Molecular basis of promiscuous chemokine binding and structural mimicry at the C-X-C chemokine receptor, CXCR2

Summary

Selectivity of natural agonists for their cognate receptors is a hallmark of G-protein-coupled receptors (GPCRs); however, this selectivity often breaks down at the chemokine receptors. Chemokines often display promiscuous binding to chemokine receptors, but the underlying molecular determinants remain mostly elusive. Here, we perform a comprehensive transducer-coupling analysis, testing all known C-X-C chemo-kines on every C-X-C type chemokine receptor to generate a global fingerprint of the selectivity and promiscuity encoded within this system. Taking lead from this, we determine cryoelectron microscopy (cryo-EM) structures of the most promiscuous receptor, C-X-C chemokine receptor 2 (CXCR2), in complex with several chemokines. These structural snapshots elucidate the details of ligand-receptor interactions, including structural motifs, which are validated using mutagenesis and functional experiments. We also observe that most chemokines position themselves on CXCR2 as a dimer while CXCL6 exhibits a monomeric binding pose. Taken together, our findings provide the molecular basis of chemokine promiscuity at CXCR2 with potential implications for developing therapeutic molecules.

Graphical abstract.

Introduction

Chemokines are small proteins secreted by immune cells that play critical roles in a myriad of physiological processes, including cellular migration and inflammatory responses, by activating chemokine receptors.^1,2 Chemokine receptors belong to the superfamily of G-protein-coupled receptors (GPCRs) with primary coupling to the Gαi subtype of heterotrimeric G-proteins and β-arrestins (βarrs).^3,4 They are expressed on a variety of immune cells with wide-ranging contributions to various aspects of our immune response mechanisms, and their aberrant signaling is implicated in multiple disease conditions, including cancer,^5,6 allergy,^7,8 psoriasis,^9,10 atherosclerosis,¹¹ and autoimmune disorders.^12,13 As a result, the chemokine receptors continue to be highly sought-after drug targets with multiple on-going clinical trials with indications for psoriasis, chronic obstructive pulmonary disease, and asthma, in addition to previously approved drugs such as C-C chemokine receptor 5 (CCR5) antagonist (Maraviroc) for AIDS¹⁴ and C-X-C chemokine receptor 4 (CXCR4) antagonist (plerixafor) for non-Hodgkin’s lymphoma.¹⁵ Moreover, a naturally encoded repertoire of more than fifty different chemokines and nearly twenty chemokine receptor subtypes also makes them an interesting system to explore the molecular principles of ligand-receptor interaction and biased signaling.

While chemokine receptors exhibit a conserved seven transmembrane architecture characteristic of prototypical GPCRs, their interaction with chemokines does not always follow the exclusive natural agonist selectivity displayed by the majority of GPCRs.¹⁶ This holds true for both C-C and C-X-C type chemokine receptors, as well as for atypical chemokine receptors. For example, CCR3 can be recognized and activated by almost a dozen different chemokines, albeit their potency and efficacy differ significantly. Similarly, the Duffy antigen receptor for chemokines (DARC), also known as the atypical chemokine receptor subtype 1 (ACKR1), displays cross-reactivity across multiple C-C and C-X-C chemokines.^17–19 At the same time, some of the chemokine receptors exhibit robust selectivity and recognize only one (e.g., CXCL12-CXCR4) or two (e.g., CXCL6/CXCL8-CXCR1) chemokines. Despite emerging structural insights into chemokine recognition by chemokine receptors,^20–31 the molecular determinants underlying the inherent ligand promiscuity remain an enigma and represent an important knowledge gap in our current understanding of GPCR activation and signaling paradigms.

In this backdrop, we set out to elucidate the molecular mechanism driving ligand promiscuity and selectivity at the C-X-C subtype chemokine receptors using a combination of biochemical, pharmacological, and structural approaches. We first performed a comprehensive transducer-coupling analysis to generate a global fingerprint of the selectivity and promiscuity of the C-X-C chemokines, followed by structure determination of CXCR2 in complex with multiple chemokines. These structural snapshots, together with site-directed mutagenesis experiments and cellular assays, allowed us to uncover the molecular mechanism underlying the striking chemokine promiscuity at CXCR2. Our study provides a conceptual framework that may be broadly applicable to understand the functional fine-tuning encoded in the chemokine-chemokine receptor system.

Results

Global profiling of chemokine selectivity and promiscuity

Considering that the notion of ligand promiscuity at chemokine receptors is based primarily on multiple scattered studies in the literature using different assays and readouts, we first measured the transducer-coupling profile of all known C-X-C chemokines on each of the C-X-C subtype chemokine receptors using G-protein recruitment, βarr2 interaction, and GRK3 recruitment assays in parallel (Figures 1A and 1B). We observed that CXCR2 exhibits the highest level of promiscuity, being activated by seven different chemokines, although the potency and efficacy vary across the ligands (Figures 1B, 1C, and S1). In fact, a direct comparison of previously reported binding affinity rank order of different CXCLs with that of the functional responses measured in our study suggests a reasonable correlation and therefore indicates that the functional responses of CXCLs primarily reflect their affinity for CXCR2. On the other hand, CXCR5 and CXCR6 display the highest degree of selectivity and are activated only by CXCL13 and CXCL16, respectively (Figure 1B). We also observed that despite the high degree of chemokine promiscuity, CXCR2 still maintains some level of selectivity and fails to exhibit any measurable functional response for several C-X-C chemokines such as CXCL4 and CXCL9-12 (Figure 1B). This is intriguing because the overall structural fold of C-X-C chemokines is highly conserved, comprising of three anti-parallel β strands followed by a carboxyl-terminal α helix.³ CXCR2 is expressed on a variety of immune cells, including neutrophils, mast cells, monocytes, and macrophages, as well as endothelial and epithelial cells,^32–35 and plays an important role in a multitude of cellular and physiological processes such as neutrophil diapedesis, mobilization of neutrophils from the bone marrow to the blood, and neutrophil recruitment in response to microbial infection and tissue injury.^36,37 Thus, it is tempting to speculate that despite chemokine binding promiscuity, there exists some level of functional specialization that fine-tunes context-dependent interaction and activation of the receptor. A better understanding of the same may help surmount the inherent challenges in selectively targeting CXCR2 under various disease conditions such as chronic inflammation, cancer progression, psoriasis, atherosclerosis, pulmonary diseases, sepsis, and neuroinflammation.^{5,10,11,32,38–40}

Figure 1. Transducer-coupling profile of C-X-C chemokines.

(A) Schematic representation of promiscuity and selectivity observed within the C-X-C chemokine receptor family.

(B) Heatmap showing functional selectivity of all C-X-C chemokines on all C-X-C receptors as measured in terms of miniGi, βarr2, and GRK3 recruitment at saturating ligand concentration. Data (mean) represent three independent biological replicates normalized with respect to signal observed with most active chemokine agonist, treated as 100%.

(C) Top: heatmap summarizing the logEC₅₀ values exhibited by different chemokines in various assays. Data (mean) represent three to six independent biological replicates, performed in duplicate, and normalized with respect to signal observed at the lowest dose, treated either as 100% (for cAMP response and Gα_oB dissociation) or 1 (βarr1/2 recruitment, βarr1/2 trafficking, and SRE-ERK assay). Bottom: heatmap summarizing the maximal response/Emax observed in the various assays. For each assay, the highest response observed has been treated as 100%, and the rest have been normalized accordingly.

See also Figures S1 and S7.

Structures of CXCR2 reveal distinct binding modalities of chemokines

Taking lead from the chemokine promiscuity fingerprint observed here, we determined the structures of CXCR2 in complex with every interacting chemokine, except CXCL7, and heterotrimeric G-protein, using cryoelectron microscopy (cryo-EM) at resolution ranging from 2.8 to 3.4 Å (Figure 2; Method S1). Although we also reconstituted the complex of CXCL7-CXCR2-G-protein, we did not observe a clear density for CXCL7 in the cryo-EM map, and therefore, we have not included it in further analysis. The overall architecture of CXCL-CXCR2-G-protein structures is quite similar to each other (Figure 2), and their superimposition indicates an overall root-mean-square deviation (RMSD) value lower than 0.5 Å for the receptor component. CXCR2 also exhibits all the typical hallmarks of receptor activation, such as outward movement of TM6 and inward movement of TM7 toward their cytosolic side, as well as rearrangement of conserved motifs (Figure S2). In addition, the G-protein interaction interface is also similar to that previously observed for other GPCR-Gαi-protein complexes and nearly identical across all the CXCR2 structures (Figure S3; Table S1). Interestingly, we observed two unanticipated features in these structures at the level of chemokine binding modality and receptor assembly. All the CXCLs except CXCL6 are positioned on the receptor as dimers, wherein one protomer engages the receptor closely while the other protomer points away without making any substantial contact with the receptor (Figure 2). The dimer interface is conserved across all CXCLs visualized here and mediated via strong hydrophobic interactions, contributed primarily by the residues from the β strand I and C-terminal helix of the individual CXCL protomers (Figures 3A, 3B, S4A, and S4B). It is interesting to note that despite having the hydrophobic residues that form the dimeric interface (Figures S4A and S4B), CXCL6 still exhibits a monomeric binding mode to CXCR2, and structural analysis offers insights into this observation. Superimposition of the structure of CXCL6 with the other CXCLs reveals that the C-terminal helix in CXCL6 swings outward by ∼78° from the core domain, which is likely to pose a steric clash with the other protomer in a dimeric arrangement and thereby precludes dimer formation (Figure 3C). While chemokines are expected to exist in monomer-dimer equilibrium under physiological conditions,^41–43 it is plausible that their relative dimerization propensity differs from one another, and it may be further fine-tuned upon their interaction with the receptor. For instance, dimeric form of CCL2 exhibits markedly reduced potency at CCR2⁴⁴ while dimeric CCL4 is inactive at CCR5.⁴⁵ On the other hand, while both monomeric and dimeric forms of CXCL8 are equally potent in activating signaling downstream of CXCR2, the monomeric form of CXCL8 is several folds more potent than the dimeric form at CXCR1.²³ The orientation of ECL2 in CXCR1 (PDB: 8IC0) has been proposed to be responsible for the same, wherein outward movement of ECL2 in CXCR1 possibly results in clash with the second protomer of CXCL8,²³ preventing engagement of dimeric CXCL8 (Figure 3D). Contrary to this, the differential orientation of ECL2 in CXCR2, as compared with CXCR1, permits the binding of dimeric CXCL8 (Figure 3D).

Figure 2. Overall structures of chemokine-CXCR2 complexes.

(A–F) Map and ribbon diagram of the ligand-bound CXCR2-mGαoA complexes (front view) are depicted. (A) CXCL1-CXCR2-mGαoA: pale violet red: CXCL1-A, light sea green: CXCL1-B, gray: CXCR2, sandy brown: miniGαoA, khaki: Gβ1, chartreuse: Gγ2. (B) CXCL2-CXCR2-mGαoA: cornflower blue: CXCL2-A, medium sea green: CXCL2-B, gray: CXCR2, sandy brown: miniGαoA, khaki: Gβ1, chartreuse: Gγ2, plum: scFv16. (C) CXCL3-CXCR2-mGαoA: Indian red: CXCL3-A, orange: CXCL3-B, gray: CXCR2, sandy brown: miniGαoA, khaki: Gβ1, chartreuse: Gγ2, plum: scFv16. (D) CXCL5-CXCR2-mGαoA: medium slate blue: CXCL5-A, salmon: CXCL5-B, gray: CXCR2, sandy brown: miniGαoA, khaki: Gβ1, chartreuse: Gγ2, plum: scFv16. (E) CXCL6-CXCR2-mGαoA: yellow green: CXCL6, gray: CXCR2, sandy brown: miniGαoA, khaki: Gβ1, chartreuse: Gγ2, plum: scFv16. (F) CXCL8-CXCR2-mGαoA: teal: CXCL8-A, rosy brown: CXCL8-B, gray: CXCR2, sandy brown: miniGαoA, khaki: Gβ1, chartreuse: Gγ2, plum: scFv16.

See also Method S1 and Figures S2 and S3.

Figure 3. Distinct binding mode of CXCL6.

(A) Structural representations of dimeric CXC ligands.

(B) Hydrophobic interactions mediating ligand dimerization.

(C) Comparison of the binding mode of CXCL6 with CXCL1, CXCL2, CXCL3, CXCL5, and CXCL8. The C-terminal helix in CXCL6 shows an outward rotation of ∼78° from the core domain, providing an explanation for its monomeric state.

(D) Superimposition of CXCL8-bound CXCR1 (PDB: 8IC0, orange) and CXCL8-bound CXCR2 (gray) highlights the outward movement of ECL2 in CXCR1 (left). ECL2 of CXCR1 exhibits steric clash with the second protomer of CXCL8 (right).

See also Figure S4.

CXCL5-binding induces a dimeric assembly of CXCR2

Interestingly, the CXCL5-CXCR2 complex forms a dimeric assembly wherein the two protomers of the ligand are arranged in a trans-configuration, with each protomer engaging their own receptor molecule characterized by a large, buried surface area (Figure 2D). This dimeric architecture displays an angle of approximately 110° between the two receptor molecules, with no direct receptor-receptor contact (Figure S4C). The overall interaction of CXCL5 with CXCR2 in each protomer is nearly identical including the interaction interface, receptor conformation, and G-protein interaction interface (Table S1). To confirm that the CXCL5-CXCR2 dimeric assembly observed is not a result of the high protein concentration used for cryo-EM analysis, we carried out single particle negative staining of CXCL5-CXCR2 complex, with CXCL8-CXCR2 as a reference, at a significantly lower protein concentration. We observed distinct dimeric classes of CXCL5-CXCR2 samples but not CXCL8-CXCR2, with the latter exhibiting primarily monomeric assembly (Figure S4D). While class B and C GPCRs are known to form obligate dimers,⁴⁶ so far only one class A GPCR, namely the Apelin receptor has been observed to display a dimer in complex with G-proteins, using cryo-EM.⁴⁷ In addition, the class F fungal GPCR Ste2 has also been visualized as a dimer in two different stoichiometries.^48,49 However, it is worth noting that these previously resolved dimers are mediated exclusively by the receptor-receptor contact interface, unlike the CXCL5-CXCR2 dimeric assembly that is mediated only through the ligand interface. Considering the inter-receptor protomer angle and orientation, it is plausible that such a dimeric arrangement represents a receptor internalizing through membrane invagination (Figure S4C) or two interacting receptor protomers from adjacent cells, although the same remains to be experimentally validated in future studies.

Interaction of C-X-C chemokines with CXCR2

The interaction of chemokines with chemokine receptors is conceptualized around a two binding site mechanism, which are referred to as chemokine recognition sites 1 and 2 (CRS1 and CRS2),⁵⁰ respectively, which is also apparent in the CXCR2 structures (Figure 4A). CRS1, constituted primarily of an interaction of the polar groove within the core domain of the chemokines with the N terminus of the receptor, is crucial for chemokine recognition⁵¹ (Figures 4B and 4C), while on the other hand, CRS2, formed via the positioning of the N terminus of chemokines in the orthosteric pocket of the receptors, is the key driver of receptor activation and signaling (Figure 4D). Additionally, the conserved Pro38 and Cys39 in the N terminus of the chemokine receptors, immediately preceding TM1, form the “PC motif” that helps impart a shape complementarity to the N-terminal loop of the chemokines, and this interaction is also referred to as CRS1.5⁵⁰ (Figure 4B). In the CXCR2 structures, the Cys39^N-term-Cys286^7.25 disulfide bridge in the receptor packs against the conserved disulfide bridges in the chemokines to facilitate the alignment of the N-terminal loop residues of the receptor with the groove residues of the CXCLs (Figure S5A). Furthermore, several hydrogen bonds and ionic contacts help stabilize the flexible N terminus of CXCR2 within the groove of CXCLs as a part of CRS1 (Table S1). Intriguingly, the N terminus of each of the chemokines is positioned in the orthosteric binding pocket at about the same depth, as measured in terms of the distance between the conserved glutamate residue in the chemokines and Trp^6.48 in CXCR2 (Figure 4E). The N terminus of the chemokines exhibits a shallow binding mode upon penetrating into the orthosteric binding pocket and makes extensive contacts within the extracellular vestibule of the TMs, forming the CRS2 (Table S1). It is interesting to note that the N terminus of the receptor-bound chemokines undergoes a conformational transition from a short and compact hook shape in the free-state structures^52–54 to a wide and extended “U-shaped” conformation at the base of the orthosteric pocket, with the N-terminal residues extending away from the pocket (Figure S5B).

Figure 4. Overall binding mode of chemokines.

(A) Representation of the two binding sites engaged by the chemokines on CXCR2. Receptors are shown as ribbons, while chemokines are shown in surface representation. CXCR2: gray; CXCL1 protomers: pink, deep cyan; CXCL2 protomers: blue, green; CXCL3 protomers: red, yellow; CXCL5 protomers: purple, salmon; CXCL6: light green; CXCL8 protomers: teal, deep pink.

(B) CRS1-focused binding modes of each chemokine-CXCR2 complex.

(C) Receptor residues in CRS1 that interact with the chemokine.

(D) Residues of CRS2 in CXCR2 interacting with residues of respective chemokine ligands.

(E) Binding of individual ligands on CXCR2 and depth, as measured from the conserved glutamate of the E-L-R motif on the chemokines, with respect to conserved W^6.48. The highly conserved W^6.48 is highlighted to help infer the depth of insertion of the chemokine N terminus into the orthosteric pocket of CXCR2.

See also Figure S5 and Table S1.

Molecular basis of chemokine promiscuity and selectivity

So, what is the underlying mechanism driving chemokine promiscuity and selectivity at CXCR2? A closer analysis of the CXCL-CXCR2 interaction interface provides important insights into this phenomenon. A set of charged residues in CRS2, namely Arg208^5.35, Arg212^5.39, Arg278^6.62, Asp274^6.58, and Asp293^7.32 of CXCR2, hereafter referred to as the R-D motif, participate in extensive contacts, through hydrogen bonds and ionic interactions, with the N-terminal E-L-R motif of CXCLs. Notably, Arg208^5.35, Arg212^5.39, and Arg278^6.62 form polar interactions with the Glu of E-L-R motif, while Asp274^6.58 and Asp293^7.32 interact with the Arg of the E-L-R motif in every interacting chemokine (Figures 5A and S5C). The three distinct residues with different properties are recognized by specific regions on the receptor—a positively charged, a hydrophobic, and a negatively charged region (Figure 5A). This spatial arrangement and interaction of the R-D motif in CXCR2 with E-L-R motif in CXCL1/2/3/5/6/8 is critical for a common recognition mechanism (Figure 5B). Interestingly, chemokines that are recognized by CXCR2 universally feature the E-L-R motif, while other CXCLs that fail to activate the receptor also lack the E-L-R motif and thus may not form stable interactions with the receptor amenable to receptor activation (Figures 5C, 5D, and S6A). These observations suggest that the spatial positioning of the E-L-R motif in the angiogenic C-X-C chemokines represents a structural mimicry that facilitates chemokine promiscuity at CXCR2. Notably, all six complexes obtained in this study show nearly identical conformations of the E-L-R motif in the C-X-C chemokines in complex with CXCR2 (Figure 5B). Interestingly, we observe that the R-D motif containing three arginine and two aspartate residues engaging the E-L-R motif in CXCLs is specific to CXCR2 (Figure S6B).

Figure 5. Structural basis of selectivity in chemokine binding to CXCR2.

(A) Left: chemokine (CXCL2) E-L-R residues interacting with CXCR2 residues. Right: recognition mechanism of E-L-R motifs in CRS2. CXCR2 is shown by electrostatic potential. CXCL1 is depicted as a ribbon model, with only the E-L-R motif shown as a stick model.

(B) Common binding modes of chemokines in CRS2. Only CXCL1-bound CXCR2 is shown.

(C and D) Alignment of all the C-X-C chemokines. Alignment was generated using the WEBLOGO tool (https://weblogo.threeplusone.com/create.cgi). Dashed line indicates lack of presence of highlighted motif.

See also Figures S5 and S6.

In order to validate these structural observations, we carried out site-directed mutagenesis and functional analysis of the R-D and E-L-R motifs in CXCR2 and C-X-C chemokines, respectively. First, in the context of the R-D motif in CXCR2, we generated two mutants, one with the three arginine residues mutated to alanine (R208A+R212A+R278A) and another with the two aspartate residues mutated to alanine (D274A+D293A). We refer to these mutants as CXCR2^RRR and CXCR2^DD, respectively. Subsequently, we first tested these mutants in G-protein assay, i.e., heterotrimer dissociation and cyclic AMP (cAMP) response upon stimulation with C-X-C chemokines, and observed a significant loss of G-protein activation for both the mutants (Figure 6A). Next, we measured βarr1/2 recruitment to these mutants upon agonist stimulation using a NanoBiT-based assay, and similar to the G-protein response, we observed a dramatic loss of βarr1/2 recruitment (Figure 6A). In addition to the R-D motif in CXCR2, we also mutated the E-L-R motif in four different CXCLs, namely CXCL1, CXCL2, CXCL3, and CXCL8, to alanine and subsequently expressed, purified, and measured their ability to activate CXCR2 in the G-protein activation and βarr1/2 recruitment assays. As presented in Figure 6B, we observed a significant loss of transducer coupling for all these CXCLs compared with their wild-type counterparts. Taken together, these data establish the key contribution of the R-D motif in CXCR2 and the E-L-R motif in CXCLs for receptor activation and transducer coupling and underscore the structural mimicry displayed by the C-X-C chemokines as the mechanism underlying their promiscuous binding (Figure 7).

Figure 6. Functional validation of R-D and E-L-R motif.

(A) Stimulation with different chemokines induces robust G-protein signaling and βarr1/2 recruitment downstream to CXCR2^WT. Mutating R208, R212, and R278 (CXCR2^RRR) and D274 and D293 (CXCR2^DD) to alanine results in a significant loss of both G-protein signaling and βarr1/2 recruitment. Data (mean ± SEM) represent three independent experiments, performed in duplicate, and have been normalized with respect to the signal observed in unstimulated condition, treated as either 100% (for G-protein assays) or 1 (for βarr recruitment).

(B) Mutating the E-L-R residues in CXCL1/2/3/8 to AAA (CXCL1/2/3/8^AAA) results in complete loss of G-protein signaling and βarr recruitment downstream to CXCR2, while the wild-type ligands exhibited robust activity in all the assays. Data (mean ± SEM) represent three independent experiments, performed in duplicate, and have been normalized with respect to the signal observed in unstimulated condition, treated as either 100% (for G-protein assays) or 1 (for βarr recruitment).

See also Figure S7.

Figure 7. A schematic representation of the R-D and E-L-R motif interaction.

C-X-C chemokines harboring the E-L-R motif interact with the R-D motif in the CXCR2, while others lacking this motif do not interact with receptor. The crosstalk of these two structural motifs in the C-X-C chemokines and CXCR2 provides a molecular mechanism underlying promiscuous binding.

Discussion

The precision of agonist-receptor pairing is a key determinant to maintain the level of noise in the signaling networks to a minimum and to ensure specificity and selectivity in downstream signaling responses. Therefore, the evolution of broad promiscuity in the chemokine-chemokine receptor system has been a perplexing phenomenon. A systematic exploration of physiological and evolutionary rationale for chemokine promiscuity to chemokine receptors remains to be explored at systems level including in vivo context. However, it can be speculated that distinct potency and efficacy of different chemokines might represent a dynamic fine-tuning mechanism in the system at the level of receptor activation and signaling. For example, depending on a specific context, a chemokine with lower potency and efficacy may be secreted and utilized to elicit a weaker response compared with another relatively more potent and efficacious chemokine to avoid oversaturation and possible harmful effects. Although we have not observed a dramatic difference in the qualitative engagement of the conventional transducers and effectors by different CXCLs on CXCR2, we still cannot rule out the possibility that different chemokines may engage additional effectors differently or modulate the conformations of canonical transducers such as βarrs to elicit differential functional responses. Future studies using conformational biosensors and global phosphoproteomics approaches may shed light on this and thereby provide additional physiological rationale for chemokine promiscuity.

While a certain level of specificity can be achieved by regulating expression patterns of the chemokines and their corresponding receptors in specific cellular contexts, this may not be sufficient to avoid undesired crosstalk entirely. Our observations reported here suggest that there may exist additional levels of regulatory mechanisms such as the oligomeric state of the interacting agonists or atypical modes of receptor clustering that also impart functional specialization in this system. For example, it is plausible that the dimeric assembly exhibited by CXCR2 observed here upon binding of CXCL5 could promote cell-to-cell contact and clustering, wherein a single agonist in dimeric state is able to activate two receptor molecules on different cells and thereby amplify the cellular responses. Similarly, concentration-dependent oligomeric state of chemokines may allow their preferential interaction with one receptor over the other, as exemplified with the binding modes of CXCL8, i.e., as a monomer to CXCR1 and dimer to CXCR2. However, future studies are essential, especially in cellular and physiological context, to probe these interesting possibilities that potentially fine-tune the functional outcomes encoded in the chemokine-chemokine receptor system.

As mentioned earlier, a similar promiscuous binding of C-C chemokines is also observed for some of the CCRs, such as CCR2 and CCR3. Although structures of some of the CCRs are available in complex with selected C-C chemokines, structural coverage is still not sufficient to identify potentially conserved structural motifs, similar to the R-D motif in CXCR2.⁵⁵ Therefore, future studies focused on visualizing some of the CCRs with multiple C-C chemokines, as executed in this study for CXCR2, may help illuminate the structural basis of promiscuous binding of C-C chemokines.

The chemokine receptor subfamily represents a rich tapestry for future studies to uncover the fundamental principles that guide naturally encoded ligand-receptor pairing and signaling bias at multiple levels. In addition, the availability of these structural snapshots and direct visualization of the conserved principles of chemokine interaction in the orthosteric binding pockets of the corresponding receptors should also facilitate the development of subtype-selective pharmacophores that has proven challenging so far. Taken together, our study provides a glimpse of the molecular mechanism based on structural mimicry to drive promiscuous binding of chemokines to CXCR2 and also offers a framework to guide ligand discovery with therapeutic potential at chemokine receptors.

Limitations of the study

We note that we did not observe a dramatic qualitative difference among the CXCLs in terms of the recruitment of canonical signal transducers such as G-protein, βarr2, and GRK3, at least in terms of the maximal response. However, there might be differences at spatio-temporal levels or in terms of additional transducers/effectors, which are not explored in this study. Therefore, the functional outcomes resulting from the distinct binding modes of CXCL5/CXCL6 remain an open question, and follow-up studies in primary cell lines may help illuminate this further. We also note that the visualization of CXCL5-induced dimeric assembly of CXCR2 in cellular context and the functional consequences of the same remain to be explored further in follow-up studies, which may provide additional interesting insights into functional specialization elicited by chemokines in cellular and physiological context.

Resource Availability

Lead contact

Further information and requests for reagents should be addressed to the lead contact, Dr. Arun K. Shukla (arshukla@iitk.ac.in).

Materials availability

All reagents described in this manuscript are available upon reasonable request from the lead contact with appropriate Materials Transfer Agreement. This study does not report the generation of any new reagents.

Star⋆Methods

Detailed methods are provided in the online version of this paper and include the following:

• KEY RESOURCES TABLE

• EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

  • ○
    Human cell line
  • ○
    Insect cell line
• METHOD DETAILS

  • ○
    General plasmids, reagents, and cell culture
  • ○
    Chemokine profiling on CXCR2
  • ○
    GloSensor assay to measure agonist induced decrease in cytosolic cAMP
  • ○
    NanoBiT-based G-protein dissociation assay
  • ○
    NanoBiT-based β-arrestin assays
  • ○
    Measuring ERK signaling using an SRE reporter assay
  • ○
    Receptor surface expression
  • ○
    Purification of chemokines
  • ○
    Expression and purification of enterokinase
  • ○
    Purification of CXCR2
  • ○
    Purification of G-proteins
  • ○
    Purification of scFv16
  • ○
    Reconstituting chemokine-chemokine receptor-G-protein complexes
  • ○
    Negative-stain electron microscopy
  • ○
    Cryo-EM grid preparation and data collection
  • ○
    Cryo-EM data processing
  • ○
    Model building and refinement

• QUANTIFICATION AND STATISTICAL ANALYSIS

Star⋆Methods

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Monoclonal ANTI-FLAG M2-HRP antibody	Sigma-Aldrich	Cat# A8592; RRID: AB_439702
Bacterial and virus strains
E. coli strain BL21(DE3)	New England Biolabs	Cat# C2527H
E. coli strain Rosetta (DE3)	Sigma-Aldrich	Cat# 70954
E. coli strain SHuffle	New England Biolabs	Cat# C3028
Chemicals, peptides, and recombinant proteins
TRIS	SRL	Cat# 71033
HEPES	SRL	Cat# 63732
NaCl	SRL	Cat# 41721
EDTA	SRL	Cat# 12070
Phenylmethanesulfonyl Fluoride (PMSF)	SRL	Cat# 84375
L-Cysteine Hydrochloride Monohydrate	Sigma Aldrich	Cat# C7880
Iodoacetamide	Sigma Aldrich	Cat# l1149
Imidazole	Sigma Aldrich	Cat# I202-500G
Benzamidine Hydrochloride	SRL	Cat# 93014 (0248255)
Lysozyme	SRL	Cat# 45822
Glycerol	SRL	Cat# 77453
Lauryl Maltose Neopentyl Glycol (L-MNG)	Anatrace	Cat# NG310, CAS no.1257852-96-2
Cholesteryl Hemisuccinate (CHS)	Sigma Aldrich	Cat# C6512
Paraformaldehyde (PFA)	Sigma Aldrich	Cat# P6148, CAS no. 30525-89-4
Poly-D-lysine	Sigma Aldrich	Cat# P0899
TMB (Tetramethylbenzidine)	Thermo Fisher Scientific	Cat# 34028
Janus Green B	Sigma Aldrich	Cat# 201677
PEI (Polyethylenimine)	Polysciences	Cat# 23966
Bovine Serum Albumin, BSA	SRL	Cat# 83803 (0140105)
FLAG peptide	GenScript	N/A
HBSS - Hank’s Balanced Salt Solution	Thermo Fisher Scientific	Cat# 14065
GIBCO Fetal Bovine Serum	Thermo Fisher Scientific	Cat# 10270-106
DMEM	Cellclone	Cat# CC3004
Phosphate-buffered saline (PBS)	Sigma Aldrich	Cat# D1283
GIBCO Penicillin-Streptomycin	Thermo Fisher Scientific	Cat# 15140122
ESF921 Insect Cell Culture Medium	Expression Systems	Cat#96-001-01
Coelenterazine	Goldbio	Cat# CZ05
D-Luciferin Sodium Salt	Goldbio	Cat# LUCNA-1G
Coomassie Brilliant Blue	SRL	Cat# 64222
Uranyl formate	Polysciences	Cat# 24762-1
Recombinant human CXCL1	Purified	N/A
Recombinant human CXCL2	Purified	N/A
Recombinant human CXCL3	Purified	N/A
Recombinant human CXCL5	Purified	N/A
Recombinant human CXCL6	Purified	N/A
Recombinant human CXCL7	Purified	N/A
Recombinant human CXCL8	Purified	N/A
Recombinant human mGαoA	Purified	N/A
Recombinant human Gβ1γ2 heterodimer	Purified	N/A
Recombinant human scFV16	Purified	N/A
Formvar/carbon coated 300 mesh copper grids	PELCO (Ted Pella)	Cat# 01753-F
Critical commercial assays
Site Directed Mutagenesis Kit	NEB	Cat# E0554
NanoBiT assay	Promega	N/A
GloSensor assay	Promega	N/A
Deposited data
CXCL1-CXCR2-Go (Receptor-Ligand Focused)	This study	PDB: 8XWA, EMD-38732
CXCL1-CXCR2-Go (Overall)	This study	PDB: 8XWV, EMD-38743
CXCL2-CXCR2-Go (Receptor-Ligand Focused)	This study	PDB: 8XVU, EMD-38719
CXCL2-CXCR2-Go (Overall)	This study	PDB: 8XXH, EMD-38749
CXCL3-CXCR2-Go (Receptor-Ligand Focused)	This study	PDB: 8XWF, EMD-38734
CXCL3-CXCR2-Go (Overall)	This study	PDB: 8XX3, EMD-38744
CXCL5-CXCR2-Go (Receptor-Ligand Focused)	This study	PDB: 8XWS, EMD-38742
CXCL5-CXCR2-Go (Overall)	This study	PDB: 8XX7, EMD-38748
CXCL6-CXCR2-Go (Receptor-Ligand Focused)	This study	PDB: 8XWM, EMD-38738
CXCL6-CXCR2-Go (Overall)	This study	PDB: 8XXR, EMD-38759
CXCL6-CXCR2-Go (composite)	This study	PDB: 8XXX, EMD-38764
CXCL8-CXCR2-Go (Receptor-Ligand Focused)	This study	PDB: 8XWN, EMD-38739
CXCL8-CXCR2-Go (Overall)	This study	PDB: 8XX6, EMD-38747
Cryo-EM structure of CXCL8 bound C-X-C chemokine receptor 1 in complex with Gi heterotrimer	Ishimoto et al.23	PDB: 8IC0
The solution structure of melanoma growth stimulating activity	Fairbrother et al.52	PDB: 1MGS
fAb complex with GroBeta	Convery et al.	PDB: 5OB5
Solution structure of CXCL5	Sepuru et al.53	PDB: 2MGS
LY3041658 Fab bound to CXCL3	Boyles et al.54	PDB: 6WZK
LY3041658 Fab bound to CXCL8	Boyles et al.54	PDB: 6WZM
Crystal structure of a class A GPCR	Liu et al.31	PDB: 6LFL
Gel images and Functional assay data	This study	Mendeley data: https://doi.org/10.17632/t3xts2g9cv.1
Experimental models: Cell lines
Human: HEK293T	ATCC	Cat# CRL-3216
Spodoptera frugiperda (Sf9) Cell line	Expression Systems	Cat# 94-001F
Oligonucleotides
CXCR2 cloning in pCAGGS vector_Forward primer: CGGGGTACCGAGGAGATCTGCCA CCATGGGGAAGACGATCATCGCC	This study	N/A
CXCR2 cloning in pCAGGS vector_Reverse primer: TCCCCCGGGCAGGGTGGTGCTG GTGTGGCC	This study	N/A
Chemokine cloning in pGEMEX vector_Forward primer: CGCGGATCC GGAAGCGGAGACGACGATGACAAG	This study	N/A
Chemokine cloning in pGEMEX vector_Reverse primer: CCCAAGCTT TGCGGCCGCTCATCATCA	This study	N/A
SDM to generate CXCL1AAA_Forward primer: GGCGTGCCAATGCCTGCAAACC	This study	N/A
SDM to generate CXCL1AAA_Reverse primer: GCCGCGGTCGCCACGCTCGCTTT	This study	N/A
SDM to generate CXCL2AAA_Forward primer: GGCGTGCCAATGCCTGCAAACCCTG	This study	N/A
SDM to generate CXCL2AAA_Reverse primer: GCCGCGGTCGCCAGCGGCGCTTT	This study	N/A
SDM to generate CXCL3AAA_Forward primer: GGCGTGCCAATGCCTGCAAACC	This study	N/A
SDM to generate CXCL3AAA_Reverse primer: GCCGCGGTCACCACGCTCGCTTT	This study	N/A
SDM to generate CXCL8AAA_Forward primer: GGCGTGCCAGTGTATCAAGACCTACTCC	This study	N/A
SDM to generate CXCL8AAA_Reverse primer: GCCGCCTTGGCGCTCCGTGGCAG	This study	N/A
SDM to generate cXCR2R208A_R212A_Forward Primer: GCTGGCCATCCTGCCTCAGTCCTTCGGC	This study	N/A
SDM to generate cXCR2R208A_R212A_Reverse Primer: AGCATGGCCCAGTTGGCGGTGTTGTTGCC	This study	N/A
SDM to generate CXCR2D293A_Forward Primer: TAACCACATCGCCAGAGCCCTGG	This study	N/A
SDM to generate CXCR2D293A_Reverse Primer: CGCCTCTCGCAGGTTTCC	This study	N/A
SDM to generate cXCR2D274A_D293A _Forward Primer: GCTGCTGGCTGCCACCCTGATGA	This study	N/A
SDM to generate cXCR2D274A_D293A _Reverse Primer: ACCAGGTTGTAGGGCAGC	This study	N/A
SDM to generate CXCR2R208A_R212A_R278A_Forward Primer: CACCCTGATGGCCACCCAGGTAATCCAGG	This study	N/A
SDM to generate CXCR2R208A_R212A_R278A_Reverse Primer: TCAGCCAGCAGCACCAGG	This study	N/A
Recombinant DNA
pcDNA3.1_CXCR2	This study	N/A
pCAGGS_CXCR2-SmBiT	This study	N/A
pCAGGS_LgBiT-βarr1	Dr. Asuka Inoue	N/A
pCAGGS_LgBiT-βarr2	Dr. Asuka Inoue	N/A
pCAGGS_LgBiT-FYVE	Dr. Asuka Inoue	N/A
pCAGGS_SmBiT-βarr1	Dr. Asuka Inoue	N/A
pCAGGS_SmBiT-βarr2	Dr. Asuka Inoue	N/A
pCAGGS_LgBiT-Ib30	Dr. Arun K. Shukla	N/A
pCAGGS_LgBiT-Ib32	Dr. Arun K. Shukla	N/A
pvL1393-FLAG-T4L-CXCR2	GenScript	N/A
pvL1392Dual-β1γ2 vector	GenScript	N/A
pET-15b(+)-miniGαo1	GenScript	N/A
pET-28-scFv16	GenScript	N/A
pET-32a(+)-Enterokinase	GenScript	N/A
pGEMEX-1-CXCL1/2/3/5/6/7/8	This study	N/A
Software and algorithms
Relion3.1.2	Zivanov et al.56–58	https://www3.mrc-lmb.cam.ac.uk/relion/index.php?title=Main_Page
UCSF ChimeraX	Pettersen et al.59	https://www.rbvi.ucsf.edu/chimerax/
UCSF Chimera	Pettersen et al.60	https://www.cgl.ucsf.edu/chimera/
PDBsum	Laskowski et al.61	http://www.ebi.ac.uk/thornton-srv/databases/pdbsum/
Graphpad Prism 10	GraphPad Software, San Diego, California, USA	https://www.graphpad.com/scientific-software/prism/
Others
100kDa Cutoff Concentrators	Cytiva	Code# 28932319
10kDa Cutoff Concentrators	Cytiva	Code# 28932296

Experimental Model and Study Participant Details

Human cell line

HEK293T cells were purchased from ATCC and used for all the cellular experiments performed in the study. The cell line was regularly inspected under the microscope for appropriate morphology, but they were not authenticated. They were cultured in DMEM supplemented with fetal bovine serum (FBS) at 37°C in 5% CO₂. In this study, any knockout or knockdown cell lines were not generated.

Insect cell line

Sf9 cells were obtained from Expression systems, and they were routinely monitored under the microscope for appropriate morphology. These cells were maintained in a shaker incubator at 27°C with 135rpm shaking, and sub-cultured in protein-free insect cell medium purchased from Expression Systems.

Method Details

General plasmids, reagents, and cell culture

Unless otherwise mentioned, the general reagents were purchased from Sigma Aldrich. Dulbecco’s Modified Eagle’s Medium (DMEM), Phosphate Buffer Saline (PBS), Trypsin-EDTA, Fetal-Bovine Serum (FBS), Hank’s Balanced Salt Solution (HBSS), and Penicillin-Streptomycin solution were purchased from Thermo Fisher Scientific. HEK293T cells (ATCC) were maintained in DMEM (Gibco, Cat. no: 12800-017) supplemented with 10% (v/v) FBS (Gibco, Cat. no: 10270-106) and 100U/mL penicillin and 100μg/mL streptomycin (Gibco, Cat. no: 15140122) at 37°C under 5% CO₂. Sf9 cells were obtained from Expression Systems and maintained in protein-free cell culture media purchased from Expression Systems (Cat. no: 96-001-01) at 27°C under shaking conditions. The cDNA coding region of CXCR2 was cloned in pcDNA3.1 vector with an N-terminal FLAG tag and in pVL1393 vector with an N-terminal FLAG tag followed by the N-terminal region of M4 receptor (residues 2-23). Constructs used for various NanoBiT assays were previously described. All DNA constructs were verified by sequencing from Macrogen.

Chemokine profiling on CXCR2

Chemokine-induced miniG-protein (engineered GTPase domain of Gα subunit),⁶² GRK3⁶³ and β-arrestin2⁶⁴ recruitment to chemokine receptors (CXCR1, CXCR2, CXCR3-A, CXCR4, CXCR5, CXCR6 and ACKR3) was monitored using a nanoluciferase complementation-based assay (NanoBiT, Promega).^65,66 4×10⁶ HEK293T cells were plated in 10cm dishes and cultured for 24h before transfection with vectors encoding for miniG-proteins, human GRK3 or human β-arrestin2 N-terminally fused with LgBiT and the chemokine receptor C-terminally fused with SmBiT. 24h after transfection, cells were harvested, incubated for 15mins at 37°C with coelenterazine H in OptiMEM, and distributed into white 96-well plates (5×10⁴ cells per well). Indicated chemokines (100nM) were then added and the luminescence generated upon nanoluciferase complementation was measured with a Mithras LB940 luminometer (Berthold Technologies) for 20mins. For each receptor, the results are represented as the percentage of the signal monitored with the most active agonist chemokine and presented as mean of three independent experiments (n = 3).

GloSensor assay to measure agonist induced decrease in cytosolic cAMP

For measuring Gi-mediated second messenger signaling, GloSensor Assay was performed to detect agonist mediated decrease in cytosolic cAMP levels, as described previously.^67–69 HEK293T cells were transfected with a mixture of 3.5μg of F22 (Promega, Cat. no: E2301) and either 3.5μg of CXCR2^WT or 3.5μg of CXCR2^RRR or 3μg of CXCR2^DD, harboring an N-terminal FLAG tag. 14-16h after transfection, cells were washed once with 1X PBS, detached by trypsinization and resuspended in assay buffer (20mM HEPES pH 7.4, 1X Hank’s Balanced Salt Solution/ HBSS and 0.5mg/mL D-luciferin (GoldBio, Cat. no: LUCNA-1G)). The resuspended cells were then seeded in 96-well plates at a density of 200,000 cells per well. The plates were then incubated for 1h 30mins at 37°C and an additional 30mins at room temperature. Using a multiwell plate reader (BMG Labtech), basal luminescence was measured for 5 cycles. As we are measuring Gi-mediated decrease in cAMP levels, we first increased the cytosolic cAMP levels by adding 5μM for-skolin and allowed the signal to stabilize. Next, we added the various ligands at the indicated final concentration and recorded luminescence for 30 cycles. The signal obtained was normalized with respect to the luminescence observed at the lowest concentration dose of each ligand, which was treated as 100%. Data was then visualized using GraphPad Prism 10 software.

NanoBiT-based G-protein dissociation assay

Agonist induced G-protein dissociation was measured using a NanoBiT-based assay as previously described.⁷⁰ HEK293T cells were transfected with a mixture of 1μg of CXCR2^WT/ 1μg of CXCR2^RRR/ 0.75μg of CXCR2^DD (all bearing an N-terminal FLAG-tag), 1μg of Gα_oB tagged with LgBiT at its N-terminus, 4μg of Gβ and 4μg of Gγ tagged with SmBiT at its N-terminus. 14-16h after transfection, the cells were washed with 1X PBS, trypsinized and seeded in 96-well plates at a density of 100,000 cells per well in the presence of assay buffer (5mM HEPES pH 7.4, 1X HBSS, 0.01% BSA and 10μM coelenterazine (GoldBio, Cat. no: CZ05)). The plates were first incubated at 37°C for 1h 30mins followed by an additional 30mins at room temperature. Basal luminescence was recorded for 3 cycles using a standard multi-plate reader (Victor X4-Perkin-Elmer). Ligand was added at the indicated final concentrations and luminescence was recorded for 20 cycles. Signal observed was normalized with respect to the luminescence observed at lowest concentration of each ligand, treated as 100%. Data was plotted and analyzed using GraphPad Prism 10 software.

NanoBiT-based b-arrestin assays

Ligand induced β-arrestin1/2 recruitment and trafficking downstream of CXCR2 was measured using a previously described NanoBiT-based assay.^71,72 For measuring recruitment of β-arrestin1/2 in direct mode, HEK293T cells were transiently transfected with a mixture of 3.5μg of N-terminally FLAG-tagged CXCR2 bearing a C-terminal SmBiT fusion and 3.5μg of N-terminally LgBiT-tagged βarr1/2, whereas for measuring β-arrestin1/2 trafficking, HEK293T cells were transfected with a mixture of 3μg of N-terminally FLAG-tagged CXCR2, 2μg of N-terminal SmBiT fused β-arrestin1/2 and 5μg of N-terminal LgBiT-fused FYVE. 14-16h after transfection, the cells were washed with 1X PBS, trypsinized and seeded in 96-well plates at a density of 100,000 cells per well in the presence of assay buffer (5mM HEPES pH 7.4, 1X HBSS, 0.01% BSA and 10μM coelenterazine (GoldBio, Cat. no: CZ05)). Basal luminescence was measured for 3 cycles followed by the addition of ligands at the indicated final concentrations. Luminescence was then measured for another 20 cycles. For, β-arrestin1/2 recruitment, average of the reading observed for cycles 5-9 was used, whereas for β-arrestin1/2 trafficking an average of the luminescence observed for cycles 10-14 was used. The observed signal was normalized with respect to the signal observed at the lowest dose of each ligand, treated as 1. Data was visualized and analyzed using GraphPad Prism 10 software.

For measuring β-arrestin1/2 recruitment in bystander mode, HEK293T cells were transiently transfected with a mixture of 2μg of CXCR2^WT/ 2μg of CXCR2^RRR/ 3μg of CXCR2^DD (all bearing an N-terminal FLAG-tag), 2μg of N-terminal SmBiT fused β-arrestin1/2 and 5μg of N-terminal LgBiT-fused CAAX.

A NanoBiT-based assay was also used for measuring Ib30 and Ib32 reactivity to β-arrestin1 upon stimulation with different ligands. The transfection mix comprised of 3μg of CXCR2 (bearing an N-terminal FLAG-tag), 2μg of N-terminal SmBiT fused β-arrestin1 and 5μg of N-terminal LgBiT-fused Ib30 or Ib32. The rest of the methodology is the same as described above.

Measuring ERK signaling using an SRE reporter assay

ERK signaling downstream to CXCR2 was measured using an SRE reporter assay, as described previously.⁷³ Briefly, HEK293T cells were transiently transfected with a mixture of 3.5μg of CXCR2 bearing an N-terminal FLAG-tag and 3.5μg of pGL4.33 (an SRE-based luciferase reporter plasmid) (Promega, Cat. no: E1340). 14-16h post-transfection, the cells were washed once with 1X PBS, detached by trypsinization and seeded in 96-well plates at a density of 100,000 cells per well in complete media. The plates were incubated for 8h following which complete media was removed and incomplete DMEM was added to the plates. The plates were then incubated overnight at 37°C. Following morning, ligands were added to the plates at the indicated final concentrations and incubated for an additional 6h. Incomplete DMEM was removed from the plates and 100μL of assay buffer (20mM HEPES pH 7.4 and 1X HBSS supplemented with 0.5mg/mL D-luciferin (GoldBio, Cat. no: LUCNA-1G)) was added to each well. Luminescence was measured immediately. The observed signal was normalized with respect to the signal observed at the lowest dose of each ligand, which was treated as 1. Data was visualized and analyzed using GraphPad Prism 10 software.

Receptor surface expression

Expression of the receptor at the cell surface was measured using whole-cell ELISA, as described previously.⁷⁴ HEK293T cells expressing the FLAG-tagged CXCR2 were seeded at a density of 0.2 million cells per well in 24-well plates and incubated overnight at 37°C. The following day, the media was removed and the wells were washed one time with 400μL of 1X TBS. The cells were then incubated with 300μL of 4% (w/v) paraformaldehyde (to allow fixation) for 20mins and the excess paraformaldehyde was then removed by washing the wells three times with 400μL of 1X TBS. Next the wells were incubated with 200μL of 1% BSA (dissolved in 1X TBS) for 1h and 200μL of anti-FLAG M2-HRP (Sigma-Aldrich, Cat. no: A8592) for an additional 1h. The wells were washed three times with 400μL of 1% BSA to remove excess unbound antibody. 200μL of TMB (Thermo Fisher Scientific, Cat. no: 34028) was then added to each well. Upon development of sufficient color, the reaction was quenched. This was done by transferring 100μL of the colored solution to a 96-well plate containing 100μL 1M H₂SO₄. Absorbance was measured at 450 nm using a multiwell plate reader (Victor X4-Perkin-Elmer). Next, Janus Green was used to quantify cell density. The extra TMB solution was removed and the wells were washed one time with 400μL of 1X TBS. The wells were then incubated with 200μL of 0.2% (w/v) Janus Green for 15-20mins. Extra stain was removed and the wells were washed three times with distilled water. 800μL of 0.5N HCl was added to each well to develop the signal. 200μL of the colored solution was then transferred to a 96-well plate and absorbance was measured at 595 nm. Receptor surface expression was normalized by dividing the signal observed at 450 nm with the signal observed at 595 nm. Surface expression of CXCR2 in all the assays is shown in Figure S7.

Purification of chemokines

Coding regions of the various chemokines were cloned in pGEMEX-1 vector with a 6X-His-tag at the N terminus followed by an enterokinase cleavage site. E.coli BL21 (DE3) competent cells were used for over-expression. Transformed cells were inoculated in 50mL TB media containing 100μg/mL ampicillin and incubated at 27°C overnight. Primary culture was then inoculated in 1L TB media containing 100μg/mL ampicillin at 27°C until OD₆₀₀ reached 1.5. The culture was then induced with 1mM IPTG and allowed to grow at 20°C for an additional 48h.

For CXCL1 / CXCL2 / CXCL3 / CXCL5 / CXCL8 /CXCL1^AAA / CXCL2^AAA / CXCL3^AAA / CXCL8^AAA, a previously published protocol was followed.⁷⁵ Harvested cells were resuspended in lysis buffer (20mM HEPES pH 7.4, 1M NaCl, 10mM Imidazole, 0.3% Triton-X, 1mM PMSF and 5% glycerol) and the resuspension was stirred for 30mins at 4°C. Complete lysis of the cells was achieved by ultrasonication for 20mins. This was followed by high-speed centrifugation at 18,000 rpm at 4°C for 30mins to remove the cell debris. Protein was enriched on Ni-NTA beads, and excess unbound/non-specific protein was removed by washing with wash buffer (20mM HEPES pH 7.4, 1M NaCl, 40mM Imidazole and 5% glycerol). Protein was eluted with elution buffer (20mM HEPES pH 7.4, 100mM NaCl, 500mM Imidazole and 5% glycerol) and the elute was dialyzed against enterokinase digestion buffer (20mM Tris-Cl pH 7.5, 150mM NaCl and 2.5% Glycerol) overnight at 4°C. Precipitated protein was removed by centrifugation at 5000 rpm at 4°C for 10mins. Digestion was set up to remove the 6X-His-tag by incubating with either homemade or store bought (NEB, Cat. no: P8070L) enterokinase in the presence of 10mM CaCl₂ at 22°C for 16h. Cleaved protein was then loaded onto the Resource S Cation Exchange Chromatography column (Loading buffer: 50mM MES pH 5.5, 50mM NaCl). Before loading, salt was diluted 3x using 100mM MES buffer pH 5.5. Gradient elution was taken by generating a linear gradient of NaCl (50-1000mM) over 16 column volumes. Peak fractions were pooled on the basis of SDS-PAGE and then dialyzed against PD-10 buffer (20mM HEPES pH 7.4, 150mM NaCl) overnight at 4°C. Protein was flash frozen and stored at -80°C in the presence of 10% glycerol.

For CXCL5, following enterokinase cleavage the protein was concentrated and loaded onto HiLoad Superdex 16/600 200 PG column (Cytiva Life sciences, Cat. no: 17517501). Fractions corresponding to cleaved CXCL5 were pooled, flash-frozen and stored at -80°C in the presence of 10% glycerol.

For purifying CXCL6, every 10g of pellet was resuspended in 50mL of Buffer A (50mM Tris-HCl pH 8.0, 6M guanidinium HCl pH 8.0 and 200mM NaCl). The cells were allowed to solubilize for a period of 1h at 4°C and then lysed by sonication. The cell lysate was then isolated via centrifugation at 25,000 rpm for 40mins and then applied to a Ni-NTA column. The beads were then washed with 2 CVs of Buffer B (6M guanidinium HCl pH 8.0 and 200mM NaCl) and eluted with Buffer C (20mM Tris-HCl pH 8.0, 200 mM NaCl and 500 mM imidazole). The eluted protein was then incubated with 20mM DTT for an hour and was then diluted dropwise in Buffer D (0.55M L-arginine hydrochloride, 20mM Tris-HCl, 200mM NaCl, 1mM EDTA, 1 mM reduced glutathione and 0.1mM oxidised glutathione pH 8.0) and incubated for 48h at 4°C. The protein solution was then concentrated with Vivaspin 10kDa MWCO concentrator (Cytiva Life sciences, Cat. no: 28932360) and dialysed against 20mM Tris-HCl pH 8.0, 200mM NaCl. The amount of protein was estimated by running SDS-PAGE and then digestion reaction was set up with homemade enterokinase, supplemented with 10mM CaCl₂. The enterokinase digested CXCL6 was then concentrated with Vivaspin MWCO 3kDa (Cytiva Life sciences, Cat. no: 28932293) and then injected into HiLoad Superdex 16/600 200 pg column (Cytiva Life sciences, Cat. no: 17517501). Fractions corresponding to the protein were pooled, flash-frozen and stored at -80°C with 10% glycerol.

Expression and purification of enterokinase

A DNA construct of bovine enterokinase catalytic light chain with N terminal-Trx tag followed by Thrombin cut site and a self-cleavable enterokinase site was cloned in pET-32a (+) vector (synthesized from GenScript). 6X-His-tag was present at the C-terminal end of the protein and a mutation was introduced in the 112^th residue to change it from C to S. The DNA was transformed in E. coli SHuffle strain and a single isolated colony from the transformed plate was inoculated in 50mL of LB media and allowed to grow overnight at 30°C. The primary culture was then transferred to 0.5L of TB media followed by induction with 70μM of IPTG at an optical density of 0.7 and allowed to grow for 16h at 16°C. Culture flasks were supplemented with a final concentration of 100μg/mL of freshly prepared ampicillin. The cells were then harvested by centrifugation after 18h and resuspended in 50mL of resuspension buffer (20mM Tris-HCl pH 7.5, 10mM EDTA, 1% Triton-X-100 and 2mM CaCl₂) and were allowed to solubilise for a period of 30mins at 4°C. Cells were lysed by sonication and the supernatant was separated by centrifugation for 30mins at 20,000 rpm at 4°C. The pellet obtained was then dissolved in 10mL of 0.1M Tris-HCl pH 8.6, 1mM EDTA, 20mM DTT and 6M guanidinium HCl. The insoluble fractions were separated by centrifugation at 25,000 rpm for 20mins at 4°C. The supernatant was collected and put up for dialysis against 3M guanidinium HCl pH 2.5 at room temperature. After dialysis the solution was mixed with 10mL of oxidation buffer (50mM Tris-HCl pH 8.3, 6M guanidinium-HCl, 0.1M oxidised glutathione) and then again dialysed against 3M guanidinium HCl pH 8.0. For initiating the refolding process, the dialysed protein solution was then dropwise diluted into 600mL of 0.7M L-arginine hydrochloride pH 8.6, 2mM Reduced glutathione and 1 mM EDTA and then incubated for 75h at 4°C. The protein was then subsequently dialysed against 0.1M Tris-HCl and 10 mM CaCl₂ and loaded onto Ni-NTA column, washed with 10mM Tris-HCl, 500mM NaCl and eluted with 500mM Imidazole containing elution buffer. The elution was then dialysed against and finally stored in 0.1M Tris-HCl pH 8.0, 500mM NaCl and 50% glycerol at -20°C.

Purification of CXCR2

CXCR2 was purified from Sf9 cells as previously described.^76–78 Briefly, Sf9 cells expressing FLAG-tagged CXCR2 were homogenized using a Dounce homogenizer in hypotonic buffer (20mM HEPES pH 7.4, 20mM KCl, 10mM MgCl₂, 1mM PMSF, 2mM benza-midine), followed by homogenization in hypertonic buffer (20mM HEPES pH 7.4, 20mM KCl, 10mM MgCl₂, 1M NaCl, 1mM PMSF, 2mM benzamidine) and lysis buffer (20mM HEPES pH 7.4, 450mM NaCl, 1mM PMSF, 2mM benzamidine, 0.1% CHS, 2mM IAA and 1% (v/v) L-MNG). The lysate was tumbled for 2h at 4°C, diluted three times (20mM HEPES pH 7.4, 8mM CaCl₂, 1mM PMSF and 2mM benzamidine), centrifuged and filtered before loading onto an M1-antibody conjugated agarose bead column at 4°C. Following loading, the beads were washed alternatively with low salt buffer (20mM HEPES pH 7.4, 150mM NaCl, 2mM CaCl₂, 0.01% (v/v) L-MNG) and high salt buffer (20mM HEPES pH 7.4, 350mM NaCl, 2mM CaCl₂, 0.01% L-MNG). Receptor was eluted using FLAG-elution buffer (20mM HEPES pH 7.4, 150mM NaCl, 0.01% (v/v) L-MNG, 250μg/mL FLAG peptide, 2mM EDTA).

Purified CXCR2 was incubated with either 1.5X molar excess (for CXCL1, CXCL2, CXCL3, CXCL5, CXCL8) or 3X molar excess (for CXCL6) of chemokine for 1h at room temperature. Ligand bound receptor was stored in the presence of 10% glycerol at -80°C till further use.

Purification of G-proteins

Gene for miniGαoA subunit was cloned in pET-15b(+) vector with an in-frame 6X-His tag at the N-terminal end and expressed in E. coli BL21 (DE3) cells.^76,79 A starter culture supplemented with 0.2% glucose was grown in LB media at 37°C for 6-8h at 220 rpm. This was followed by growing a primary culture overnight at 30°C, supplemented with 0.2% glucose. 15mL primary culture was used to inoculate 1.5L TB media. The secondary culture was induced with 50μM IPTG at an O.D₆₀₀ of 0.8 and cultured at 25°C for an additional 18-20h. Following this, the culture was harvested. Cells were lysed in lysis buffer (40mM HEPES pH 7.4, 100mM NaCl, 10mM Imidazole, 10% Glycerol, 5mM MgCl₂, 1mM PMSF, 2mM Benzamidine) in the presence of 1 mg/mL lysozyme, 50μM GDP and 100μM DTT. The lysate was then centrifuged at 18000 rpm for 30mins at 4°C. The clarified supernatant was loaded onto pre-equilibrated Ni-NTA beads. Subsequently the beads were washed extensively with wash buffer (20mM HEPES pH 7.4, 500mM NaCl, 40mM Imidazole, 10% Glycerol, 50μM GDP and 1mM MgCl₂), and bound protein was eluted with elution buffer (20mM HEPES pH 7.4, 100mM NaCl, 10% Glycerol, 500mM Imidazole). Eluted protein was pooled and incubated with TEV protease overnight at a ratio of 1:20 (TEV:Protein) at room temperature to facilitate removal of His-tag. The protein was then concentrated and injected into HiLoad Superdex 200 PG 16/600 column (Cytiva, Cat. no: 17517501) to separate the cleaved fraction from the uncleaved fraction. Fractions corresponding to our protein of interest were pooled, supplemented with 10% glycerol, quantified and stored at -80°C till further use.

The genes encoding Gβ1 with an in-frame C-terminal 6X-His tag and Gγ2 were expressed in Sf9 cells.^76,79 Post 72h of infection, cells were harvested and resuspended in lysis buffer (20mM Tris-Cl pH 8.0, 150mM NaCl, 10% Glycerol, 1mM PMSF, 2mM Benzamidine, 1mM MgCl₂). Cells were lysed by douncing and centrifuged at 18000 rpm for 40mins at 4°C. The pellet was resuspended in solubilization buffer (20mM Tris-Cl pH 8.0, 150mM NaCl, 10% Glycerol, 1% DDM, 5mM β-ME, 10mM Imidazole, 1mM PMSF and 2mM Benzamidine) and dounced again. Solubilization was allowed to proceed for 2h at 4°C under constant stirring. The lysate was clarified by centrifugation at 4°C and filtered. Filtered supernatant was loaded onto Ni-NTA resin. This was followed by washing of the beads with wash buffer (20mM Tris-Cl pH 8.0, 150mM NaCl, 30mM Imidazole, 0.02% DDM) and elution of bound protein with elution buffer (20mM Tris-Cl pH 8.0, 150mM NaCl, 300mM Imidazole, 0.01% MNG). Eluted protein was concentrated, quantified, and stored at -80°C in presence of 10% glycerol.

Purification of scFv16

scFv16 was purified as previously described.^76,78,79 Briefly, E. coli Rosetta (DE3) cells were grown in 2XYT media, supplemented with 0.5% glucose and 5mM MgSO₄ till O.D₆₀₀ reached 0.6, and induced with 250μM IPTG. Following induction, cells were cultured for an additional 16-18h at 18°C and harvested. The cell pellet was resuspended in 20mM HEPES pH 7.4, 200mM NaCl, 10mM Imidazole, 2mM Benzamidine, 1mM PMSF and incubated at 4°C for 1h with constant stirring. The lysate was clarified by centrifugation and the filtered supernatant was enriched on Ni-NTA beads. This was followed by extensive washing with wash buffer (20mM HEPES pH 7.4, 200mM NaCl, 10mM Imidazole) and subsequent elution with elution buffer (20mM HEPES pH 7.4, 200mM NaCl, 300mM Imidazole). The eluate was further enriched on amylose resin (NEB, Cat. no: E8021L), washed (20mM HEPES pH 7.4, 200mM NaCl) and the bound protein was eluted with 10mM maltose (prepared in 20mM HEPES pH 7.4, 200mM NaCl). For removing His-MBP tag, the eluate was incubated overnight with TEV protease at a ratio of 1:20 (TEV:Protein) at room temperature. The mixture was then passed through Ni-NTA beads to remove the tag. Eluted protein was concentrated and cleaned by size exclusion chromatography on Hi-Load Superdex 200 preparative grade 16/600 column (Cytiva Life sciences, Cat. no: 17517501). Purified protein was supplemented with 10% glycerol and stored at -80°C.

Reconstituting chemokine-chemokine receptor-G-protein complexes

Purified chemokine-receptor complex was incubated with 1.2-fold molar excess of mGαoA, Gβ1γ2, and scFv16, in the presence of 25mU/mL apyrase (NEB, Cat. no: M0398S) and 5mM CaCl₂, for 2h at room temperature. The mixture was concentrated using 100 MWCO concentrator (Cytiva, Cat. no: GE28-9323-19) and injected into Superdex 200 Increase 10/300 GL SEC column. Peak fractions containing the complex were pooled and concentrated to roughly 12-18mg/mL using the same concentrator and stored at -80°C (in absence of glycerol) until further use.

Negative-stain electron microscopy

Negative staining of the various CXCR2-G-protein complexes was performed to verify complex formation and sample homogeneity, as per previously published protocol.^71,80 Complexes were diluted to 0.02mg/mL and immediately dispensed onto freshly glow discharged carbon/formvar coated 300 mesh Cu grids (PELCO, Ted Pella). The sample was allowed to incubate for 1min following which excess sample was blotted off with a filter paper. The grid was first touched on a drop of freshly prepared 0.75% (w/v) uranyl formate solution and the excess stain was blotted off using a filter paper. This was followed by rotating the grid on a second drop of stain for 30s and again the excess stain was removed by blotting on a filter paper. The grid was then allowed to air dry and placed on a TEM specimen grid holder. Imaging and data collection was performed at 30,000x magnification with a FEI Tecnai G2 12 Twin TEM (LaB6) operating at 120kV and equipped with a Gatan CCD camera (4k x 4k). Data processing of the collected micrographs was performed with Relion 3.1.2.^56–58 More than 10,000 particles were autopicked, extracted with a box size of 280 px and subjected to reference free 2D classification to generate 2D class averages.

Cryo-EM grid preparation and data collection

3.0 μl of the purified CXCR2-Go complexes were dispensed onto glow discharged Quantifoil holey carbon grids (R1.2/1.3, Au, 300 mesh) at a concentration of approximately 15.0 mg/mL (CXCL1-CXCR2-Go), 16.7 mg/mL (CXCL2-CXCR2-Go), 12.1 mg/mL (CXCL3-CXCR2-Go), 16.6 mg/mL (CXCL5-CXCR2-Go), 18.4 mg/mL (CXCL6-CXCR2-Go), and 23.4 mg/mL (CXCL8-CXCR2-Go). The grids were blotted for 4 seconds at 4°C and 100% humidity with a blot force of 10 using a Vitrobot Mark IV (Thermo Fischer Scientific) and immediately plunge frozen in liquid ethane (-181°C).

Data collection of all samples was performed on a Titan Krios G3i (Thermo Fisher Scientific) operating at an accelerating voltage of 300 kV equipped with a Gatan K3 direct electron detector and BioQuantum K3 imaging filter. Movie stacks were acquired in counting mode at a pixel size of 0.83 Å /pix and a dosage rate of approximately 15.6 e^-/Ų/s using EPU software over a defocus range of -0.8 to -1.6μm. Each movie was fractionated into 48 frames with a total dose of 50.1 e^-/Ų that was obtained throughout the 2.3 s exposure period. In total 8,273, 3,555, 2,108, 3,752, 4,722, and 4,509 movie stacks were acquired for CXCL1-CXCR2-Go, CXCL2-CXCR2-Go, CXCL3-CXCR2-Go, CXCL5-CXCR2-Go, CXCL6-CXCR2-Go, and CXCL8-CXCR2-Go respectively.

Cryo-EM data processing

All datasets of the CXCR2-Go complexes were processed following a similar pipeline. Briefly, raw movies were aligned with MotionCor2 in RELION 4.0,⁵⁸ imported into cryoSPARC v4.4⁸¹ and subjected to CTF estimation using Patch CTF (multi).

For the CXCL1-CXCR2-Go dataset, 4,437,786 autopicked particles (template based) were extracted using a box size of 280 pix (fourier cropped to 70 pix) and then cleaned using reference-free 2D classification and heterogeneous refinement to remove ice contamination and distorted particles. 317,394 particles were re-extracted with a box size of 280 pix (fourier cropped to 180 pix), and subjected to heterogeneous refinement into three classes. 115,169 particles that were curated via many rounds of heterogeneous refinement were imported into RELION v4.0. Following Bayesian polishing with a box size of 300 pix (fourier cropped to 240 pix), the polished particles were imported into cryoSPARC and subjected to CTF refinement and NU refinement, providing a reconstruction with a nominal resolution of 3.07 Å at fourier shell correlation of 0.143. To further improve the density of the receptor region, local refinement was performed using the receptor-focused mask, providing a reconstruction with a nominal resolution of 3.48 Å.

For the CXCL2-CXCR2-Go dataset, 1,927,680 template based autopicked particles were extracted using a box size of 280 pix (fourier cropped to 70 pix) and then cleaned using heterogeneous refinement to remove ice contamination and distorted particles. 623,954 particles were re-extracted with a box size of 280 pix (fourier cropped to 180 pix), and subjected to heterogeneous refinement into three classes. 285,884 particles corresponding to the best 3D class were imported into RELION v4.0. Following Bayesian polishing with a box size of 300 pix (fourier cropped to 240 pix), the resultant polished particles were imported into cryoSPARC and subjected to CTF refinement and NU refinement (in cryoSPARC), providing a reconstruction with a nominal resolution of 2.8 Å at 0.143 fourier shell cut-off. To further improve the density of the receptor region, local refinement was performed using the receptor-focused mask, providing a reconstruction with a nominal resolution of 3.09 Å.

For the CXCL3-CXCR2-Go dataset, 1,133,660 template picked particles were extracted using a box size of 280 pix (fourier cropped to 70 pix) and then cleaned using heterogeneous refinement to remove ice contamination and distorted particles. 307,000 particles were re-extracted with a box size of 280 pix (fourier cropped to 180 pix), and subjected to heterogeneous refinement into three classes. RELION v4.0 was used to import the particles curated via many rounds of heterogeneous refinement, and subjected to 3D classification without alignment followed by Bayesian polishing with a box size of 300 pix (fourier cropped to 240 pix). 46,110 particles that resulted were imported into cryoSPARC and subjected to non-uniform refinement using estimated CTF values, providing a reconstruction with a nominal resolution of 3.38 Å at a fourier shell correlation of 0.143. Local refinement using a mask on the receptor was performed to improve the features corresponding to the receptor, providing a reconstruction with a nominal resolution of 3.65 Å.

For the CXCL5-CXCR2-Go dataset, 2,047,293 particles were automatically picked using the template picker subprogram, extracted with a box size of 560 pix (fourier cropped to 140 pix) and subjected to several rounds of 2D classification. Following reextraction with a box size of 560 pix (fourier cropped to 320 pix), and heterogeneous refinement with a C2 symmetry constraint were performed to remove fuzzy particles, yielding a total of 131,780 particles. The clean particle stack was imported into RELION v4.0 subjected to Bayesian polishing with a box size of 560 pix (fourier cropped to 440 pix), particles were imported back into cryoSPARC. Imported particles were subjected to CTF refinement and NU refinement with a C2 symmetry constraint to produce a map with a global indicated resolution of 3.32 Å at fourier shell correlation of 0.143. Local refinement with a mask on the receptor with a C2 symmetry constraint was performed to improve the interpretability of the map, yielding a reconstruction with a global resolution of 3.06 Å

For the CXCL6-CXCR2-Go dataset, 1,609,421 particles were picked and extracted with a box size of 280 pix (fourier cropped to 70 pix), and subjected to several rounds of heterogeneous refinement to eliminate carbon edges and ice contaminations in cryoSPARC. Following re-extraction with a box size of 280 pix (fourier cropped to 180 pix), and subjected to heterogeneous refinement into three classes. A total of 193,262 particles were imported and curated in RELION using 3D classification without alignment followed by Bayesian polishing with a box size of 300 pix (fourier cropped to 240 pix). Finally, the best-class consisting of 61,539 particles were imported and reconstructed in cryoSPARC using CTF refinement and non-uniform refinement, yielding a reconstruction with an overall resolution of 3.17 Å at 0.143 FSC criterion. In addition, the features of the reconstruction were improved following local refinement with a mask on the receptor resulting in a reconstruction with a nominal resolution of 3.71 Å. Since CXCL6 was not clearly discernible in the overall reconstruction, we prepared a composite map using the combine-focused-maps sub-module in Phenix⁸² with the overall reconstruction and the receptor-ligand focused map as inputs.

For the CXCL8-CXCR2-Go dataset, 2,152,291 particles were autopicked, extracted using a box size of 280 pix (fourier cropped to 70 pix), and subjected to heterogeneous refinement. Following re-extraction with a box size of 280 pix (fourier cropped to 180 pix), and subjected to heterogeneous refinement into three classes. 99,138 particles corresponding to the best class following heterogeneous refinement were imported into RELION v4.0, subjected to Bayesian polishing with a box size of 300 pix (fourier cropped to 240 pix). The polished particles were then re-imported into cryoSPARC and were subjected to CTF refinement and non-uniform refinement to yield a map with a global resolution of 2.99 Å according to the gold-standard FSC cut-off of 0.143. Local refinement with a mask on the receptor and ligand was performed to yield a 3D reconstruction with a nominal resolution of 3.29 Å.

Local resolution of all maps was calculated using Blocres included within the cryoSPARC package⁸¹ with the half maps as input. Final maps were sharpened with phenix.auto_sharpen^82,83 to enhance features for model building. Protein-protein interactions were determined using PDBsum.⁶¹ Detailed pipelines for data processing and refinement are included in Method S1.

Model building and refinement

Coordinates of CXCR2 were generated in AlphaFold (https://alphafold.ebi.ac.uk/entry/P25025), while the atomic coordinates of min-iGo, and other component of G-protein (Gβ, Gγ and scFv16) were obtained from the cryo-EM structure of EP54-C3aR-Go complex (PDB: 8I95).⁷⁶ The initial model of the chemokines was obtained from the Swiss-model using previously solved CXCL8 structure as template (PDB: 6WZM).⁵⁴ These initial models were docked into the individual EM maps with Chimera,^59,60 followed by flexible fitting of the docked models with the “all atom refine” module in COOT.⁸⁴ The models obtained were refined with phenix.real_space_refinement with secondary structural restraints against the EM maps after several rounds of manual readjustment in COOT. The final models were evaluated using Molprobity and the “Comprehensive Validation (cryo-EM)” sub-module within Phenix. Data collection, processing, and model refinement statistics are included in Table S1. All figures in the manuscript were prepared using either Chimera or ChimeraX packages.^59,60 We note that all chemokines are numbered starting from the first residue after removal of the signal sequence.

Quantification and Statistical Analysis

All experiments were conducted with a minimum of three biological replicates to ensure reproducibility. Data analysis and visualization were performed using GraphPad Prism v10. The precise number of replicates and specific details about data normalization are included in the corresponding figure legends.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.molcel.2025.01.024.

Highlights.

• Global fingerprint of C-X-C chemokine promiscuity and selectivity at CXCRs

• Cryo-EM structures of CXCR2 in complex with chemokines and G-proteins

• Distinct structural motifs in the chemokines and CXCR2 suggest structural mimicry

• Structure-guided experiments uncover the molecular basis of chemokine promiscuity

In brief.

Saha et al. present a global fingerprint of C-X-C chemokine promiscuity and selectivity at the C-X-C chemokine receptors and determine cryo-EM structures of the most promiscuous receptor, CXCR2, in complex with six different C-X-C chemokines and G-proteins. These structural snapshots, combined with extensive biochemical and pharmacological analysis, elucidate the molecular basis of promiscuous chemokine binding and structural mimicry at CXCR2, mediated via distinct structural motifs in the receptor and chemokines.

Data and code availability

• All three-dimensional cryo-EM density maps, coordinates for the atomic models, and local-refined maps generated in this study have been deposited in the Protein Data Bank (PDB) and Electron Microscopy Data Bank (EMDB) and are publicly available as of the date of publication. Accession numbers (EMDB and PDB IDs) are listed in the key resources table. Original gel images and a copy of the data plotted for various functional assays have been deposited to Mendeley Data, and they are publicly available after publication. The DOI is listed in the key resources table.

• This paper does not report any original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.