10.4.3 Specific versus Non-Specific Complexes
We have discussed the energeticdifferences between specific and non-specific complexes (Figure 10.9),
but what structuralfeatures distinguish them? One example of a non-specific complex is between the
glucocorticoid receptor DNA-binding domain and non-target DNA. The natural target of this protein is a
palindrome in which the hexameric half-sites are separated by three bases, and the protein binds this as a
homodimer with a self-complementary protein–protein interface (Section 10.3.3). In the non-specific com-
plex, the separation of the half-sites is increased to four base pairs. The protein still forms a dimer on this
DNA, and each monomer interacts with its half-site, but the second monomer is pulled out of alignment
with its recognition sequence. There are fewer direct hydrogen bonds to the bases, but just as many inter-
actions with the phosphate backbone. The surface match is not so good for the non-specific complex, so
that the buried surface area is only about half that the specific interaction. In another non-specific complex,
a layer of bound water molecules at the protein–nucleic acid interface is observed that must contribute an
entropic penalty to the binding and disfavours the non-specific complex.^20
The non-specific complex of the BamHI restriction enzyme illustrates the role of both surface electro-
static and shape complementarity (Figure 10.10). Thus, target recognition is here associated with struc-
tural changes in the protein that alter both the surface charge distribution and the shape.
10.4.4 Electrostatic Effects
For most RNA- and DNA-binding proteins, the affinity for both specific and non-specific nucleic acid
decreases sharply as the ionic strength increases. This arises from the polyelectrolyte character of DNA
and RNA, which affect the binding of polycationic proteins and ligands. The salt concentration also affects
the ability of proteins to slide along the longer target.
One curious exception to this salt-dependence rule occurs in the case of the transcription factor TBP
from the hyperthermophilic archaea Pyrococcus,^21 where the binding affinity increases with ionic strength.
This may be explained by a putative cation-binding site that bridges the surfaces of the protein and the
DNA. The protein surface that is in proximity to the DNA is electronegative, rather than the more usual
electropositive (e.g.the surface of BamHI; Section 10.4.3), and the counter-ion binding helps to overcome
the charge repulsion.
Subtle electrostatic switching may also explain how the met repressor (Figure 10.6a) is activated by its
ligand, S-adenosylmethionine, to gain a 1000-fold increase in its affinity for the ‘met box’ target DNA. The
met repressor undergoes little apparent structural change upon binding of ligand. But because the ligand
is charged, it changes the surface potential of the repressor, which now makes a more favourable match to
the surface of the DNA. Perhaps this effectively neutralizes the charges on the DNA, but does so in an asym-
metrical fashion, to favour its conformational adjustment, not only to the bound protein, but also to its neigh-
bour at an adjacent met box. Attractive protein–protein interactions also favour the binding of adjacent
repressors, with the net result that they bind co-operatively to the DNA element (Section 10.4.6).
10.4.5 DNA Conformabilit y
Duplex DNA can be treated globally as a moderately flexible rod, with two principal degrees of freedom
(Section 2.3.1). One mode involves twisting along the local helical axis, but structural restraints to the
twisting impart a torsional stiffness to the molecule. The energy of twistingis proportional to the square
of the change of twist angle per unit length. The other mode is bending, and stereochemical restraints to
this mode are associated with axial stiffness. The energy for bendingis roughly proportional to the square
of the curvature, where curvature is a bend-angle per unit length, which is the same as the reciprocal of the
local radius of curvature. Bending is usually achieved by changing the roll angle at the base steps. The
energy required for bending may be anisotropic, with a preferred azimuthal angle, according to the ability
of the base steps to roll, and the projection of the roll with respect to the plane of curvature of the bend.
When the stacking is maintained, the bending is smooth, as shown for the nucleosomal DNA (Figure
10.1d), but the DNA becomes kinked if the stacking is disrupted (e.g.the base step in Figure 10.4f).
400 Chapter 10