HU And Related Proteins 
 
The bacterial type-II DNA-binding proteins include the abundant HU and integration host factor (IHF) proteins and the phage-encoded TF1. They are believed to condense their cognate genomes by binding at multiple sites and inducing coherent bends. HU and TF1 possess little, if any, binding sequence specificity, but IHF, which is also required for site-specific recombination, DNA replication, and transcription, binds at specific sites characterized by a limited consensus sequence.  
All proteins of this type bind to DNA as dimers and bend the DNA to a substantial extent. In particular, IHF induces a bend angle of at least 160°, and possibly in excess of 180°, within the 2 1/2  turns of double helix that comprise its binding site. IHF is normally a heterodimer, while HU can exist either as a homo- or as a heterodimer. In both IHF and HU, the two ~10 kDa subunits intertwine to form a compact core, from which two long beta-sheet ribbon arms extend. These arms track along the minor groove from the inside to the outside of the wrapped DNA, where they terminate at the two substantial kinks. In addition to these interactions via the b arms, IHF also clamps the hairpin by minor-groove contacts with alpha-helices from both subunits in the core of the dimer. All the contacts to the DNA are either in the minor groove or are part of an extensive network of electrostatic interactions with the backbone phosphate groups.
 The two kinks are induced by the partial intercalation between adjacent base pairs of an absolutely conserved proline sidechain located at the tips of the b arms. In the IHF–DNA complex, the DNA bend is maintained by two distinct mechanisms. On the outside of the bend, the hydrophobic intercalation stabilizes the opening of the minor groove. On the inside, charge neutralization counteracts the enhanced repulsion between the phosphates on opposite sides of the narrowed grooves.
The sequence dependency of IHF binding is determined by the conformation of DNA rather than bybase-specific contacts. The “consensus” sequence consists of two short elements separated by approximately half a turn in only one of the two half-sites. The conserved sequence CAA at the kink site can accommodate the severe distortion induced by the protein better than other short sequences, and so is favored.