High-resolution, three-dimensional modeling of human leukocyte antigen class I structure and surface electrostatic potential reveals the molecular basis for alloantibody binding epitopes

Abstract

The potential of human leukocyte antigens (HLA) to stimulate humoral alloimmunity depends on the orientation, accessibility and physiochemical properties of polymorphic amino acids. We have generated high-resolution structural and physiochemical models of all common HLA class I alleles and analyzed the impact of amino acid polymorphisms on surface electrostatic potential. Atomic resolution three-dimensional structural models of HLA class I molecules were generated using the MODELLER computer algorithm. The molecular surface electrostatic potential was calculated using the DelPhi program. To confirm that electrostatic surface topography reflects known HLA B cell epitopes, we examined Bw4 and Bw6 and ascertained the impact of amino acid polymorphisms on their tertiary and physiochemical composition. The HLA protein structures generated performed well when subjected to stereochemical and energy-based testing for structural integrity. The electrostatic pattern and conformation of Bw4 and Bw6 epitopes are maintained among HLA molecules even when expressed in a different structural context. Importantly, variation in epitope amino acid composition does not always translate into a different electrostatic motif, providing an explanation for serologic cross-reactivity. Mutations of critical amino acids that abrogate antibody binding also induce distinct changes in epitope electrostatic properties. In conclusion, high-resolution structural modeling provides a physiochemical explanation for serologic patterns of antibody binding and provides novel insights into HLA immunogenicity.

Introduction

Elucidation of the crystallographic structure of human leukocyte antigen (HLA) class I and II more than 20 years ago contributed immeasurably to understanding of the relationship between structure and function of HLA [1], [2]. Since then, advances in molecular sequencing technology have enabled resolution at the amino acid sequence level of more than 2000 different HLA alleles and this has led to better understanding of the role of HLA in health and disease, and in the field of tissue transplantation has allowed improved definition of HLA incompatibilities between transplant donors and recipients.

Knowledge of the amino acid sequence of an HLA allele allows insight into its potential peptide binding repertoire and in the context of transplantation, comparison of the sequences of different HLA alleles enables prediction of immunogenicity of a particular HLA mismatch [3], [4], [5]. Amino acid sequence alone, although it is useful, provides limited insight into the molecular basis of protein–protein interactions that are mediated in large part by electrostatic properties; [6], [7], [8] electrostatic forces are particularly important determinants of the specificity and affinity of alloantibody binding [9], [10]. The electrostatic properties of HLA are determined by the number and distribution of polar and charged amino acid residues, and integration of this information in structural models of HLA combined with application of electrostatic theory to biologic macromolecules enable the three-dimensional (3D) topographic distribution of electrostatic potential to be determined [11], [12].

We describe here the use of comparative protein structure modeling, based on available crystallographic data and amino acid sequence data, to generate high-resolution structural and physiochemical models of all the common HLA class I alleles. The structural HLA class I models generated allowed the impact of single and multiple amino acid polymorphisms on surface electrostatic potential to be visualized and this provided an explanation at the molecular level for alloantibody binding patterns to known HLA epitopes, as well as to epitopes not previously explicable by amino acid sequence alone.