Tracing the evolution of novel features of human Toll-like receptor 4

Abstract

Toll-like receptor 4 (TLR4) is a critical innate immune protein that activates inflammation in response to extracellular cues. Much of the work to understand how the protein works in humans has been done using mouse models. Although human and mouse TLR4 have many shared features, they have also diverged significantly since their last common ancestor, acquiring 277 sequence differences. Functional differences include the extent of ligand-independent activation, whether lipid IVa acts as an antagonist or agonist, and the relative species cross-compatibility of their MD-2 cofactor. We set out to understand the evolutionary origins for these functional differences between human and mouse TLR4. Using a combination of phylogenetics, ancestral sequence reconstruction, and functional characterization, we found that evolutionary changes to the human TLR4, rather than changes to the mouse TLR4, were largely responsible for these functional changes. Human TLR4 repressed ancestral ligand-independent activity and gained antagonism to lipid IVa. Additionally, mutations to the human TLR4 cofactor MD-2 led to lineage-specific incompatibility between human and opossum TLR4 complex members. These results were surprising, as mouse TLR4 has acquired many more mutations than human TLR4 since their last common ancestor. Our work has polarized this set of transitions and sets up work to study the mechanistic underpinnings for the evolution of new functions in TLR4.

Keywords: Toll-like receptor 4; ancestral sequence reconstruction; comparative immunology; innate immunity; opossum; protein evolution.

Figure 1

Functional differences between human and mouse TLR4 arose by an evolutionary process. (a) The cartoon shows how LPS activates the TLR4/MD‐2 complex. In the absence of LPS, TLR4 and MD‐2 form individual complexes (pink and blue). When the cofactor CD14 delivers LPS to the hydrophobic pocket in MD‐2, a dimer of TLR4/MD‐2 forms and activates NF‐κB signaling. (b) Phylogenetic tree schematic showing the relationship between human (h), mouse (m), opossum (o), and the human/mouse ancestral TLR4 proteins (a). The table to the right shows several functional features that differ between the human and mouse proteins. The information in the gray region was filled in by the current study. (c) Rates of sequence change for the extracellular domain of TLR4 (top), the intracellular domain of TLR4 (middle), and MD‐2 (bottom). The numbers give the average number of substitutions per site along each evolutionary branch.

Figure 2

The TLR4/MD‐2 complex uses one conserved and one variable protein–protein interface. (a) Schematic of the TLR4/MD‐2 complex indicating the views shown in panel b (view 1) and panel c (view 2). Points of interest are indicated on the structure as numbers from 1 to 3. (b) The primary TLR4/MD‐2 interface, taken from PDB 3FXI.30 The top figure shows the orientation of the structure below using the same color scheme as the schematic in panel a. LPS is shown as spheres, with oxygen indicated in brown. The pink subunits are not shown in the lower subpanel. The bottom subpanel shows MD‐2 pulled away from the TLR4 interface. The protein surfaces are colored from white (low conservation, score = 0) to red (high conservation, score <−1.5) as calculated using consurf.31 The blue points indicate atoms that are within 5 Å of the partner protein. (c) The second view shows the interface between a dimer of TLR4/MD‐2 heterodimers. The “view 2” (shown in panel a) is now oriented directly into the page. The pink subunits are not shown in the lower subpanel. The colors in the lower subpanel (red, white, and blue) are as in panel b.

Figure 3

Human TLR4 acquired low ligand‐independent activation relative to other mammals. Panels show the ligand‐independent activation of TLR4 alone (panel a) and co‐transfected with MD‐2 and CD14 (panel b) as a function of amount of transfected plasmid. The colors denote different species: human (blue), mouse (orange), the human/mouse ancestor (red), and opossum (green).

Figure 4

Lipid IVa antagonism evolved along the human lineage. (a) Bar plots show relative activity of TLR4 measured for combinations of TLR4 and MD‐2 from different species, challenged with buffer (gray), LPS (red), lipid IVa (blue), or LPS and lipid IVa together (purple). Conditions are indicated schematically below each bar. Bars for each TLR4 protein (h, m, o, or a) are normalized to the species‐matched TLR4/MD‐2 + LPS value. For the ancestral protein, values are normalized to the ancTLR4/hMD‐2 + LPS value. Error bars are standard error on the mean of three biological replicates. (b) Heat map showing the log₂ fold activation induced by LPS above the buffer control for each pairwise combination of TLR4 and MD‐2. (c) Heat map showing the log₂ fold activation induced by lipid IVa above the buffer control for each pairwise combination of TLR4 and MD‐2. (d) Heat map showing the log₂ fold inhibition of LPS activity (LPS + IVa sample divided by the LPS sample) for each pairwise combination of TLR4 and MD‐2. For panels b–d, the color scale is indicated below the plot. *P < 0.05; **P < 0.005 using a one‐tailed t‐test.

Figure 5

Mutations to MD‐2 loop determine cross‐compatibility between MD‐2 and TLR4 between species. (a) Bar plots show activation of TLR4/MD‐2 complexes with buffer (gray) or LPS (red). Conditions are indicated schematically below each bar. Bars for each TLR4 protein (h, m, o, or a) are normalized to the species‐matched TLR4/MD‐2 + LPS value. For the ancestral protein, values are normalized to the ancTLR4/hMD‐2 + LPS value. Error bars are standard error on the mean of three biological replicates. (b) Heat map showing the log₂ fold activation induced by LPS above the buffer control for each pairwise combination of TLR4 and MD‐2. *P < 0.05; **P < 0.005 using a one‐tailed t‐test. (c) Sequence comparison between the 118–127 loops in human, mouse, and opossum MD‐2. (d) The MD‐2 loop populates a slightly different conformation in crystal structures of human and mouse TLR4/MD‐2/LPS complexes (3FXI 30 and 3VQ2 33). The colors of each subunit are the same as the cartoon in Figure 1(a). The blue loop is taken from the human structure; the yellow loop is taken from the mouse structure. The locations of positions 118 and 127 are indicated. (e) Snapshots of this loop taken at 5 ns intervals from 300 ns MD simulations of human MD‐2/LPS complexes with Ser (left) or Pro (right) at positions 118 and 127. The spheres indicate the mutated positions, the blue or yellow region the intervening loop. The gray structure is from the start of the simulations.