Solution structure of the isolated histone H2A-H2B heterodimer.

During chromatin-regulated processes, the histone H2A-H2B heterodimer functions dynamically in and out of the nucleosome. Although detailed crystal structures of nucleosomes have been established, that of the isolated full-length H2A-H2B heterodimer has remained elusive. Here, we have determined the solution structure of human H2A-H2B by NMR coupled with CS-Rosetta. H2A and H2B each contain a histone fold, comprising four α-helices and two ß-strands (α1-ß1-α2-ß2-α3-αC), together with the long disordered N- and C-terminal H2A tails and the long N-terminal H2B tail. The N-terminal αN helix, C-terminal ß3 strand, and 310 helix of H2A observed in the H2A-H2B nucleosome structure are disordered in isolated H2A-H2B. In addition, the H2A α1 and H2B αC helices are not well fixed in the heterodimer, and the H2A and H2B tails are not completely random coils. Comparison of hydrogen-deuterium exchange, fast hydrogen exchange, and {(1)H}-(15)N hetero-nuclear NOE data with the CS-Rosetta structure indicates that there is some conformation in the H2A 310 helical and H2B Lys11 regions, while the repression domain of H2B (residues 27-34) exhibits an extended string-like structure. This first structure of the isolated H2A-H2B heterodimer provides insight into its dynamic functions in chromatin.

C-terminal acidic domain of histone chaperone human NAP1 is an efficient binding assistant for histone H2A-H2B, but not H3-H4.

Nucleosome assembly protein 1 (NAP1) binds both the (H3-H4)2 tetramer and two H2A-H2B dimers, mediating their sequential deposition on DNA. NAP1 contains a C-terminal acidic domain (CTAD) and a core domain that promotes dimer formation. Here, we have investigated the roles of the core domain and CTAD of human NAP1 in binding to H2A-H2B and H3-H4 by isothermal calorimetry and native mass spectrometry and compared them with the roles of yeast NAP1. We show that the hNAP1 and yNAP1 dimers bind H2A-H2B by two different modes: a strong endothermic interaction and a weak exothermic interaction. A mutant hNAP1, but not yNAP1, dimer lacking CTAD loses the exothermic interaction and shows greatly reduced H2A-H2B binding activity. The isolated CTAD of hNAP1 binds H2A-H2B only exothermically with relatively stronger binding as compared with the exothermic interaction observed for the full-length hNAP1 dimer. Thus, the two CTADs in the hNAP1 dimer seem to provide binding assistance for the strong endothermic interaction of the core domain with H2A-H2B. By contrast, in the relatively weaker binding of hNAP1 to H3-H4 as compared with yNAP1, CTAD of hNAP1 has no significant role. To our knowledge, this is the first distinct role identified for the hNAP1 CTAD.

Novel structural and functional mode of a knot essential for RNA binding activity of the Esa1 presumed chromodomain.

Chromodomains are methylated histone binding modules that have been widely studied. Interestingly, some chromodomains are reported to bind to RNA and/or DNA, although the molecular basis of their RNA/DNA interactions has not been solved. Here we propose a novel binding mode for chromodomain-RNA interactions. Essential Sas-related acetyltransferase 1 (Esa1) contains a presumed chromodomain in addition to a histone acetyltransferase domain. We initially determined the solution structure of the Esa1 presumed chromodomain and showed it to consist of a well-folded structure containing a five-stranded beta-barrel similar to the tudor domain rather than the canonical chromodomain. Furthermore, the domain showed no RNA/DNA binding ability. Because the N-terminus of the protein forms a helical turn, we prepared an N-terminally extended construct, which we surprisingly found to bind to poly(U) and to be critical for in vivo function. This extended protein contains an additional beta-sheet that acts as a knot for the tudor domain and binds to oligo(U) and oligo(C) with greater affinity compared with other oligo-RNAs and DNAs examined thus far. The knot does not cause a global change in the core structure but induces a well-defined loop in the tudor domain itself, which is responsible for RNA binding. We made 47 point mutants in an esa1 mutant gene in yeast in which amino acids of the Esa1 knotted tudor domain were substituted to alanine residues and their functional abilities were examined. Interestingly, the knotted tudor domain mutations that were lethal to the yeast lost poly(U) binding ability. Amino acids that are related to RNA interaction sites, as revealed by both NMR and affinity binding experiments, are found to be important in vivo. These findings are the first demonstration of how the novel structure of the knotted tudor domain impacts on RNA binding and how this influences in vivo function.