Improved thermodynamic parameters for prediction of RNA duplex formation are derived from optical melting studies of 90 oligoribonucleotide duplexes containing only Watson-Crick base pairs. To test end or base composition effects, new sets of duplexes are included that have identical nearest neighbors, but different base compositions and therefore different ends. Duplexes with terminal GC pairs are more stable than duplexes with the same nearest neighbors but terminal AU pairs. Penalizing terminal AU base pairs by 0.45 kcal/mol relative to terminal GC base pairs significantly improves predictions of ∆G°3 7 from a nearest-neighbor model. A physical model is suggested in which the differential treatment of AU and GC ends accounts for the dependence of the total number of Watson-Crick hydrogen bonds on the base composition of a duplex. On average, the new parameters predict ∆G°3 7 , ∆H°, ∆S°, and T M within 3.2%, 6.0%, 6.8%, and 1.3°C, respectively. These predictions are within the limit of the model, based on experimental results for duplexes predicted to have identical thermodynamic parameters.The thermodynamics of secondary structure formation are important for unraveling structure-function relationships for RNA. For example, these thermodynamics provide a foundation for predicting secondary structure and stability, both of which can correlate with function. Moreover, predicting secondary structure is a crucial intermediate step toward predicting three-dimensional structure (1, 2). In addition, differences between the thermodynamics of secondary structure formation and of overall folding can provide insight into the thermodynamics of tertiary structure formation (3-7).Watson-Crick base pairs are one of the most important motifs in RNA secondary structures. The thermodynamics of Watson-Crick base pair formation have been studied in short RNA duplexes (8, 9). The results are well-represented by a nearest-neighbor model in which the thermodynamic stability of a base pair is dependent on the identity of the adjacent base pairs. This model has been termed an individual nearest-neighbor (INN) model (10, 11). The pioneering implementation by Borer et al. (8) employed 6 nearest-neighbor parameters and separate initiation parameters for duplexes with and without a GC base pair. Due to advances in oligoribonucleotide synthesis (12), Freier et al. (9) were able to determine all 10 nearest-neighbor parameters in the INN model and the initiation parameter for duplexes with at least one GC base pair. The initiation parameter for duplexes with only AU base pairs was not determined.It has been suggested that a nearest-neighbor model that treats terminal base pairs differently from internal base pairs (8) or treats terminal GC base pairs differently from terminal AU base pairs (10, 11, 13) may improve modeling of duplex stability. The model proposed by Gray (10) has been termed an independent short sequence (ISS) model because the 14 sequence-dependent parameters of the model must be combined into 12 "short sequence" p...

Thermodynamic parameters for prediction of RNA duplex stability are reported. One parameter for duplex initiation and 10 parameters for helix propagation are derived from enthalpy and free-energy changes for helix formation by 45 RNA oligonucleotide duplexes. The oligomer sequences were chosen to maximize reliability of secondary structure predictions. Each of the 10 nearest-neighbor sequences is wellrepresented among the 45 oligonucleotides, and the sequences were chosen to minimize experimental errors in AGI at 37°C. These parameters predict melting temperatures of most oligonucleotide duplexes within 5°C. This is about as good as can be expected from the nearest-neighbor model. Free-energy changes for helix propagation at dangling ends, terminal mismatches, and internal G-U mismatches, and free-energy changes for helix initiation at hairpin loops, internal loops, or internal bulges are also tabulated.Stabilities of RNA duplexes and secondary structures of RNAs are often predicted by using free-energy parameters from a nearest-neighbor model (1-5). Sometimes, however, predictions are inconsistent with experimental data (6-10). One factor hindering successful predictions is that the reliability of parameters was limited by the availability of model oligonucleotides (2). Recent breakthroughs in synthesis of RNA oligoribonucleotides (11-16) permit design of oligonucleotides to provide improved parameters. This paper presents thermodynamic parameters derived from data on 45 complementary RNA duplexes. The parameters are able to predict the stabilities of RNA duplexes within the limits ofthe nearest-neighbor model. MATERIALS AND METHODSChoice of Sequences. Sequences were selected to minimize errors in the free-energy change for duplex formation at 37°C, AG97 (17,18). Thus, as much as possible, melting temperatures at 0.1 mM are near 37°C to minimize extrapolation. The oligomers were also chosen to independently represent all 10 nearest-neighbor sequences comprising Watson-Crick base pairs.Oligonucleotide Synthesis. Oligonucleotides not reported elsewhere were synthesized on solid support using phosphoramidite procedures and purified as described (11,19). Purities were confirmed by high-performance liquid chromatography for all oligomers.Thermodynamic Parameters. Absorbance vs. temperature melting curves were measured in 1 M NaCl/0.005 M Na2HPO4/0.5 mM EDTA (disodium salt), pH 7, as described (11). Concentrations were determined from the high-temperature absorbance using extinction coefficients calculated as described (20). In units of 0.1 mM-1 cm-1, calculated hightemperature extinction coefficients at 280 nm not reported elsewhere are as follows: GUGCAC, 2.77; GUCUAGAC, 3.66; GAUAUAUC, 3.05; GUAUAUAC, 3.00. Thermodynamic parameters of helix formation were obtained by two methods. (t) Individual melting curves were fit to a two-state model with sloping baselines and the enthalpy and entropy changes derived from the fits were averaged (21), and (ii) reciprocal melting temperature, tm-1, vs. log (CT) was plot...

Molecular dynamics (MD) simulations for RNA tetramers r(AAAA), r(CAAU), r(GACC), and r(UUUU) are benchmarked against 1H–1H NOESY distances and 3J scalar couplings to test effects of RNA torsion parametrizations. Four different starting structures were used for r(AAAA), r(CAAU), and r(GACC), while five starting structures were used for r(UUUU). On the basis of X-ray structures, criteria are reported for quantifying stacking. The force fields, AMBER ff99, parmbsc0, parm99χ_Yil, ff10, and parmTor, all predict experimentally unobserved stacks and intercalations, e.g., base 1 stacked between bases 3 and 4, and incorrect χ, ϵ, and sugar pucker populations. The intercalated structures are particularly stable, often lasting several microseconds. Parmbsc0, parm99χ_Yil, and ff10 give similar agreement with NMR, but the best agreement is only 46%. Experimentally unobserved intercalations typically are associated with reduced solvent accessible surface area along with amino and hydroxyl hydrogen bonds to phosphate nonbridging oxygens. Results from an extensive set of MD simulations suggest that recent force field parametrizations improve predictions, but further improvements are necessary to provide reasonable agreement with NMR. In particular, intramolecular stacking and hydrogen bonding interactions may not be well balanced with the TIP3P water model. NMR data and the scoring method presented here provide rigorous benchmarks for future changes in force fields and MD methods.