Structural basis of chemokine and receptor interactions: Key regulators of leukocyte recruitment in inflammatory responses

Abstract

In response to infection or injury, the body mounts an inflammatory immune response in order to neutralize pathogens and promote tissue repair. The key effector cells for these responses are the leukocytes (white blood cells), which are specifically recruited to the site of injury. However, dysregulation of the inflammatory response, characterized by the excessive migration of leukocytes to the affected tissues, can also lead to chronic inflammatory diseases. Leukocyte recruitment is regulated by inflammatory mediators, including an important family of small secreted chemokines and their corresponding G protein‐coupled receptors expressed in leukocytes. Unsurprisingly, due to their central role in the leukocyte inflammatory response, chemokines and their receptors have been intensely investigated and represent attractive drug targets. Nonetheless, the full therapeutic potential of chemokine receptors has not been realized, largely due to the complexities in the chemokine system. The determination of chemokine–receptor structures in recent years has dramatically shaped our understanding of the molecular mechanisms that underpin chemokine signaling. In this review, we summarize the contemporary structural view of chemokine–receptor recognition, and describe the various binding modes of peptide and small‐molecule ligands to chemokine receptors. We also provide some perspectives on the implications of these data for future research and therapeutic development.

Importance Statement

Given their central role in the leukocyte inflammatory response, chemokines and their receptors are considered as important regulators of physiology and viable therapeutic targets. In this review, we provide a summary of the current understanding of chemokine: chemokine–receptor interactions that have been gained from structural studies, as well as their implications for future drug discovery efforts.

Abbreviations

CCL

C–C motif chemokine ligand

CCR

C–C motif chemokine receptor

CRS1/1.5/2

chemokine receptor site 1/1.5/2

CS1/1.5/2

chemokine site 1/1.5/2

CXC

C–X–C motif chemokine ligand

CXCR C–X–C

motif chemokine receptor

ECL

extracellular loop

GAG

glycosaminoglycan

GPCR

G protein‐coupled receptor

ICL

intracellular loop

NMR

intracellular loop

1. INTRODUCTION

Our bodies respond to infection or injury by mounting inflammatory immune responses, whose primary functions are to neutralize pathogenic microbes and promote tissue repair. The key effector cells for these responses are the leukocytes (white blood cells), which include monocytes (and their derived macrophages and dendritic cells), granulocytes (neutrophils, eosinophils, and basophils), and lymphocytes (T and B cells). The effectiveness of an inflammatory response is critically dependent on the appropriate deployment of these functionally specialized leukocytes, that is, leukocyte recruitment to the right tissues at the right time. Alternatively, dysregulation of leukocyte recruitment can lead to an inadequate immune response or failure to resolve the response, that is, chronic inflammatory disease.

Leukocyte recruitment in inflammation is regulated by the coordinated action of numerous inflammatory mediators. Critical among these are the chemokines and their receptors, the focus of this review. Chemokines are a family of about 50 small proteins that are secreted in the tissues as part of the immediate response to tissue damage.1, 2, 3 Chemokine receptors are a complementary family of about 20 G protein‐coupled receptors (GPCRs) located in the plasma membranes of leukocytes.2, 3, 4 Chemokines secreted into the vasculature activate their receptors, stimulating leukocyte adhesion to the vascular endothelium, reorganization of the actin cytoskeleton, and eventually migration of the leukocytes into the tissues (extravasation). Moreover, within the tissue extracellular matrix, chemokines bind to glycosaminoglycans (GAGs), forming defined concentration gradients that further direct the leukocytes to the location of chemokine production and tissue damage.

The selective interactions of chemokines and chemokine receptors play a fundamental role in controlling the selective recruitment of leukocyte subsets in beneficial immune responses, as well as essentially all inflammatory diseases. Each type of leukocyte expresses a subset of chemokine receptors and each receptor responds to a subset of chemokines.5 Consequently, leukocytes migrate only to tissues expressing chemokine ligands for their expressed receptors. Thus, the chemokines and their receptors represent a fascinating example of two complementary protein families whose complex pattern of selectivity has profound biological consequences. Successful therapeutic targeting of chemokines or their receptors, which has proven challenging to date, will benefit from improved understanding of chemokine–receptor interactions, particularly the factors influencing selectivity and the molecular mechanisms of receptor activation.

In this review, we summarize the current state of knowledge on the structural basis of chemokine–receptor recognition, the diverse binding modes of peptide and small‐molecule ligands to chemokine receptors, and finally discuss the implications of these data for future research and therapeutic development.

2. THE CHEMOKINE AND CHEMOKINE–RECEPTOR PROTEIN FAMILIES

Chemokines are small (typically <10 kDa) soluble proteins that contain four conserved cysteine residues forming two disulfide bonds. They are classified according to the spacing (and presence) of the first two of these Cys residues, which may be: adjacent (CCL1‐28; L indicates ligand, i.e., chemokine); separated by one or three residues (CXCL1‐17 or CX₃CL1, respectively); or missing the first conserved Cys (XCL1 and XCL2). Prior to the establishment of this systematic nomenclature,6, 7 most chemokines were given common names related to their activity. For example, human CCL2, CCL8, CCL7, and CCL13 were named monocyte chemoattractant proteins (MCP‐1 to ‐4, respectively). Chemokine receptors are named (CCR1‐10, CXCR1‐8, CX₃CR1, and XCR1) according to the class of chemokines to which they predominantly respond.6, 7

Numerous nuclear magnetic resonance (NMR) and X‐ray structures of chemokines8, 9 have shown that they have a conserved 3D architecture (Figure 1A) consisting of: a highly flexible N‐terminal region (usually ∼10 residues) preceding the first conserved Cys; an irregular loop (the N‐loop; ∼8–10 residues); a single helical turn; three β‐strands forming an antiparallel β‐sheet; and a C‐terminal α‐helix lying across one face of the β‐sheet. The disulfide bonds tether the N‐terminus and N‐loop to the secondary structure core. Most chemokines can form dimers or higher oligomeric structures, which are usually required for binding to GAGs and activity in vivo.10 However, the monomeric form is sufficient (and, in the case of CC chemokines, required) for receptor binding and activation.11, 12, 13, 14, 15, 16

Figure 1.

Chemokines, chemokine receptors, and chemokine: receptor interactions. A, Key features of chemokine proteins (shown in orange) include two disulfide bonds (yellow), N‐terminus (green), N‐loop (magenta), and β3‐region (cyan). CCL2 (PDB: http://firstglance.jmol.org/fg.htm?mol=1DOK) is shown as a representative chemokine. B, Chemokine receptors (e.g., CCR2, PDB: http://firstglance.jmol.org/fg.htm?mol=5T1A) are class A GPCRs consisting of seven transmembrane α‐helices (TM1‐7) linked by three intracellular and three extracellular loops (ICL1‐3 and ECL1‐3, respectively). C, CCR2 has a similar overall fold to other archetypal GPCRs, such as β₂‐adrenoceptor. Overlay depicts CCR2 (blue) and β₂‐AR (magenta; PDB: http://firstglance.jmol.org/fg.htm?mol=2RH1). D, Two‐site, two‐step model for chemokine: chemokine–receptor binding and activation. E, Three‐step model for chemokine: chemokine‐receptor binding and activation. CS1/2: chemokine site 1 or 2; CRS1/2: chemokine receptor site 1 or 2. GPCR, G protein‐coupled receptor

The biological effects of chemokines in humans are mediated through a family of GPCRs. Chemokine receptors, whose structures are discussed in detail below, have the canonical GPCR architecture (Figure 1B), consisting of seven‐transmembrane helices (7TM) connected by three intracellular loops (ICL1‐3) and three extracellular loops (ECL1‐3), a flexible, extracellular N‐terminal region, and an intracellular C‐terminal region that includes an additional α‐helix (helix 8). Conserved disulfide bonds connect the N‐terminus to ECL3 and ECL1 to ECL2. Indeed, chemokine receptors share a similar overall fold with other class A GPCRs, such as the archetypical aminergic β₂‐adrenoceptor,17 albeit with a larger, more‐open extracellular ligand binding pocket, consistent with their ability to bind larger protein ligands (Figure 1C).

3. THE TWO‐SITE, TWO‐STEP MODEL, AND ELABORATIONS

Mutational studies and synthetic modification studies of chemokines have clearly identified two regions that are critical for receptor interactions. Chemokines with modified N‐terminal regions are often impaired in receptor activation (signaling) but can still bind to their receptors, thereby acting as antagonists.18, 19, 20 On the other hand, modifications of the N‐loop region or the spatially adjacent β3‐strand commonly result in reduced binding affinity for receptors.21, 22, 23 Similarly, mutations in the 7TM region of chemokine receptors often affect transmembrane signaling,24, 25 whereas the receptor N‐terminal region is critical for high‐affinity binding of cognate chemokines.26, 27 The importance of the receptor N‐terminal region in chemokine recognition is supported by the observation that blocking tyrosine sulfation in this region can reduce chemokine binding and activation.28, 29, 30 Moreover, peptides (and tyrosine‐sulfated peptides) derived from the receptor N‐terminal region commonly bind to the N‐loop/β3 regions of chemokines.28, 31, 32, 33, 34, 35

Based on the mutational and initial peptide binding data, two groups proposed the “Two‐site Model” for the interactions of chemokines with their receptors.26, 36 In this model (Figure 1D), the chemokine N‐loop (chemokine site 1, CS1) was proposed to bind to the receptor N‐terminal region (chemokine receptor site 1, CRS1). Subsequently, the chemokine N‐terminal region (CS2) was proposed to associate with the receptor ECLs causing receptor activation; later, it became clear than the receptor 7TM region was the major contributor to this second receptor site (CRS2).

The major structural features of the two‐site model have stood the test of time following the structural determination of chemokine receptors in complex with chemokine ligands. However, as originally proposed, the two‐site model assumed (or strongly suggested) that the CS1–CRS1 interaction occurred first and was the major contributor to binding affinity and that the subsequent CS2–CRS2 interaction was concurrent with the receptor conformational change leading to G protein activation. These aspects of the model, sometimes referred to as the “Two‐site, Two‐step Model” to emphasize the presumed sequence of the interactions, became dogma in the field. However, at least three lines of evidence suggest that the CS1–CRS1 interactions may be relatively weak and that CS2–CRS2 interactions may be critical for high‐affinity binding. Firstly, mutations in CS2 can affect receptor binding affinity.20, 22, 24, 36 Secondly, some chemokines (or modified chemokines) act as antagonists, binding tightly to receptors and forming CS2–CRS2 interactions (vide infra) without causing receptor activation.18, 19, 20, 37 And thirdly, the binding affinities of different chemokines to receptor N‐terminal (CRS1) peptides to do not correlate with the affinities of the same chemokine to the intact receptor.38 These observations led Sanchez et al.38 to propose the “Three‐step Model” (Figure 1E), in which the steps are: (a) low affinity, reversible, binding of CS1 to CRS1; (b) association of CS2 with CRS2 to form a high affinity but not yet activated complex; and (c) receptor conformational rearrangement and activation.

Although the “Three‐step Model” accounts for separation of high‐affinity binding and activation, and is generally consistent with the high resolution structural data discussed below, it makes the same assumption as the two‐site, Two‐step model regarding the sequence of the interactions. Verification of this sequence will require kinetics experiments, which are technically challenging. Moreover, as detailed by Volkman and colleagues,39 a variety of additional observations point to the need for further refinement of the mechanistic model. Possible factors to consider include: involvement of additional structural elements in the interactions21, 40, 41, 42; allosteric cooperativity between the two sites43, 44; preferential interactions of chemokines with different parts of the same receptor24, 45, 46; “biased” stimulation of different intracellular signaling pathways by different chemokines at same receptor47, 48, 49, 50, 51; receptor activation by some CXC chemokine dimers11, 52, 53; receptor homo‐ or hetero‐dimerization54; and the effects of receptor or chemokine posttranslational modifications.28, 30, 46, 55 In summary, while the models shown in Figures 1D,E capture key features of chemokine–receptor interactions, the details will need to be further experimentally elaborated and may be specific to each chemokine–receptor pair.

4. CHARACTERIZATION OF CS1–CRS1 INTERACTIONS BY NMR

The N‐terminal region of chemokine receptors (CRS1) is highly flexible and not well resolved in crystal structures, even with bound chemokines (discussed below). However, peptides derived from this region bind to chemokines in solution, enabling the interaction surfaces to be mapped by NMR spectroscopy.31, 56, 57 Although such peptides typically bind with low affinity, tyrosine sulfation, or other chemical modifications can stabilize these complexes.28, 33, 34 These approaches have enabled structures to be determined for CXCL8 bound to a peptidomimetic based on CXCR1 and for CXCL12, CCL11, and CCL5 bound to peptides or sulfopeptides derived from their respective receptors CXCR4, CCR3, and CCR5 (Figure 2).29, 40, 52, 58, 59

Figure 2.

Chemokine structures determined in complex with receptor N‐terminal peptides by solution NMR. A, CCL5:CCR5 N‐terminal sulfopeptide (PDB: http://firstglance.jmol.org/fg.htm?mol=6FGP). B, CCL11:CCR3 N‐terminal sulfopeptide (PDB: http://firstglance.jmol.org/fg.htm?mol=2MPM). C, CXCL8:CXCR1 N‐terminal peptidomimetic (PDB: http://firstglance.jmol.org/fg.htm?mol=1ILQ). D, Monomeric CXCL12 bound to CXCR4 N‐terminal peptide (PDB: http://firstglance.jmol.org/fg.htm?mol=2N55). E, Trapped dimeric CXCL12 bound to two molecules of a CXCR4 N‐terminal sulfopeptide (PDB: http://firstglance.jmol.org/fg.htm?mol=2K05). In (A)–(D), chemokines are colored as in Figure 1A and peptides are shown as blue ribbons, with aromatic side chain sticks. In (E), one protomer and sulfopeptide have the same color scheme as in (A)–(D), and the second protomer and sulfopeptide are shown in gray

The unifying feature of these structures is that the CRS1 peptide always interacts with the N‐loop/β3 region of the chemokine (CS1). This is consistent with NMR chemical shift mapping data for other chemokine–receptor pairs and with much mutational data, supporting the importance of CS1–CRS1 interactions. However, there are also substantial differences between the positions and orientations of the receptor peptides in these complexes (Figure 2). It remains to be determined whether these structural differences accurately reflect distinct modes of interaction between these chemokines and their intact receptors.

5. CHEMOKINE–RECEPTOR STRUCTURES IN COMPLEX WITH CHEMOKINES, PEPTIDES, AND SMALL MOLECULES

5.1. Chemokine–receptor structures

At present, there are distinct structures for seven different chemokine receptors—CXCR1, CXCR4, CCR2, CCR5, CCR7, CCR9 and the viral chemokine–receptor CX3CR1 homologue US28 (Table 1). As these receptor structures have been variously solved in complex with chemokine ligands, modified peptides, and small‐molecule antagonists, the structures demonstrate the diversity of ligand binding sites and provide valuable insights into the structural basis for chemokine–receptor activation (Figure 3). Here, we first discuss the challenges associated with the elucidation of GPCR structures and then present the chemokine‐bound structures to highlight the important chemokine: receptor interactions. We then compare and contrast the binding modes for small‐molecule antagonists, which may have important implications for the therapeutic targeting of chemokine receptors.

Table 1.

Summary of available chemokine‐receptor structures and their associated ligands

Receptor	Ligand type	Ligand	Technique	PDB entry	Reference
CXCR1	Ligand free		Solid‐state NMR	http://firstglance.jmol.org/fg.htm?mol=2LNL	Park et al.60
US28	Ligand free		X‐ray	http://firstglance.jmol.org/fg.htm?mol=5WB1	Miles et al.61
CXCR4	Viral chemokine antagonist	vMIP‐II	X‐ray	http://firstglance.jmol.org/fg.htm?mol=4RWS	Qin et al.41
CCR5	Chemokine antagonist	[5P7]CCL5	X‐ray	http://firstglance.jmol.org/fg.htm?mol=5UIW	Zheng et al.62
US28	Chemokine agonist	CX3CL1	X‐ray	http://firstglance.jmol.org/fg.htm?mol=4XT3	Burg et al.63
US28	Chemokine agonist	CX3CL1	X‐ray	http://firstglance.jmol.org/fg.htm?mol=4XT1	Burg et al.63
US28	Engineered chemokine agonist	CX3CL1.35	X‐ray	http://firstglance.jmol.org/fg.htm?mol=5WB2	Miles et al.61
CXCR4	Cyclic peptide antagonist	CVX15	X‐ray	http://firstglance.jmol.org/fg.htm?mol=3OE0	Wu et al.64
CXCR4	Small‐molecule antagonist	IT1t	X‐ray	http://firstglance.jmol.org/fg.htm?mol=3ODU	Wu et al.64
CXCR4	Small‐molecule antagonist	IT1t	X‐ray	http://firstglance.jmol.org/fg.htm?mol=3OE6	Wu et al.64
CXCR4	Small‐molecule antagonist	IT1t	X‐ray	http://firstglance.jmol.org/fg.htm?mol=3OE8	Wu et al.64
CXCR4	Small‐molecule antagonist	IT1t	X‐ray	http://firstglance.jmol.org/fg.htm?mol=3OE9	Wu et al.64
CCR2	Small‐molecule orthosteric/allosteric antagonists	BMS‐681 and CCR2‐RA‐[R]	X‐ray	http://firstglance.jmol.org/fg.htm?mol=5T1A	Zheng et al.65
CCR2	Small‐molecule antagonist	MK‐0812	X‐ray	http://firstglance.jmol.org/fg.htm?mol=6GPS	Apel et al.66
CCR2	Small‐molecule antagonist	MK‐0812	X‐ray	http://firstglance.jmol.org/fg.htm?mol=6GPX	Apel et al.66
CCR5	Small‐molecule antagonist	Maraviroc	X‐ray	http://firstglance.jmol.org/fg.htm?mol=4MBS	Tan et al.67
CCR5	Small‐molecule antagonist	Compound 21	X‐ray	http://firstglance.jmol.org/fg.htm?mol=6AKX	Peng et al.68
CCR5	Small‐molecule antagonist	Compound 34	X‐ray	http://firstglance.jmol.org/fg.htm?mol=6AKY	Peng et al.68
CCR7	Small‐molecule allosteric antagonist	Cmp2105	X‐ray	http://firstglance.jmol.org/fg.htm?mol=6QZH	Jaeger et al.69
CCR9	Small‐molecule allosteric antagonist	Vercirnon	X‐ray	http://firstglance.jmol.org/fg.htm?mol=5LWE	Oswald et al.70
CCR5	HIV‐1 envelope spike, CD4 complex		Cryo‐electron microscopy	http://firstglance.jmol.org/fg.htm?mol=6MEO	Shaik et al.71

Abbreviation: NMR, nuclear magnetic resonance.

Figure 3.

Chemokine–receptor structures bound to their respective ligands. Shown are ribbon representations of receptors with overlaid space‐filling representations of all ligands for which bound structures are available. A, Ligand‐free CXCR1. B, CXCR4 bound with viral chemokine vMIP‐II (orange), cyclic peptide antagonist CVX15 (yellow), and a small‐molecule antagonist IT1t (red). C, CCR2 with small‐molecule orthosteric agonists MK‐0812 (yellow), BMS‐681 (red), and allosteric antagonist CCR‐2A [R] (magenta). D, CCR5 with chemokine antagonist [5P7]CCL5 (orange), and small‐molecule antagonists Maraviroc (yellow), compound 21 (red), and compound 34 (cyan). E, CCR9 with allosteric antagonist vercirnon (magenta). F, US28 with chemokine agonist CX3CL1 (orange) and engineered chemokine CX3CL1.35 (yellow). In addition to the receptors shown, the structure of CCR7 bound to allosteric ligand Cmp2105 was reported recently but has not yet been released by the PDB69

5.2. Challenges associated with structural determination of GPCRs

According to the generalized understanding of GPCR activation, the receptor N‐terminus, extracellular loops and the exposed surface of the transmembrane domains are involved in ligand binding, whereas the intracellular regions and loops are involved in G protein‐coupling and signal transduction. Indeed, the conformational heterogeneity related to the flexibility of the receptor N and C‐termini and the loops that connect the transmembrane domains, combined with low natural protein abundance, low thermal stability and membrane‐localization all have posed fundamental challenges to the successful GPCR crystallization. However, with the development and refinement of techniques to facilitate crystallization in lipidic cubic phase,72 as well as truncating flexible receptor regions, introducing stabilizing mutations and receptor‐fusion proteins such as T4 lysozyme73 or stabilizing antibodies,74 the number of GPCR crystal structures has surged in the past decade (currently there are structures for 62 unique receptors, https://www.gpcrdb.org/structure/statistics).75 Indeed, as indicated below, structural elucidation of chemokine receptors has been possible in the context of these technological advances. In addition, the recent application of cryo‐electron microscopy to investigate the functional interactions of GPCRs, heterotrimeric G proteins and other associated proteins will likely provide further structural insights on chemokine receptors.76

5.3. Structure of ligand‐free CXCR1

The only ligand‐free structure of a chemokine receptor reported to date is that of full‐length, unmodified CXCR1 in liquid crystalline phospholipid bilayers, which was determined by Park and colleagues using solid‐state NMR spectroscopy.60 This structure confirmed that CXCR1 adopts a similar overall fold to other GPCRs and also revealed well‐defined structural features in flexible regions that are often modified for crystallization, including two disulfide bonds, the three intracellular loops, and the C‐terminal helix 8 (Figure 3A). Although the CXCR1 structure provides a methodological basis for structure determination of other GPCRs by solid‐state NMR, the approach is quite laborious and has not been widely adopted.77, 78

5.4. Chemokine receptors in complex with chemokine ligands

To date, three structures have been reported for chemokine receptors bound to chemokines. In two of these structures the chemokine is an antagonist so the receptor is in an inactive state, whereas the third structure contains a chemokine agonist and activated receptor. A key aspect of these interactions is the ability of different ligands to make contact with two subpockets (designated major and minor) within the binding site of the receptor.

5.5. CXCR4 + vMIP‐II

The first human chemokine–receptor structure crystallized with a chemokine ligand was CXCR4, determined in complex with a viral chemokine antagonist, vMIP‐II at 3.1 Å resolution (Figures 3B and 4A).41 This structure was one of several ground‐breaking advances in the field reported by Handel, Kufareva and colleagues. vMIP‐II was selected as it is a high‐affinity antagonist of CXCR4, yet further manipulations were required to overcome the substantial challenges relating to the cocrystallization of a protein ligand and receptor in lipidic cubic phase. In addition to the insertion of T4 lysozyme in the third intracellular loop, and thermostabilizing mutations, a non‐native disulfide bond was used to trap an irreversible complex, enabling structural determination. The structure revealed an extensive CXCR4:vMIP‐II interaction interface (1,330 Ų), with every residue in the chemokine N‐terminus and N‐loop making contacts with the receptor (Figure 4A). This interaction spanned the CRS1 and CRS2 sites of the paradigmatic two‐site model for chemokine: receptor binding and also incorporated an intermediate region, named CRS1.5, where the receptor and chemokine form a short antiparallel β‐sheet (Figure 4A). The vMIP‐II N‐terminus formed hydrogen bonds with D97^2.63 and E288^7.39 in the minor subpocket.¹ There are two other important inferences from this chemokine‐bound structure that shape our understanding of chemokine–receptor function. Firstly, in keeping with previous reports about the potential dimerization of chemokine receptors, the overall chemokine binding pose implies that a receptor dimer could accommodate two monomeric chemokine ligands. Secondly, the CXCR4:vMIP‐II provides a structural rationale for the observation that dimeric CXC but not CC chemokines can bind and activate their receptors.11, 14

Figure 4.

Chemokine‐bound chemokine receptors show three distinct regions of interaction, CRS1 (green), CRS1.5 (cyan), and CRS2 (blue). A, CXCR4 (yellow ribbon structure) bound to vMIP‐II (grey surface representation). B, CCR5 bound to [5P7]CCL5. C, US28 bound to CX3CL1. D, Superposition of CXCR4:vMIP‐II (gray) (PDB: http://firstglance.jmol.org/fg.htm?mol=4RWS), CCR5:[5P7]CCL5 (yellow; PDB: http://firstglance.jmol.org/fg.htm?mol=5UIW) and US28:CX3CL1 (blue; PDB: http://firstglance.jmol.org/fg.htm?mol=4XT1), showing the outward movement of TM6 in the active receptor conformation

5.6. CCR5 + [5P7]CCL5

The structure of CCR5 has also been solved in complex with a chemokine antagonist [5P7]CCL5 (at 2.2 Å), representing the first structure where both the ligand and receptor are specific to a single (C–C) subfamily (Figure 3D and 4B).62 Consistent with the CXCR4:vMIP‐II structure, the disulfide of [5P7]CCL5 interacts with residues in the intermediate CRS1.5 region of CCR5, including the conserved Pro19‐Cys20 motif of the receptor. More strikingly, the [5P7]CCL5 globular core is oriented toward ECL2 so that the chemokine resides ∼7 Å deeper than vMIP‐II in the receptor binding pocket (Figure 4B). [5P7]CCL5 is also unique among the chemokine: chemokine–receptor structures in its ability to occupy both the major and minor subpockets of its receptor (Figure 4B). Importantly from a drug design perspective, the [5P7]CCL5 interacts with CCR5 in a very similar manner to the HIV inhibitor Maraviroc (see below).

5.7. US28 + CX3CL1

The viral chemokine–receptor US28 structure has been solved in complex with the human chemokine agonist CX3CL1 both in the presence and absence of a nanobody bound to the intracellular face of the receptor (Figure 3F and 4C).63 These structures are essentially identical in terms of the CX3CL1 binding mode, with the only differences a slightly improved resolution with nanobody (2.9 vs. 3.1 Å in nanobody‐free) and a different orientation of helix 8. Similar to the CXCR4:vMIP‐II structure, the globular body of CX3CL1 sits on top of the US28 extracellular vestibule, whereas the chemokine N‐terminus interacts with the transmembrane domains in the central core of US28 where it occupies the minor subpocket of the orthosteric binding site. More specifically, the four N‐terminal residues of CX3CL1 form a hook‐like conformation that makes contacts with residues on TM1, TM3, TM7, and ECL2, in particular with Glu^7.39 that is considered important for chemokine–receptor signaling (Figure 4C). It is therefore possible that the structure of CX3CL1's N‐terminal hook could serve as a structural template for the design of small molecules to modulate chemokine receptors.

In contrast to the other chemokine–receptor structures determined, US28 adopts the features of an active‐state–like conformation (Figure 4C,D). In particular, the outward movement of TM6 (∼9 Å relative to its position in the CCR5 and CXCR4 structures) and the inward movement of the intracellular portion of TM7 toward the center of the transmembrane bundle, closely resemble the active β₂‐adrenergic structure.80 In addition, the positions of the DRY motif (Asp128^3.49, Arg129^3.50, and Tyr130^3.51) in TM3 and the NPXXY motif (Asn287^7.49, Pro288^7.50, and Tyr291^7.53) in TM7, which are both important for G protein activation, are consistent with an active receptor conformation.25 Interestingly, in a follow up study using a G protein‐biased modified form of CX3CL1 (named CX3CL1.35), a marginally different US28 binding mode was observed.61 As these chemokines occupy overlapping and spatially segregated regions of the US28 orthosteric pocket, these data suggest that G protein activation may be mediated as a consequence of steric bulk distorting the walls of the major pocket of the receptor.61 Consistent with this interpretation, in an additional structure without a chemokine bound, US28 displayed a constricted extracellular domain with an inward collapse of ECL1 and ECL2.61

It is worth noting that the chemokine–receptor N‐terminus was not resolved in these three chemokine‐bound structures. In order to address this shortfall, Ziarek et al. produced an experimentally‐validated hybrid model, combining molecular modeling with NMR and X‐ray structural data.40 Interestingly, this model suggested that chemokine residues outside of sites CS1 and CS2 may contribute to receptor binding and activation.

5.8. Chemokine receptors in complex with peptides and small molecules demonstrate various modes of inhibition

5.8.1. CXCR4

The first chemokine–receptor crystal structure to be determined was CXCR4, both in complex with the small‐molecule isothiourea derivative antagonist IT1t, as well as bound to a cyclic peptide, CVX15 (Figures 3B and 5A).64 Analogous to the approach used to enhance crystallization of the β₂‐adrenoceptor,73 insertion of T4 lysozyme in the third intracellular loop, in combination with thermostabilizing mutations were used to improve stability of the receptor for crystallization. The derived CXCR4 models lacked the 26 N‐terminal residues of the receptor, as there was no interpretable density for this region. As the first receptor capable of binding to small proteins, there were appreciable differences in the positioning and orientation of the TM helices 1, 2, and 7 compared to the other crystallized GPCRs. Likewise, presumably to accommodate chemokine ligands, the binding cavity in CXCR4 is larger, more open and is also located closer to the extracellular surface, in comparison to other GPCRs that bind small molecules. These features have been observed in subsequent chemokine–receptor structures. The IT1t occupies the minor subpocket, where it makes contacts with residue side chains from TM1 (E32^1.26), TM2 (W94^2.60, D97^2.63), ECL1 (W102), TM3 (Y116^3.32), ECL2 (R183, I185, C186, D187, R188), and TM7 (E288^7.39; Figure 5A). There is some overlap between the binding sites for CVX15 and IT1t (e.g., residues D187 and R188; Figure 5A), although, CVX15 also forms additional contacts with TM5 and 6 due to its larger size. Whereas IT1t binds within the minor subpocket, reminiscent of the chemokine vMIP‐II, CVX15 binds deeper into the major subpocket (Figure 5A).41, 64 Importantly, these differences between the chemokine/IT1t and CVX15 peptide binding sites in CXCR4 may provide the basis for designing improved modulators that occupy both the minor and major subpockets.

Figure 5.

Chemokine receptors bound to orthosteric small‐molecule ligands. A, CXCR4 showing different mode of binding to CVX15 (yellow) and IT1t (red). Blue sticks depict interacting residues to IT1t (minor subpocket), green sticks show interacting residues to CVX15 (major subpocket), and cyan‐colored sticks (largely obscured) show residues interacting to both IT1t and CVX15. The dashed line represents the boundary between the major and minor subpockets. B, CCR2 bound to MK‐0812 (yellow) and BMS‐681 (red). MK‐0812 and BMS‐681 bind in a similar way to CCR2 and mostly occupy the minor subpocket. C, CCR5 bound to Maraviroc, which binds deeper into the structure in comparison to CXCR4 and CCR2 ligands. Maraviroc also interacts with E283^7.39 at the intersection of major and minor subpocket and W86^2.60 (equivalent to E288 and W94 in CXCR4, respectively). Note: Ballesteros–Weinstein receptor residue numbering has been omitted for clarity

5.8.2. CCR2

CC chemokine receptor 2 (CCR2) is predominantly expressed in monocytes, as well as in dendritic cells and T cell subpopulations. The CCR2 signaling pathway has been broadly implicated in inflammatory and neurodegenerative diseases including atherosclerosis, multiple sclerosis, neuropathic pain, and allergic rhinitis. Accordingly, CCR2 has been the target of numerous clinical trials, although no therapies have reached the market to date (see http://www.clinicaltrials.gov). In separate studies, CCR2 was crystallized with the orthosteric antagonist, MK‐0812 (Figures 3C and 5B),66 and in ternary complex with the small‐molecule antagonists BMS‐681 and CCR2‐RA (Figure 3C).65 In each case, to enable crystallization, CCR2 was truncated at the flexible N‐ and C‐termini and was fused to rubredoxin66 or T4 lysozyme.65 Moreover, in the Zheng et al. study, the simultaneous addition of two antagonists markedly stabilized the protein to facilitate crystallization.65 There was a high degree of similarity between the CCR2 structures (RMSD of 0.46 Å between equivalent Cα atoms).66

MK‐0812 is a small‐molecule CCR2/CCR5 dual antagonist developed by Merck as a candidate to treat multiple sclerosis, although it was terminated following a Phase II clinical trial where it did not show efficacy. MK‐0812 and BMS‐681 are orthosteric antagonists that occupy overlapping positions in the orthosteric pocket, predominantly in the minor subpocket (Figure 5B), where they presumably compete with chemokine binding.65 The MK‐0812 bound structure confirmed the key interacting binding site residues (Y49^1.39, W98^2.60, Y120^3.32, and E291^7.39) and highlight the important contribution of residue E291^7.39 to high‐affinity binding of CCR2 antagonists (Figure 5B).66 Furthermore, based on structure‐guided molecular modeling, the authors postulated that chemically distinct, selective CCR2 antagonists such as Ex15 (patent WO2012/171863) largely superimpose with MK‐0812, but also form a direct hydrogen bond with residue H121^3.33 and extend deeper into the lipid bilayer beyond the helical bundle. These observations provide an elegant structural explanation for CCR2 selectivity, as H121 is nonconserved in CCR5, where the corresponding residue (F109^3.33) does not form favorable interactions with the sulfonamide moiety of Ex15.

In contrast, CCR2‐RA‐[R] binds to a novel allosteric pocket, more than 30 Å away from the orthosteric site, on the intracellular aspect of the receptor (Figure 6A).65 With a balanced combination of hydrophobic and polar features, the CCR2‐RA‐[R] site represents a favorable site for drug development, ideally positioned to modulate the CCR2–G protein interaction. The authors also noted that there is evidence for similar intracellular pockets in other chemokine receptors, including CCR1, CCR5, CCR4, CXCR1, and CXCR2.65, 81, 82 Intriguingly, there seems to be positive cooperativity between the two antagonist binding sites, despite the distance that separates them, whereby BMS‐681 enhances the binding of [³H]CCR2‐RA by >30%. This is consistent with the observation that the double antagonist bound CCR2 is the “most inactive” GPCR structure observed to date, sharing the conformational microswitches in the intracellular portion of TM3 and TM7 that disrupt the receptor–G protein interface.65 Taken together, these CCR2 structures may help to guide the development of more selective and efficacious novel CCR2 antagonists.

Figure 6.

Chemokine receptors bound to allosteric small‐molecule ligands. A, CCR2 bound to CCR2‐RA[R] (magenta), with interacting resides shown in blue. B, CCR9 bound to vercirnon (magenta). These allosteric ligands bind to similar intracellular binding sites within the transmembrane bundle where they interact with conserved residues (CCR2: T77^2.39, Y305^7.53, G309^8.47, E310^8.48, K311^8.49 and F312^8.50; CCR9: T83^2.39, Y317^7.53, G321^8.47, E322^8.48, R323^8.49, and F324^8.50). Note: Ballesteros–Weinstein receptor residue numbering has been omitted for clarity

5.8.3. CCR5

The receptor CCR5 has been a priority target for structural determination, particularly given its function as a coreceptor for HIV‐1 viral entry. The first CCR5 structure was solved at 2.7 Å resolution in complex with the marketed HIV drug Maraviroc.67 Maraviroc bound at a distinct site buried deeper within the CCR5 transmembrane bundle compared to other chemokine receptors (Figure 5C). Spaning the major and minor subpockets, Maraviroc sits in a pocket formed by residues from TM1, 2, 3, 5, 6, and 7, including E283^7.39 and W86^2.60 that have been implicated in CXCR4 ligand binding (Figure 5A,C).64, 67 Of particular note, the phenyl group of Maraviroc makes contact with five aromatic residues (Y108^3.32, F109^3.33, F112^3.36, W248^6.48, and Y251^6.51) deep within the pocket that have been shown to be critical for ligand binding in mutagenesis studies.67

As a fine demonstration of the utility of GPCR crystallization, Peng and colleagues optimized the binding and functional properties of Maraviroc based on the CCR5 structure.68 Thus, using structure‐based drug design they developed analogs with improved anti‐HIV potency, bioavailability and reduced cytochrome P450 inhibition (e.g., compound 21 and 34; Figure 3D).68 These compounds represent promising drug candidates for the treatment of HIV‐infection.

5.8.4. CCR9

The activation of CCR9 promotes leukocyte recruitment in the gut. Thus CCR9 antagonists have been developed for the treatment of inflammatory bowel disease. One such small‐molecule antagonist, vercirnon progressed to Phase 3 clinical trials for Crohn's disease, although the efficacy was limited and the compound did not reach the market. Vercirnon has been crystalized in complex with CCR9 at 2.8 Å, using an alternate thermostabilization approach called StaR (Figure 3E).70 Along with truncated N‐ and C‐termini, this StaR CCR9 construct contains eight amino acid substitutions that did not alter the vercirnon binding properties, but abrogated the requirement for fusion partners during crystallization. Nevertheless, the CCR9 structure shares a high degree of similarity in the transmembrane domains with other chemokine receptors. Remarkably, in the crystal structure vercirnon binds within the helical bundle to the intracellular side of the receptor, contacting residues from TM1, 2, 3, 6, 7, and helix 8 (Figure 3E and 6B). In this position, approximately 33 Å from the orthosteric site in an allosteric pocket with cytoplasmic access, vercirnon can exert allosteric antagonism and prevent G protein‐coupling or arrestin binding, analogous to the simultaneously published CCR2‐RA‐[R] structure (Figure 6A).65

6. REMAINING QUESTIONS AND FUTURE DIRECTIONS

Our knowledge of the ligand binding and activation of chemokine receptors has been transformed by structural studies in the past decade. Mechanistic interpretation of these structures is guided by numerous previous studies on chemokine structure and dynamics, chemokine–receptor pharmacology and the effects of both chemokine and receptor mutations. By combining the insights amassed from these studies, it is becoming possible to design drugs with desirable properties and improved receptor selectivity.

Nevertheless, there are many outstanding questions that must be addressed to yield a fuller understanding of chemokine–receptor activation. For example, what are the “rules” governing chemokine: receptor selectivity? What are the critical structural requirements that enable transition of a receptor from inactive to active state? Alternatively, what structural features result in stabilization of the inactive state (receptor antagonism)? What are the details of interactions with G proteins and other intracellular effectors? What active‐state conformations bias the receptor toward signaling via specific pathways? And how are chemokine: receptor interactions regulated by posttranslational modifications or by receptor dimerization?

While the studies discussed above have begun to address these questions, it will be critical to assess whether the conclusions are broadly applicable across the chemokine and receptor families or are specific to each chemokine: receptor pair. As additional structures are obtained, such as the recent cryo‐electron microscopy structure of CCR5 in complex full‐length HIV envelope spike and CD4,71 and complemented with other powerful experimental and computational approaches, there is reason to be optimistic that the next few years will yield substantial progress in both fundamental understanding as well as therapeutic benefits in the treatment of inflammatory diseases.

Bhusal RP, Foster SR, Stone MJ. Structural basis of chemokine and receptor interactions: Key regulators of leukocyte recruitment in inflammatory responses. Protein Science. 2020;29:420–432. 10.1002/pro.3744