Back to JDK's publications.

Biochemistry, 43 (18), 5388 -5405, 2004.

DOI: 10.1021/bi035976d

Sequence-Dependent Thermodynamic Parameters for Locked Nucleic Acid (LNA)-DNA Duplex Formation

Patricia M. McTigue,

Jason D. Kahn*

Department of Chemistry and Biochemistry, University of Maryland, College Park, College Park, Maryland 20742-2021, and Celadon Laboratories, Inc., Technology Growth Center, 6525 Belcrest Road, Suite 500, Hyattsville, Maryland 20782

Received November 5, 2003

Revised Manuscript Received February 26, 2004

Abstract:

The design of modified nucleic acid probes, primers, and therapeutics is improved by considering their thermodynamics. Locked nucleic acid (LNA) is one of the most useful modified backbones, with incorporation of a single LNA providing a substantial increase in duplex stability. In this work, the hybridization H, S, and melting temperature (T_M) were measured from absorbance melting curves for 100 duplex oligonucleotides with single internal LNA nucleotides on one strand, and the results provided H, S, , and T_M relative to reference DNA oligonucleotides. LNA pyrimidines contribute more stability than purines, especially A^L, but there is substantial context dependence for each LNA base. Both the 5' and 3' neighbors must be considered in predicting the effect of an LNA incorporation, with purine neighbors providing more stability. Enthalpy-entropy compensation in H and S is observed across the set of sequences, suggesting that LNA can stabilize the duplex by either preorganization or improved stacking, but not both simultaneously. Singular value decomposition analysis provides predictive sequence-dependent rules for hybridization of singly LNA-substituted DNA oligonucleotides to their all-DNA complements. The results are provided as sets of H, S, and parameters for all 32 of the possible nearest neighbors for LNA+DNA:DNA hybridization (5' MX^L and 5' X^LN, where M, N, and X = A, C, G, or T and X^L represents LNA). The parameters are applicable within the standard thermodynamic prediction algorithms. They provide T_M estimates accurate to within 2 C for LNA-containing oligonucleotides, which is significantly better accuracy than previously available.