Ghydby using algorithms derived from semi-empirical relationships based on heats of solvent transfer of
small molecules and protein folding–unfolding equilibria. Record has shown that
(9.15)From this equation, it is clear that Ghyd can be estimated in a simple and direct manner by measurement
of the heat capacity change associated with a binding interaction. A near-complete thermodynamic analysis,
including a detailed partitioning of free energy, has been presented recently for a range of DNA intercalators^23
and reveals some interesting thermodynamic features of DNA intercalation by small molecules. The binding
of most intercalators is enthalpically driven. For ethidium and daunomycin this is opposed by an unfavourable
entropy term while for propidium and adriamycin, there is a small favourable entropy term. These differences
in T Sare likely to be due to differences in hydration of the ligands involved. Heat capacity changes for
ethidium, propidium, daunomycin, and adriamycin fall within a narrow range of about 630 J mol^1 K^1.
In general, heat capacity changes are smaller for intercalators than they are for groove binders. In add-
ition, the heat capacity change for actinomycin D binding is much larger. Both these observations make sense
when one considers that heat capacity change correlates with changes in nonpolar and polar solvent access-
ible surface area. For drug–DNA interactions, the following empirical relationship has been established:
(9.16)where Anpand Apare binding-induced changes in nonpolar and polar surface areas, respectively. The
change in nonpolar surface area for ethidium is 407 Å^2 , whereas for actinomycin D it is 1046 Å^2. Intercalation
of positively charged ligands gives rise to a small favourable polyelectrolyte term. For actinomycin D,
which is uncharged, the value of Gpeis negligible.
For intercalation, the overwhelming energetic driving forces are hydrophobic interactions and other weak,
non-covalent molecular interactions, such as van der Waals and hydrogen bonding. These free energy terms
are large enough to overcome the79.5 kJ mol^1 energetic cost that arises from conformational changes
and losses in degrees of rotational and translational freedom that accompanies binding. In the case of acti-
nomycin D, molecular interactions make a small unfavourable contribution to binding free energy. This is
also the case for the groove binder, Hoechst 33258 (Section 9.7.5). Actinomycin D contains cyclic peptide
moieties that, like Hoechst 33258, occupy the DNA minor groove. Binding of a ligand in the minor groove has
the effect of expulsion of site-specifically bound water or cation molecules. Hence water-base pair hydro-
gen bonds are broken to form ligand-base pair hydrogen bonds. Therefore, from a thermodynamic point-
of-view, a favourable energetic contribution arising from ligand hydrogen bonding may be limited.
9.6.5 Bisintercalation
Bisintercalatorsare bifunctional molecules that possess two planar intercalating aromatic ring systems
covalently linked by chains of varying length. It is also possible to link three or more ring systems together
using linkers. A good reason for designing and synthesising a bisintercalator is that it should have a sig-
nificantly higher affinity and much slower dissociation kinetics than the monointercalator equivalents. The
binding constant for a bisintercalator should be approximately the square of the monomer binding constant.
Since biological activity is often closely correlated with binding affinity, bisintercalators should also have
enhanced medicinal application.
Bisintercalators have a larger site size than their monointercalator counterparts, which can lead to increased
sequence selectivity. Many simple monointercalators have a binding site size of 1 in 3 base pairs. Therefore,
to avoid violation of the neighbour exclusion principle, a bisintercalator made from such monomerscan be
expected to span a site of at least 6 base pairs, which is the same size as the recognition sites of many sequence-
specificrestriction endonuclease enzymes (Section 5.3.1).
Bisintercalators are also ideal model systems for probing and evaluating neighbour exclusion effects,
because they can be prepared with variable linker lengths. Ligands with short linkers may be induced to
CAApnp0 382 0 026.(.) 0 121 0 077.(.)p
GChyd ()80 10 pReversible Small Molecule–Nucleic Acid Interactions 355