Effects of H1 Histone Variant Overexpression on Chromatin Structure*

The importance of histone H1 heterogeneity and total H1 stoichiometry in chromatin has been enigmatic. Here we report a detailed characterization of the chromatin structure of cells overexpressing either H10 or H1c. Nucleosome spacing was found to change during cell cycle progression, and overexpression of either variant in exponentially growing cells results in a 15-base pair increase in nucleosome repeat length. H1 histones can also assemble on chromatin and influence nucleosome spacing in the absence of DNA replication. Overexpression of H10 and, to a lesser extent, H1c results in a decreased rate of digestion of chromatin by micrococcal nuclease. Using green fluorescent protein-tagged H1 variants, we show that micrococcal nuclease-resistant chromatin is specifically enriched in the H10 variant. Overexpression of H10 results in the appearance of a unique mononucleosome species of higher mobility on nucleoprotein gels. Domain switch mutagenesis revealed that either the N-terminal tail or the central globular domain of the H10 protein could independently give rise to this unique mononucleosome species. These results in part explain the differential effects of H10 and H1c in regulating chromatin structure and function.

The fundamental repeating unit of the eukaryotic chromatin is the nucleosome, which consists of an octamer of two molecules each of the core histones H2A, H2B, H3, and H4, and in higher eukaryotes, at least a single molecule of H1 (or linker) histone (1,2). The precise location and the number of H1 histones in the nucleosome are not known (3)(4)(5)(6). About 166 base pairs (bp) 1 of DNA are wrapped around the H1-containing nucleosome in two complete turns. Histone H1 is believed to make contacts with the linker DNA, and facilitate further condensation of the nucleosomal template into higher order structures such as the 30-nm chromatin fibers (7,8).
All histones, except H4, are composed of multiple primary sequence variants (9 -11) and show distinct patterns of expression during development and differentiation in a variety of organisms (9,(12)(13)(14). In the mouse, at least seven linker his-tone variants exist. These include the somatic variants H1a, H1b, H1c, H1d, H1e, and H1 0 and the testis-specific H1t variant. All these variants have the same general structure consisting of a central globular domain, a short N-terminal tail, and a longer C-terminal tail (1,2). H1 histones have been studied in some detail and found to differ in their timing of synthesis, rates of synthesis, turnover rates, phosphorylation status, and ability to compact chromatin, but a satisfactory description of their functional significance is still missing (1,2,10). H1 0 has been termed an "extreme" somatic variant, as its sequence is considerably diverged from other somatic H1 histones (15,16). It accumulates to high levels in non-dividing and terminally differentiated cells (17).
Over the past three decades, reports have appeared in the literature correlating the expression of particular H1 histone subtypes with changes in chromatin structure and gene expression in accordance with the developmental status of many organisms (18 -21). The most dramatic effect of a H1 histone variant, histone H5, has been observed in chicken erythrocytes, where it is progressively deposited onto the chromatin as the erythrocyte matures (22,23). This process is marked by the concomitant loss of H1 histones already bound to the chromatin, condensation of the chromatin, and an increase in nucleosome spacing. When histone H5 was inducibly overexpressed in a rat cell line, the cell cycle slowed and expression of certain genes were altered in a direct correlation with the amount of excess H5 bound to the chromatin (24). The chromatin from histone H5-overexpressing cells was found to be resistant to cleavage by micrococcal nuclease (MN) (25).
To gain an insight into the functional significance of H1 histone variants, our laboratory developed a system for the inducible overexpression of individual mouse H1 histones in cultured mouse fibroblasts (26,27). Overexpression of individual H1 variants using this system disrupts the normal stoichiometry of the H1 variants, making it feasible to study the functional significance of individual variants in vivo. Using this system, we focused on two of the mouse H1 histone variants, H1 0 and H1c, and demonstrated the differential effects of their overexpression on gene expression and cell cycle progression (27). Overexpression of the H1 0 variant lead to reduced steadystate transcript levels for all RNA polymerase II genes studied, and a transient delay in the re-entry of G 0 -arrested cells into the S phase of the cell cycle following release from arrest. This is not surprising, given the similarity of histone H1 0 to the avian histone H5 in sequence, size, and expression patterns (15). Surprisingly, overexpression of the H1c variant led to either a dramatic increase or no change in the steady-state transcript levels of all genes tested, and had no effect on the re-entry of G 0 cells into the S-phase. We further demonstrated that the differential effects of these two variants are due to differences in their globular domains (28). In this study we took advantage of the H1 variant overexpression system to see if global differences in the structure of chromatin from cells producing either H1 0 or H1c could be detected and if these differences could explain the observed functional differences between these two variants.

Generation of Expression Vectors and Cell
Lines-The H1 0 -and H1coverexpressing cell lines, MTH1 0 and MTH1c, were described previously (26,27). Plasmid pMTAneo was constructed by deleting the H1 coding sequences from plasmid MT43MslAneo (27). Construction of cell lines overexpressing "domain switch" mutants of H1 0 and H1c have been described elsewhere (28). Plasmids pMTH1 0 GFPAneo and pMTH1cGFPAneo were constructed using standard cloning methodology (29). Briefly, the coding sequence for the mammalian codon optimized, red-shifted, enhanced green fluorescent protein (GFP) gene was obtained from plasmid pEGFP-C1 (CLONTECH) and cloned in frame after the codon for the last lysine residue of the H1 0 and H1c genes within the plasmids pMTH1 0 Aneo and pMTH1cAneo, respectively (27). During the subcloning steps, a single alanine residue was inserted between the last lysine of the H1 proteins and the first methionine residue of the GFP protein in the H1GFP fusion proteins. Cell lines MTA, MTH1 0 GFP, and MTH1cGFP were created by transfecting BALB/c 3T3 cells with plasmids pMTAneo, pMTH1 0 GFPAneo, and pMTH1cGFPAneo, respectively, as described previously (26,27). Although we present data from only a few cell lines in this study, many stable cell lines obtained from independent transfections were analyzed to avoid complications arising out of clonal variation.
Cell Culture and Induction Protocols for H1 Variant Overexpression-All experiments were initiated from stocks of stable cell lines stored in liquid nitrogen and were maintained as described previously (26,27). For the overexpression of H1 histone variants during exponential growth conditions, cells were seeded at a low density; typically, less than 10% of the total surface area of the flask was covered. Twelve h after the initial seeding, cells were treated with 50 M ZnCl 2 for 12 h. This was followed by another 84 h of induction with 100 M ZnCl 2 . The medium was replaced every 24 h. Cells were harvested prior to confluence, i.e. no more than 85% of the surface area of the flask was covered by cells at the time of harvesting. For the overexpression of H1 histone variants under density arrest, cells were grown to confluence prior to induction with ZnCl 2 as described above. Total histones were extracted and separated by high performance liquid chromatography (HPLC) as described previously (28). To label DNA, exponentially growing cells were treated with 0.5 Ci/ml [ 3 H]thymidine (90 Ci/mmol, NEN Life Science Products) for 72 h. The labeled cells were then trypsinized and seeded into fresh flasks at low or high densities and induced with ZnCl 2 in the absence of [ 3 H]thymidine, as described above.
MN Digestions-MN was from Sigma (catalogue no. N8630). Nuclei were prepared and digested with MN as described previously (30), except that the wash buffer was supplemented with 200 M CaCl 2 and 1.5 units/ml MN to make the digestion buffer. Digestion was allowed to proceed at 25°C. For nucleosome linker length determinations nuclei were lysed by the addition of EDTA, EGTA, and SDS to final concentrations of 2 mM, 1 mM, and 0.5%, respectively. Total DNA was isolated after proteinase K digestion by phenol extraction. For all other experiments, MN-digested chromatin was fractionated as follows. After MN digestion, nuclei were pelleted for 2 min at 16,000 ϫ g at 4°C. The supernatant was collected and termed the soluble S1 chromatin fraction, consisting mainly of mononucleosomes (31). The pellet was resuspended in nuclei lysis buffer containing 2 mM EDTA and 1 mM EGTA and incubated on ice for 15 min. The lysed nuclei were pelleted for 10 min at 16,000 ϫ g at 4°C. The supernatant, containing mono-and polynucleosomes (31), was termed the soluble S2 fraction. The term MN-soluble chromatin refers to the combined S1 and S2 fractions. The pelleted material is referred to as the MN-insoluble fraction.
Quantitation of GFP-tagged H1 Variants in the Micrococcal Nuclease-soluble and -insoluble Chromatin Fractions by Fluorometry-Nuclei from density-arrested cells expressing either H1 0 GFP or H1cGFP were mixed with a 20-fold excess of control nuclei, digested with MN and fractionated as described above. The MN-insoluble chromatin pellet and the combined S1 and S2 fractions were extracted with high salt buffer (0.5 M KCl, 10 mM Tris-HCl, pH 7.2, 5 mM MgCl 2 , 0.5 mM phenylmethylsulfonyl fluoride, 2 mM EDTA, and 1 mM EGTA) for 30 min to extract all chromatin-bound H1 histones. Insoluble material was removed by centrifugation at 16,000 ϫ g for 5 min. Histone H1 extracted in the high salt buffer supernatant was used directly for fluorometry, as the presence of 0.5 M KCl had no effect on GFP fluorescence. Individual chromatin fractions were excited at 488 nm, and the fluorescence emitted from GFP was measured at 509 nm in an Aminco Bowman Series 2 luminescence spectrophotometer (Spectronics Instruments). GFP fluorescence observed in the different chromatin fractions was taken to be a measure of the relative amounts of the particular GFP-tagged H1 variant present.
Nucleoprotein (NP) Gels-Chromatin samples for nucleoprotein analysis were prepared and run on composite NP gels as described previously (31). Nuclei were digested for 15 min as described above, except that the digestion buffer was supplemented with 0.3 M sucrose, 4 mM MgCl 2 , 1% thiodiglycol, and 1 mM phenylmethylsulfonyl fluoride. The soluble S2 chromatin fraction was mixed with an equal volume of glycerol and run immediately on NP gels. Bands were visualized after staining the gels with ethidium bromide (EtBr). For the NP patterns of chromatin from cells expressing H1GFP fusion proteins, the gels were first scanned in the blue fluorescence mode on a Storm fluorimager (Molecular Dynamics) to visualize GFP fluorescence from nucleosome species carrying bound H1GFP, and then stained with EtBr and scanned on the fluorimager to visualize all nucleosome species.

Overexpression of H1 Histone Variants Increases Nucleosome
Spacing-MN digestion of chromatin from H1-overexpressing cells or control cells reveals that the vast majority of material is organized in a regularly repeating nucleosomal ladder (Fig.  1A). Under these conditions H1 0 or H1c are about 70 -75% of FIG. 1. Nucleosome spacing in cells overproducing histone H1 0 or H1c under exponential growth conditions. A, exponentially growing MTA, MTH1 0 , and MTH1c cells were induced with ZnCl 2 for a total of 96 h as described under "Experimental Procedures." Overexpression levels of H1 0 and H1c were 70% and 77% of total chromatinbound H1 histones, respectively, as revealed by HPLC analysis. Nuclei were digested with 1.5 units/ml MN at 25°C for 5, 10, and 15 min. DNA was isolated and electrophoresed on a 1.8% Metaphor agarose gel which was subsequently stained with EtBr. Lanes marked 100 bp contained 1 g of a 100-bp DNA ladder (Life Technologies, Inc.). B, densitometric scans of the 10-min MN digestion patterns of MTA, MTH1 0 , and MTH1c cell lines shown above in panel A (lanes 4 -6). The x axis is the distance migrated (mm) in the gel, and the y axis is the absorbance. The arrows indicate the densitometric peak absorbance to which the size measurements for oligonucleosomes were made and correspond to mononucleosomes on the far right and oligonucleosomes of increasing size from right to left. the total H1 species in their respective overexpressing cell line as determined by HPLC (28,32). In addition, the total amount of H1 is estimated to be 1.2-1.4 times that of control cells (32). These results indicate that overexpression of H1 histones to high levels does not lead to the formation of aberrant chromatin structures. Overexpression of either variant does result in a clear 15-16-bp increase in nucleosome spacing from 175 bp in control MTA cells to 189 or 190 bp in the H1 histone-overexpressing cells (Fig. 1, Table I). This observed increase in nucleosome spacing agrees with results obtained with in vitro cell-free chromatin assembly systems where addition of increasing amounts of H1 histone was found to progressively increase nucleosome spacing (33)(34)(35). No increase in nucleosome spacing was observed in MTH1 0 and MTH1c cells not treated with ZnCl 2 (data not shown).
We also investigated the effect of MN digestion of chromatin obtained from cultures induced to overexpress H1 variants under density-arrested conditions ( Fig. 2A and Table I). We noted little difference among the cell lines in apparent nucleosome spacing. Upon closer inspection, we noted that the nucleosomal repeat length was close to 190 bp in all cell lines, including control cells not overexpressing H1 (Fig. 2B, compare lanes 1-3 to lanes 4 -6; also see Table I). Nucleosome spacing appears to change with the cell cycle, being maximal in quiescent cultures and minimal in exponentially growing cells.
To investigate if the increase in nucleosome spacing caused by H1 histone overexpression in exponentially growing cells was coupled to DNA replication, H1 histones were overexpressed for 36 h under these conditions in the presence of 5 mM hydroxyurea (HU). Inhibition of DNA synthesis by HU was almost total, as revealed by the negligible incorporation of tritiated thymidine in these cells (data not presented). The overall amounts and percentages of overproduced H1s were slightly less due to the shorter incubation times. Nevertheless, H1 overexpression and incorporation into chromatin was observed in the absence of DNA synthesis and resulted in an increase in nucleosome spacing over control cells similar to that of exponentially growing cells in the absence of HU treatment (Fig. 2B, compare lanes 7-9 to lanes 4 -6). H1 histones appear to assemble correctly on the chromatin and influence nucleosome spacing even in the absence of DNA replication ( Table I).
Overexpression of H1 Variants Results in Chromatin That Is Increasingly Resistant to Cleavage by Micrococcal Nuclease-In the previous experiments, we noted that cleavage of chromatin from H1-overexpressing cell lines required more MN or longer incubation times to achieve the level of digestion of control chromatin. Formation of MN-resistant chromatin has been observed previously, during in vitro chromatin reconstitution and assembly experiments carried out in the presence of H1 histones (33,36), as well as for the in vivo overexpression of the avian H5 variant (25). To investigate this effect quantitatively, exponentially growing cells were labeled with tritiated thymidine. The labeled cells were then induced to overexpress H1 histones under both exponential growth conditions and density arrest conditions, as described under "Experimental Procedures." To normalize the MN/chromatin ratio between aliquots, a small number of labeled nuclei from control, MTH1 0 , or MTH1c cells was mixed with a 20-fold excess of unlabeled nuclei from control cells prior to digestion with MN. Fig. 3A shows that overexpression of H1 0 and, to a lesser degree, H1c in exponentially growing cultures results in a quantitatively FIG. 2. Cell cycle dependence of nucleosome spacing in cells overexpressing histone H1 0 or H1c. A, density-arrested cultures of MTA, MTH1 0 , and MTH1c cells were induced with ZnCl 2 for a total of 96 h as described under "Experimental Procedures." Overexpression levels of H1 0 and H1c were 75% and 82% of total chromatin bound H1 histones, respectively, as revealed by HPLC analysis. Nuclei were digested with 1.5 units/ml MN at 25°C for 5, 10, and 15 min. DNA was isolated and electrophoresed on a 1.8% Metaphor agarose gel, which was subsequently stained with EtBr. Lanes marked 100 bp contained 1 g of a 100-bp DNA ladder (Life Technologies, Inc.). B, exponentially growing (lanes 1-3) and density-arrested (lanes 4 -6) cultures were induced with ZnCl 2 for a total of 96 h as described under "Experimental Procedures." Exponentially growing cultures were treated with 5 mM hydroxyurea for 12 h, followed by a 36-h induction with 100 M ZnCl 2 in the continued presence of hydroxyurea (lanes 7-9). Overexpression levels of H1 0 and H1c in cells treated with hydroxyurea were 60% and 64% of total chromatin bound H1 histones, respectively. Nuclei were isolated and processed as described in A, except that the MN digestion was carried out for 10 min.  (37,38). Also, H1 0 is known to accumulate naturally to high levels in non-dividing and terminally differentiated cells (17), which have compact chromatin and show reduced levels of transcription. Interestingly, overexpression of either variant in densityarrested cells resulted in an equal, slower digestion of chromatin by MN (Fig. 3B). Density-arrested cells may tend to organize the chromatin into more compact structures, which may be facilitated by the presence of "extra" overproduced histone H1, irrespective of the predominant H1 histone variant. It should be noted that in all cases prolonged digestion with MN resulted in the release of nearly all the chromatin as soluble, mono-or oligonucleosomes.
Micrococcal Nuclease-resistant Chromatin Is Enriched in the H1 0 Variant-HPLC analysis of total histones isolated from the MN-soluble and resistant chromatin fractions of control and H1 variant-overexpressing cells indicated that the MNresistant fraction was specifically enriched in the H1 0 variant (data not shown). To extend this observation, we designed a sensitive assay that avoids potential bias that may arise due to the perturbation of the normal stoichiometry of the H1 variants in vivo. For this assay, we used stable cell lines that express the chimeric H1 variant-GFP constructs H1 0 GFP (cell line MTH1 0 GFP) or H1cGFP (cell line MTH1cGFP). These cell lines express low levels of the H1GFP fusion proteins (5-7% of the total H1 histones present), and do not significantly perturb the natural in vivo stoichiometry of the H1 variants. GFP fluorescence from these chimeric proteins allows us to easily compare the amounts of individual H1 variants in the MN-soluble and -insoluble chromatin fractions. These H1GFP fusion proteins behave identically to their respective native parent H1 variants in that they are localized specifically to the chromatin in the nucleus, give rise to regular MN digestion and NP patterns (see below), and appear to bind chromatin with the same affinity as the parent H1 variants (data not shown). Because of the low levels of expression of the H1GFP fusion proteins, the rates of MN digestion of chromatin from these lines were identical to control MTA cells. The amount of H1 0 GFP and H1cGFP present in the MN-soluble and -insoluble chromatin fractions obtained from density-arrested cells was quantitated by fluorometry (Fig. 4). After 15 min of digestion with MN, 77% of H1 0 GFP was retained in the MN-resistant chromatin pellet compared with only 34% of H1cGFP.
The Mono-and Dinucleosome Banding Patterns of Chromatin from H1 0 -and H1c-overexpressing Cells Are Different-To carry out a more detailed structural characterization of the chromatin from H1 0 -and H1c-overexpressing cells, we employed the technique of composite NP gels (31). These high resolution agarose-polyacrylamide-glycerol gels can resolve individual mono-, di-, and higher order nucleosomes based on the conformation of these particles, the length of DNA, and the number of histone H1 and HMG molecules associated with them. When MN-soluble S2 chromatin was run on these gels, five bands (M I through M V) corresponding to mononucleosomes and three bands (D 1 through D 3 ) corresponding to dinucleosomes were resolved (Fig. 5A). This banding pattern com- Total input radioactivity was measured by lysing undigested nuclei with 0.5% SDS prior to liquid scintillation counting. Values are the average from three independent experiments carried out simultaneously. Overexpression levels of H1 0 and H1c in exponential cultures were 69% and 77%, respectively. Overexpression levels of H1 0 and H1c in density-arrested cultures were 74% and 85%, respectively. pares well with the pattern obtained by Garrard et al., and we therefore used their nomenclature (31,39). The mononucleosome pattern of chromatin from cells overexpressing H1c is very similar to that of the control chromatin (compare lanes 1 and 3 of Fig. 5A). Overexpression of H1 0 leads to the appearance of a faster migrating M III fraction of mononucleosomes (Fig. 5A, lane 2, indicated by the arrow). The M III fraction of mononucleosomes reportedly contains 1 molecule of H1 histone per particle (39). The faster migrating M III species is not observed in chromatin obtained from control MTA cells or H1c-overexpressing cells, and hence is unique to the chromatin of H1 0 -overexpressing cells. The difference in the calculated molecular mass (28) and pI of H1 0 and H1c proteins is not sufficient to explain the high mobility M III species. The binding of H1 0 to a mononucleosome may give rise to a compact nucleosome conformation, and this may explain why the unique M III band migrates faster. This may relate to the proposed role of H1 0 in the compaction of chromatin. This increased compaction may very well start at the level of mononucleosomes. We extracted the proteins and confirmed by SDSpolyacrylamide gel electrophoresis that the fast migrating M III band contained predominantly the H1 0 variant (Fig. 5B).
Another difference between the chromatin of H1-overexpressing cells and the control chromatin lies in the bands corresponding to the dinucleosomes (D 1 , D 2 , and D 3 ). D 1 contains one molecule of H1 histone per dinucleosome particle, whereas D 2 contains two and D 3 contains two to three H1 molecules (31,39). The control chromatin from the MTA cell line shows approximately equal amounts of D 1 and D 2 dinucleosomes, and negligible amounts of D 3 (Fig. 5A, lane 1). Overexpression of either H1 0 or H1c variant leads to a shift from the D 1 species to the D 2 and D 3 species, suggesting that the majority of the dinucleosomes in these cells carry two or three H1 histones (Fig. 5A, lanes 2 and 3). Preliminary compositional analyses of these dinucleosome species using SDSpolyacrylamide gel electrophoresis indicate that they contain increasing amounts of H1 histones from D 1 to D 3 , suggesting that more than one H1 molecule can be bound per nucleosome (data not shown).
The NP patterns of chromatin from cells expressing the H1GFP fusion proteins was also analyzed. Fluorescence from GFP allows direct visualization and identification of nucleosome species containing bound H1GFP. Images showing GFP and EtBr fluorescence from the same gel are shown for both H1 0 GFP-and H1cGFP-expressing cell lines (Fig. 6). The major mononucleosome bands containing H1GFP fluorescence probably represent M III-like species, which, due to the additional mass of the GFP moiety (ϳ28 kDa), migrate more slowly on NP gels. Interestingly, this band migrates faster in H1 0 GFP samples than in H1cGFP samples, suggesting a more compact nucleosomal conformation.
Both the N-terminal Tail and the Central Globular Domain of the H1 0 Protein Are Independently Capable of Giving Rise to FIG. 5. NP patterns of mono-and dinucleosomes obtained from the chromatin of cells overexpressing histone H1 0 or H1c. Nuclei were isolated from density-arrested MTA, MTH1 0 , and MTH1c cells following induction with ZnCl 2 , and digested with 1.5 units/ml MN for 15 min at 25°C. Overexpression levels of H1 0 and H1c were 77% and 85% of total chromatin-bound H1 histones, respectively. A, the MNsoluble S2 chromatin fraction from control and H1 variant-overexpressing cells was resolved on composite NP gels as described under "Experimental Procedures," and the NP pattern was visualized by EtBr staining. The unique faster migrating M III mononucleosome species observed upon overexpression of H1 0 is indicated by the arrow. B, the mononucleosome bands from the M I, M II, and M III bands were excised and total histones were extracted as described previously (31). Aliquots were run on 17% polyacrylamide-SDS gels, and histones were visualized by staining with Coomassie Blue.
FIG. 6. Direct visualization and identification of mono-and oligonucleosomes carrying bound H1 histones using GFPtagged H1 0 and H1c on NP gels. Density-arrested MTH1 0 GFP and MTH1cGFP cells were induced with 100 M ZnCl 2 as described under "Experimental Procedures." Overexpression levels of H1 0 GFP and H1cGFP were 19% and 13% of total chromatin-bound H1 histones, respectively. Nuclei were isolated and digested with 1.5 units/ml MN for 15 min at 25°C. MN-soluble S2 chromatin was run on NP gels as described under "Experimental Procedures." The gels scanned on a Molecular Dynamics Storm fluorimaging unit in the blue fluorescence mode to visualize the GFP fluorescence from NP bands carrying bound H1GFP fusion proteins and then stained with EtBr to visualize all the NP bands. Note that bleed-through of GFP fluorescence from bands carrying H1GFP fusions contributes to the fluorescent signal observed in the lanes marked EtBr. The lane labeled MTA contains material from control cells that do not express H1-GFP hybrids.
the Unique High Mobility M III Mononucleosome Species-Previously, we generated a set of domain switch mutants of H1 0 and H1c to identify the domain responsible for the differential effects of these two proteins on gene expression (28). We identified the globular domains of H1 0 and H1c to be responsible for their differential effects on gene expression. We used the same mutants in this study to determine which domains of the H1 0 were responsible for giving rise to the high mobility M III mononucleosome species that is observed on NP gels (Fig.  5A, lane 2). The NP banding patterns of chromatin from cells overexpressing any of the different domain switch mutants clearly show a downward shift in their M III species (Fig. 7,  lanes 3-7), as is observed for chromatin from the MTH1 0 cell line (Fig. 7, lane 2), with the sole exception of MTCC0 (Fig. 7,  lane 8). No significant downward shift is observed in the M III species of the MTA (Fig. 7, lanes 1 and 10) and MT H1c (Fig. 7, lane 9) cell lines. The apparent "supershift" observed in the M III species of chromatin obtained from MTH1C0Cdel (Fig. 7,  lane 3) is likely to be due to a 30-amino acid deletion that is present in the C-terminal tail (which was obtained from H1c) of this hybrid protein. This deletion should result in the formation of a H1C0Cdel carrying mononucleosome with a significantly reduced molecular mass resulting in higher mobility on NP gels. Nevertheless, this deletion does not affect our results because the full-length H1c protein does not give rise to the high mobility M III species (Fig. 7, compare M III species in lanes 9 and 2). Further, with the exception of H1C0Cdel, there is no correlation between the mobility of domain switch mutants on NP gels and differences in their molecular mass (28). In fact, the domain switch mutant with the highest molecular mass, H1C00, clearly shows the downward shift of its M III species (Fig. 7, lane 5), indicating that a more compact mononucleosome conformation probably results in the high mobility M III species. Our data indicate that all the domain switch constructs carrying either the short N-terminal tail or the central globular domain of the H1 0 protein alone, or both these domains together, are capable of giving rise to the higher mobility M III species. The C-terminal tail of H1 0 does not appear to have this property, as indicated by the absence of any downward shift in the M III species obtained from the MTCC0 cell line (Fig. 7, lane 8). DISCUSSION The role played by linker histones in regulating chromatin structure and gene expression is controversial (4, 40 -42). Early studies, with a few notable exceptions (24,25,43), were unable to show a clear and direct correlation between individual H1 variants, chromatin structure and chromatin function. We developed a system for overexpressing individual H1 genes in homologous mouse cells, thereby perturbing the normal stoichiometry of variants (26,27). Initially, we showed that overexpression of H1 0 and H1c variants lead to different effects on gene expression and cell cycle progression, the former having an inhibitory effect, and the latter having either a stimulatory effect or no effect at all (27). We also demonstrated that the differential effects of H1 0 and H1c on gene expression are due to differences in the central globular domains of these two proteins (28).
In this study we demonstrated alterations in chromatin structure upon the overexpression of H1 0 and H1c. Some of the alterations were variant-specific, while others occurred upon overproduction of either variant. The latter class includes an increase in nucleosome spacing in exponentially dividing cells, increased MN resistance in density-arrested cells, and increased levels of slow migrating dinucleosome species in NP gels. These effects may be due to increased levels of total H1 per nucleosome associated with overproduction of either variant. Several reports have argued that two or more H1 histones can bind per nucleosome (5,6,44,45). Whether these "extra" H1 histones bind to these nucleosomes with the same affinity as the first H1 is not known. Also, it is not clear whether the extra H1 histones associate with the core nucleosome in the same manner as the initial H1, or mainly with the linker DNA with an affinity equal to the its affinity for naked DNA (6). The binding of more than one H1 histone per nucleosome is likely to play important roles in gene regulation, especially if this occurs in only a subset of the chromatin.
We also observed alterations in chromatin structure that were associated specifically with the overexpression of the H1 0 variant. These include increased MN-resistance of chromatin from exponentially growing cells, preferential distribution of H1 0 in MN-resistant chromatin fractions, and the presence of a unique faster migrating mononucleosome species in NP gels. In light of our earlier studies involving the effects of H1 variant overexpression on gene expression and cell cycle progression (27), we propose a simple, coherent model for the differential effects of H1 0 and H1c variants.
We suggest that histone H1 0 functions as a "replacement variant" (46) in vivo and replaces other H1 variants in G 0arrested cells. H1 0 may then result in the progressive increase in condensed chromatin in quiescent cells. This is supported by the observation that H1 0 accumulates naturally in non-dividing, terminally differentiated cells, which exhibit compact chromatin, along with reduced levels of transcription and no replication (17). 2 Overexpression of H1 0 during any phase of the cell cycle simply mimics the natural accumulation of this variant in growth-arrested cells, leading to the formation of condensed chromatin. Due to the inaccessibility of binding sites for both FIG. 7. NP patterns of mono-and dinucleosomes obtained from the chromatin of cells overexpressing histone H1 0 , H1c, and their "domain switch" mutants. Density-arrested cultures of the indicated cell lines were induced with 100 M ZnCl 2 . Hybrids are designated by the variant from which the N-terminal, globular, and C-terminal domains were derived, i.e. 00C is composed of the N-terminal and globular domains of H1 0 and the C-terminal domain of H1c.
MTH1C0Cdel carries a small deletion in the C-terminal tail. Nuclei were prepared, and the MN-soluble chromatin was resolved on composite NP gels as described under "Experimental Procedures." The NP pattern was visualized by EtBr staining. Overexpression levels were as follows: H1 0 ϭ 78%; H1C0Cdel ϭ 80%; H100C ϭ 77%; H1C00 ϭ 74%; H10C0 ϭ 81%; H10CC ϭ 78%; H1CC0 ϭ 65%; H1c ϭ 86%. the basal transcriptional machinery and specific transcription factors within this condensed chromatin, there is a general repression of gene expression, as observed in our earlier studies (27). Compensatory mechanisms probably help maintain a minimal level of gene expression, to assure survival of the cell. The natural accumulation of H1 0 in quiescent cells is mediated in part by the replication-independent mode of expression of the endogenous gene (17). However, properties of the H1 0 protein, most notably the preferential association with MN-resistant, presumably condensed (47), chromatin observed in this study may also contribute to the accumulation of this variant and its proposed role in stabilizing the quiescent state.
As shown in our earlier studies, histone H1c is functionally opposed to H1 0 (27,28). Histone H1c appears to be deficient in condensed chromatin when compared with H1 0 . Overexpression of H1c to high levels in exponential cells does not lead to a very high degree of chromatin compaction as assayed by its sensitivity to MN cleavage (Fig. 3A). H1c therefore appears to be associated in vivo with chromatin regions that have a relatively "open" architecture (37,38), which might facilitate transcription and replication by allowing easy access to the factors involved in these processes. Overexpression of H1c might accentuate the normal functions of H1c by making larger quantities of this protein available for binding to regions not normally occupied by this variant. This would "open up" extensive regions of the chromatin, making it readily accessible for the binding of trans-activating factors and leading to the enhanced expression of some genes, as demonstrated in our previous studies (27).
The identification that both the N-terminal tail and the central globular domain of the H1 0 protein can independently give rise to the unique high mobility M III mononucleosome species is interesting (Fig. 7). The involvement of the globular domain of H1 0 in the formation of the high mobility M III species was to be expected, based on our earlier results on the effects of this domain on gene expression (28). That the short N-terminal tail of H1 0 also affects mononucleosome mobility, possibly by affecting its conformation, implies that the tails of the H1 proteins may play important roles in some of their functions.