Report

Structure of the CCR5 Chemokine Receptor–HIV Entry Inhibitor Maraviroc Complex

See allHide authors and affiliations

Science  20 Sep 2013:
Vol. 341, Issue 6152, pp. 1387-1390
DOI: 10.1126/science.1241475

CCR5-Maraviroc Structure

The chemokine receptor CCR5, a G protein–coupled receptor best known as a co-receptor during HIV-1 infection, is important in a variety of physiological processes. Tan et al. (p. 1387, published online 12 September; see the Perspective by Klasse) now report the high-resolution crystal structure of CCR5 bound to the HIV-1 entry inhibitor, Maraviroc. The structure suggests that Maraviroc acts as a noncompetitive inhibitor by binding to a region of CCR5 that is distinct from the binding site of HIV-1 and chemokines. Comparison of the structure of CCR5 with the other HIV-1 co-receptor, the chemokine receptor CXCR4, provides insight into the co-receptor selectivity of the virus.

Abstract

The CCR5 chemokine receptor acts as a co-receptor for HIV-1 viral entry. Here we report the 2.7 angstrom–resolution crystal structure of human CCR5 bound to the marketed HIV drug maraviroc. The structure reveals a ligand-binding site that is distinct from the proposed major recognition sites for chemokines and the viral glycoprotein gp120, providing insights into the mechanism of allosteric inhibition of chemokine signaling and viral entry. A comparison between CCR5 and CXCR4 crystal structures, along with models of co-receptor–gp120-V3 complexes, suggests that different charge distributions and steric hindrances caused by residue substitutions may be major determinants of HIV-1 co-receptor selectivity. These high-resolution insights into CCR5 can enable structure-based drug discovery for the treatment of HIV-1 infection.

Chemokine receptors and their peptidic ligands, chemokines, are the main organizers of leukocyte trafficking and are validated therapeutic targets owing to their involvement in many physiopathological disorders (1, 2). The chemokine receptor CCR5 binds and responds to four endogenous chemokine agonists: RANTES, macrophage inflammatory protein–1α (MIP-1α), MIP-1β, and monocyte chemotactic protein2 (MCP-2) (3). CCR5 and another chemokine receptor, CXCR4, are required for human immunodeficiency virus type 1 (HIV-1) infectivity, acting as co-receptors of the viral envelope glycoprotein gp120 (4). The structure of CXCR4 has been determined (5), but the details of chemokine recognition and viral infectivity are poorly understood. HIV can infect a variety of CD4-expressing immune cells and evolves inside the infected organism to encompass a wider range of susceptible cells by changing its co-receptor specificity, a phenomenon related to HIV tropism. HIV-1 strains using the co-receptor CCR5 are termed R5, and the strains using CXCR4 are termed X4, and those using either co-receptor are R5X4 (6). Intense research into the development of inhibitors capable of blocking HIV entry by targeting the co-receptor CCR5 has led to approval of the allosteric CCR5 inhibitor maraviroc for treatment of HIV-1 infection (711).

Structural studies were carried out using an engineered human CCR5 construct, purified and crystallized in complex with maraviroc (figs. S1 and S2) (12). The crystal structure of the CCR5-maraviroc complex was determined at 2.7 Å resolution with two complexes per asymmetric unit (ASU) (Fig. 1, figs. S3 and S4, and table S1). Molecule A will be used for discussion purposes.

Fig. 1 Overall fold of the CCR5-maraviroc complex and comparison with CXCR4.

(A) Overall structure of the two CCR5-rubredoxin molecules related by pseudo-translational symmetry in one ASU. The receptor is colored blue, and the rubredoxin is light cyan. The ligand maraviroc is shown in orange sphere representation. The disulfide bonds are shown as yellow sticks. Zinc ions are shown as gray spheres. (B to D) Structure comparison between CCR5 (blue) and CXCR4 (PDB ID: 3ODU, green). The ligands are shown in stick representation. maraviroc in CCR5 and IT1t in CXCR4 have orange and magenta carbons, respectively. (C) Top view of the extracellular side of CCR5 and CXCR4; (D) bottom view of the intracellular side of CCR5 and CXCR4.

The overall CCR5 fold shares a similar architecture with previously solved class A G protein–coupled receptor (GPCR) structures, containing seven transmembrane (7TM) α helices (I to VII) connected by three extracellular loops (ECL1-3) and three intracellular loops (ICL1-3) (Fig. 1A). CCR5 is structurally similar to the chemokine receptor CXCR4 [Cα root mean square deviation within the 7TM bundle between CCR5-maraviroc and CXCR4-IT1t is 1.8 Å (sequence identity = 34%)]. The largest loop in CCR5, ECL2, forms a β-hairpin structure; the conformations of the N-terminal segment (residues 19 to 26) and ECLs are constrained by two disulfide bonds, one linking Cys-1013.25 (13) with Cys-178 of ECL2, and another one connecting Cys-20 at the N terminus with Cys-2697.25 (Fig. 1, B and C). Despite the overall similarity, CCR5 and CXCR4 structures differ substantially in a number of regions. In the CXCR4 crystal structures, the C terminus after helix VII adopted an extended disordered conformation (5), whereas in CCR5, a short α helix VIII is observed. This difference could be explained by the presence of an α-helical sequence motif F(RK)xx(FL)xxx(LF) in CCR5’s helix VIII, whereas in CXCR4 this motif is partially modified (FKxxAxxxL) (5, 14). However, it cannot be excluded that this difference might also be due to different crystal-packing interactions. Helix IV in CCR5 is tilted by about 15° with respect to the corresponding helix in CXCR4; its intracellular portion is 1.5 turns shorter than in CXCR4 and forms a classical α helix, in contrast with a distorted π helix in CXCR4. ICL2, which is unstructured in CXCR4, contains a two-turn α helix in CCR5 running parallel to the membrane (Fig. 1D). Phe-135 and Ala-136 in the α helix of ICL2 form a hydrophobic cluster with Leu-1283.53 and Ala-1293.54 at the intracellular end of helix III to stabilize the conformation of ICL2.

In the CCR5-maraviroc structure, the ligand occupies the bottom of a pocket defined by residues from helices I, II, III, V, VI, and VII (Fig. 2A and fig. S5). The nitrogen of the tropane group is likely protonated and engaged in a salt-bridge interaction with Glu-2837.39. The carboxamide nitrogen forms a hydrogen bond with Tyr-2516.51. The length of the carbon chain between the above-mentioned two nitrogens was reported to be critical for the anti-HIV infection activity of the inhibitors (15), which correlates with the spatial locations of Glu-2837.39 and Tyr-2516.51. The amine moiety of the triazole group hydrogen bonds with Tyr-371.39 and with a water molecule. Another two hydrogen bonds are formed by one of the fluorines in the cyclohexane ring with Thr-1955.39 and Thr-2596.59. In addition, the phenyl group reaches deep into the pocket to form hydrophobic interactions with five aromatic residues: Tyr-1083.32, Phe-1093.33, Phe-1123.36, Trp-2486.48, and Tyr-2516.51. The triazole, tropane, and cyclohexane groups also fit into small subpockets and make hydrophobic contacts with CCR5. The above interactions are supported by previous mutagenesis studies (Fig. 2B) (11, 16).

Fig. 2 CCR5 ligand-binding pocket for maraviroc.

(A) Key residues in CCR5 for maraviroc binding. maraviroc (orange carbons) and receptor residues (blue carbons) involved in ligand binding are shown in stick representation. Other elements are colored as follows: oxygen, red; nitrogen, dark blue; sulfur, yellow; fluorine, light cyan. The water molecule involved in interacting with maraviroc is shown as a red sphere. (B) Schematic representation of interactions between CCR5 and maraviroc. Mutations reported to be critical for maraviroc binding are indicated with blue squares (11, 16).

Compared to the CXCR4–IT1t structure, the binding site for maraviroc in CCR5 is deeper and occupies a larger area of the pocket, making no contacts with the ECLs (Fig. 3, A, B, D, and E). In CCR5, the extracellular end of helix VII shifts away from the central axis of the receptor by ~3 Å compared with CXCR4, leading to a corresponding shift of CCR5’s N-terminal fragment connected to helix VII by the disulfide bond Cys-20–Cys-2697.25. Also in CXCR4, Asp-972.63 and Arg-183 in ECL2 form a salt bridge, which is absent in CCR5 because of substitutions to Tyr-892.63 and His-175, resulting in a 6 Å shift at the β-hairpin tip of ECL2 of CXCR4 toward the ligand-binding pocket compared with CCR5. Consequently, the entrance to the CXCR4 ligand-binding pocket is partially covered by its N terminus and ECL2, whereas the CCR5 ligand binding pocket is more open.

Fig. 3 Comparison of the ligand-binding pockets between CCR5-maraviroc and CXCR4-IT1t.

(A and D) Top views of the ligand-binding pockets in CCR5 [(A), blue] and CXCR4 [(D), green], showing a more open ligand-binding pocket in CCR5. The receptors are shown in both cartoon and molecular surface representations. The ligands are shown in stick representation. maraviroc in CCR5 and IT1t in CXCR4 have orange and magenta carbons, respectively. (B and E) Side views of the ligand binding pockets in CCR5 (B) and CXCR4 (E), showing that maraviroc binds deeper in CCR5 than IT1t in CXCR4. (C and F) Top views of the ligand-binding pockets in CCR5 (C) and CXCR4 (F). Both CCR5 and CXCR4 surfaces are colored according to their electrostatic potential from red (negative) to blue (positive), showing different charge distribution within the ligand-binding pockets of these two receptors.

Maraviroc has been characterized as an inverse agonist of CCR5 (17), suggesting that maraviroc stabilizes CCR5 in an inactive conformation. The major evidence for an inactive state of the CCR5-maraviroc structure is the conformation of the highly conserved class A GPCR residues Trp-2486.48 and Tyr-2446.44, which are involved in relaying ligand-stabilized conformational changes in the binding pocket into conformational changes in the cytoplasmic domain. In the CCR5-maraviroc structure, these residues are in conformations similar to those observed in other inactive structures and distinct from their active-state conformations (18, 19). Moreover, the phenyl group of maraviroc forms a hydrophobic interaction with Trp-2486.48, preventing its activation-related motion. The inactive state of the CCR5-maraviroc structure is also manifested by the close packing of helix VI with other helices of the 7TM bundle at the intracellular side of receptor that precludes G-protein binding. The inactive conformation of helix VI can be stabilized by an “ionic lock” formed by a salt-bridge interaction between the conserved Arg3.50 in helix III and Asp/Glu6.30 at the intracellular end of helix VI (20, 21). Although wild-type CCR5 lacks any acidic residue at position 6.30, the thermostabilizing mutation Ala-2336.33Asp makes a salt-bridge contact with Arg3.50, potentially locking the receptor in an inactive conformation (fig. S2).

Mutagenesis and biochemical studies suggest that maraviroc and some other small-molecule CCR5 inhibitors are allosteric modulators (911, 16, 17). The N-terminal region of CCR5, together with ECL2, have been identified as the major binding determinants for its chemokine ligands (22). In the so-called two-site model, this region is viewed as a chemokine recognition site (site 1), which interacts with the chemokine core (16). Several residues in the 7TM region were found to be important for CCR5 activation upon binding to the chemokine ligand, such as Tyr-371.39 and Trp-2486.48 (11), which are involved in maraviroc binding. This region is considered as an activation site (site 2), which interacts with the flexible N terminus of the chemokine (23). In the CCR5 structure, maraviroc is buried in a cavity within the 7TM domain, which is distinct from site 1 of chemokine recognition, but potentially overlaps with site 2. Thus, the CCR5-maraviroc structure indicates that maraviroc most likely inhibits chemokine function by blocking receptor activation through interactions with site 2, which explains the allosteric inhibition of chemokine signaling by maraviroc. It has also been reported that CCR5’s N terminus and ECL2 play an important role in gp120 binding and HIV-1 infection (11, 24, 25), suggesting that maraviroc interferes with the effects of gp120 binding in a similar allosteric fashion. The above conclusion does not rule out the possibility that maraviroc may reduce chemokine and gp120 binding in an allosteric inverse agonism manner by stabilizing CCR5 in an inactive conformation. Thus, maraviroc may alter chemokine signaling and gp120 binding by allosterically blocking agonist recognition and/or inhibiting receptor activation.

The third variable region, V3 loop, of gp120 forms a β-hairpin structure and has been identified as the major determinant of cellular tropism and co-receptor specificity (24, 26, 27). The stem region of the V3 loop has been reported to be responsible for gp120 binding to the co-receptor’s N terminus, whereas the V3 crown interacts with the co-receptor’s ECL2 and residues inside the ligand-binding pocket (24, 25, 28). Sequence analysis of the V3 region shows that it is more positively charged in the X4-tropic viruses than in the R5-tropic viruses (29, 30). Several acidic residues in CXCR4—Asp-972.63, Asp-1714.60, Asp-187 (ECL2), Asp-1935.32, and Asp-2626.58—are key for ligand binding in the CXCR4 structures (5) and have been reported to be critical for HIV-1 infectivity (28). In CCR5, these acidic residues are substituted by uncharged residues Tyr-892.63, Gly-1634.60, Ser-179, Gln-1885.32, and Asn-2586.58, respectively. Additionally, the N terminus of CXCR4 contains nine acidic residues, whereas CCR5 only has three (Fig. 3, C and F, and fig. S6A). These differences may correlate with the different charge distribution in the V3 loops of X4- and R5-tropic viruses. It was reported that the binding of gp120 to CCR5 was sensitive to mutations of some uncharged residues of CCR5, such as Trp-862.60, Trp-94, Tyr-1083.32, Trp-2486.48, and Tyr-2516.51 (11), providing additional evidence for the importance of the net charge in the V3 loop for co-receptor selectivity. These residues form a cluster within the CCR5 ligand-binding pocket, which composes a potential binding site for gp120 (fig. S7).

To further understand the mechanism of co-receptor selectivity, we have built models of CCR5–R5-V3 and CXCR4–X4-V3 complexes based on the CCR5 structure and previous studies (5, 25, 30) (fig. S6B). The models suggest that the different charge distributions in the co-receptor ligand binding pockets and steric hindrances caused by residue substitutions may be major determinants of HIV-1 co-receptor selectivity (fig. S8) (12). These models initiate our understanding of HIV-1 tropism in a structural perspective; however, additional structures of the co-receptors in an apo state or complemented with the same or similar allosteric or orthosteric ligands, and complexes between the co-receptors and gp120-CD4, are needed to fully understand the mechanisms of HIV-1 tropism.

Supplementary Materials

www.sciencemag.org/content/341/6152/1387/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S8

Table S1

References (3140)

References and Notes

  1. Materials, methods, and discussion are available as supplementary materials on Science Online.
  2. In Ballesteros-Weinstein numbering, a single most-conserved residue among the class A GPCRs is designated x.50, where x is the transmembrane helix number. All other residues on the helix are numbered relative to this conserved position.
  3. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; Y, Tyr; and x, any amino acid.
  4. Acknowledgments: We thank I. Wilson, V. Katritch, and T. Handel for careful review and scientific feedback on the manuscript; A. Walker for assistance with manuscript preparation; C. Wang and D. Wacker for help on collection of x-ray diffraction data; and E. Kellenberger for providing the CCR5 model. Atomic coordinates and structure facors have been deposited in the Protein Data Bank with identification code 4MBS. This work was supported by “National Basic Research Program of China” grants 2012CB518000 and 2012CB910400; NIH grant R01 AI100604; National Science Foundation of China grants 31270766, 81161120425, and 81025017; and Shanghai Science and Technology Committee grants 11JC1414800 and 12PJ1410500. Additionally, V.C. and R.C.S. acknowledge support from NIH grant U54 GM094618 (Target GPCR-28); I.K. acknowledges support from U01 GM094612, U54 GM094618, and R01 GM071872. The data presented in this paper are tabulated in the main paper and in the supplementary materials.
View Abstract

Navigate This Article