Selective interactions of cationic porphyrins with G-quadruplex structures.

G-quadruplex DNA presents a potential target for the design and development of novel anticancer drugs. Because G-quadruplex DNA exhibits structural polymorphism, different G-quadruplex typologies may be associated with different cellular processes. Therefore, to achieve therapeutic selectivity using G-quadruplexes as targets for drug design, it will be necessary to differentiate between different types of G-quadruplexes using G-quadruplex-interactive agents. In this study, we compare the interactions of three cationic porphyrins, TMPyP2, TMPyP3, and TMPyP4, with parallel and antiparallel types of G-quadruplexes using gel mobility shift experiments and a helicase assay. Gel mobility shift experiments indicate that TMPyP3 specifically promotes the formation of parallel G-quadruplex structures. A G-quadruplex helicase unwinding assay reveals that the three porphyrins vary dramatically in their abilities to prevent the unwinding of both the parallel tetrameric G-quadruplex and the antiparallel hairpin dimer G-quadruplex DNA by yeast Sgs1 helicase (Sgs1p). For the parallel G-quadruplex, TMPyP3 has the strongest inhibitory effect on Sgs1p, followed by TMPyP4, but the reverse is true for the antiparallel G-quadruplex. TMPyP2 does not appear to have any effect on the helicase-catalyzed unwinding of either type of G-quadruplex. Photocleavage experiments were carried out to investigate the binding modes of all three porphyrins with parallel G-quadruplexes. The results reveal that TMPyP3 and TMPyP4 appear to bind to parallel G-quadruplex structures through external stacking at the ends rather than through intercalation between the G-tetrads. Since intercalation between G-tetrads has been previously proposed as an alternative binding mode for TMPyP4 to G-quadruplexes, this mode of binding, versus that determined by a photocleavage assay described here (external stacking), was subjected to molecular dynamics calculations to identify the relative stabilities of the complexes and the factors that contribute to these differences. The DeltaG(o) for the external binding mode was found to be driven by DeltaH(o) with a small unfavorable TDeltaS(o) term. The DeltaG(o) for the intercalation binding model was driven by a large TDeltaS(o) term and complemented by a small DeltaH(o) term. One of the main stabilizing components of the external binding model is the energy of solvation, which favors the external model over the intercalation model by -67.94 kcal/mol. Finally, we propose that intercalative binding, although less favored than external binding, may occur, but because of the nature of the intercalative binding, it is invisible to the photocleavage assay. This study provides the first experimental insight into how selectivity might be achieved for different G-quadruplexes by using structural variants within a single group of G-quadruplex-interactive drugs.