The prevalence of both receptor tyrosine O-sulfation and the ubiquitous chemokine fold suggests the possibility of a conserved sulfotyrosine-binding site. In the extracellular N
-terminus of most chemokine receptors, a tyrosine is present approximately ten residues away from a highly conserved cysteine [9
] (e.g., Tyr21 of CXCR4). Using the CXCL12/CXCR4 complex as a guide, our structure-based homology analysis suggested that only the recognition site for sTyr21 is likely to be conserved in other chemokines. The sTyr21-binding cleft is located between the N-loop and β3 strand, which possesses both polar and hydrophobic contacts. This pocket, termed the chemokine groove, has been previously identified as a key mediator of receptor binding in CC, CX3C, and CXC subfamilies through mutagenesis and NMR binding experiments [31
]. Overall, our analysis revealed few globally conserved sequence positions beyond the structurally essential cysteine residues located in the N
-terminus and β3 strand that distinguish the four chemokine subfamilies [48
]. However, several apolar and basic residues were found to cluster in and around the sulfotyrosine-binding pocket. Members from all four subfamilies, composing more than 70% of chemokines, possessed a basic residue at position 60. Further, 90% of all human chemokines possess an apolar residue at position 62.
Structural alignment not only confirmed the proper orientation and spatial position of residues identified by primary sequence alignment, but revealed compensatory amino acids in chemokines where the key sulfotyrosine recognition residues were not conserved. Absence of a basic residue at position 60 was remedied mainly through three alternative mutations. Most of the chemokines contained arginine or lysine residues at other positions within the β3 strand that position toward the pocket; several proteins also contained charged amino acids in the β2-β3 turn. The other deficient proteins, including CCL20, CCL21, and CCL24 all contain positively charged residues in the N-loop oriented toward the binding cleft. Indeed, CCL24 Arg15 experiences complete line broadening in an NMR titration with a CCR3 N
-terminal peptide suggesting that this residue participates directly in receptor binding [49
]. CCL25, CCL28, CXCL16, and CXCL17, which lack experimentally determined structures, could not be modeled due to low sequence homology with other family members. Interestingly, all four proteins still retain an apolar residue at position 62, and three of them (all except CXCL16) possess a basic residue at site 58 or 60 suggesting the presence of a cleft compatible with sulfotyrosine binding.
The problem of identifying the receptor sulfotyrosine binding pocket on chemokines has usually been solved by titrating sulfated peptides representing the N
-terminal fragment of the receptor of interest into a chemokine sample and monitoring backbone amide chemical shift changes by HSQC NMR [37
]. Several methods, such as solid-phase synthesis [50
] and in vitro
enzymatic sulfation [17
], have been utilized to sulfate these peptides. However, these techniques are challenging due to low yields, the labile nature of the sulfate modification, and difficulties associated with separating complex mixtures of products. In an effort to produce a simpler probe, we titrated CXCL12, CCL5, CX3CL1, and XCL1 with free sulfotyrosine (H2
). These proteins were chosen as representatives of each subfamily because the binding site of each chemokine, with the exception of XCL1, had previously been probed with receptor peptides [31
]. Each protein exhibited localized chemical shift perturbations in the chemokine groove correlating with previous peptide binding studies. In addition, although no XCL1/XCR1 binding information is published, the localization of perturbations to the analogous cleft suggests a similarly conserved sulfotyrosine-binding site. Interestingly, the strong perturbation in CCL5 Tyr3 suggests sulfotyrosine is a powerful probe for binding pocket identification regardless of whether the chemokine is in the native oligomeric state for receptor binding. Tyr3 is only located near the sulfotyrosine pocket when CCL5 is dimeric; however, only monomeric CCL5 interacts with the CCR5 N
]. Although the sulfotyrosine probe binds too weakly (Kd
M) for structural characterization, these results suggest that short sulfopeptides, which should be easier to produce, may possess sufficient specificity and affinity to enable structure determination by NMR [53
Identification of specific sulfotyrosine binding pockets on chemokines could define a new category of targets for structure-based drug discovery. Currently, only a single chemokine/sulfopeptide structure, CXCL12/CXCR4, has been determined [17
]. Using this structure we recently performed a high-throughput computational docking study to identify small molecules with the propensity to bind the sTyr21 site. The top ranking candidate ligands were then tested for binding by NMR, and one compound with a 64 μM affinity was found to inhibit CXCR4-mediated calcium flux signaling of THP-1 monocytes [18
]. Similar docking studies against the XCL1 site defined by sulfotyrosine chemical shift mapping have identified small molecule ligands that bind the target site and are currently being tested for inhibition of XCL1/XCR1 interactions [54
Overall, our results show that the chemokine superfamily possesses a conserved sulfotyrosine binding site, critical for high-affinity interactions that can potentially be targeted for the design of specific or broad-spectrum inhibitors. Since the discovery of the first chemokine, CXCL8, almost 25 years ago a total of 43 human chemokines have been identified. As only three new chemokines have been isolated in the last decade [48
], it is generally believed that most human chemokines have been discovered. Thus, we conclude that the chemokines lacking a potential sulfotyrosine recognition site represent exceptions to the general rule. The chemokines and their receptors are broadly expressed and relatively promiscuous with two or more partners in most cases. As a consequence, inhibitors targeting a specific receptor may not have the optimal specificity, since they may interfere with signaling of multiple chemokine ligands, or a second receptor could coordinate chemotaxis toward a given site of chemokine secretion. If the chemokine ligands can instead be inhibited by blocking the sulfotyrosine-mediated receptor interaction, novel inhibitors might be designed with favorable therapeutic properties. In addition, high-affinity ligands could also be adapted for use as diagnostic molecules for imaging of chemokine levels in either research or clinical settings. Although the structural similarities outlined in this article and elsewhere suggest that sulfotyrosine-directed selective blocking of individual chemokines could be challenging, the potential for identifying broad-spectrum inhibitors represents a powerful complementary strategy.