Dear Editor,
CXC chemokine receptor 3 (CXCR3) and three interferon-induced CXC chemokines, specifically CXCL9 (Mig), CXCL10 (IP-10), and CXCL11 (I-TAC), are strongly associated with the migration of CD4+ Th1 cells and CD8+ cytotoxic T lymphocytes in immune responses1,2. The physiological and pathological functions of CXCR3 have been studied in infection, cancer, autoimmune diseases, and transplant rejection2,3,4,5. Only one CXCR3 antagonist, AMG487, has been evaluated in clinical trials for psoriasis and rheumatoid arthritis. AMG487 is a quinazolinone derivative that could prevent the binding of CXCL10 and CXCL11 to CXCR3 with high selectivity6. To date, the mechanism of the antagonism of AMG487 remains unclear. Here we determined the structure of CXCR3 complexed with AMG487 and the structure of the CXCR3–DNGi complex activated by CXCL10. The molecular mechanism of CXCR3 inhibition by AMG487 is elucidated, and we believe that our study will provide insightful perspectives for developing CXCR3-targeting antagonists.
The method of coupling a nanobody Nb6 to the receptor was used to facilitate the cryo-EM analysis. The ICL3 of CXCR3 was replaced by the ICL3 of the kappa opioid receptor. The chimeric receptor retained comparable response to CXCL10 (Supplementary Fig. S1). CXCR3κOR was purified in the presence of AMG487 and coupled to nanobody Nb6 (Supplementary Fig. S2a, b). Single particle analysis of the purified complex yields a density map at 3.0 Å resolution (Supplementary Fig. S2c–g and Supplementary Table S1). In the density map, the transmembrane helices could be distinguished, and residues 54–338 could be traced (Fig. 1a, b and Supplementary Fig. S2h). A density in the central pocket was found to be suitable for accommodating AMG487 (Fig. 1b).
In the CXCR3κOR–AMG487–Nb6 complex, AMG487 is trapped in a negatively charged pocket in CXCR3, with a buried surface area of 481.6 Å2 (Fig. 1c). The binding pocket is open to the lipid bilayer through a cleft between TM1 and TM7. The trifluoromethoxy (OCF3) group is buried in the cleft and faces toward the lipid bilayer (Fig. 1c, d). AMG487 is sandwiched between TM7 on one side and TM1 and TM2 on the other side (Fig. 1d). The aza-quinazolinone group is mainly stabilized by stacking with Trp1092.60 and is hydrogen bonded to the side chain of Tyr3087.43 (Fig. 1d and Supplementary Fig. S3a). The ethoxyphenyl group is stabilized by hydrophobic interactions with Tyr601.39, Leu1062.57, and Trp1092.60, while the benzene ring of the ethoxyphenyl group stacks with the benzene ring of Tyr3087.43 (Fig. 1d and Supplementary Fig. S3a). The pyridinylmethyl group is sandwiched between the side chains of Lys3007.35 and Ser3047.39 in TM7 and forms a hydrogen bond with the side chain of Ser3047.39 (Fig. 1d and Supplementary Fig. S3a). The hydroxy group of Tyr601.39 points towards the center of the benzene ring in the benzeneacetamide group, and the carbonyl group is hydrogen bonded to the side chain of Ser3017.36 (Fig. 1d and Supplementary Fig. S3a).
NBI-74330 is a CXCR3 antagonist that shows therapeutic potential in animal models of atherosclerosis and arthritis7,8. In AMG487, the 4’ position of the benzene ring is occupied by an OCF3 group (Fig. 1e). In NBI-74330, a fluorine atom occupies the 4’ position and a trifluoromethyl (CF3) group occupies the 3’ position of the benzene ring (Fig. 1e). When docking NBI-74330 into the density of AMG487 in Coot, the posture was very similar to that of AMG487 (Fig. 1f). Therefore, we believe that NBI-74330 may inhibit CXCR3 in a way similar to AMG487. By comparing the structures of chemokine receptors complexed with antagonists, we found that the binding pocket of AMG487 in CXCR3 is largely overlapped with that of MK-0812 and BMS-681 in CCR2 (Supplementary Fig. S3b)9,10. In comparison, the binding pocket of AMG487 only partially overlaps with that of the CCR5 antagonists maraviroc, compound 21, and compound 34, and the CXCR4 antagonist IT1t (Supplementary Fig. S3c, d). The CF3 groups in AMG487, MK-0812, and BMS-681 occupy similar positions between TM1 and TM7 (Supplementary Fig. S3b). Residues 1.39, 2.60, 7.36, 7.39, and 7.43 involved in interactions with AMG487 are also in contact with MK-0812 and BMS-681 (Supplementary Fig. S3e, f). Among these residues, 1.39 and 2.60 are identical in CXCR3 and CCR2, and residues 7.36, 7.39, and 7.43 on TM7 are Ser301, Ser304, and Tyr308 in CXCR3, but Gln288, Glu291, and Met295 in CCR2 (Fig. 1d and Supplementary Fig. S3e, f). The comparison suggests that AMG487 adopts an antagonistic mechanism similar to MK-0812 and BMS-681 but possesses some distinct features.
The CXCR3–CXCL10–DNGi complex was stabilized by the NanoBit tethering strategy. The complex was purified and further stabilized by scFv16 (Supplementary Fig. S4a, b). A density map at 3.2 Å resolution was obtained through single particle analysis (Supplementary Fig. S4c–g and Supplementary Table S1). The densities of the transmembrane helices could be distinguished, and the atom model was built accordingly (Fig. 1g, h and Supplementary Fig. S4h). For CXCL10, however, only eight residues in the proximal N-terminus could be recognized (Fig. 1h), suggesting that CXCL10 binds to CXCR3 with high flexibility.
The N-terminus of CXCL10 is surrounded by TM1, TM2, TM3, TM6, and TM7 (Fig. 1i). An uncharged “VPLS” motif and a positively charged “RTVR” motif could be distinguished. In the “VPLS” motif, Val1CXCL10 and Pro2CXCL10 insert most deeply and form hydrophobic interactions with Tyr601.39, Trp1092.60, Phe1313.32, Tyr2716.51, and Tyr3087.43 (Fig. 1i). The main chain carbonyl group of Val1CXCL10 is hydrogen bonded to the hydroxyl group of Ser3047.39 (Fig. 1i). Tyr205ECL2 is involved in hydrophobic stacking with Leu3CXCL10, while Ser4CXCL10 is coordinated by Asp521.31 and Lys3007.35 (Fig. 1i). In the “RTVR” motif, the hydroxyl group of Thr6CXCL10 is hydrogen bonded to the side chain of Gln204ECL2 (Fig. 1i). Arg8CXCL10 is in contact with the side chain of Glu2937.28 and the main chain of Cys2907.25 (Fig. 1i). Mutation of Asp521.31, Tyr601.39, Trp1092.60, Phe1313.32, Gln204ECL2, Cys2907.25, Glu2937.28, Ser3047.39, and Tyr3087.43 resulted in decreased potency of CXCL10 (Fig. 1j and Supplementary Table S2), suggesting these residues are crucial for CXCL10 binding and receptor activation. Truncation of two amino acids in the N-terminus of CXCL10 resulted in reduced CXCR3-binding properties, loss of calcium signaling capacity, and a 30-fold reduction in chemotactic activity11. Mutation of Arg5 and Arg8 in the N-terminus of CXCL10 resulted in a 7- and 60-fold increase in IC5012. Therefore, both the uncharged VPLS motif and the positively charged RTVR motif in the N-terminus of CXCL10 are critical for the ligand binding and receptor activation. Due to the sequence similarity of the N-terminus of CXCL9, CXCL10, and CXCL11 (Supplementary Fig. S5a), the structure of the CXCR3–CXCL10–DNGi complex may provide insight into the binding pattern of the N-terminus of CXCL9 and CXCL11.
Compared to AMG487-coupled CXCR3, the extracellular region of CXCL10-coupled CXCR3 is more compact with the inward movement of TM2 and TM5 (Supplementary Fig. S5b). On the intracellular side, the outward wing of TM6 releases the packing between TM3 and TM6, exposing the binding cavity for the Gαi protein (Supplementary Fig. S5c, d). The conformational change of TM6 is accompanied by the inward displacement of TM3, TM5, and TM7 (Supplementary Fig. S5c, d). MD simulations indicate that the apo CXCR3 generally undergoes greater movement than the AMG487-bound CXCR3 (Supplementary Fig. S5e–g). In a snapshot of the apo CXCR3 simulation (Supplementary Fig. S5h), movements on the extracellular side of TM1, TM6, and TM7, and small shifts on the intracellular side of TM7 are observed. Therefore, we suggest that the binding of AMG487 may stabilize the inactive conformation. In addition, the binding pockets of AMG487 and CXCL10 partially overlap (Fig. 1k). Several residues (Tyr601.39, Trp1092.60, Ser3047.39, and Tyr3087.43) that are critical for chemokine binding and the activation of CXCR3 are occupied by AMG487 in the inactive state. The observation is consistent with the previous study that AMG487 could compete with CXCL10 by binding to the orthosteric pocket6.
According to the mass spectrum analysis (Supplementary Fig. S6a, b) and electron densities (Supplementary Fig. S6c–e), several lipids were identified in the density map of the CXCR3κOR–AMG487–Nb6 complex. Firstly, a cholesterol molecule is trapped in the cavity surrounded by TM2, TM3, and TM4 (Supplementary Fig. S6c). Allosteric antagonists AZ3451 and ORG27569 were found to bind to PAR2 and CB1 receptors in similar sites (Supplementary Fig. S6c). Secondly, in the cavity defined by TM3, TM4, and TM5, a lysophosphatidylcholine molecule is found (Supplementary Fig. S6d). This cavity is also well-known for allosteric antagonist development, such as NDT9513727/avacopan targeting C5aR and AS408 targeting β2-AR (Supplementary Fig. S6d). In addition, a phosphatidylcholine binding site surrounded by TM3, TM5, and TM6 was identified (Supplementary Fig. S6e). The entrance to the phosphatidylcholine binding pocket between TM5 and TM6 is guarded by Leu2285.51 and Ala2656.45. Compared to the inactive structure of CXCR2 and CXCR413,14, the distances between the α-carbon atoms of 5.51 and 6.45 are comparable (Supplementary Fig. S6f–h). However, the side chains of 5.51 and 6.45 in CXCR3 are smaller and the distance between the side chains is obviously larger. Therefore, we suggest that the smaller residues in CXCR3 make the cavity able to accommodate the lipids. This lipid-binding site has not been characterized in other GPCRs. Further studies are required to verify the binding and functions of these lipids in the wild-type receptor and to explore the druggability of these lipid-binding sites.
Data availability
The coordinates and maps for the CXCR3κOR–AMG487–Nb6 and CXCR3–CXCL10–DNGi–scFv16 complex structure have been deposited in the Protein Data Bank/Electron Microscopy Data Bank under accession code 8K2W/36841 and 8K2X/36842, respectively.
References
Groom, J. R. et al. Immunity 37, 1091–1103 (2012).
Hickman, H. D. et al. Immunity 42, 524–537 (2015).
Chow, M. T. et al. Immunity 50, 1498–1512.e5 (2019).
Lee, J. H. et al. Arthritis Res. Ther. 19, 163 (2017).
Croudace, J. E. et al. Blood 120, 4246–4255 (2012).
Johnson, M. et al. Bioorg. Med. Chem. Lett. 17, 3339–3343 (2007).
Ahmad, S. F. et al. Int. Immunopharmacol. 118, 110099 (2023).
van Wanrooij, E. J. et al. Arterioscler. Thromb. Vasc. Biol. 28, 251–257 (2008).
Zheng, Y. et al. Nature 540, 458–461 (2016).
Apel, A. K. et al. Structure 27, 427–438.e5 (2019).
Proost, P. et al. Blood 98, 3554–3561 (2001).
Campanella, G. S., Lee, E. M., Sun, J. & Luster, A. D. J. Biol. Chem. 278, 17066–17074 (2003).
Wu, B. et al. Science 330, 1066–1071 (2010).
Liu, K. et al. Nature 585, 135–140 (2020).
Acknowledgements
This work was in part supported by the National Natural Science Foundation of China, the Youth Science Fund (32100963 to H.L.H.), Shenzhen Science and Technology Innovation Committee (JCYJ20210324131802008 to H.L.H.), and the Shenzhen-Hong Kong Cooperation Zone for Technology and Innovation (HZQB-KCZYB-2020056). This work was also supported by the Kobilka Institute of Innovative Drug Discovery and Presidential Fellowship and University Development Fund at the Chinese University of Hong Kong, Shenzhen (H.L.H., H.Z.J., Q.C.). We thank the Kobilka Cryo-EM Center at the Chinese University of Hong Kong, Shenzhen, for supporting EM data collection. We thank BGI for the mass spectrometry.
Author information
Authors and Affiliations
Contributions
H.Z.J. purified the complexes and conducted the cryo-EM analysis. B.P. accomplished the cAMP assay. Y.C.C. conducted the MD simulation. Q.C. contributed to data collection. Q.P. helped in baculovirus preparation. H.Z.J., B.P., R.B.R., and H.L.H. wrote the manuscript.
Corresponding authors
Ethics declarations
Conflict of interest
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Jiao, H., Pang, B., Chiang, YC. et al. Structure basis for the modulation of CXC chemokine receptor 3 by antagonist AMG487. Cell Discov 9, 119 (2023). https://doi.org/10.1038/s41421-023-00617-0
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41421-023-00617-0