Positive Selection on Mammalian Immune Genes-Effects of Gene Function and Selective Constraint

Abstract

Genome-wide analyses of various taxa have repeatedly shown that immune genes are important targets of positive selection. However, little is known about what factors determine which immune genes are under positive selection. To address this question, we here focus on the mammalian immune system and investigate the importance of gene function and other factors such as gene expression, protein-protein interactions, and overall selective constraint as determinants of positive selection. We compiled a list of >1,100 immune genes that were divided into six functional categories and analyzed using data from rodents. Genes encoding proteins that are in direct interactions with pathogens, such as pattern recognition receptors (PRRs), are often expected to be key targets of positive selection. We found that categories containing cytokines, cytokine receptors, and other cell surface proteins involved in, for example, cell-cell interactions were at least as important targets as PRRs, with three times higher rate of positive selection than nonimmune genes. The higher rate of positive selection on cytokines and cell surface proteins was partly an effect of these categories having lower selective constraint. Nonetheless, cytokines had a higher rate of positive selection than nonimmune genes even at a given level of selective constraint, indicating that gene function per se can also be a determinant of positive selection. These results have broad implications for understanding the causes of positive selection on immune genes, specifically the relative importance of host-pathogen coevolution versus other processes.

Keywords: Rodentia; coevolution; gene expression tissue specificity; molecular evolution; pN/pS; pathogen-mediated selection.

Fig. 1.

a) Differences in rates of positive selection between functional categories of immune genes could be a direct effect of their function (arrow 1) or an indirect effect of differences in selective constraint (arrows 2 and 3). b) Immune genes were divided into six different categories depending on the function of the encoded protein: (1) PRRs recognize “pathogen-associated molecular patterns” (PAMPs; i.e. molecular structures unique for microbes). Sensors of “damage-associated molecular patterns” (DAMPs, i.e. endogenous molecules reflecting tissue damage) were not included. (2) Cytokines, chemokines, and their receptors, also including anaphylatoxin receptors. (3) Other cell surface proteins include receptors and ligands involved in cell–cell interactions (e.g. CD80–CD28, Fas–FasL), and Fc and complement receptors. (4) Intracellular signaling proteins are involved in intracellular signal transduction and include adaptors, enzymes such as kinases, ubiquitinases, and proteases, transcription factors and cofactors, etc. (5) Extracellular protease activity includes secreted proteases and protease inhibitors, mainly involved in the complement and coagulation pathways. (6) Effectors are proteins involved in killing of pathogens by direct interaction, including antimicrobial peptides, restriction factors, and various other proteins.

Fig. 2.

Selection on different categories of immune genes and nonimmune control genes. a) Box plot of dN/dS. b) The proportion of genes with signatures of positive selection. c) Box plot of pN/pS. d) The proportion of genes with signatures of positive selection against pN/pS for three of the categories of immune genes and nonimmune control genes. Lines are predictions from a binomial general linear model with positive selection against gene function and pN/pS.