Geometric Patterns for Neighboring Bases Near the Stacked State in Nucleic Acid Strands

Structural
variation in base stacking has been analyzed frequently in isolated
double helical contexts for nucleic acids, but not as often in nonhelical
geometries or in complex biomolecular environments. In this study,
conformations of two neighboring bases near their stacked state in
any environment are comprehensively characterized for single-strand
dinucleotide (SSD) nucleic acid crystal structure conformations. An
ensemble clustering method is used to identify a reduced set of representative
stacking geometries based on pairwise distances between select atoms
in consecutive bases, with multiple separable conformational clusters
obtained for categories divided by nucleic acid type (DNA/RNA), SSD
sequence, stacking face orientation, and the presence or absence of
a protein environment. For both DNA and RNA, SSD conformations are
observed that are either close to the A-form, or close to the B-form,
or intermediate between the two forms, or further away from either
form, illustrating the local structural heterogeneity near the stacked
state. Among this large variety of distinct conformations, several
common stacking patterns are observed between DNA and RNA, and between
nucleic acids in isolation or in complex with proteins, suggesting
that these might be stable stacking orientations. Noncanonical face/face
orientations of the two bases are also observed for neighboring bases
in the same strand, but their frequency is much lower, with multiple
SSD sequences across categories showing no occurrences of such unusual
stacked conformations. The resulting reduced set of stacking geometries
is directly useful for stacking-energy comparisons between empirical
force fields, prediction of plausible localized variations in single-strand
structures near their canonical states, and identification of analogous
stacking patterns in newly solved nucleic acid containing structures.