|Home | About | Journals | Submit | Contact Us | Français|
We previously reported that reconstituted nucleosomes undergo sequence-dependent translational re-positioning upon removal of the core histone tail domains in physiological conditions, indicating that the tails influence choice of position. We report here that removal of the core histone tail domains increases the exposure of the DNA backbone in nucleosomes to hydroxyl radicals, a non-biased chemical cleavage reagent, indicative of an increase in motility of the DNA on the histone surface. Moreover, we demonstrate that the divalent cations Mg2+ and Ca2+ can replace the role of the tail domains with regard to stabilization of histone-DNA interactions within the nucleosome core and restrict re-positioning of nucleosomes upon tail removal. However, when nucleosomes were incubated with Mg2+ after tail removal, the original distribution of translational positions was not re-attained, indicating that divalent cations increase the energy barrier between translational positions rather than altering the free energy differences between positions. Interestingly, other divalent cations such as Zn2+, Fe2+, Co2+, and Mn2+ had little or no affect on the stability of histone-DNA interactions within tailless nucleosomes. These results support the idea that specific binding sites for Mg2+ and Ca2+ ions exist within the nucleosome and play a critical role in nucleosome stability that is partially redundant with the core histone tail domains.
In the eukaryotic cell, an entire genome’s worth of DNA is condensed by the binding of counterions and basic histone proteins into a highly compacted structure known as chromatin. The initial stage of compaction involves the wrapping of 147 bp of DNA 1¼ turns around an octamer of the core histone proteins to form the nucleosome core. The histone octamer is composed of two copies each of H2A, H2B, H3, and H4, which consist of an N-terminal tail domain that constitutes 25–30% of the mass of each of the core histones and a C-terminal histone-fold domain that participates in extensive protein-protein interactions and forms a ramp of positively charged residues onto which the DNA is wrapped (van Holde, 1989) (1). Linker histones (H1s) bind to the nucleosome core and the linker DNA between cores and promote formation of chromatin secondary structures such as the 30nm chromatin fiber and higher order structures in the ionic environment of the nucleus (2–4).
The highly condensed nature of native chromatin greatly restricts access to the DNA. However, both energy-intensive and passive processes facilitate access to DNA within chromatin, including ATP-dependent remodeling activities, which alter nucleosome structure and/or move or mobilize nucleosomes to allow increased exposure of internal DNA binding sites, incorporation of histone variants, and posttranslational modifications of the core histone tail domains, which signal binding of factors that remodel chromatin (5). Posttranslational modifications also can directly alter stability of chromatin structures. For example, acetylation of core histone tails increases access of specific trans-acting factors such as Gal4 and TFIIIA to nucleosomal DNA (6, 7) and can lead to destabilization of histone-DNA interactions in nucleosomes (8, 9) and reduce stability of secondary and tertiary chromatin structures (10). In addition, fully wrapped and inaccessible nucleosomes are in rapid equilibrium with states in which the DNA is accessible to DNA binding factors both in vitro and in vivo (11, 12).
Genome-wide nucleosome mapping has demonstrated that nucleosomes are precisely positioned in the vicinity of the start site of transcription of most promoters in numerous organisms and bracket a ‘nucleosome free region’ (13). This positioning allows access of transacting factors and the transcription machinery to promoter elements, and may play a role in the generation of paused polymerases (14). Moreover, precisely positioned nucleosomes also function to block access of trans-acting factors and contribute to gene repression in specific instances (15, 16). Importantly, evidence indicates that DNA sequence is a primary contributor to nucleosome positioning in vivo (17). Given the import of sequence-dependent nucleosome positioning in the regulation of gene expression, (14) it is therefore important to understand the molecular determinants of nucleosome positioning.
Early evidence from nuclease mapping techniques suggested that the core histone tail domains do not influence choice of nucleosome position on specific DNA fragments (18, 19). However, a later study showed that nucleosomes reconstituted with a DNA fragment containing a heat shock protein gene adopted altered translational positions upon reconstitution with histones lacking the core histone tails, especially the H2B tails (20). In addition, more recent data employing high-resolution protein-DNA crosslinking demonstrated that core histone tails do indeed contribute to selection of nucleosome positions. Specifically, it was demonstrated that nucleosomes can spontaneously move to new translational positions on selected DNA fragments upon removal of the tail domains (21). Likewise, an examination of the interferon-γ gene showed that the movement of nucleosomes required for transcription induced by ATP-dependent nucleosome remodeling complexes did not occur if acetylation of the core histone tails was blocked (22), suggesting that core histone tails play an important role in the mobility of nucleosomes in vivo. These results suggest the core histone tail domains can influence both nucleosome positioning and mobility, likely by interacting with nucleosomal or linker DNA (23).
The core histone tail domains contribute to the thermal stability of the nucleosome core (24). The stability and mobility of nucleosomes is also affected by specific metal ions, which bind in chromatin in multiple modes. Concentrations of Mg2+ and/or Ca2+ typically found in transcription extracts are sufficient to prevent nucleosome mobilization in the absence of ATP-dependent remodeling activities (11). Moreover, Ca2+ can stabilize the nucleosome against the increased invasion by micrococcal nuclease that results upon removal of the core histone tail domains (25). These effects may be due to specific binding sites observed by X-ray crystallography for cations such as Mg2+ and Mn2+ within the nucleosome core particle (26, 27). Here we report that removal of the tail domains from pre-formed nucleosomes results in a global enhancement in the accessibility of the DNA backbone to cleavage by hydroxyl radicals, indicating a previously unrecognized and substantial increase in the motility of the DNA on the histone surface. While divalent cations did not appear to affect stability or translational positions of nucleosomes containing intact histones, both Ca2+ and Mg2+ stabilized histone-DNA interactions and inhibited translational re-positioning of nucleosomes upon tail removal. In contrast, the cations Na+ Zn2+, Fe2+, Co2+, and Mn2+ had little effect on the stability of histone-DNA interactions within tailless nucleosomes, suggesting that overall these ions bind in nucleosomes in a fundamentally different manner. These results highlight the specific biological roles of Ca2+ and/or Mg2+ in nucleosome structure.
A 215 bp DNA fragment containing Xenopus borealis somatic-type 5S RNA gene sequences from −78 to +137, was obtained from the plasmid pXP-10 as described (21). The DNA was radiolabeled with T4 polynucleotide kinase (New England Biolabs (NEB)) and [γ-32P]ATP after cleavage with EcoRI (NEB), then the 215 bp fragment liberated by cleavage with DdeI (NEB). The radiolabeled DNA fragment was PAGE purified by standard procedures (21). A 182bp DNA fragment containing (TATAAACGCC)12 repeats was obtained from the plasmid pHCn41 (28) as described (21).
Core histone proteins H3 and H4 were obtained from chicken erythrocytes by standard methods (29). H2A, H2B and the mutant H2A-A45C were expressed in E. coli, purified and modified with the bi-functional cross-linking agent, 4-azidophenacyl bromide (APB), for 1 hour at RT in the dark as described (21). Nucleosomes were reconstituted by standard salt-dialysis (29).
Nucleosomes containing native H2A or H2A-A45C-APB were reconstituted by salt-dialysis in large scale (5 ml, 200 μg each of DNA and histone octamer), then concentrated to 0.5ml using a microtube filter apparatus (Millipore YM-50). The concentrated nucleosomes were incubated with 0.04 ml trypsin-agarose beads (Sigma) for 15min at room temperature, and centrifuged to remove the beads (30). In our hands, no further degradation of proteins was observed after bead removal while addition of trypsin inhibitors was not necessary and sometimes led to further degradation. The extent of trypsinization of the core histone proteins was analyzed by 18% SDS-PAGE.
Nucleosomes were reconstituted on the radiolabeled 5S DNA fragment in a large scale, concentrated to 1 ml as described above, then incubated with 200 units of ml BamHI (New England Biolabs) for 15 min at 37°C (31). The Bam HI-resistant nucleosomes were purified by sucrose gradient and the buffer exchanged to 10 mM Tris.Cl (pH 8.0) in a final volume of 0.5 ml using the filtration units (30). Translational positions were analyzed by electrophoresis on native 5% polyacrylamide gels (20mM HEPES, pH 7.5), with at 106V for 2 hours. A portion of the sample was treated to remove the core histone tail domains as described (30).
Intact or tailless nucleosomes (0.2 pmol; 5 in 50 μl final volume in 10 mM Tris, pH 8.0, 0.1 mM EDTA) were incubated with the salts at concentrations as indicated in the figure legends and 10 min at 25°C, then probed with hydroxyl radical by adding 5 μl of each of the three reagents (1mM Fe/2mM EDTA, 10mM sodium ascorbate, 0.4% H2O2) for 3 minutes as described (32). The reaction was stopped by the addition of glycerol to 5% and nucleosomes isolated in 0.7% agarose nucleoprotein gels (33). The wet gels were exposed to X-ray film and radiolabeled DNA from bands in the gels was purified, denatured and the separated on denaturing 6% sequencing gels. The gels were dried and cleavage patterns analyzed by phosphoimagery (Molecular Dynamics).
Gradient-purified nucleosomes containing H2A-A45C-APB 5S were incubated in the presence or absence of 2 mM Mg2+ for 10 min at 25°C then treated with trypsin agarose beads to remove the c ore histone tail as above. The samples were then irradiated at 365 nm for 30 seconds as described (21), separated on 0.7% agarose nucleoprotein gels, the wet gels exposed to X-ray film, and DNA purified from the nucleosome band. Cross-linked complexes were separated from uncross-linked DNA by electrophoresis on 6% SDS polyacrylamide gels (21). DNA was isolated from the polyacrylamide gel, treated with NaOH to effect base elimination and strand cleavage at crosslink sites, then products analyzed by sequencing gel electrophoresis as described (21).
We previously reported that the core histone tail domains influence choice of nucleosome translational positions on DNA fragments. Specifically, we found that removal of the tail domains from nucleosomes results a redistribution of translational positions. Repositioning occurred for nucleosomes assembled with two different DNA sequences that exhibit moderate affinity for histones but did not occur with a high-affinity sequence (21). Others have shown that removal of the tail domains results in nucleosomes more susceptible to digestion by micrococcal nuclease in low concentrations of the divalent cation cofactor (0.1 mM Ca2+), indicative of a weakening of histone-DNA contacts at the periphery of the nucleosome core (25). However, the effect of tail domain removal is not observed when digestions were carried out in 1 mM Ca2+ indicating that this ion can stabilize histone-DNA interactions at the edge of the core in the absence of the tail domains (25). Moreover, the divalent cation Mg2+ restricts thermal repositioning of nucleosomes containing intact histones (11). Thus we wished to determine whether divalent cations affected nucleosome repositioning upon tail removal in low-salt solutions and the extent to which the tails and counterions contribute to the stability of histone-DNA interactions throughout the nucleosome core regions.
To this end, we first examined histone-DNA interactions by hydroxyl radical footprinting in nucleosomes before and after tail removal and the effect of Mg2+ on these patterns. Nucleosomes reconstituted with a DNA fragment containing the Xenopus 5S nucleosome-positioning element adopt several translational positions distinguishable on a 4.5% polyacrylamide gel (Fig. 1, A and B). As shown previously, ~75% of the nucleosomes adopt two closely related translational positions with the center of dyad symmetry near the 5S RNA gene transcription start site, while ~25% have a translational position further downstream (Fig. 2B) (21, 34). We selected for the major translational positions by Bam HI digestion (Fig. 1B, lanes 2 and 3) then purified the nucleosomes on sucrose gradients (31). A portion of the selected 5S nucleosomes was treated with trypsin-linked agarose beads to remove core histone tails and proper proteolysis confirmed by SDS-PAGE (Fig. 1C).
The intact or tailless 5S nucleosomes were incubated without or with 2 mM Mg2+ then treated with hydroxyl radicals and the cleavage patterns analyzed. Nucleosomes reconstituted with intact histones exhibited a classic sinusoidal pattern of protection in standard buffer lacking Mg2+ (Fig. 2, lane 4). Interestingly, the hydroxyl radical cleavage pattern of these nucleosomes was not altered when 2 mM Mg2+ was added before the cleavage reaction; nucleotides were similarly protected against hydroxyl radical cleavage in the presence or absence of the cation (Fig. 2, lanes 4 and 5, see also Fig. 3). Likewise, the cleavage pattern of naked DNA was unaffected by the presence of Mg2+ as shown previously (32) (results not shown). Also as previously reported (21), removal of the core histone tail domains by limited trypsin proteolysis results in a shift in the position of the nucleosome approximately 20 nt upstream and is apparent in the hydroxyl radical cleavage pattern (Fig. 2, compare lanes 4 and 6). This movement of the nucleosome was still evident when Mg2+ was added after removal of core histone tail domains but before the hydroxyl radical digestion (Fig. 2, lane 7). Moreover, removal of the core histone tail domains resulted in a striking alteration in the hydroxyl radical cleavage pattern throughout the region occupied by the nucleosome. Specifically, the short regions (2–4 bp) of strong protection between the peaks of cleavage seen with intact nucleosomes were clearly more digested after tail removal (Fig. 2, compare lanes 4 and 6; see also Fig. 3). This increase in cleavage indicates a reduction in the stability of histone-DNA interactions and/or a greater motility of the DNA on the nucleosome surface. Importantly, the addition of Mg2+ induced a remarkable ‘reversion’ in the hydroxyl radical cleavage pattern of the tailless nucleosomes, such that the short regions between cleavage peaks were protected to approximately the levels observed before tail removal (Fig. 2, lane 7, Fig. 3). However, the tailless nucleosomes apparently do not shift back to their original location upon addition of Mg2+(Fig. 2, lane 7, Fig. 3). Thus, the presence of 2 mM Mg2+ can compensate for the lack of the core histone tail domains and stabilize histone-DNA interactions throughout the nucleosome core but does not result in reversal of the repositioning of the nucleosome that occurs upon tail removal.
A DNA sequence consisting of twelve tandem TATAAACGCC repeats was selected as having one of the highest binding affinities for core histone proteins in the mouse genome (28). We previously showed that, in contrast to observations with nucleosomes containing the 5S sequence, translational positioning of nucleosomes containing the high-affinity (TATAAACGCC)12 sequence is not altered upon removal of core histone tail domains (21). To determine whether nucleosomes containing the (TATAAACGCC)12 sequence exhibit destabilization similar to 5S nucleosomes upon the removal of core histone tails, nucleosomes were reconstituted with this sequence, purified by sucrose gradients then a portion of the sample incubated with trypsin agarose beads to remove the tail domains. Intact or tailless nucleosomes were exposed to hydroxyl radicals, nucleosomes isolated on 0.7% agarose nucleoprotein gels and the cleavage patterns were analyzed as above. With intact nucleosomes, the typical hydroxyl radical cleavage pattern was observed, with peaks in cleavage every 10 bp, while the DNA backbone between these peaks was well protected by the core histone octamer (Fig. 4A, lane 4). However, these regions were clearly less protected upon removal of the tail domains (Fig. 4A, lane 11 and 4B, top scans) indicating that core histone tails stabilize histone-DNA interactions in the (TATAAACGCC)12 nucleosomes similarly to the 5S nucleosomes. In particular, destabilization upon tail removal was most evident in regions near the periphery of the nucleosome (Fig. 4A, red bar). To determine whether Mg2+ can stabilize histone-DNA interactions in the tailless nucleosomes containing TATAAACGCC repeats, nucleosomes were incubated with increasing amounts of Mg2+ then treated with hydroxyl radicals and cleavage patterns analyzed as above. As before, the hydroxyl radical cleavage pattern of intact nucleosomes or naked DNA was not influenced by any concentration of Mg2+ examined (Fig. 4A, lanes 5–8, results not shown). However, the DNA in tailless nucleosomes was significantly more protected against hydroxyl radical cleavage in the presence of Mg2+ at all concentrations analyzed, compared to the cleavages of tailless nucleosomes in the absence of Mg2+ (Fig. 4A, lanes 11–15; 4B, middle trace).
Monovalent salts such as Na+ can partially drive condensation of chromatin (35), and Na+ also binds to the specific regions in the nucleosome core (26). To investigate whether Na+ also affects the stability of tailless nucleosomes, intact or tailless nucleosomes containing (TATAAACGCC)12 repeats were incubated with 50, 100, 150 mM Na+, respectively, for 10 minutes, treated with hydroxyl radicals, then analyzed as above. We observed that the extent of protection against hydroxyl radical cleavage in the native as well as the tailless nucleosomes was virtually unchanged by the inclusion of Na+ (Fig. 4A, lanes 16–18; 4B, bottom trace), indicating that Na+ does not recapitulate the effect of Mg2+ or the core histone tail domains in stabilizing histone-DNA interactions.
To determine whether other divalent cations can replace the role of the tail domains in stabilizing histone-DNA interactions, intact and tailless nucleosomes containing (TATAAACGCC)12 repeats were prepared and incubated with Mg2+, Ca2+, Zn2+, Fe2+, Mn2+ or Co2+ at a final concentration of 2 mM then treated with hydroxyl radicals and the cleavage patterns analyzed as above. The hydroxyl radical cleavage pattern of free DNA was not affected by the incubation of Mg2+, Ca2+, Zn2+, Mn2+, or Co2+, respectively (data not shown). Interestingly, regions between cleavage peaks in tailless nucleosomes were more protected against hydroxyl radicals after the addition of Ca2+, in a manner similar to that observed with Mg2+ (Fig. 5A, lane 5, 5B). However, the hydroxyl radical cleavage pattern of tailless nucleosomes was not altered by the addition of Zn2+, Fe2+, Mn2+, or Co2+ (Fig. 5A, lanes 6–9, 5B). We conclude that Mg2+ and/or Ca2+ play specific roles in stabilizing histone-DNA interactions that cannot be fulfilled by Zn2+, Fe2+, Mn2+, or Co2+.
As mentioned above, Mg2+ inhibits thermally-induced translational repositioning of nucleosomes (11, 36) but its effect on tail-dependent repositioning has not been determined. As shown above, Mg2+ added after removal of the core histone tail domains did not alter the positions of tailless 5S nucleosomes (Figs. 2 and and3).3). To determine whether Mg2+ affects the mobility of 5S nucleosomes when added before removal of the tails, nucleosomes were prepared, incubated without or with 2 mM Mg2+ then treated with trypsin agarose beads to remove core histone tails. Interestingly, the 5S nucleosomes treated with trypsin to remove tails in the absence of 2 mM Mg2+ migrated more slowly than the intact nucleosomes on PAGE gels containing 1× TBE buffer (Fig. 6A, lanes 2 and 3), while nucleosomes treated with trypsin in the presence of Mg2+ exhibited a more rapid migration (Fig. 6A, lane 4). The migration of tailless 5S nucleosomes on polyacrylamide gels buffered with 20 mM HEPES was also similar to that with TBE (Fig 6A and B). These data suggest that the distribution of translational positions resulting after removal of the core tail domains from pre-formed 5S nucleosomes depends on the presence or absence of Mg2+ during tail removal.
We next determined whether Mg2+ inhibited repositioning of 5S nucleosomes upon removal of core histone tails by crosslink mapping. Nucleosomes were reconstituted with a modified H2A (H2A-A45C-APB) to allow accurate mapping of translational positions, which crosslinkes to the DNA template about 39 bp to either side of the nucleosome dyad (21, 37). 5S nucleosomes were treated with trypsin agarose beads to remove the core histone tail domains in the presence or absence of 2 mM Mg2+, then nucleosomes were irradiated with UV light for 30 seconds to induce protein → DNA crosslinking (21). Crosslinked species formed with approximately equivalent efficiency in intact nucleosomes, tailless nucleosomes and tailless nucleosomes generated in the presence of Mg2+ (results not shown). The positions of crosslinks were mapped as described (21) and were detected within intact nucleosomes at nucleotides +95, +85, +6, +16, and +45 (Fig. 6C, lane 4), consistent with previous work (21) and correspond to translational positions centered at +55 (+95/+16), +45 (+85/+6) and +8 (+45) (Fig 6D, top). When the core histone tails were removed in the absence of Mg2+ a drastic alteration in the pattern was observed, with crosslinks primarily detected at nucleotides at +63, and +23 (Fig. 6C, lane 5), again consistent with previous data (21), and corresponding to nucleosomes centered at −17 and +24 (Fig. 6D, bottom). However, when 2 mM Mg2+ was present prior to tail removal, the cross-linking pattern of tailless nucleosomes was not significantly altered compared to intact nucleosomes (Fig. 6C, compare lanes 4 and 6). SDS-PAGE analyses indicate that nucleosomes were equivalently trypsinized in the presence or absence of Mg2+ (results not shown). We conclude that Mg2+ inhibits the spontaneous movement of 5S nucleosome when present before removal of core histone tail domains.
In this work we demonstrate a unique role for Mg2+ and Ca2+ ions in nucleosome stability. This role is uncovered by removal of the core histone tail domains of canonical nucleosomes and is manifest as 1) stabilization of histone-DNA interactions within the nucleosomes core as reported by hydroxyl radical cleavage patterns and 2) restriction of equilibration between alternative translational positions in the absence of the core histone tail domains. Both of these effects are likely directly related in that exchange between translational positions likely involves a transition state in which a number of histone-DNA interactions are lost. The stabilization afforded by Mg2+ and Ca2+ increases the energy barrier for attaining this transition state and therefore prevents the facile inter-conversion between translational positions that occurs upon tail removal (21).
Our previous work demonstrated that the tail domains influence ‘choice’ of translational positions on certain DNA sequences. We demonstrated that removal of the tail domains from intact nucleosomes resulted in re-positioning of nucleosomes assembled on DNA sequences with low to moderate affinity for the core histones. However, no repositioning was observed upon tail removal from nucleosomes containing a high-affinity sequence (21). In this work we find that removal of the tail domains in the presence of Mg2+ does not lead to a repositioning of nucleosomes, as discussed above. However, it is also worth noting that addition of Mg2+ to tailless, repositioned nucleosomes does not result in recovery of the original distribution of translational positions. Thus, while the divalent cations Mg2+and Ca2+stabilize histone-DNA interactions in a manner similar to the core histone tails, they do not recapitulate the influence of the tails on the choice of nucleosome translational positioning.
It is likely that the stabilization of Ca2+ and Mg2+ have physiological relevance. First the free concentration of Mg2+ in the nucleus is in the range studied in our experiments. Second, in native chromatin the core histone tail domains are involved in a number of interactions other than contacting the core DNA and thus the native chromatin environment resembles to some degree that represented modeled tailless nucleosomes and therefore, an appropriate constellation of non-core tail interactions might recapitulate the tailless state studied here (38, 39). Thus a model emerges wherein reallocation of tail interactions, either by posttranslational modifications, interactions with ancillary factors, and/or inter-nucleosomal interactions (40), may lead to repositioning of nucleosomes, and thus exposure of cognate DNA sites for trans-acting factors. Moreover, micrococcal digestions performed at low cofactor concentrations (0.1 mM Ca2+) reveal that even when the core histone tails are present, divalent cations play a role in the stability of histone-DNA interactions at the edge of the nucleosome core region (25). These results support the idea that binding of Mg2+ or Ca2+ to specific sites in the nucleosome serve to stabilize the overall binding of DNA within the native nucleosome structure (26). Interestingly, nucleosomes containing (TATAAACGCC)12 repeats, a DNA sequence having one of the strongest binding affinities for core histone proteins (41), were also detectably destabilized upon tail removal, especially near edges of the nucleosomes (Fig. 4A). Thus, regardless of whether the tails contribute to positioning for a particular sequence, they appear to contribute to the overall stability of nucleosomes containing all sequences.
We interpret the alteration in the hydroxyl radical cleavage pattern of nucleosomes upon tail removal as an indication of greater DNA motility on the nucleosome surface. Previous work by us and others shows that tail removal results in a modest 3–5 fold increase in spontaneous DNA unwrapping from the histone surface (9, 42), likely insufficient to account for the entire effect we observe in the current work. In addition, the effect of tail removal on the hydroxyl radical protection pattern cannot be due to a randomization of translational positions in the absence of tail domains. Indeed, alternative translational positions adopted after tail removal are typically overlap with the same rotational orientation, and thus would add constructively. For example the addition of Mg2+ to trypsinized nucleosomes does not re-establish the original distribution of translational positions but does significantly alter the hydroxyl radical cleavage pattern. Importantly, tail removal causes the same effect with nucleosomes containing (TATAAACGCC)12 repeats, but these do not undergo re-positioning upon tail removal. Our results agree work showing the core histone tail domains are important for the thermal stability of the nucleosomes (24). In addition, our results correspond well with previous work showing that divalent cations bind to specific sites within nucleosomes (26) and can restrict thermally-induced nucleosome repositioning along DNA fragments (11, 36) stabilize histone-DNA interactions at the edge of the nucleosome core region (25).
In this work we demonstrate that histone-DNA interactions in tailless nucleosomes are stabilized by the presence of Mg2+ or Ca2+ but not by the cations Zn2+, Fe2+, Mn2+, or Co2+. This suggests that stabilization of histone-DNA interactions is not sequence-specific and that only specific divalent cations such as Mg2+ or Ca2+ can stabilize these interactions. One explanation for the specific effect of these cations is that specific binding sites for Mg2+ and Ca2+ exist within the nucleosome. Interestingly, these ions do not bind specifically to DNA or alter naked DNA solution structures, while several transition metal cations (Mn2+, Co2+, Ni2+, Cu2+, Pd2+, and Cd2+) bind specifically to sites on purines and pyrimidines and alter the structure of B-DNA (43, 44). While transition metal cations may indeed form similar specific interactions with nucleosome DNA, our data suggests that only the binding of Mg2+ or Ca2+ elicits stabilization of the nucleosome structure. Moreover, previous studies have demonstrated a role for mono and divalent alkaki earth cations in promoting condensation of native and reconstituted chromatin (3, 45, 46), and X-ray crystallographic studies reveal specific binding sites for Mn2+ or Mg2+ within a core nucleosome particle (26). Thus, potential nucleosome binding sites for Mg2+ or Ca2+ may be comrprised of both protein and DNA components. Interestingly, tailless nucleosome arrays cannot fold into chromatin fibers even in the presence of Mg2+ or Ca2+ indicating that the tail domains cannot be completely replaced by these divalent cations. We expected that Mn2+ might stabilize the tailless nucleosome because Mg2+ or Mn2+ binds to the nucleosome core to stabilize nucleosome structure (26). However, Mn2+ did not have any detectable effect on stability of tailless nucleosomes containing TATAAACGCC repeats. It will be of interest in the future to determine the location and salient characteristics of the metal binding sites involved in the effects observed in this study.
This work was supported by NIH Grant R01-52426.