POCKETCHECK: a comprehensive contact calculation for distinct peptide-HLA structures

Abstract

Data quality of diffraction experiments depends on several factors: a) The diffractometer and the area detector hardware, b) the sample, c) the experimental procedure and d) the data reduction approach and software.The talk will highlight key aspects of each of these factors. The hardware revolves around the notions of absolute detectivity, overhead, minimizing systematic errors and diffractometer access.The sample choice, mounting, protection environment is controlled within reason by the user.The experimental procedure comprises the choice of wavelength, the geometric strategy, the mode of scan and detector operation and the decision on absolute detectivity vs. redundancy.The data reduction software has to be optimized at extracting consistently area detector data not only under good conditions, but also under real life flaws of the practical experimental procedure. There have been several attempts over the years to define which positions in the HLA binding groove (pockets) influence the specificity of bound amino acids at each position in the peptide. The structural determination of the HLA molecule by X-ray crystallography has provided valuable information for understanding how peptides bind to HLA. Originally, six pockets (A-F) were defined by calculating the surface of the binding groove based upon the crystal structure of HLA-A2 [1]. Since then, x-ray crystallography has been performed for a variety of HLA alleles bound to a range of peptides, which has lead to broader pocket definitions [2]. KeywordsSeveral studies performed peptide sequencing for allelic variants to understand the magnitude of certain mismatches on peptide specificity. We have previously described the ability of distinct HLA variants to present peptides of >10 amino acids in length [3]. In these cases it is especially important to be able to define which positions within the HLA binding cleft are in contact with a given peptide and thus influence the sequence of the peptides selected. The knowledge about individual peptide features allowed for the crystallographic analysis of selected pMHC complexes.The Protein data bank was searched for all deposited structures of peptide:HLA-complexes http://rcsb.org/pdb and those were submitted to the contact map analysis webserver http://ligin.weizmann.ac.il/cma. The output html was piped into an InterSystems Caché "post-relational" database allowing objectoriented data storage. Tabular data with residue position, amino acid and buried surface that contact a particular peptide position were compared for >100 HLA class I structures.These new dynamic definitions increase the precision of peptide prediction and support the characterization of individual weights for individual amino acids. This knowledge facilitates a rating of the allogenicity of mismatches and will be a further step towards intelligent HLA mismatching. Despite the success of macromolecular crystallography (MX) and the enormous advances in the field, one of the major hurdles still to be surmounted is the availability of well-diffractin...