Numerous studies have investigated the location of linker histone H1 in the nucleosome. It is generally accepted that the globular domain of H1 is located in the vicinity of the pseudo-dyad axis, where it contacts both the DNA entering/exiting the nucleosome as well as the central gyre of the nucleosomal DNA. Although the binding of the primary DNA binding site to DNA has been experimentally characterised, the precise location of the secondary DNA binding site is less-well defined. In this research program we aim to identify the secondary DNA binding site residues involved in contacting the nucleosomal DNA and the precise position of these contacts. We have approached this problem by defining the most likely model of the linker histone-nucleosome complex, and then testing the validity of this model experimentally. This will be performed by a combination of residue mapping and cryo-electronmicroscopy.
In order to model the DNA binding of the primary DNA-binding site of GH5, the crystal structure of the A-chain of the globular domain of GH5 (Ramakrishnan et al., 1993) was superimposed onto the co-crystal structure of the DNA-binding domain of the structural analogue, CAP (Parkinson et al., 1996). The most important residues contacting the DNA-backbone in the CAP-DNA-complex was identified by analysing the ionic and hydrogen-bonded interactions involved. By comparing secondary structural elements of both proteins (GH5 and CAP), areas with similar secondary structure were identified. In this way, structurally equivalent residues were identified in GH5. A Root Mean Square (RMS) superposition of atoms corresponding to the location of the side-chain charge of structurally equivalent residues (as defined by the Arg CZ atom, the Lys NZ atom and the CG atom of His) was superimposed to create a "charge-centre model". In order to remove steric clashes, the conformation of the proposed DNA-binding site of GH5 was optimised by nudging the conformation of the protein and evaluating the potential energy associated with that particular conformation. The model of GH5 bound via helix III to double-stranded DNA that had been optimised accordingly, also showed no significant signs of steric hindrance.
The potential locations of the secondary DNA-binding site of GH5, given the dimensions of nucleosomal DNA and the distance between the GH5 binding sites, was modeled next. DNA already associated with GH5 as part of the model of the primary DNA-binding site of GH5 as linker DNA was "ligated" to the nucleosomal DNA termini of the crystal structure of the nucleosome (Luger et al. 1997) in silico. It was possible to attach this linker DNA-GH5 construct to the nucleosomal core particle terminus in a manner consistent with the recent cross-linking evidence supplied by Zhou et al. (1998), where the putative secondary DNA-binding site contacts DNA near the nucleosomal dyad. In this orientation, the b-hairpin of GH5 extends into the nucleosome. It was clear that the secondary DNA binding site does not consist of a distinct site, but rather an interaction surface generated by the C-terminal end of helix I, the inter-helical segment connecting helix I and II, and distinct contact points in both strands of the C-terminal b-sheet. These regions co-localize over the central gyre of nucleosomal DNA and straddles DNA in the vicinity of the dyad axis. The highly conserved basic residues, Arg 42(inter-helical segment), Lys 40(helix I) and Arg 94 (C-terminal b-strand), are in close contact with nucleosomal DNA near the nucleosomal dyad. In addition, Gly 44 (inter-helical segment), which is also highly conserved among linker histones, donate a strong (2.68Å) hydrogen bond to the DNA-backbone. Other less conserved residues belonging to the aforementioned GH5 segments, also contribute to GH5 binding, mainly via hydrogen-bonded interactions.
We have designed, expressed and purified GH5 mutants that will aid in the characterisation of the secondary DNA-binding site by site-directed metal affinity cleavage or chemical cross-linking. A molecular model of the hydroxyl radical donor, cysteaminyl-EDTA, coupled to appropriate cysteine residues, will be used to identify the potential cutting sites of hydroxyl radicals liberated from the donor. Three residues, chosen for their proximity to the DNA contacted by the secondary DNA-binding site of GH5, were mutated (A96C, I36C, R37C). The single cysteine residues in these recombinant proteins are currently being modified with the EDTA adduct to allow metal chelation mapping of the secondary DNA binding site of gH5.
We plan to isolate H1-containing chromatin filaments, and to determine the structure of the resulting chromatosome by cryo-electronmicroscopy. We may also make use of EM images supplied by the laboratory of Christopher Woodcock. We will then make use of the solved crystal structures of the nucleosome core particle, and of the globular domain of chicken histone H5, and predict, from first principles, the expected electron density distributions that is expected from a cryo-electronmicroscopic visualization. This prediction will require the refinement of current physical models. The predicted images will then be compared to the observed images, and the level of molecular insight enhanced by making use of this molecular docking approach. This will provide the first macromolecular structural evidence for the location of the linker histone in a nucleosome, and provide additional insight into the structural and regulatory role of the linker histones.