The rotational positioning of DNA about the histone octamer appears to be determined
by certain sequence-dependent modulations of DNA structure. To establish the detailed
nature of these interactions, we have analysed the sequences of 177 different DNA
molecules from chicken erythrocyte core particles. All variations in the sequence
content of these molecules, which may be attributed to sequence-dependent preferences
for DNA bending, correlate well with the detailed path of the DNA as it wraps around
the histone octamer in the crystal structure of the nucleosome core. The sequence-dependent
preferences that correlate most closely with the rotational orientation of the DNA,
relative to the surface of the protein, are of two kinds: ApApA/TpTpT and ApApT/ApTpT,
the minor grooves of which face predominantly in towards the protein; and also GpGpC/GpCpC
and ApGpC/GpCpT, whose minor grooves face outward. Fourier analysis has been used
to obtain fractional variations in occurrence for all ten dinucleotide and all 32
trinucleotide arrangements. These sequence preferences should apply generally to many
other cases of protein-DNA recognition, where the DNA wraps around a protein. In addition,
it is observed that long runs of homopolymer (dA) X (dT) prefer to occupy the ends
of core DNA, five to six turns away from the dyad. These same sequences are apparently
excluded from the near-centre of core DNA, two to three turns from the dyad. Hence,
the translational positioning of any single histone octamer along a DNA molecule of
defined sequence may be strongly influenced by the placement of (dA) X (dT) sequences.
It may also be influenced by any aversion of the protein for sequences in the "linker"
region, the sequence content of which remains to be determined.