Adaptive Evolution in TRIF Leads to Discordance between Human and Mouse Innate Immune Signaling

Abstract

The TIR domain-containing adapter inducing IFN-β (TRIF) protein is an innate immune system protein that mediates the MyD88-independent toll-like receptor response pathway in mice and humans. Previously, we identified positive selection at seven distinct residues in mouse TRIF (mTRIF), as compared with human and other mammalian orthologs, thus predicting protein functional shift in mTRIF. We reconstructed TRIF for the most recent common ancestor of mouse and human, and mutated this at the seven sites to their extant mouse/human states. We overexpressed these TRIF mutants in immortalized human and mouse cell lines and monitored TRIF-dependent cytokine production and gene expression induction. We show that optimal TRIF function in human and mouse is dependent on the identity of the positively selected sites. These data provide us with molecular data relating observed differences in response between mouse and human MyD88-independent signaling in the innate immune system with protein functional change.

Keywords: ancestral gene resurrection; innate immune signaling; positive selection.

Fig. 1.

Signatures of adaptive evolution identified in TRIF and TRIF-dependent signaling pathways. (A) Schematic adapted from the Kyoto Encyclopedia of Genes and Genomes (KEGG) illustrating TRIF-dependent pathways in toll-like signaling in innate immunity. Selective pressure variation analysis was performed for both lineage-specific (model A) and site-specific (model 8) selection. Genes boxed in gray represent those that are predicted to have signatures of positive selection; gene names in bolded black contain residues that are predicted to be under site-specific selection, and genes in red are genes under selection in the mouse lineage only. Highlighted in yellow are TLR3 and TRIF predicted to be under selection in mouse by our earlier analysis (Webb et al. 2015). (B) A portion of the PRANK amino acid alignment of TRIF homologues showing the human, mouse, and ancestral TRIF protein sequences. The positions of the positively selected sites are indicated, as well as the position (relative to human—O'Neill and Bowie 2007) of the three known functional TRIF domains: Traf6 binding motif (T6), Toll/interleukin-1 receptor (TIR) domain, and the receptor-interacting protein (RIP) homotypic interaction motif (RHIM). (C) Schematic illustrating the different protein sequences of TRIF used in this study. Residues highlighted in the mouse protein (mTRIF) are those predicted to be under positive selection. Their position relative to the three functional domains is indicated. The position and identity of the homologous residues in the hTRIF, the reconstructed ancestral TRIF (aTRIF), the humanized ancestral TRIF protein (haTRIF), and the murinized ancestral TRIF (maTRIF) are also shown.

Fig. 2.

Amino acid residues under positive selection in TRIF, modulate species-specific function. Functional analysis of TRIF gene constructs. TRIF-induced RANTES production for human HEK 293 cells (A + C) and mouse 3T3 cells (B + D). Cell lines were transiently transfected with plasmids constitutively expressing each of the versions of the TRIF transgenes as indicated, or an empty vector (EV) control. In (A) and (B), cells were stimulated with LPS 12 h post-infection. At 24 h post transection, supernatants were recovered and assayed for the presence of the RANTES cytokine by ELISA. Each graph represents data from three biological replicates assayed in triplicate from a single day’s experiment. Similar trends were observed on different days. Error bars denote SD. *P < 0.05 compared with hRANTES value in HEK cells. **P < 0.01 compared with mRANTES value in 3T3 cells. Quantitative RT-PCR analysis as outlined in figure indicating the expression of IP-10 in human HEK 293 TLR4 cells (E) and in murine 3T3 cells (F) expressing the indicated TRIF constructs. Baseline of 0 is set to the expression of IP-10 in cells transfected with the empty vector control.

Fig. 3.

In silico analysis highlights the structural importance of sites under positive selection in mTRIF. (A) The predicted structure of a portion of the mTRIF protein (residues Pro110 to Gln394) generated using MODELLER (Fiser and Sali 2003). Residues are highlighted based on their dynamic flexibility index, (dfi) score, as per the indicated scale, and the position of four mouse residues under positive selection are indicated. (B) A close up of His 388 and Asp 338, two mTRIF residues under positive selection, in the predicted TRIF protein structure. The shortest distance between their two side chains is indicated (2.7 Å) and suggests a potential interaction between these two residues. Both (A) and (B) were generated using pymol (Delano 2002). (C) A summary of the in silico mutagenesis analysis done on mTRIF protein structure using dezyme software (https://soft.dezyme.com). The number of mutations that led to either a decrease (stabilizing) or increase (destabilizing) in the folding free energy (△△G) of mTRIF is indicated.