Diversification of CD1 Molecules Shapes Lipid Antigen Selectivity

Abstract

Molecular studies of host-pathogen evolution have largely focused on the consequences of variation at protein-protein interaction surfaces. The potential for other microbe-associated macromolecules to promote arms race dynamics with host factors remains unclear. The cluster of differentiation 1 (CD1) family of vertebrate cell surface receptors plays a crucial role in adaptive immunity through binding and presentation of lipid antigens to T-cells. Although CD1 proteins present a variety of endogenous and microbial lipids to various T-cell types, they are less diverse within vertebrate populations than the related major histocompatibility complex (MHC) molecules. We discovered that CD1 genes exhibit a high level of divergence between simian primate species, altering predicted lipid-binding properties and T-cell receptor interactions. These findings suggest that lipid-protein conflicts have shaped CD1 genetic variation during primate evolution. Consistent with this hypothesis, multiple primate CD1 family proteins exhibit signatures of repeated positive selection at surfaces impacting antigen presentation, binding pocket morphology, and T-cell receptor accessibility. Using a molecular modeling approach, we observe that interspecies variation as well as single mutations at rapidly-evolving sites in CD1a drastically alter predicted lipid binding and structural features of the T-cell recognition surface. We further show that alterations in both endogenous and microbial lipid-binding affinities influence the ability of CD1a to undergo antigen swapping required for T-cell activation. Together these findings establish lipid-protein interactions as a critical force of host-pathogen conflict and inform potential strategies for lipid-based vaccine development.

Keywords: adaptive immunity; antigen presentation; positive selection; protein evolution.

Fig. 1.

Diversity of the CD1 gene family in primates. (A) Ribbon diagrams of human CD1a (PDB ID: 5J1A) and MHC (PDB ID: 1AKJ) with alpha helices highlighted in red, beta-sheet in purple, loops in yellow. (B) Illustration of the CD1-TCR interaction where CD1 bind lipid tails in a hydrophobic pocket with polar head groups typically exposed. The TCR (gray) “reads” the displayed antigen leading to T-cell activation. (C) Cladogram representing phylogenetic relationship of primate CD1a-e paralogs used in this study with surfaces generated in PyMol and antigen-binding pockets outlined in yellow (PDB IDs: 5J1A, 4ONO, 4MQ7, 3S6C). TCR types that recognize each CD1 paralog are also indicated. (D) Primate CD1a diverges most in the lipid-binding domain, which may alter pocket morphology and TCR interactions. Most of the sequence divergence in the primate CD1a proteins is predicted to exist in the beta-sheet that transects the center of the protein with some variation in the central surface region. RMSD, root mean square deviation. (E) Structural features of the CD1a receptor. Labels showing location of A′ Roof, F′ portal.

Fig. 2.

Evidence of repeated positive selection among primate CD1 orthologs. (A) Amino acid sites exhibiting strong signatures of positive selection (elevated dN/dS) are highlighted in teal and mapped onto corresponding crystal structures. Alpha helices are denoted in red, beta-sheet in purple. (PDB IDs: 5J1A, 4ONO, 4MQ7, 3S6C). Table summarizes positions in CD1 paralogs contributing to signatures of positive selection as well as statistics from PAML M7-M8 model comparisons. (B) Sites with elevated dN/dS indicative of positive selection (teal) cluster in the MHC domain of CD1a protein (PDB ID: 5J1A). Alpha helices denoted in red, beta-sheet in purple. (C) Multiple sequence alignment of primate species used to calculate dN/dS ratios for CD1a paired with phylogenetic species tree highlighting the branches (teal) predicted by aBSREL to be undergoing episodic positive selection.

Fig. 3.

Structural modeling illustrates diversity at the CD1a T-cell interaction interface. (A) Predicted attributes of various primate CD1a structures. Surface characteristics across selected primates reveal differences in portal size, number of portals, and pocket morphology. Portals where T-cell receptor “reads” head group are highlighted with gray/yellow outlines. Pocket morphologies and electrostatic properties are shown below surface models. (B) PyMol generated top-view of human CD1a bound to dideoxy-mycobactin (PDB ID: 1XZO). Rapidly-evolving positions 73 and 153 coordinate head groups of antigenic ligand. Note hydrogen bonding between head group and 153E. (C) Primate CD1a A′ roof predicted structures where CD1a interacts with TCR. Notably, the orangutan model does not form roof structure due to mutation at site under selection. Olive baboon and crab-eating macaque form A′ roof with residue of differing property at site 153.

Fig. 4.

Divergence of CD1a shapes predicted endogenous and exogenous lipid antigen affinities. (A) Plot of relative Gibbs free energy values for all ligands tested by ligand docking predictions using AutoDock Vina. Lowest energy values for each set are plotted. Sulfatide, sphingomyelin, lysophosphatidylcholine are endogenous lipid ligands. Dideoxymycobactin (DDM) is a synthetic lipid analog of Mycobacterium tuberculosis siderophore mycobactin. Urushiol is the etiological agent of poison ivy rash. (B) Lipid-swapping predictions based on predicted Kd (dissociation constant) from docking studies. (C) Predicted pocket volume for CD1a orthologs. Legend: Hum, Human; Chp, Chimpanzee; Bon, Bonobo; Gor, Gorilla; Orn, Orangutan; Oli, Olive Baboon; Grn, Green Monkey; Mng, Mangabey; Crb, Crab-eating macaque; SNM, Snub-nosed monkey; Col, Colobus;Mrm, Marmoset;Cpn, Capuchin.

Fig. 5.

Rapidly-evolving positions in CD1a are sufficient to modulate predicted affinity for lipid antigens. (A) Mutation of site 98 to tryptophan in human CD1a (olive baboon and crab-eating macaque share this amino acid at this position) results in increased predicted binding affinity to DDM, with overall fold increase between endogenous ligand and DDM. Mutation of tryptophan at site 98 in crab-eating macaque to ancestral glycine results in higher binding affinity for all endogenous ligands tested, and loss of feasible lipid-swapping and DDM presentation. (B) Mutation of site 98 to tryptophan in snub-nosed monkey CD1a results in increased predicted binding affinity to DDM. Colobus, which is not predicted to swap endogenous ligand for DDM, also increases spread between binding affinities. In colobus, however, it is a decrease in affinity for endogenous lipid rather than increase in DDM affinity that is responsible for the fold change.

Fig. 6.

Conceptual framework for lipid-driven diversification of CD1 molecules. (A) Crab-eating macaque CD1a, which encodes a tryptophan in position 98, is predicted to lose the ability to present self-lipid lysophosphatidylcholine when this position is mutated to the consensus at this site, glycine. An overlay of the differences in pocket morphology shows how the tryptophan limits access to the deeper chambers of the pocket. (B) Humans possess a valine at position 98, which has been proposed to act as a barrier limiting larger ligands access to the pocket. When this residue is mutated to a tryptophan in silico, further decreasing access to the deeper chambers of the pocket, the ability to swap out endogenous for exogenous ligand is improved, suggesting that a large hydrophobic residue in this position may be beneficial in the context of Mycobacterium tuberculosis infection in primates. Cartoons were informed by analysis of Autodock Vina docking results analyzed in PyMol. (C) Conceptual framework for lipid-driven evolution of CD1, resulting in accelerated evolution and rapid diversification of host immune receptors. Lipid biosynthesis pathways are complex and interdependent, thereby adding levels of complexity that may slow the rate at which pathogens can successfully evolve new lipid antigens. Figure created using Biorender.com.