Mapeamento de epítopos e desimunização de uma L-asparaginase de Bacillus subtilis por meio de algoritmo genético

Abstract

L-asparaginase is a therapeutic protein used in the treatment of acute lymphoblastic leukemia, and its use may result in neutralization by the immune system or even produce hypersensitivity reactions. Components of adaptive immunity such as MHC-II (Major Histocompatibility Complex II) are responsible for presenting exogenous peptides processed by antigen-presenting cells. Immunoinformatics tools can be used to predict the binding affinity of a peptide with the MHC-II cleft to localize epitopes, as well as perform epitope deletion deimmunization. These technologies may be employed in the generation of biobetters, however, the predictive tools available do not allow the identification of promiscuous amino acids. The aim of this work was to develop computational tools for protein deimmunization, improve an in silico immunogenicity prediction tool, construct a protein deimmunization genetic algorithm (GA) and later apply these tools to identify and remove the most immunogenic epitopes of a Bacillus subtilis L-asparaginase sequence. The TEPITOPE peptide-MHC-II affinity prediction method had its performance improved by summation of immunogenicity scores of overlapping epitopes and smoothing of these data through a spline. This protocol was applied to identify immunogenic regions in a Bacillus subtilis L-asparaginase (BsA) and this information was used by a protein deimmunization genetic algorithm. The algorithm uses a set of 100 BsA sequences and performs recombination and mutation operations at each cycle only in regions whose immunogenicity scores are among the top 20% of the protein and considering only neutral mutations (|DDG| between wild-type and mutant proteins is less than 0.5 kcal/mol), substituting sequences of higher immunogenicity for those with lower ones. At the end of execution cycles, the algorithm was able to, with only three mutations, generate a protein whose immunogenicity is 63.5 % lower than that of the original sequence, having two of the most immunogenic epitopes removed, maintaining structure and stability intact (RMSD=0.01Å and DDG = -0.2935 kcal/mol)). The protein deimmunization genetic algorithm allows parameterization and implementation of new immunogenicity and stability assessments that may increase its efficiency. The smoothing prediction method, the epitope visualization script and the genetic algorithm can be applied in an automated manner to a large number of amino acid chains aiding protein engineering projects.