The histone-fold motif is a protein structure feature common to all four histones that was first revealed in the structure of the core histone octamer (1). The mode of interaction of this motif with DNA, suggested by the octamer structure, is shown in detail in the structure of the nucleosome core particle (2). The histone fold has since been recognized in the structures of several other proteins, spanning the evolutionary range from archaebacteria to mammals, that are involved, like the histones, in DNA:protein recognition and protein-protein interactions . Some of the histone-fold proteins thus far identified show little or no sequence homology with the histones (3).
The histone fold consists of a long central a-helix connected by loops to two shorter a-helices. It may have arisen by gene duplication of a helix-strand-helix motif, with helix fusion resulting in the long central a-helix (4). The histones in the nucleosome core particle occur as heterodimers through extensive interactions between the two histone folds in a "handshake motif," with extensive hydrophobic interactions between the nonpolar faces of the two long amphipathic a-helices. The same basic fold occurs in the homodimeric archaeal histone HMfB (from the hyperthermophile Methanothermus fervidus) (5), which shows some sequence similarity to the histones, and in the presence of DNA forms tetramers similar to the H3 2H42 tetramer (see Histones). In HMfB, however, there are no regions similar to the extensions to the basic fold in H3 that contribute to the organization of the DNA in the nucleosome core particle.
Strikingly, the histone fold has also been found to occur in other proteins. The first reports were for TATA box-binding protein-associated factors (TAFs), components of the transcription initiation factor TFIID. The 2 A X-ray crystallography structure of the complex of TAF(II)42 and TAF(II)62 from Drosophila showed that it was a tetramer (a dimer of heterodimers), with the interacting regions (which constitute a relatively small part of the total protein) organized in histone folds similar to those of H3 and H4 (6) (Fig. 1). The finding that human TAF(II)20 (homologous to Drosophila TAF(II)30a) probably also contains a histone-fold motif similar to that in H2B (7) (no H2A homologues have been found) led to the suggestion that there is a histone octamer-like TAF complex within human TFIID, composed of a tetramer of TAF(II)80 and TAF(II)31 and two homodimers of TAF(II)20 (in Drosophila a tetramer of TAF(II)62 and TAFII42) and two homodimers of TAF(II)30a) (6, 7). Surprisingly, three histone-like TAFs also occur amongst the 20 subunits of the P/CAF acetyltransferase complex (see Histone Acetylation), namely, TAF(II)31, TAF(II)20 / 15 and PAF65a (similar to human TAF(II)80), which resemble H3, H2B, and H4, respectively, and an octamer-like structure has again been suggested (8). The yeast 1.8-MDa SAGA (Spt-^da- GCN5-acetyltransferase) acetyltransferase complex, which is similar to the human P/CAF complex, also contains three histone-fold TAFs with homology to H2B, H3, and H4 as integral components (9), and the Spt3 protein appears to have histone folds in its N- and C-terminal regions, which have been suggested to interact intramolecularly (10). X-ray crystallography has recently shown that yet another pair of TAFs, human TAF(II)18 and TAF(II)28, which show only weak homology to histones, also form a heterodimer through a histone fold (9). The histone-fold motif thus appears to be widespread, and it may simply be a robust structural motif particularly well-suited to protein dimerization in large complexes. It remains to be seen, however, whether histone octamer-like TAF structures have a particular role to play in processes involving histones. Despite the presence of the histone fold, any interaction of TAFs with DNA is likely, a priori, to be different from that in the nucleosome core because the arginine side chains of the core histones that insert into the minor groove of the surrounding DNA are absent; in the case of Drosophila TAF(II)42 and TAF (II)62, they are replaced by serine and glutamate. The four-helix-bundle interface and the orientation of the two heterodimers within the TAFII tetramer are also different from those in the H32H42 tetramer, although this in itself might not necessarily preclude DNA binding.
Figure 1. Drawings of the histone folds in H3 and H4, and Drosophila TAF(II)42 and TAF(II)62. (a) The TAF(II)42/TA heterodimer from which the H32H42 tetramer in chromatin is formed; (b) the tetramer TAF(II)42 2.TAF(II)622 (residues TAF62).
Although no high-resolution structural data are yet available, it seems likely, on the basis of sequence homology with H2A and H2B, that the histone-fold motif also occurs in two subunits of the trimeric mammalian transcription factor CBF [CCAAT-binding factor (11)] and its yeast homologue HAP (CCAAT boxes are widespread promoter elements; see CAAT Box) and in the human protein NC2 [negative cofactor 2 (12); also known as DRAP1] and its yeast homologue, which repress transcription of class II genes by binding to TBP and to DNA and inhibiting preinitiation complex formation by preventing TFIIB binding. It is striking that the histone fold already present in the Archaea has been adapted not only for DNA packaging in eukaryotes but also apparently as a dimerization motif in gene regulatory proteins, which sometimes show little, if any, sequence similarity. This is reminiscent of the "winged helix" DNA-binding motif in the globular domain of histone H5 (see Histones), which is also found in bacterial cyclic AMP receptor protein and the liver transcription factor HNF3g, a member of a large family of transcription factors (the HNF/forkhead family) from several organisms.