Structure-activity relationship study of antioxidative peptides by QSAR modeling: the amino acid next to C-terminus affects the activity.

Screening, isolation and in vitro assays have been used for characterization of antioxidative peptides derived from food proteins, and incompatible deductions of structural characteristics derived from the isolated peptides have been brought forward. However, there is still little information concerning the structure-activity relationship of antioxidative peptides. QSAR modeling was performed, respectively, on synthetic tripeptides and tetrapeptides related to LLPHH. According to cumulative squared multiple correlation coefficients (R(2)), cumulative cross-validation coefficients (Q(2)) and relative standard deviation for calibration set (RSD(c)), two credible models for tripeptide and tetrapeptide databases, respectively, have been built with partial least squares (PLS) regression (R(2) for models of tripeptide and tetrapeptide are 0.744 and 0.943, Q(2) are 0.631 and 0.414, and RSD(c) are 0.323 and 0.111, respectively). Meanwhile, according to the cumulative multiple correlation coefficient for the predictive set (R2(ext)) and the relative standard deviation for the predictive set (RSD(p)), the predictive ability of the model for tripeptides also is excellent (R2(ext)) and RSD(p) are 0.719 and 0.450, respectively). Hydrogen bond property and hydrophilicity of the amino acid residue next to the C-terminus, and the hydrophobicity as well as electronic propertyof the N-terminus are more significant; meanwhile, the electronic property of the C-terminus is beneficial for antioxidant activity. The structural characteristics we found are very useful in understanding and predicting the peptide structures responsible for activity and development of functional foods with peptides as active compounds, or antioxidative peptides as alternatives to other antioxidants.