Importance of oligoelectrolyte end effects for the thermodynamics of conformational transitions of nucleic acid oligomers: A grand canonical Monte Carlo analysis

Abstract

Effects of salt concentration on the stabilities of oligonucleotide helices are analyzed directly in terms of ΔΓ_N→yN ≡ Γ − Γ_N^nat, the difference in the salt–nucleotide phosphate preferential interaction coefficients for the denatured state, having yN phosphate charges, and for the native state, having N phosphate charges (y = 1 for hairpin denaturation and y = 0.5 for dimer denaturation). Previous experimental studies of the denaturation of hairpin oligo-nucleotides (having 18 < N < 44) indicate significant differences between ΔΓ_N→N and ΔΓ_∞, the value determined for the denaturation of the corresponding polynucleotide. These differences are thermodynamic manifestations of the oligoelectrolyte end effect. In contrast, the available data on the denaturation of oligonucleotide dimer helices (N ⩽ 22) imply that differences between ΔΓ_∞ and ΔΓ_N→0.5N, and hence oligoelectrolyte end effects, are small or negligible. To determine the origin of these apparently conflicting implications concerning the importance of oligoelectrolyte end effects, we have calculated the N dependence of Γ_N from grand canonical Monte Carlo simulations for an idealized model of the structure and charge distribution of each oligomer conformation. Our calculations are in quantitative agreement with the experimental finding for d(TA) hairpin oligomers that − ΔΓ_N→N decreases linearly as N⁻¹ increases, and with the extant experimental determinations of ΔΓ_N→0.5N. These results provide an illustration of how the large electrostatic end effects exhibited by the hairpin denaturation data are masked when ΔΓ_∞ is compared with values of ΔΓ_N→0.5N for short dimer helices (N ⩽ 22). For 0.5N > 24, − ΔΓ_N→0.5N is predicted to be a linear function of N⁻¹ whose slope has the opposite sign from, and is more salt-concentration dependent than, the corresponding slope of − ΔΓ_N→N as a function of N⁻¹. Our calculations also yield predictions about the N dependences of the individual values of Γ_N that can be tested by determining Donnan coefficients from membrane dialysis equilibrium experiments. For long enough hairpin and dimer oligonucleotides (yN ⩾ 24), in either native or denatured forms, we predict that the (positive) difference Γ_∞ − Γ_N increases linearly with increasing N⁻¹. For smaller values of N the difference Γ_∞ − Γ_N continues to increase with increasing N⁻¹.