Determining the structural amino acid attributes which are important in both protein thermostability and alkalophilicity: a case study on xylanase

Xylanases are used in the recycling of biomass and have other industrial applications including pulp bleaching. These enzymes are also applied in the baking industry and for the manufacture of animal feed. Such technologies as, for example, pulp bleaching entail high temperatures and high pHs. As a result, there is great demand from industry for thermostable and halostable forms of xylanase. Due to the relatively high variation in the thermo- and halo-stability of xylanases, feature selection was employed as a model to discover the important attributes of their amino acid sequences affecting the thermo- and halo-stability of the enzyme. A data set containing the amino acid sequences of xylanases with different thermo- and halostabilities was collected. Seventy-four amino acid attributes were obtained for each enzyme sequence. After running a feature selection algorithm for each of the thermo- and halostablity variables, features were classified as either important, unimportant or marginal. The results showed a significant correlation between structural amino acid attitudes and stability in harsh temperatures or alkaline conditions. Features such as lysine, glutamic acid, and positively/negatively charged residues showed a positive correlation with both the thermostability and alkalophilicity attributes of the protein. For the first time, we found attributes which were important for both stability at high temperatures as well as in alkaline conditions by mining sequence-derived amino acid attributes using data mining.