Prediction of DNA duplex stability and thermodynamics is invaluable for many molecular biology applications involving sequence dependent hybridization reactions. Sequence dependent stability of duplex DNA plays a major role in fundamental processes of the living cell, such as replication, transcription, and recombination.
The analytical methods for calculation of DNA duplex thermodynamics from sequence and DNA concentration are based on the nearest-neighbor (n-n) model. There are 10 n-n base pair doublets in DNA duplexes. However, when the ends are taken into account, 12 n-n parameters are required to describe the thermodynamics of melting of short duplex DNAs in terms of the n-n model.
At least 11 sets of n-n sequence dependent thermodynamic parameters for DNA have been published. These sets were compared in their ability to predict free energies and melting temperatures of linear DNA duplexes 10 base pair long. Some of the sequences showing the biggest discrepancies in predictions from different sets of n-n parameters were investigated experimentally. The n-n set published from our lab, Doktycz et al. (1992), and two recently reported sets of n-n thermodynamic parameters (Allawi and SantaLucia, 1997; Sugimoto et al., 1996) provided more accurate predictions of experimental melting temperatures of short duplex DNA oligomers, than the older, more commonly used set of n-n parameters (Breslauer et al., 1986). However, none of the n-n sets was able to predict melting temperatures of several selected 10 base pair sequences that displayed the most discordance between predictions of the various n-n sets. These and other observations suggested that the general characterization of DNA thermodynamic stability in terms of a n-n model may be inadequate, and significant sequence dependent interactions in DNA duplexes may extend over distances beyond nearest-neighbors. To investigate such sequence dependent interactions and their dependence on length and salt, we performed UV-melting studies on 39 DNA dumbbells. We employed DNA dumbbells because they offer many advantages for studies of sequence dependent effects. Since DNA dumbbells consist of a duplex stem effectively crosslinked on both ends by single strand loops, the melting process of the dumbbell stem is unimolecular and concentration independent. The end-loops stabilize a dumbbell stem compared to the same sequence without loops. For these reasons the dumbbell system provides a more realistic mimic of short sequences in a long DNA environment, without the concentration dependence and other anomalies associated with melting of short duplex DNAs. Dumbbells used in the melting studies had constant end sequences and variable central sequences of four to eight base pair long. From melting data of 39 dumbbells in buffers containing total sodium ion concentrations of 25, 55, 85, 115 mM, nearest-neighbor and next-nearest-neighbor (triplet) interactions in duplex DNA were evaluated. Rigorous statistical analysis revealed that melting data of dumbbells in 85 and 115 mM Na+ can be adequately described by the n-n model. However, melting data of dumbbells in 25 and 55 mM Na+ cannot be adequately fitted in terms of the nearest-neighbor model within the errors of the measurements. If next-nearest-neighbor (triplet) interactions are considered, a reasonable fit of the melting data is obtained even at these low Na+ concentrations. This indicates that next-nearest-neighbor (n-n-n) interactions in duplex DNAs are significant in solutions below 55 mM Na+. To test the evaluated nearest-neighbor and next-nearest-neighbor parameters, two additional DNA dumbbell molecules were prepared and melted in 25 mM Na+. When melting temperatures predicted with the n-n and n-n-n parameters were compared to experimental melting temperatures, more accurate predictions of melting temperatures of the dumbbells were obtained when the next-nearest-neighbor parameters were employed.
A correction for the nucleation enthalpy that is required to apply the evaluated n-n parameters to predict thermodynamics of short linear DNA duplex oligomers, was evaluated in 115 mM Na+. The correction for the nucleation enthalpy depends on percentage of GC base pairs and length of the linear DNA duplex.