Identification of Key Residues That Confer Rhodobacter sphaeroides LPS Activity at Horse TLR4/MD-2

et al. (2014) Identification of Key Residues That Confer Rhodobacter sphaeroides LPS Activity at Horse TLR4/MD-2. PLoS ONE 9(5): e98776. doi:10.1371/journal.pone.0098776 Identification of Key Residues That Confer Rhodobacter sphaeroides LPS Activity at Horse TLR4/MD-2 Katherine L. Irvine 0 Monique Gangloff 0 Catherine M. Walsh 0 David R. Spring 0 Nicholas J. Gay 0 Clare E. Bryant 0 Sangdun Choi, Ajou University, Republic of Korea 0 1 Department of Veterinary Medicine, University of Cambridge, Cambridge, United Kingdom, 2 Department of Biochemistry, University of Cambridge, Cambridge, United Kingdom, 3 Department of Chemistry, University of Cambridge , Cambridge , United Kingdom The molecular determinants underpinning how hexaacylated lipid A and tetraacylated precursor lipid IVa activate Toll-like receptor 4 (TLR4) are well understood, but how activation is induced by other lipid A species is less clear. Species specificity studies have clarified how TLR4/MD-2 recognises different lipid A structures, for example tetraacylated lipid IVa requires direct electrostatic interactions for agonism. In this study, we examine how pentaacylated lipopolysaccharide from Rhodobacter sphaeroides (RSLPS) antagonises human TLR4/MD-2 and activates the horse receptor complex using a computational approach and cross-species mutagenesis. At a functional level, we show that RSLPS is a partial agonist at horse TLR4/MD-2 with greater efficacy than lipid IVa. These data suggest the importance of the additional acyl chain in RSLPS signalling. Based on docking analysis, we propose a model for positioning of the RSLPS lipid A moiety (RSLA) within the MD-2 cavity at the TLR4 dimer interface, which allows activity at the horse receptor complex. As for lipid IVa, RSLPS agonism requires species-specific contacts with MD-2 and TLR4, but the R2 chain of RSLA protrudes from the MD-2 pocket to contact the TLR4 dimer in the vicinity of proline 442. Our model explains why RSLPS is only partially dependent on horse TLR4 residue R385, unlike lipid IVa. Mutagenesis of proline 442 into a serine residue, as found in human TLR4, uncovers the importance of this site in RSLPS signalling; horse TLR4 R385G/P442S double mutation completely abolishes RSLPS activity without its counterpart, human TLR4 G384R/S441P, being able to restore it. Our data highlight the importance of subtle changes in ligand positioning, and suggest that TLR4 and MD-2 residues that may not participate directly in ligand binding can determine the signalling outcome of a given ligand. This indicates a cooperative binding mechanism within the receptor complex, which is becoming increasingly important in TLR signalling. - Funding: This work was supported by a project grant from the Horserace Betting Levy Board to CEB and a Horserace Betting Levy Board Veterinary Research Training Scholarship to KLI. This work was also supported by a Wellcome Trust program grant to NJG and CEB. CEB is a BBSRC Research Development Fellow. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors would like to confirm that Nicholas J. Gay is a PLOS ONE Editorial Board member, and that this does not alter their adherence to PLOS ONE Editorial policies and criteria. . These authors contributed equally to this work. The Toll-like receptor (TLR) family is the most widely studied pattern recognition receptor family, comprising 13 mammalian members, of which 10 are expressed in humans [1]. Toll-like receptor 4 (TLR4) is expressed on the plasma membrane and endosomes to sense lipopolysaccharide (LPS) from Gram-negative bacteria, requiring the co-receptor MD-2 to bind the lipid component of LPS (lipid A) and form the active complex [2,3]. The crystal structure of human and mouse TLR4/MD-2 bound to E. coli LPS (ECLPS) demonstrates a heterotetrameric structure, comprising two TLR4, two MD-2 and two LPS molecules [4,5]. Five of the six lipid A acyl chains sit deep within the MD-2 pocket, and the sixth (R2 chain) protrudes from the MD-2 pocket to form part of the dimerisation interface with the opposing TLR4 (denoted thereafter TLR4*). TLR4/MD-2 from all domestic mammalian species recognises hexaacylated lipid A from E. coli (figure 1A) as an agonist, but structural variability in lipid A from other Gram-negative organisms alters their efficacy (maximum stimulation, Emax) and potency (half maximum effective concentration, EC50) at the receptor complex. A reduction in acyl chain number from six to four in the lipid A synthesis intermediate lipid IVa (figure 1B) makes this compound an antagonist at TLR4/MD-2, but only in humans. At horse and mouse TLR4/MD-2, lipid IVa is an agonist, with residues R385 in horse TLR4, and K367 and R434 in mouse TLR4, being important for agonist activity [6,7,8,9]. Eritoran, a tetraacyl chain lipid A analogue derived from the structure of Rhodobacter sphaeroides LPS (RSLPS), is an antagonist at human, mouse and horse TLR4/MD-2. Antagonist-conformation crystal structures of human TLR4/ MD-2 proteins bound to lipid IVa and eritoran demonstrate gross similarities to each other and significant differences to lipid A [10,11]. Their relative orientations in the MD-2 lipid-binding pocket are inverted, with either the 49-phosphate (49-PO4) in the antagonist conformation or the 1-PO4 in the agonist conformation facing the LPS-binding motif of MD-2 [4,5,10,11,12]. Both ligands have only four acyl chains, which are fully accommodated within the MD-2 binding pocket. A recent crystal structure of lipid IVa bound to mouse TLR4/MD-2 demonstrates that lipid IVa can nevertheless form similar interactions to E. coli lipid A in the presence of appropriate surface charge distribution on TLR4/ MD-2 [5]. This results in a 47 A difference in glucosamine backbone position along a 180u rotation between human- and mouse-bound lipid IVa conformations. The positioning of the ligand triggers conformational switching in MD-2 and dimerisation of TLR4 for agonists, and apparently fails to do so in the case of antagonists. It is not clear, however, whether novel lipid A structures will also share similar interacting properties. Subtle differences exist in the interaction patterns of lipid IVa and lipid A at the TLR4/MD-2 complex, and these probably underlie the small differences in efficacy between the two ligands at mouse TLR4/MD-2 [5]. The mouse lipid IVa-TLR4-MD-2 complex is monomeric in solution, but dimeric within crystals as a result of crystal packing. At the molecular level, the mouse TLR4 dimer interface residues K367* and R434* both contact the lipid IVa and lipid A 1-PO4 groups, whereas an extra hydrogen bond forms between lipid A/TLR4. This extra bond is, therefore, most likely required for maximal ligand efficacy. Lipid A from Rhodobacter sphaeroides (RSLA; figure 1C) has five acyl chains, with an extended and unsaturated R20 acyl chain, two shortened chains (...truncated)