Deciphering the in vitro homo and hetero oligomerization characteristics of CXCL1/CXCL2 chemokines

Abstract

Chemokines share the fundamental property of oligomerization and regulate leukocyte migration via interacting with glycosaminoglycans and G-protein coupled receptors. Our studies on murine neutrophil activating chemokines CXCL1(mKC)/CXCL2(MIP2) deciphered their differential homo oligomerization potentials and heterodimer forming capabilities, thus adding another layer of regulatory mechanism for leukocyte trafficking during infection/inflammation.

---

Chemokines are small 8–10 kD chemotactic cytokines that perform diversified biological activities including leukocyte trafficking, organogenesis, wound healing, inflammation, angiogenesis, tumorigenesis etc.^1–3 They regulate these molecular signaling events by interacting with G-protein coupled receptors⁴ on the leukocytes (GPCRs) and glycosaminoglycans⁵ (GAGs) in the extracellular matrix and endothelial cell surface. Chemokines are sub-divided into four different classes (CXC, CC, CX3C, C) based on the arrangement of cysteine residues at the N-terminus.⁶ Despite their significant differences in chemokine sequences and functionalities, they adopt a common tertiary fold encompassing an extended disordered N-terminal domain, followed by 3₁₀ helix, three anti parallel β strands and a C-terminal α-helix.⁷

Oligomerization is considered to be an indispensable regulatory mechanism by which chemokines generate differential/sustainable chemotactic gradients on the endothelial cell surface during the influx of the migrating leukocytes.⁴ Chemokines commonly adapt two different types of dimers (CXC/CC-type), however XC chemokine lymphotactin adopts a non-canonical dimeric structure.^8,9 CXC type dimerization involves interactions between the residues from first beta strand from each monomer thus forms a six stranded β-sheet topped by two C-terminal α-helices whereas CC-type dimerization involves interactions between the residues present on the unstructured N-terminus of each monomer that forms a β-sheet in a dimer.^10,11 Further, several chemokines such as CXCL4, CXCL7, CXCL10, CXCL12 and CCL5 are reported to form tetramers and higher order oligomers by involving both CXC and CC dimeric surfaces.^12–16 Recent studies have identified that some chemokines (CCL3–CCL4, CXCL4–CCL5, CXCl1–CXCL7, CXCL4–CXCL8, CCL21–CXCL13, CXCL9–CXCL12, CCL2–CCL8) also undergo hetero-oligomerization thus adding another layer of complexity to the mechanism of leukocyte trafficking.^17–25

Neutrophil activating chemokines (NACs), involved in providing the first line of defense by recruiting neutrophils at the site of infection, belongs to the CXC family and contains the conserved ELR–CXC motif for receptor recognition.^26,27 Studies revealed that NACs have an intrinsic tendency to form higher oligomerization states. Despite belonging to same sub-family and sharing high sequence/structural identity, a significant diversity in their oligomerization potencies is reported.²⁸ However, limited knowledge is available regarding the formation of hetero oligomerization within the NACs. In order to shed light on the homo/hetero oligomerization potencies of NACs, we have chosen murine CXCL1(mKC) and CXCL2(MIP2) that belongs to growth related oncogene (GRO) subfamily of NACs. We explored their oligomerization potencies using biophysical and biochemical experimental analysis. Our studies indicated that mCXCL1 forms a weaker oligomeric species as compared to its paralogue mCXCL2, and they also interact with each other to form hetero-oligomers. To the best of our knowledge, this is the first experimental study providing the formation of hetero-dimers within the members of GRO chemokines.

Oligomerization potentials of mCXCL1 and mCXCL2 were examined using glutaraldehyde cross linking assay under same experimental conditions including protein concentrations. As shown in Fig. 1A, under the influence of the crosslinking agent both proteins forms higher order oligomers. At lower concentration of glutaraldehyde (0.0005% and 0.001%) the dimeric species are formed, and a further increase of the cross-linker concentration (0.005% and 0.01%) resulted in higher order tetrameric species. In order to delineate their oligomerization efficacies, we quantified the intensity profiles of the various oligomeric states (Fig. 1B). The intensity analysis suggested that mCXCL2 has a higher affinity to form the oligomers as compared to mCXCL1. At 0.0005% of glutaraldehyde an excess of monomeric species of mCXCL1 is left over and at higher concentration (0.01% glutaraldehyde), although overall dimeric content of both proteins are similar, the ratio of the left over monomer to tetramer varies, suggesting that mCXCL1 has a weak oligomerization tendency as compared to its paralogue mCXCL2 (Fig. 1B).

Fig. 1 (A) 15% SDS-PAGE analysis of glutaraldehyde cross linking of mCXCL1 and mCXCL2; (B) normalized intensities of mCXCL1, and mCXCL2 oligomeric forms generated by 0.005% and 0.01% glutaraldehyde; M, D, T denotes monomer, dimer, and tetramer oligomeric states; ‘s’ and ‘e’ represents starting and ending concentrations of glutaraldehyde; (C) ¹H–¹⁵N HSQC spectra of mCXCL1 and mCXCL2 (∼150 μM) at 25 °C. Few dimeric resonances in mCXCL2 spectrum were annotated for clarity, and the corresponding spectral region of mCXCL1 comprising of both monomeric and dimeric peaks was encircled.

The oligomerization profiles observed in the crosslinking experiment depends on two factors (a) the association constant, which indeed is a ratio of the forward and backward kinetic rate constants and; (b) non-specific nature and availability of the residues that are involved in covalent crosslinking. In order to confirm that the observed differential homo oligomerization behavior is not due to the influence of covalent cross-linker, but an inherent and unique feature of their association rates, we performed the NMR ¹H–¹⁵N HSQC experiment of mCXCL1 and mCXCL2 at same concentration (∼150 μM) (Fig. 1C). Clearly, mCXCL1 displayed two set of peaks corresponding to its dimeric (major) and monomeric (minor) populations in contrast to single set of dimeric resonances in mCXCL2 spectrum, establishing their intrinsic differential homo dimerization capabilities. Such a differential oligomerization can be attributed to the dynamic nature of the C-terminal helix in mCXCL1 as compared to mCXCL2 in the dimeric conformation.^29,30

Further, recent structural studies of the mCXCL1 dimer evidenced the similarity of structural features of mCXCL1/2 in their monomeric and dimeric forms.^29,31 Upon confirming the differential homo dimerization and similar structural features, we were intrigued to understand the plausibility of heterodimer formation within these two entities sharing high sequence and structural similarity. First, we carried out the comparative analysis of Cα contacts for mCXCL1 and mCXCL2 by constructing their contact maps (Fig. 2A). We observed 82% of the contact maps of CXCL1 and CXCL2 are overlapping and the remaining 18% of the differences in the contacts are due to the differences in their local amino acid sequences. Further our sequence analysis suggested that, at the β1–β1′ strand of dimeric interface (Fig. 2B), only a single residue is different (L30 in mCXCL1 and T29 in mCXCL2). Thus, all the in silico analysis is pointing towards a favorable heterodimer formation.

Fig. 2 (A) Overlay of contact maps of mCXCL1 (green) and mCXCL2 (pink) where N and C represents N-terminal and C-terminal of the protein sequences, the structural overlay of the monomeric subunits are also shown; (B) structure of mCXCL1 dimer highlighting the residues at the dimer interface β-strand termini; (C) overlay of ¹H–¹⁵N HSQC spectra [encircled spectral region (Fig. 1C)] of ¹⁵N-mCXCL1 (blue) and ¹⁵N-mCXCL1 + ¹⁴N-mCXCL2 (red) in the ratio of 1 : 1 showing the molecular interaction of mCXCL1 with mCXCL2. The notations in the inset are M-monomer, D-dimer, and HD-hetero dimer; (D) normalized intensities of the NH resonances of Q25 and L30; M, D, H represents monomer, dimer, hetero-dimer respectively; ‘a’ and ‘c’ represents alone (mCXCL1) and complex (mCXCL1 + mCXCL2) respectively; * denotes the intensities of hetero-dimer peaks specific to the mCXCL1/mCXCL2 complex spectra.

In order to establish the formation of mCXCL1/2 heterodimers experimentally, we mixed equimolar concentrations of both proteins and monitored the ¹H–¹⁵N resonances of mCXCL1 (Fig. 2C, Fig. S1†). The addition of unlabeled mCXCL2 resulted in new set of dimer-interface peaks (Fig. 2C) along with the existing two sets of resonances (mCXCL1 monomer/homo-dimer), accompanied by a significant attenuation in the peak intensities. We quantified the intensities of the dimer interface residues Q25 and L30 of mCXCL1 (Fig. 2D) before and after the addition of mCXCL2, in order to calculate the populations of the three species (monomer, homo and hetero dimers). Our analysis suggested the presence of 30–35% of monomeric species and 60–65% of dimer species (in terms of monomeric population) in the apo spectra, which is in good agreement with the reported Kd (∼30 μM) for mCXCL1 monomer–dimer equilibrium.²⁹ The homo monomer–dimer equilibrium is now distributed in the ratio of ∼25% monomer, ∼35% homodimer and 40% heterodimer for both the resonances Q25 and L30 in the complex spectra (Fig. 2D), thus directly demonstrating the potential formation of mCXCL1–mCXCL2 heterodimers.

Growth related oncogene chemokines that are formed as a resultant of the gene duplication events and includes three NACs (CXCL1, CXCL2 and CXCL3) (Fig. 3A) with a high level of sequence and structural similarity with varied functional potencies in all the mammalian species.^29,31–35 The varied functional behaviors can be attributed to; (i) the differential receptor/GAG binding surfaces, where the key residues are altered during evolution thus creating the same structure with differential binding surfaces. In the case of mCXCL1, these residues are recognized to be H20, K22, K62 and K66 respectively.²⁹ Alteration of these residues in the homodimers will significantly influence the GAG binding and recruitment process within the CXCL1/2/3 proteins (Fig. 3A and B). (ii) The variability in the oligomerization equilibrium also significantly contribute to the differential activity/recruitment profile; as the net accumulation of the individual chemokine oligomeric variants potentially regulates the steepness and sustainability of chemotactic gradients.⁴ (iii) Further, the recruitment profile is also altered by the generation of hetero-dimeric/oligomeric species in which both the oligomerization propensities and the receptor/GAG binding efficacies attenuate. Such heterodimeric species can generate more contrasting binding surfaces on each monomeric counterpart (Fig. 3C), thus contributing to the specificity and differential binding characteristics. These molecular regulatory mechanisms can certainly add another layer of regulatory mechanism in governing the influx of leukocyte migration during infection and injury (Fig. 4). Moreover, such hetero-species are less susceptible to protease degradation compared to their homo-oligomers as the site of cleavage and degradation kinetics varies in both monomeric counterparts. Indeed studies on hetero-oligomers between the CXC chemokine PF4/CXCL4 and the CC chemokine RANTES/CCL5 depicted the enhancement of leukocyte arrest on endothelial cells as compared to CCL5 alone.²³ On a similar note, heterodimers of CXCL4 and CXCL8 had direct implications with an enhancement of CXCL4 anti-proliferative effect on endothelial cells and CXCL8 mediated migration of hCXCR2 cells.¹⁹

Fig. 3 (A) Amino acid sequences of murine GRO chemokines (mCXCL1, mCXCL2 and mCXCL3) in which conserved residues are highlighted (ELR motif – cyan, cysteine involved in disulfide bridges – yellow), positive charge residues H20, K22, K62, K66 crucial for GAG binding (blue) and their alterations (red). The secondary structural elements are shown with arrows (β1 essential for dimerization – green, β2 and β3 – red) and cylinders (3₁₀ helix – blue, α-helix – purple); surface representation of homodimers (B) and heterodimers (C) of mCXCL1 (light pink), mCXCL2 (pale green) and mCXCL3 (slate blue). The monomeric fold of the CXCL3 was generated through homology modeling, and the homo/hetero dimeric structures of all the three chemokines were obtained by performing the symmetry operations in Pymol molecular graphics system as described in supplementary methods. The essential residues for GAG binding on the helical surface and their alterations are annotated and highlighted. Positive charge residues (blue), negative charge residues (red) and hydrophobic residues (grey).

Fig. 4 Schematic showing the different oligomeric species (monomers, homodimers and heterodimers) of mCXCL1/mCXCL2 (pink/green spheres) chemokines driving neutrophil recruitment process at the infected site through their interactions with cell surface glycosaminoglycans (GAGs) and G-protein coupled receptors (GPCRs) on the neutrophils.

Concluding remarks

In summary, we reported the first line of experimental evidence for the formation of heterodimeric complex within GRO genes. Our study established that the murine CXCL1 (mKC) and CXCL2 (MIP2) forms homo and hetero oligomers under in vitro conditions with different efficacies. Such a phenomenon of chemokine oligomerization is very important while considering them for therapeutic targets, as different oligomeric forms will have different functional characteristics, expression profiles and activation/signaling mechanisms. Moreover, the monomer–oligomer equilibrium will alter upon binding to the GAGs, a phenomenon known to facilitate/enhance the chemokine oligomerization which is in contrast to receptor binding, where specificity arises from few variants of oligomeric species for GPCR signaling.^26,28,36,37 Still there is a lot of gap in our understanding of homo/hetero oligomerization with respect to GAG/receptor binding and their regulatory role in leukocyte trafficking. Future studies involving molecular and cellular insights into this complex network of interactions will aid us in formulating chemokine variants with altered biological properties as therapeutics for a number of inflammatory and infectious diseases.

Acknowledgements

This work is supported by the DBT-IYBA fellowship – BT/07/IYBA/2013-19, SERB – SB/YS/LS-380/2013 and startup aid from MHRD-IITR, Government of India to KMP. Authors acknowledge the NMR Instrumentation center at IIT-Roorkee and CBMR Lucknow for spectrometer time. We thank Mr Krishnakant Gangele, IIT-Roorkee and Dr Sri Rama Koti Ainavarapu, TIFR-Mumbai for technical support.

References

1. M. Baggiolini, Nature, 1998, 392(6676), 565–568.
2. C. Gerard and B. J. Rollins, Nat. Immunol., 2001, 2(2), 108–115.
3. D. Raman, T. Sobolik-Delmaire and A. Richmond, Exp. Cell Res., 2011, 317(5), 575–589.
4. C. L. Salanga and T. M. Handel, Exp. Cell Res., 2011, 317(5), 590–601.
5. K. Gulati and K. M. Poluri, Glycoconjugate J., 2016, 33(1), 1–17.
6. A. Zlotnik and O. Yoshie, Immunity, 2012, 36(5), 705–716.
7. E. J. Fernandez and E. Lolis, Annu. Rev. Pharmacol. Toxicol., 2002, 42, 469–499.
8. X. Wang, J. S. Sharp, T. M. Handel and J. H. Prestegard, Prog. Mol. Biol. Transl. Sci., 2013, 117, 531–578.
9. R. L. Tuinstra, F. C. Peterson, S. Kutlesa, E. S. Elgin, M. A. Kron and B. F. Volkman, Proc. Natl. Acad. Sci. U. S. A., 2008, 105(13), 5057–5062.
10. G. M. Clore, E. Appella, M. Yamada, K. Matsushima and A. M. Gronenborn, Biochemistry, 1990, 29(7), 1689–1696.
11. T. M. Handel and P. J. Domaille, Biochemistry, 1996, 35(21), 6569–6584.
12. X. Zhang, L. Chen, D. P. Bancroft, C. K. Lai and T. E. Maione, Biochemistry, 1994, 33(27), 8361–8366.
13. G. J. Swaminathan, D. E. Holloway, R. A. Colvin, G. K. Campanella, A. C. Papageorgiou, A. D. Luster and K. R. Acharya, Structure, 2003, 11(5), 521–532.
14. M. G. Malkowski, J. Y. Wu, J. B. Lazar, P. H. Johnson and B. F. Edwards, J. Biol. Chem., 1995, 270(13), 7077–7087.
15. X. Wang, C. Watson, J. S. Sharp, T. M. Handel and J. H. Prestegard, Structure, 2011, 19(8), 1138–1148.
16. J. W. Murphy, H. Yuan, Y. Kong, Y. Xiong and E. J. Lolis, Proteins: Struct., Funct., Bioinf., 2010, 78(5), 1331–1337.
17. J. Carlson, S. A. Baxter, D. Dreau and I. V. Nesmelova, Biochim. Biophys. Acta, 2013, 1834(1), 158–168.
18. S. E. Crown, Y. Yu, M. D. Sweeney, J. A. Leary and T. M. Handel, J. Biol. Chem., 2006, 281(35), 25438–25446.
19. I. V. Nesmelova, Y. Sham, A. Z. Dudek, L. I. van Eijk, G. Wu, A. Slungaard, F. Mortari, A. W. Griffioen and K. H. Mayo, J. Biol. Chem., 2005, 280(6), 4948–4958.
20. I. V. Nesmelova, Y. Sham, J. Gao and K. H. Mayo, J. Biol. Chem., 2008, 283(35), 24155–24166.
21. S. Paoletti, V. Petkovic, S. Sebastiani, M. G. Danelon, M. Uguccioni and B. O. Gerber, Blood, 2005, 105(9), 3405–3412.
22. D. Venetz, M. Ponzoni, M. Schiraldi, A. J. Ferreri, F. Bertoni, C. Doglioni and M. Uguccioni, Int. J. Cancer, 2010, 127(10), 2300–2312.
23. P. von Hundelshausen, R. R. Koenen, M. Sack, S. F. Mause, W. Adriaens, A. E. Proudfoot, T. M. Hackeng and C. Weber, Blood, 2005, 105(3), 924–930.
24. A. Z. Dudek, I. Nesmelova, K. Mayo, C. M. Verfaillie, S. Pitchford and A. Slungaard, Blood, 2003, 101(12), 4687–4694.
25. E. Guan, J. Wang and M. A. Norcross, J. Biol. Chem., 2001, 276(15), 12404–12409.
26. M. W. Nasser, S. K. Raghuwanshi, D. J. Grant, V. R. Jala, K. Rajarathnam and R. M. Richardson, J. Immunol., 2009, 183(5), 3425–3432.
27. P. M. Murphy, Semin. Hematol., 1997, 34(4), 311–318.
28. K. Rajarathnam, C. M. Kay, B. Dewald, M. Wolf, M. Baggiolini, I. Clark-Lewis and B. D. Sykes, J. Biol. Chem., 1997, 272(3), 1725–1729.
29. K. M. Poluri, P. R. Joseph, K. V. Sawant and K. Rajarathnam, J. Biol. Chem., 2013, 288(35), 25143–25153.
30. W. Shao, L. F. Jerva, J. West, E. Lolis and B. I. Schweitzer, Biochemistry, 1998, 37(23), 8303–8313.
31. D. Rajasekaran, C. Keeler, M. A. Syed, M. C. Jones, J. K. Harrison, D. Wu, V. Bhandari, M. E. Hodsdon and E. J. Lolis, Biochemistry, 2012, 51(28), 5642–5654.
32. L. A. Al-Alwan, Y. Chang, A. Mogas, A. J. Halayko, C. J. Baglole, J. G. Martin, S. Rousseau, D. H. Eidelman and Q. Hamid, J. Immunol., 2013, 191(5), 2731–2741.
33. S. Haskill, A. Peace, J. Morris, S. A. Sporn, A. Anisowicz, S. W. Lee, T. Smith, G. Martin, P. Ralph and R. Sager, Proc. Natl. Acad. Sci. U. S. A., 1990, 87(19), 7732–7736.
34. H. R. Luttichau, J. Biol. Chem., 2010, 285(12), 9137–9146.
35. D. F. Smith, E. Galkina, K. Ley and Y. Huo, Am. J. Physiol.: Heart Circ. Physiol., 2005, 289(5), H1976–H1984.
36. A. E. Proudfoot, T. M. Handel, Z. Johnson, E. K. Lau, P. LiWang, I. Clark-Lewis, F. Borlat, T. N. Wells and M. H. Kosco-Vilbois, Proc. Natl. Acad. Sci. U. S. A., 2003, 100(4), 1885–1890.
37. J. H. Tan, M. Canals, J. P. Ludeman, J. Wedderburn, C. Boston, S. J. Butler, A. M. Carrick, T. R. Parody, D. Taleski, A. Christopoulos, R. J. Payne and M. J. Stone, J. Biol. Chem., 2012, 287(18), 14692–14702.

---

Footnote

† Electronic supplementary information (ESI) available: Complete details of the experimental methods. See DOI: 10.1039/c6ra01884j

---