concentration stabilises the Z-conformation because it has a much smaller separation between the phos-
phate anions in opposing strands than is the case for B-DNA, 8 Å as opposed to 11.7 Å. A detailed stereo-
chemical examination of this conformational change shows that it calls for an elaborate mechanism and
this has posed a problem known as the chain–sense paradox: ‘How does one reverse the sense of direction
of the chains in a B-helix (↑↓) to its opposite in a Z-helix (↓↑) without unpairing the bases?’ Further
consideration to this problem will be given later (Section 2.5.4).
The scanning tunnelling microscope has the power to resolve the structure of biological molecules with
atomic detail (Section 11.5.2). Much progress has been made with dried samples of duplex DNA, in
recording images of DNA in wet state, and in revealing details of single-stranded poly(dA). Such STM
microscopy has provided images of poly(dG-me^5 dC)poly(dG-me^5 dC) in the Z-form. Both the general
appearance of the fibres and measurements of helical parameters are in good agreement with models
derived from X-ray diffraction data.
2.3 Real DNA Structures
2.3.1 Sequence-Dependent Modulation of DNA Structure
So far we have emphasised the importance of hydrogen bonds in base-pairing and DNA structure and have
said little about base stacking. We shall see later that both these two features are important for the ener-
getics and dynamics of DNA helices (Section 2.5), but it is now time to look at the major part played by
base stacking in real DNA structures. Two particular hallmarks of B-DNA, in contrast to the A- and
Z-forms, are its flexibility and its capacity to make small adjustments in local helix structure in response
to particular base sequences.^14
Different base sequences have their own characteristic signature: they influence groove width, helical
twist, curvature, mechanical rigidity and resistance to bending. It seems probable that these features help
proteins to read and recognise one base sequence in preference to another (Chapter 10), possibly only
through changes in the positions of the phosphates in the backbone. What do we know about these
sequence-dependent structural features?
One surprise to emerge from single-crystal structure analyses of synthetic DNA oligomers has been the
breadth of variation of local helix parameters relative to the mean values broadly derived from fibre dif-
fraction analysis and used for the standard A- and B-form DNA structures described earlier. Dickerson has
compared eight dodecamer and three decamer B-DNA structures.^15 The mean value of the helical twist
angle between neighbouring base pairs is 36.1° but the standard deviation (SD) is 5.9° and the range is
from 24° to 51°. Likewise, the mean helical riseper base pair is 3.3 6Å with a SD of 0.4 6Å but with a
range from 2.5 to 4.4 Å. (NB: because rise is a parameter measured between the C-1atoms of adjacent
base-pairs, it can be smaller than the thickness of a base pair if the ends of the two base pairs bow towards
each other. Such bowingis also defined as ‘positive cup’.) Roll angles between successive base pairs aver-
age0.6° but with a SD of 6.0° and a range from –18° to16°. These variations in twist and roll have the
effect of substantially re-orienting the potential hydrogen bond acceptors and donors at the edges of the
bases along the floor of the DNA grooves, so they may well be a significant component of the sequence-
recognition process used by drugs and proteins (Chapters 9 and 10). These and other modes of local
changes in the geometry of base pairs are illustrated in Figure 2.19.
The major irregularities in the positions of the bases in real DNA structures contrast with only second-
ary, small conformational changes in their sugar–phosphate backbones. The main characteristic of these
sequence-dependent modulations is propeller twist. This results when the bases rotate by some 5° to 25°
relative to their hydrogen-bonded partner around the long axis through C-8 of the purine and C-6 of the
pyrimidine (Figure 2.19b, centre). Sections of oligonucleotides with consecutive A residues, as in
d(CGCAAAAAAGCG)d(CGCTTTTTTGCG) have unusually high propeller twist (approximately 25°)
and these permit the formation of a three-centred hydrogen bonding network in the major groove between
adenine-N-6 and two thymine-O-4 residues, the first being the Watson–Crick base pair partner and the second
DNA and RNA Structure 33