Histone H5 promotes the association of condensed chromatin fragments to give pseudo-higher-order structures.

We describe two distinct situations in which chicken erythrocyte chromatin fragments associate in solution. The erythrocyte-specific histone H5 is implicated since chromatins that do not contain H5 do not show this behaviour. Well-defined oligomers of between approximately 6 and approximately 18 nucleosomes prepared at low ionic strength condense and associate when the ionic strength is raised to 75 mM, forming pseudo-higher-order structures. The associated forms, probably predominantly dimers, are stabilized by migration of about 10% of the H5, and of the minor lysine-rich histone H1, from the non-associated forms, probably reflecting the preference of H5 for higher-order structures observed previously [Thomas, J. O. and Rees, C. (1983) Eur. J. Biochem. 134, 109-115]. Since the final (H1 + H5) content of the aggregate at 75 mM is never higher than that of the fragment prepared at low ionic strength, migration is probably to a small proportion of sites that have inevitably become vacant due to handling losses at the higher (but not at low) ionic strength. H5 thus maximizes its interactions in the condensed state of chromatin and even maintains the association of two or more fragments without continuity of the DNA. Aggregates of oligomers larger than about 18 nucleosomes may be too long to withstand hydrodynamic shear forces in the absence of such continuity. During nuclease digestion of nuclear chromatin, H5 and, to a lesser extent, H1, are released from the ends of very short fragments and bind to larger oligomers of various sizes leading to heterogeneous aggregates that survive exposure to low ionic strength. These aggregates, in contrast to those described above, have up to 60% more H5 and 20% more H1 than chromatin prepared at low ionic strength. Whether the excess H5 and H1 bind non-specifically or to a second low-affinity binding site on each nucleosome is not known. The associated forms described above (1) are well defined and potentially useful for structural studies, whereas the other aggregates (2) seem less likely to be directly relevant to the native structure of chromatin.