The histone H1 family is the most divergent subgroup of the highly conserved chromosomal histone proteins (1, 2). H1 histones are bound to the outer surface of nucleosomes near the entry/exit point of the linker DNA and are also known as linker histones (2–4). They have been implicated to participate in determining the higher-order folding states of chromatin and thus in control of gene activity (5).
In higher eukaryotes, H1 is a heterogeneous family and at least nine subtypes have been found in mammals (6–10). Although it seems that each subtype may have a distinct function, the specific role played by linker histones is still enigmatic. Moreover, individual subtypes or some groups of subtypes are not essential for viability in some systems studied (11–13), but on the other hand it seems now clear that linker histones in general are essential for proper development of higher organisms (14).
In general, H1 histones consist of a conserved globular domain and variable N- and C-tails responsible for the H1 family heterogeneity (6, 15). The tails, highly enriched in positively charged lysine and arginine residues, stabilize chromatin folding by shielding the negative charges on the DNA backbone (16, 17) and also by adopting specific secondary structure upon interaction with DNA (18, 19). They are targets for several postsynthetic modifications, particularly phosphorylation. H1 phosphorylation increases during cell cycle progression and has been observed in a number of different organisms and cell types (20–22). It occurs at specific serine and threonine residues located in the tail domains (18, 23). This modification is believed to alter linker histone interaction with DNA and thus to modulate chromatin structure (24, 25). High level of H1 phosphorylation was observed during mitosis and suggested that it may play an active role in mitotic chromosome condensation (26). In some cell systems, however, H1 phosphorylation was uncoupled from mitosis and highly condensed chromatin was enriched in unphosphorylated H1 (25).
A specific subtype of linker histones, histone H5, has been found to accumulate in nucleated avian erythrocytes (27, 28). H5 is a counterpart of mammalian histone H1° and both of them are considered to be differentiation-specific H1 subvariants. In immature cells, H5 is phosphorylated and then it becomes dephosphorylated during erythrocyte maturation (29). H5 shows strong preference for higher-order chromatin structures (30–32) and is more tightly bound to DNA or chromatin compared with other H1 subvariants (33–36), most probably due to the higher Arg/Lys ratio in its tails.
In principle, H1 phosphorylation should neutralize the positive charges and weaken the binding to DNA, resulting in more open, decondensed, chromatin structure (24, 25, 37–42). However, differences between N- and C-tail domains in the binding of phosphorylated H1 histones to DNA have been described. For instance, Hill et al. (24) found that the phosphorylation of isolated N-terminal domains of sea urchin sperm-specific linker histones abolished their binding to DNA. In contrast, phosphorylation in the C-terminus had little effect on its overall affinity for DNA. Talasz et al. (43) reported that linker histones with different levels of phosphorylation, isolated from cells in different phases of the cell cycle (22), did not show any differences in their binding to mononucleosomes in vitro.
Most of the previous in vitro studies have been carried out on isolated chromatin fragments, mononucleosomes, or naked DNA. However, it seems clear that linker histone affinity may be substrate-dependent, which indicates that in vitro binding studies should preferably be performed on chromatin templates that are as intact as possible. Linker histones are bound to chromatin mainly through ionic interactions and they can be selectively extracted using salt concentrations in the range of 0.3 to 0.7 M NaCl. This property of linker histones was used to develop a method to investigate H1-chromatin interactions in situ using the DNA-binding fluorochrome DAPI as an indirect probe (44–46). Furthermore, our method was extended to study the affinity of a particular H1 subfraction reconstituted into nuclei after depletion of the endogenous linker histones (47). We have now applied this method to study how phosphorylation affects histone H5 affinity for chromatin in situ in normal human H1-depleted fibroblasts.
Materials and Methods
DAPI, digitonin, and Trizma (Tris base) were purchased from Sigma. Recombinant human cdk5/p35 active kinase (lot no. 22420AU, specific activity 1,830 U/mg) was purchased from Upstate (Lake Placid, NY). All other chemicals were purchased from Fluka (Buchs, Switzerland) if not otherwise indicated.
Preparation of H5 Histone
Chicken blood was obtained from a poultry slaughter house under the regulations of the Bulgarian Veterinary Medical Activity Law. Linker histones were extracted with 5% HClO4 and histone H5 was purified by gel exclusion chromatography on a Bio Gel P100 column as detailed by Srebreva and Zlatanova (48).
In Vitro Phosphorylation of H5 Histones
Four milligrams of purified histone H5 was phosphorylated in vitro by cdk5/p35 kinase, according to the recommendations of the manufacturer, with the following modifications: final H5 concentration, 2 mg/ml; total amount of kinase, 4 μg in 2 ml reaction solution; no BSA was added. The phosphorylation was performed at 30°C. The extent of phosphorylation was monitored by capillary electrophoresis after 2, 4, and 6 hours. After 7 hours, the reaction was stopped by precipitation of H5 proteins with TCA (final concentration 20%). The mixture was left on ice for 1 h, centrifuged, washed, and lyophilized as detailed previously (49). The same relative conditions (enzyme/substrate ratio 1:1,000) were used to phosphorylate a small batch of highly pure recombinant H5 (kindly provided by Professor Jean Thomas, University of Cambridge), which was used in control experiments to determine the N- and C-terminal phosphorylation pattern by mass spectrometry. Such control experiments were also performed using a tenfold increase in the enzyme/substrate ratio.
High performance capillary electrophoresis (HPCE) was performed on a Beckman system P/ACE 2100. Data collection and postrun data analyses were carried out using P/ACE and System Gold software (Beckman Instruments). The capillary cartridge used was fitted with 75 μm internal diameter fused silica of 67 cm total length (60 cm to the detector). In all experiments an untreated capillary was used. Protein samples were injected by pressure and detection was performed by measuring UV absorption at 200 nm. Separation of H5 was performed as described (50–53). The linker histones were analyzed in 0.1 M sodium phosphate buffer (pH = 2.0) containing 0.02% HPMC. All runs were carried out at a constant voltage (12 kV) and at a capillary temperature of 25°C.
Enzymatic Cleavage and Mass Spectrometry
In vitro phosphorylated H5 was digested with α-chymotrypsin [EC 126.96.36.199] (Sigma type I-S, 1/150 w/w) in 100 mM sodium acetate buffer (pH = 5.0) for 40 min at room temperature. The peptides obtained were separated using a Nucleosil 300-5 C18 column (150 mm × 4 mm I.D.; 5 μm particle pore size; end-capped; Macherey-Nagel, Düren, Germany). Samples of ∼50 μg were injected onto the column. Chromatography was performed within 50 min at a constant flow of 0.5 ml/min with a two-step acetonitrile gradient starting at solvent A—solvent B (87:13) (solvent A: water containing 0.1% TFA; solvent B: 85% acetonitrile and 0.1% TFA). The concentration of solvent B was increased linearly from 13% to 20% during 25 min and from 20% to 50% during 28 min. Fractions obtained in this way were collected and, after adding 20 μl 2-mercaptoethanol (0.2 M), lyophilized and stored at −20°C.
Determination of the molecular masses of the N- and C-terminal fragments of H5 obtained by RP-HPLC was carried out by electrospray ion-mass-spectrometry (ESI-MS) technique using a Finnigan LCQ ion trap instrument (San Jose, CA). Samples (5–10 μg) were dissolved in 50% aqueous methanol containing 0.1% formic acid, and injected into ion source.
Determination of the phosphorylation sites of the N-and C-terminal fragments of H5 was carried out by further digestion with trypsin [188.8.131.52] (Roche, sequencing grade, 1/50 w/w) in 5 mM NH4HCO3 (pH = 8.5) for 1 h at 37°C followed by LC-ESI-MS as described previously (54).
Human diploid foreskin fibroblasts (AG 1523, passages 14–19) were cultured in Earle's Minimal Essential Medium supplemented with 10% fetal bovine serum, penicillin (50 IU/ml), streptomycin (50 μg/ml), and L-glutamine (2 mM). The cells were kept at 37°C in a 5% CO2 atmosphere and serially passaged at a 1:2 split ratio every 4th day. Before the experiments, the cells were plated on coverslips in 12-well plates and used when reaching confluence at a cell density of about 2 × 105 cells/coverslip.
The fibroblasts were rinsed in KRG buffer (120 mM NaCl; 4.9 mM KCl; 1.2 mM MgSO4 × 7H2O; 1.7 mM KH2PO4; 8.3 mM Na2HPO4 × 2H2O; 10 mM glucose) and permeabilized with 40 μg/ml digitonin in Tris buffered saline (TBS; 10 mM Tris-HCl, 150 mM NaCl, pH 7.4) containing 0.5 mM MgCl2 for 10 min. The cells were then extracted with 0.7 M NaCl in TBS for 5 min to remove all native linker histones. The salt extraction buffer contained also 1 M sucrose to prevent cellular disruption. Thereafter, the cells were washed in TBS and reconstituted by incubation in either phosphorylated or nonphosphorylated H5 histones (20 μg/ml, dissolved in TBS) for 1 h. The reconstitution solution was supplemented with the protease inhibitors AEBSF (69 μg/ml; Calbiochem, San Diego, CA), pepstatin (2 μg/ml; Boehringer Mannheim, Mannheim, Germany), and leupeptin (5 μg/ml; Boehringer Mannheim). The reconstituted cells were extracted for 5 min with different concentrations of NaCl ranging from 0.15 to 0.7 M and then fixed in 4% paraformaldehyde for 2 days. All preparations were performed on ice.
DAPI Staining and Image Cytofluorometry
The fixed cells were stained with 50 nM DAPI and the fluorescence intensity (FI) was measured by image cytofluorometry as detailed previously (47). The number of G1 cells per frame was about 100, and their mean integrated fluorescence was used for the calculations of H5 affinity. Four frames on each coverslip were analyzed, and each experiment included data from duplicate glasses. Thus, about 800 G1 cells were measured for each data point within a series of measurements.
The analysis of salt extraction curves was performed as described previously (44, 45) using a least-squares curve fitting to a linker histone binding equation (33). The salt extraction curves were normalized using an average of the three highest FIs after extraction as 100%. At this point, all linker histones were considered to be extracted from chromatin. The average of the three lowest FIs then represented the level at which all reconstituted linker histones were considered to be bound to chromatin. The NaCl concentration required to induce a 50% increase in FI from this level was then calculated from the fitted equation and used as a measure of average apparent linker histone affinity for chromatin in situ. The results are, unless otherwise indicated, expressed as mean ± S.D. The statistical significance of differences between results was analyzed using unpaired Student's t-test.
Results and Discussion
The electropherogram obtained from purified H5 histones showed N-terminally nonacetylated and acetylated H5 peaks (Fig. 1A) as expected (55). After H5 in vitro phosphorylation by cdk5/p35, a number of slower migrating peaks representing phosphorylated forms of both nonacetylated and acetylated forms of H5 appeared (Fig. 1B), indicating that almost all H5 molecules contained one or more phosphate groups.
Histone H5 was suggested to become phosphorylated in vivo at four major sites in a cell cycle-dependent manner and two of those sites were found to be located in Ser-Pro-X-Basic motifs in the highly basic C-terminal domain (CTD) (29, 56). The H5 phosphorylation in the present experiments was carried out using cdk5/p35, which is predicted to phosphorylate five Ser-residues in histone H5 located in SPXK motifs in the C terminus. To check the enzyme specificity, a small batch of recombinant H5 was phosphorylated in vitro using the same phosphorylation conditions. After chymotrypsin cleavage and subsequent analysis of the fragments using tandem mass spectrometry four phosphorylation sites in the C terminus were detected (Fig. 2A). A large majority of these molecules were mono- or diphosphorylated. We also found that the N-terminal fragment was a mixture of nonphosphorylated and mono-phosphorylated molecules (Fig. 2B), indicating that the in vitro phosphorylated H5 contained about 10% H5 where the N terminus was monophosphorylated in combination with the C-terminal phosphorylation pattern. To further check the specificity of the enzyme, we performed in vitro phosphorylation of recombinant H5 using a tenfold higher enzyme to substrate ratio. After chymotrypsin cleavage and mass spectrometric analyses, we found that the C-terminal phosphorylation pattern was clearly shifted to di-, tri-, and tetraphosphorylated forms and that also pentaphosphorylated molecules were present (Fig. 2C). The N- and C-terminal fragments were further isolated by RP-HPLC and digested using trypsin. The predicted SPXK phosphorylation sites in the CTD were then verified by MS analysis (data not shown). Concomitantly, the N-terminal phosphorylation pattern showed only a small increase in the number of monophosphorylated molecules (Fig. 2D), indicating a substantially lower specificity for the enzyme to phosphorylate this site in the N terminus. This may be explained by the absence of SPXK motifs in the N terminus of H5, and the present phosphorylation site in the N terminus was also proved to be Ser7, which is the only nonmotif site in H5 where serine is followed by a proline. In conclusion, the phosphorylation conditions used in our reconstitution experiments resulted in a balanced mixture of phosphorylated H5 molecules (Fig. 1B) in accordance with a typical average phosphorylation pattern present in exponentially growing cells containing multiple H1 subtypes (57, 58).
Permeabilized fibroblasts were treated with 0.7 M NaCl to extract endogenous linker histones. Thereafter, they were reconstituted by incubation with either nonphosphorylated or in vitro phosphorylated H5 histones at physiological ionic strength. The salt extraction of linker histones leads to an increase in FI which is proportional to the sum of DAPI binding sites that become available when H1 is detached from chromatin (44–46). We have recently applied this method to study linker histone–chromatin interactions in normal human fibroblast nuclei after depletion of the endogenous linker histones and reconstitution with H1 subfractions (47). The presence of exogenous protein in nuclei after reconstitution was verified by Alexa-labeled H1. The exogenous linker histones showed a slightly reduced affinity for chromatin compared with the native linker histones (47), indicating that the reconstituted proteins did not bind exactly in the same manner as in the native state. However, this system allows linker histone–chromatin interactions to be studied in situ using a chromatin template that structurally is as close as possible to the intact cell nucleus.
Histone H5 reconstituted nuclei stained with DAPI were similar in appearance to native fibroblast nuclei. Background cytoplasmic fluorescence was negligible in accordance with previous results (47). The reconstituted cells were then extracted with NaCl concentrations in the range of 0.15 to 0.7 M. When the relative FIs were plotted against the NaCl concentration (Fig. 3), they showed a close fit to the linker histone binding equation described by Kumar and Walker (33). The salt concentration needed to induce a half-maximal increase in FI in nuclei reconstituted with nonphosphorylated H5 (Fig. 3A) was 0.52 ± 0.01 M (n = 5), which was slightly, but significantly (P < 0.01), lower than the corresponding value from native H5 in chicken erythrocytes (0.55 ± 0.02 M) reported previously (35). This reduction in affinity for the exogenously applied H5 is in accordance with our previous results with other H1 subfractions used for reconstitution (47). In contrast, the corresponding salt concentration needed to induce a 50% increase in FI in nuclei reconstituted with phosphorylated H5 (Fig. 3B) was substantially lower, 0.41 ± 0.02 M (n = 4, P < 0.0001). Examples of fluorescence images and their corresponding fluorescence intensity data are presented in Figure 4.
Thus, we observed that phosphorylated H5 molecules had a considerably reduced affinity for chromatin in situ compared with nonphosphorylated ones. This finding is consistent with the proposal that phosphorylation of H1 tails loosens H1-DNA interactions, leading to relaxed chromatin structure (25). However, a number of different phosphorylation sites have been described and they may have differential effects on the DNA-binding properties of linker histones. Data in this respect are controversial. The discrepancies could be due to differences in linker histone subtypes, the distribution of phosphorylation sites within them, the number of phosphate groups incorporated, and the template systems used (43, 59). Phosphorylation has thus been reported to reduce DNA-binding of histone H5 (60). However, these authors used a cAMP-dependent kinase isolated from calf thymus to introduce phosphate groups in H5. This enzyme probably phosphorylated H5 at multiple sites, mainly in the globular domain, as shown previously using a cAMP-dependent kinase isolated from pig brain (61). In experiments with sea urchin sperm-specific linker histones, Hill et al. (24) found that phosphorylation of six sites in the N-terminal domain almost abolished its binding to DNA. Moreover, SPKK motifs in sea urchin sperm H1 are localized in N-terminal domain (62) and phosphorylation of peptides, containing SPKK sequences was shown to weaken their binding to DNA (37). In contrast, phosphorylation of three residues in the isolated CTD from sea urchin sperm H1 had a negligible effect on binding of this domain to DNA (24), probably because of the extended lysine rich C-terminal in this species. Interestingly, the same pattern of phosphorylation in the whole H1 molecule (nine sites) did not significantly change its binding to DNA, whereas it showed a clearly reduced affinity for chromatin in solution (24). This is probably explained by the heterogeneous pattern of phosphorylation sites in the extended CTD of sea urchin sperm H1, the most distal part containing all phosphorylation sites while the major remaining part of the highly charged CTD, lacking phosphorylation sites, determines the binding to DNA of the whole CTD as well as the entire molecule. Moreover, Talasz et al. (43) found that linker histones isolated from cells in G1, S, or mitosis, containing one to five phosphate groups (22), did not show any differences in their binding to mononucleosomes in vitro. These authors thus observed very high binding affinity to a mononucleosome, but on the other hand a low chromatin aggregation capability, in the case of highly phosphorylated H1 histones. However, since mononucleosomes lack linker DNA, the affinity of H1 is probably not significantly affected by H1 phosphorylation in such preparations.
A remaining paradox concerns the relation between H1 affinity for chromatin, H1 phosphorylation, and chromatin condensation (25). The various H1 subtypes were shown to have different inherent affinities for chromatin and different chromatin condensing capacities, and it was concluded that the CTD was the main determinant of these properties (63). Interestingly, partial phosphorylation of the H1° CTD did not cause neither a substantial DNA condensation nor a large reduction in affinity for naked DNA, whereas the fully phosphorylated CTD showed increased DNA condensation and reduced affinity for DNA (64). However, our findings, using a chromatin template, clearly show that phosphorylation of SPXK-motifs in the CTD is a strong modulator of H5 binding to chromatin. Along this line, recent data challenged the commonly accepted idea that the binding, and function, of the linker histone CTD is mainly regulated by charge neutralization. For example, C-terminal phosphorylation at two sites directly modulated the affinity of histone H1.1 for chromatin in vivo without influencing the charge distribution or the overall net charge of the tail domain (65). Moreover, histone H1 binds dynamically to chromatin (66, 67) and phosphorylation of the tails facilitated its mobility (41, 42, 68).
In conclusion, the present results verify that linker histone affinity for chromatin in situ can be measured using DAPI as a fluorescent probe. Linker histone-depleted normal fibroblast nuclei represent relatively intact chromatin templates well suited for reconstitution experiments. Phosphorylation by cdk5/p35 resulted in a substantial reduction of H5 affinity for chromatin in this template. In line with in vivo results using green fluorescent protein-tagged H1.1 (65), our results indicate that phosphorylation of SPXK-motifs in the CTD is a strong modulator of linker histone binding properties, which may be responsible for the dynamic regulation of chromatin structure. Our data thus supply further evidence for the importance of the CTD in the determination of linker histone biological functions.
The authors thank A. Devich, Innsbruck Medical University, and A. Lönn, Linköping University, for excellent technical assistance. They also thank Professor Jean Thomas, University of Cambridge, for providing recombinant H5 protein.