PhD Lecture by Ilka Hoof, CBS

Prediction and analysis of MHC class I binding specificities beyond humans

Thursday June 4, 2009 at 14:00
CBS, DTU, Lyngby, Building 208, Auditorium 062
Assessment Committee:    Associate professor Susanne Brix Pedersen, DTU (Chairman)
Professor Anders Krogh, KU
General director Ronald Edward Bontrop, BPRC
Chair of defense:Associate professor Morten Nielsen, DTU
Supervisor:Professor Ole Lund, DTU

The major histocompatibility complex (MHC) plays a crucial role in the generation of adaptive immune responses in vertebrate species. MHC class I molecules, in humans known as human leukocyte antigen (HLA) class I molecules, present intracellular pathogen-derived protein fragments to CD8+ T cells. Even though MHC molecules have been shown to have a major impact on disease progression and outcome, binding specificities have been characterized for only a minor fraction (around 5%) of the huge set of known HLA molecules. Nonhuman primates serve as important model organisms for studies of infectious diseases in humans, such as those caused by human immunodeficiency virus (HIV) and hepatitis C virus. Knowledge of the binding specificity of nonhuman primate MHC class I molecules may facilitate a more comprehensive evaluation of vaccine studies in these animals, but the binding specificity of only a small number of nonhuman primate MHC molecules has been characterized. The focus of my PhD project has been on improving methods to predict the binding specificity of MHC class I molecules and on comparing the (predicted) peptide binding repertoire between species and between individuals of the same species. In this thesis, I present an updated version of the pan-specific neural network based prediction method NetMHCpan. By considering both the amino acid sequence of the ligand and the polymorphic contact environment of the MHC molecule, the method is able to extrapolate to MHC molecules it has not encountered in the training process. This allows for the prediction of the binding specificity of any MHC class I molecule for which the full-length protein sequence is available. One of the major achievements of NetMHCpan is its applicability beyond human MHC molecules, as demonstrated by its remarkable prediction performance for chimpanzee and macaque MHC class I molecules. In order to make the NetMHCpan predictions easily accessible, we established the MHC motif viewer website, which allows any user to browse the (predicted) binding specificity space of human, chimpanzee and macaque MHC class I molecules. I present the results of a study, in which I analyzed the CD8+ T cell response data of an HIV-1 infected patient cohort. A group of HIV-1 infected patients had been tested for CD8+ T cell responses against a set of potential HIV-1 epitopes. Given the HLA type and the autologous virus sequence for each patient, I was able to compare HLA-matched immune responses against tested and autologous epitope sequences from 1517 patient-peptide pairs and, thereby, to quantify how genetic variation within the epitopes influenced the ability of the T cell receptor (TCR) to recognize an epitope. Epitopes recognized in ELISPOT studies were found to be significantly more similar to the autologous virus than those that did not elicit a CD8+ T cell response, suggesting that peptide similarity might be a major determinant of cross-recognition by the TCR. Further, I present the results of two studies, in which I applied NetMHCpan to establish the binding specificity of nonhuman primate MHC class I molecules. Using these predictions, I compared the peptide-binding repertoire of chimpanzee MHC molecules and HLA molecules associated with slow progression to AIDS in HIV-1 infected humans. Like human long-term nonprogressors, HIV-1 infected chimpanzees are relatively resistant to the development of AIDS. Interestingly, our results show that protective HLA molecules and chimpanzee MHC molecules share peptide binding repertoires as opposed to HLA molecules that are associated with fast disease progression, implying that the MHC repertoire of chimpanzees may have been formed under selective pressure induced by an HIV-1 like virus. The aim of the second study was to identify potential HLA analogs among Chinese rhesus macaque MHC class I alleles. This was achieved by predicting the peptide binding specificity for a large set of HLA alleles and for a set of available Chinese rhesus macaque alleles. Potential HLA analogs were identified by clustering these alleles with respect to their binding specificity. In summary, the results of this thesis demonstrate the capability of NetMHCpan to contribute to immunological studies, which require allele-specific detail of the peptide-binding repertoire of MHC class I molecules, with the potential to go beyond species borders.

Everybody is welcome. Registration is not necessary.