Pairwise amino acid secondary structural propensities

We investigate the propensities for amino acids to form a specific secondary structure when they are paired with other amino acids. Our investigations use molecular dynamics (MD) computer simulations, and we compare the results to those from the Protein Data Bank (PDB). Proper comparison requires weighting of the MD results in a manner consistent with the relative frequency of appearance in the PDB of each possible pair of amino acids. We find that the propensity for an amino acid to assume a secondary structure varies dramatically depending on the amino acid that is before or after it in the primary sequence. This cooperative effect means that when selecting amino acids to facilitate the formation of a secondary structure in peptide engineering experiments, the adjacent amino acids must be considered.

Figure 1

Cumulative Ramachandran plots for the 20 amino acids. The cumulative plot for each amino acid is generated by combining the 40 different Ramachandran plots that an amino acid will appear in out of all possible X-Z permutations of the sequence GGXZGG. Darker shading represents higher relative probability. Relative probabilities are color coded (online) in the following manner: $0\le \mathrm{white}<0.01, 0.01\le \mathrm{gray}<0.1, 0.1\le \mathrm{green}<0.4, 0.4\le \mathrm{blue}<0.7, 0.7\le \mathrm{red}<0.9, 0.9\le \mathrm{black}$.

Figure 2

Cooperative secondary structure propensities from the MD simulations: the probability for each amino acid to have (Φ,Ψ) consistent with either α-helix structure (left bar for each amino acid), β-structure (middle bar), or neither (coil/loop; right bar), averaged over all 40 possible X-Z pairs in which an amino acid can appear. The height of a bar represents the average. The dispersion for each probability is calculated from the 40 different amino acid pairs that are combined to calculate each average. For almost all probabilities, the dispersion is almost as large as the size of the average probability, showing that the propensity for any amino acid to assume a specific secondary structure depends strongly on the amino acid that is next to it in the primary sequence.

Figure 3

Comparison of PDB data with MD simulations normalized with the weighting factors for amino acid pairs based upon their appearance in the PDB data. Darker shading represents higher relative probability.