Nucleic acid reactivity: challenges for next-generation semiempirical quantum models.

Area of Science:

• Computational Chemistry
• Biochemistry
• Quantum Mechanics

Background:

• Semiempirical quantum models are crucial for studying RNA catalysis and phosphoryl transfer reactions.
• Combined quantum mechanical (QM)/molecular mechanical (MM) methods are widely used in this field.
• Accurate models are needed to understand nucleic acid structure and reactivity.

Purpose of the Study:

• To broadly assess the performance of existing semiempirical quantum models for nucleic acid structure and reactivity.
• To identify limitations of current models and guide the development of next-generation quantum models.
• To quantify the accuracy of semiempirical models against high-level QM benchmark calculations.

Main Methods:

• Evaluation of neglect of diatomic differential overlap (NDDO) and self-consistent density-functional tight-binding (DFTB) models.
• Comparison against high-level QM benchmark calculations for seven biologically relevant datasets.
• Benchmarking against density-functional models like M062X and B3LYP.

Main Results:

• AM1/d-PhoT demonstrated the most robust proton affinity predictions.
• AM1/d-PhoT and DFTB3-3ob/OPhyd reasonably reproduced RNA phosphoryl transfer model reactions.
• Semiempirical models underestimated polarizabilities and inadequately described torsion profiles and phosphorus modeling.

Conclusions:

• Current semiempirical models have significant limitations in accurately describing nucleic acid structure and reactivity.
• Next-generation quantum models should focus on improving relative conformational energies, barriers, and nonbonded interactions.
• Further development is needed for accurate modeling of pentavalent phosphorus, especially with thio substitutions.