Molecular basis for the recognition of 24-(S)-hydroxycholesterol by integrin αvβ3

Abstract

A growing body of evidence suggests that oxysterols such as 25-hydroxycholesterol (25HC) are biologically active and involved in many physiological and pathological processes. Our previous study demonstrated that 25HC induces an innate immune response during viral infections by activating the integrin-focal adhesion kinase (FAK) pathway. 25HC produced the proinflammatory response by binding directly to integrins at a novel binding site (site II) and triggering the production of proinflammatory mediators such as tumor necrosis factor-α (TNF) and interleukin-6 (IL-6). 24-(S)-hydroxycholesterol (24HC), a structural isomer of 25HC, plays a critical role in cholesterol homeostasis in the human brain and is implicated in multiple inflammatory conditions, including Alzheimer's disease. However, whether 24HC can induce a proinflammatory response like 25HC in non-neuronal cells has not been studied and remains unknown. The aim of this study was to examine whether 24HC produces such an immune response using in silico and in vitro experiments. Our results indicate that despite being a structural isomer of 25HC, 24HC binds at site II in a distinct binding mode, engages in varied residue interactions, and produces significant conformational changes in the specificity-determining loop (SDL). In addition, our surface plasmon resonance (SPR) study reveals that 24HC could directly bind to integrin αvβ3, with a binding affinity three-fold lower than 25HC. Furthermore, our in vitro studies with macrophages support the involvement of FAK and NFκB signaling pathways in triggering 24HC-mediated production of TNF. Thus, we have identified 24HC as another oxysterol that binds to integrin αvβ3 and promotes a proinflammatory response via the integrin-FAK-NFκB pathway.

Figure 1

2D structures and metabolic pathways of cholesterol and its two metabolites. 25-hydroxycholetserol (25HC) and 24(S)-hydroxycholesterol (24HC) are produced by enzymatic metabolism of cholesterol cholesterol-25-hydroxylase (C25H) and cholesterol-24-hydroxylase also known as CYP46A1, respectively.

Figure 2

Oxysterols such as 25-hydroxycholesterol bind to integrin αvβ3 at site II, which is distinct from the primary RGD-binding site. (A) A full-length integrin αvβ3 (secondary structure representation) bound to an oxysterol (licorice representation) at site II is shown embedded in the lipid bilayer consisting of POPC and cholesterol. The αv and β3 subunits of integrin are shown in light green and pink colors, respectively. Site II, denoted by an arrow, is 6 Å away from the primary “RGD” binding site, where extracellular matrix proteins such as fibronectin are known to bind. The polar headgroups of the lipid bilayer are represented as the vdW surface model (nitrogen in blue, phosphate in orange, and oxygen in red colors), and the alkyl chains are shown as a transparent stick model. The extracellular domain is embedded in the bilayer through two transmembrane helices. (B) Site II is formed at the interface between the β-propeller domain (light green) of the αv subunit and the βI domain (light pink) of the β3 subunit. Critical binding site residues from each domain are shown in respective darker colors. The specificity-determining loop (SDL) connects site II with the primary “RGD” binding site.

Figure 3

Molecular interactions of 24HC at site II of integrin αvβ3. (A) Molecular docking and MD simulations revealed that 24HC binds to integrin αvβ3 at site II in an orientation that is distinct from that of 25HC. In this orientation, the 3-OH group engages in polar interactions with S162 and A263 of the βI domain, and the 24-OH group is near S399 of the β-propeller domain. The two domains of integrin αvβ3 are shown as secondary structure representation, and the binding site residues are shown in licorice. (B) The distance between the center-of-mass (COM) of the binding site residues and COM of 24HC through the entire simulation time (200 ns) indicates the stability of the ligand within the binding site. (C) Major polar interactions between the two hydroxyl groups of 24HC and various binding site residues are tracked as distances between the interacting functional groups. (D–E) Radar charts showing the polar and hydrophobic interactions between 24HC and various binding site residues quantified as % occupancy, the fraction of the simulation time during which 24HC is within 5 Å of the listed residues from the β-propeller domain (light green), and the βI domain (light pink), respectively.

Figure 4

The free energy surface (FES) for 24HC’s access and binding to the integrin αVβ3 site II. The FES was characterized by two collective variables: (1) the distance between the center-of-mass (COM) of the ligand and COM of the binding site residues in nm (X-axis), and (2) the orientation angle of the ligand, defined as the angle between 3-OH and C23 of 24HC and the backbone carbon atom of Ile265 in degrees (Y-axis). The minimum energy path of 24HC’s access to the binding site is given as black bold connected points. A and B represent one of the intermediate states and a final bound state, respectively.

Figure 5

24HC produces significant conformational changes in the specificity-determining loop (SDL) in the βI-domain of integrin αvβ3. (A) SDL undergoes extensive conformational changes during 200 ns MD simulations. The time evolution of the entire loop (residues 158–190) is shown at various time intervals from the start (red color) to the end (blue color). (B) The root-mean-square-fluctuations (RMSF) indicate the extent of conformational changes in the βI-domain observed during the simulations. In addition to SDL, both α1 and α7 helices undergo moderate fluctuations. (C–D) 24HC binding disrupts H-bond networks within the SDL residues and between SDL and the β-propeller domain. SDL is shown at the start and end of the simulation in pink and cyan colors, respectively. The H-bonds between Y122 and T182 and T182 and K125 broke after 30 ns. The H-bond between Y122 and K125 broke after around 100 ns. The H-bond between the residues P169 from SDL (βI) and Q120 from the β-propeller domain quickly moved apart from 2.9 to 16.6 Å. The β-propeller domain (with Q120) is shown at the start and end of the simulation in green and light blue colors, respectively.

Figure 6

The binding affinities of 24HC and 25HC to integrin αvβ3 were determined by surface plasmon resonance (SPR). (A–B) 24HC and (C–D) 25HC. Increasing concentrations of the ligand (as indicated in the figures) were injected into both the integrin-immobilized and control blank surface of the CM5 chip, as described in “Materials and methods” Section. The analyte injection was terminated at 180 s and allowed to dissociate in the running buffer for 600 s. Representative SPR data were quantified using the steady-state equilibrium binding model to calculate the affinity constant (K_D), as shown in the curve-fitting graphs.

Figure 7

24HC converting enzyme CYP46A1 is induced by Tumor Necrosis Factor-alpha (TNF) in macrophages, and 24HC activates a proinflammatory response in macrophages. (A) RT-PCR analyses of CYP46A1 expression in RAW 264.7 macrophages treated with TNF (100 ng/ml) for 2 h. (B) RAW 264.7 macrophages were treated with either DMSO (vehicle) or 24HC (50 µM) for 16 h. TNF production was analyzed by ELISA. The ELISA values are represented as mean ± standard deviation (n = 16 technical replicates from two independent experiments). *p ≤ 0.05 using Student’s t-test.

Figure 8

Integrin-FAK-NFκB signaling is essential for 24HC-mediated proinflammatory response in macrophages. (A) FAK is required for 24HC-mediated proinflammatory response. RAW 264.7 macrophages were treated with 24HC (50 µM) for 16 h in the presence of either DMSO (vehicle control) or FAK inhibitor PF-431396 (5 mM). TNF secretion was analyzed by ELISA. (B) NFκB is required for 24HC-mediated proinflammatory response. RAW 264.7 macrophages were treated with 24HC (100 µM) for 16 h in the presence of either DMSO (vehicle control) or NFκB inhibitor BAY-11 (1 μM). TNF secretion was analyzed by ELISA. The ELISA values are represented as mean ± standard deviation (n = 14–24 technical replicates from two–three independent experiments). *p ≤ 0.05 using Student’s t-test.