Specificity and stringency in DNA triplex formation.

Triple-helix formation can in principle serve as a general method for sequence-specific recognition and physical separation of duplex DNA molecules. Realization of this goal depends on how much the triplex is destabilized by mismatches and other defects (specificity) and on finding conditions in which perfect complexes are stable and defect complexes are not (stringency). We have addressed the question of specificity by determining the difference in free energy between perfect and defect complexes by using UV melting curves and equilibrium competition experiments. We find that third strands that bind with either single-base bulges or single mismatches are destabilized relative to the perfect triplex by 2.5-2.9 and 3.2-4.0 kcal/mol (1 cal = 4.184 J), respectively, essentially equivalent to the corresponding values determined for duplex DNA and RNA. Also, we present a method, referred to as stringency clamping, which maintains specific binding under conditions far from normal stringency. To do this, we provide for the formation of a competing structure involving the third strand with stability between that of the perfect and imperfect complexes; the competitive interaction effectively prevents triplex formation at imperfect sites even far below their melting temperature. We illustrate the phenomenon with three different stringency clamps, two of which compete for the all-pyrimidine third strand through Watson-Crick pairing and one that competes through all-pyrimidine pairing at acidic pH. We demonstrate physical separation of two duplex DNA molecules differing by a single base pair in their target sequence for triple-helix formation.