Probing receptor binding activity of interleukin-8 dimer using a disulfide trap.

Interleukin-8 (IL-8, also known as CXCL8), a member of the chemokine superfamily, plays a pivotal role in recruiting neutrophils under conditions such as tissue injury, wound healing, and bacterial infection (1). IL-8 belongs to the class of proteins now identified as ‘weak’ transient homodimers that exist as both monomers and dimers under physiological conditions (2). During active neutrophil recruitment, IL-8 is translocated from the injury site in the tissue to the vasculature, and so its concentration will vary spatially and temporally and cannot be defined by a single physiological concentration. Local concentration under these conditions could reach levels high enough so that both monomers and dimers exist. IL-8 function involves binding to neutrophil G protein-coupled receptors (GPCR)1 and cell-surface glycosaminoglycans (GAG) (3, 4). Therefore, knowledge of the relative binding affinities and functional activities of the monomer and dimer for their GPCR receptors and GAGs is critical for understanding the role of monomer-dimer equilibrium in neutrophil recruitment. IL-8 dimerizes at μM concentrations (Kd ∼10 μM), and binds its receptors (CXCR1 and CXCR2) at nM concentrations (Kd ∼ 1 nM), suggesting that a monomer is sufficient for receptor function (4-6). An IL-8 ‘trapped’ monomer and also monomer mutants show native binding affinities (7, 8), and various other structure-function studies have also shown that mutating dimer-interface residues does not affect binding indicating that dimer-interface residues are not essential for receptor function (4, 8, 9). However, measuring binding affinities of the dimer is less straight forward, as the ∼1000-fold difference between the IL-8 monomer-dimer equilibrium constant (μM) and receptor binding constant (nM) precludes binding studies of the native IL-8 dimer. In principle, the IL-8 dimer, compared to the monomer, could bind with higher, lower or same affinity, or could be completely inactive, and also the IL-8 dimer could bind the two receptors with different affinities. Previous studies have used the strategy of designing non-dissociating disulfide-linked IL-8 dimers, but these variants show both native-like and reduced receptor function (10-12). All these studies used the criterion of proximity in the IL-8 dimer structure for mutating a pair of residues to cysteine for introducing disulfides across the dimer interface (13). Leong et al. designed a single chain fusion protein by joining the C-terminus of one monomer to the N-terminus of another monomer using a flexible linker, and created four different variants by introducing a pair of disulfides across the dimer interface between Glu29 or Thr37 in one monomer and Ala69 or Ser72 in the other monomer (10). Compared to native IL-8, they observed the disulfide variants to bind CXCR1 with 6 to 21-fold lower affinities and CXCR2 with native to 4-fold lower affinities, and concluded that dimer dissociation is not essential for IL-8 biological activity. Williams et al. observed that a disulfide-linked dimer created by introducing a pair of disulfides (E29C/A69C) across the dimer interface, was as active as the native protein in a neutrophil Ca2+ release assay (11). We had reported in a preliminary study that a R26C disulfide-linked dimer, produced by introducing a single disulfide at the site of two-fold symmetry, was 15-fold less active in a neutrophil elastase activity (12). The latter two studies did not report binding affinities, and also neutrophils express both CXCR1 and CXCR2 receptors, and so reduced function could be due to reduced binding to one or both receptors. Most importantly, the consequence of introducing disulfides in these variants is not known, except Williams et al. have characterized their E29C/A69C mutant by nuclear magnetic resonance (NMR) spectroscopy (11). They observed chemical shift changes for a number of helical residues, including those of Phe65 and Leu66 that play an important role in stabilizing the native dimer, suggesting that the packing interactions in the dimer interface are perturbed; so its functional properties do not correspond to that of the native dimer. Whereas Arg26 is solvent exposed, residues Glu29, Thr37 and Ala69 are largely buried in the dimer structure (13); so mutating the latter set of residues for introducing new disulfides, if not optimally accommodated, would perturb the dimer interface structure, as observed by Williams et al. for the E29/A69 disulfide mutant. Therefore, these disulfide-linked dimers could have differential structural perturbations of the dimer interface and so have varied function, and therefore it is not clear which of these disulfide-linked variants mimic native dimer structure. We recently exploited the knowledge that IL-8 binding to the CXCR1 N-domain peptide is in the same μM range as the dimer dissociation constant, and observed that in presence of both monomers and dimers, only the monomer preferentially binds a receptor N-domain peptide (14, 15). We measured the binding constant of the trapped monomer for the receptor N-domain using ITC to be ∼ 5 μM, and a NMR study has shown that the native dimer binds the N-domain with much lower affinity ∼170 μM (16). Though these studies show that the dimer has a lower binding potency, it can be argued that these results do not exclude the possibility of native dimer binding to the intact receptor with the same affinity as the monomer. Further, we could not carry out similar binding studies with a CXCR2 N-domain peptide due to experimental constraints, and so the relative binding affinity of the native dimer for CXCR2 remains unanswered. In this report, we have addressed some of these questions by studying the structural and receptor binding properties of the ‘trapped’ R26C IL-8 dimer containing a single disulfide at the two-fold symmetry axis. NMR studies indicate that mutating Arg26 → Cys and introducing a new disulfide does not perturb the native dimer interface, and that the structure of this disulfide-linked trapped dimer is indistinguishable from the native dimer. Therefore, the binding affinity of the trapped dimer should reflect the binding affinity of the native dimer. The R26C trapped dimer, compared to a trapped monomer, binds CXCR1 and CXCR2 receptors with ∼70 fold and ∼20 fold lower affinities, respectively. We could not detect any binding of the trapped dimer to a CXCR1 N-domain peptide indicating that the significantly reduced affinity of the dimer should be due to reduced binding of the N-loop residues. These results emphasize that only the monomer is a high-affinity ligand for both CXCR1 and CXCR2 receptors, and also provide a structural mechanism for the lower binding affinity of the dimer.