Increased accuracy of vibrational circular dichroism calculations for isotopically labeled helical peptides

Abstract

Vibrational circular dichroism (VCD) spectra have been computed with qualitatively correct sign patterns for α-helical peptides using various methods, ranging from empirical models to ab initio quantum mechanical computations. However, some details, such as deuteration effects and isotope substitution shifts and sign patterns for the resultant amide I' band shape, have remained a predictive challenge. Fully optimized computations for a 25-residue Ala-rich peptide, including implicit solvent corrections and explicit side chains that experimentally stabilize these model helical peptides in water, have been carried out using density functional theory (DFT). These fully minimized structures show minor changes in the (ϕ,ψ) torsions at the termini and yield an extra negative band to the low energy side of the characteristic amide I' couplet VCD, in agreement with experiments. Additionally, these calculations give the right sign and relative intensity patterns, as compared to experimental results, for several ¹³C=O substituted variants. The differences from previously reported computations that used ideal helical structures and vacuum conditions imply that inclusion of distorted termini and solvent effects can have an impact on the final detailed spectral patterns. Inclusion of side chains in these calculations had very little effect on the computed amide I' IR and VCD. Tests of constrained geometries, varying dielectric, and different functionals indicate that each can affect the band shapes, particularly for the ¹²C=O components, but these aspects do not fully explain the difference from previous spectral simulations. Inclusion of long-range amide coupling, as obtained from DFT computation of the full structure, or transfer of parameters from a somewhat longer peptide model, rather than shorter model, seems to be more important for the final detailed band shape under isotopic substitution. However, these corrections can also induce other changes, suggesting that previously reported, limited calculations may have been qualitatively useful due to a balance of errors. This may also explain the success of simple empirical IR models.