Infectious diseases, imposing density-dependent mortality on MHC/HLA variation, can account for balancing selection and MHC/HLA polymorphism

TL;DR

This paper presents a model where infectious diseases impose density-dependent regulation on HLA haplotypes, explaining the extensive polymorphism and balancing selection observed in the human MHC system.

Contribution

The study introduces a novel model linking pathogen-driven density regulation to HLA haplotype diversity and balancing selection, supported by empirical HLA data.

Findings

High-frequency haplotypes show positive selection.

HLA alleles exhibit positive epistasis and pleiotropy.

HLA haplotypes form a network with quasi-species properties.

Abstract

The human MHC transplantation loci (HLA-A, -B, -C, -DPB1, -DQB1, -DRB1) are the most polymorphic in the human genome. It is generally accepted this polymorphism reflects a role in presenting pathogen-derived peptide to the adaptive immune system. Proposed mechanisms for the polymorphism such as negative frequency-dependent selection (NFDS) and heterozygote advantage (HA) focus on HLA alleles, not haplotypes. Here, we propose a model for the polymorphism in which infectious diseases impose independent density-dependent regulation on HLA haplotypes. More specifically, a complex pathogen environment drives extensive host polymorphism through a guild of HLA haplotypes that are specialised and show incomplete peptide recognition. Separation of haplotype guilds is maintained by limiting similarity. The outcome is a wide and stable range of haplotype densities at steady-state in which effective Fisher fitnesses are zero. Densities, and therefore frequencies, emerge theoretically as alternative measures of fitness. A catalogue of ranked frequencies is therefore one of ranked fitnesses. The model is supported by data from a range of sources including a Caucasian HLA dataset compiled by the US National Marrow Donor Program (NMDP). These provide evidence of positive selection on the top 350-2000 5-locus HLA haplotypes taken from an overall NMDP sample set of 10E5. High-fitness haplotypes drive the selection of 137 high-frequency alleles spread across the 5 HLA loci under consideration. These alleles demonstrate positive epistasis and pleiotropy in the formation of haplotypes. Allelic pleiotropy creates a network of highly inter-related HLA haplotypes that account for 97% of the census sample. We suggest this network has properties of a quasi-species and is itself under selection. We also suggest this is the origin of balancing selection in the HLA system.