, core histones, and highly phosphorylated linker histones obtained from mitotic cells, there was a distinct negative effect according to the formation of oligonucleosome aggregates, which was neutralized after removal of the phosphate groups with alkaline phosphatase. Our findings therefore support the idea that phosphorylation at mitosis relieves constraints due to loosening of protein-DNA interaction, thereby permitting mechanisms other than H1 histone phosphorylation to condense the chromatin (21).