NMR mapping of RANTES surfaces interacting with CCR5 using linked extracellular domains

Abstract

Chemokines constitute a large family of small proteins that regulate leukocyte trafficking to the site of inflammation by binding to specific cell-surface receptors belonging to the GPCR superfamily. The interactions between N-terminal (Nt-) peptides of these GPCRs and chemokines have been studied extensively using NMR spectroscopy. However, due to lower affinities of peptides representing the three extracellular loops (ECLs) of chemokine receptors to their respective chemokine ligands, information concerning these interactions is scarce. To overcome the low affinity of ECL peptides to chemokines, we linked two or three CCR5 extracellular domains by either biosynthesis in Escherichia coli or by chemical synthesis. Using such chimeras, CCR5 binding to RANTES was followed using ¹H-¹⁵N-HSQC spectra to monitor titration of the chemokine with peptides corresponding to the extracellular surface of the receptor. Nt-CCR5 and ECL2 were found to be the major contributors to CCR5 binding to RANTES, creating a nearly closed ring around this protein by interacting with opposing faces of the chemokine. A RANTES positively charged surface involved in Nt-CCR5 binding resembles the positively charged surface in HIV-1 gp120 formed by the C4 and the base of the V3. The opposing surface on RANTES, composed primarily of β2-β3 hairpin residues, binds ECL2 and was found to be analogous to a surface in the crown of the gp120 V3. The chemical and biosynthetic approaches for linking GPCR surface regions discussed herein should be widely applicable to investigation of interactions of extracellular segments of chemokine receptors with their respective ligands.

Introduction

Chemokines constitute a large family of small proteins (8–10 kDa) that regulate leukocyte trafficking to the site of inflammation [1]. Recent experimental evidence suggests a broadening repertoire of functions for these intriguing molecules, including involvement in angiogenesis, metastasis and leukocyte maturation and development [1–3]. Chemokines exert their biological activity by binding to cell surface receptors, which belong to the G protein-coupled receptor (GPCR) superfamily. Chemokine receptors belong to class-A GPCRs and like all GPCRs they consist of seven helical trans-membrane domains. Unlike most other class-A GPCRs, which bind small molecules or peptides, chemokine receptors bind small protein ligands. To date, 19 receptors and 50 different chemokines have been identified [4].

Two chemokine receptors, CCR5 and CXCR4, serve as co-receptors for human immunodeficiency virus type 1 (HIV-1). CCR5, which binds macrophage inflammatory proteins 1a and 1β (MIP-1a and MIP-1b) and RANTES, is used by HIV-1 viruses designated R5 viruses, and CXCR4, which binds the stromal cell-derived factor-1 (SDF-1) as a sole ligand, is used by HIV-1 strains designated X4 viruses. CCR5 is of special importance in HIV-1 research [5, 6], because R5 strains are responsible for HIV-1 transmission in the vast majority of HIV-1 infections [6].

Structures of a large number of chemokines have been solved by NMR and X-ray crystallography, among them MIP-1a, MIP-1b and RANTES [7–10]. All chemokines share a conserved tertiary fold composed of an N-terminus which forms a β-strand in CC-chemokine dimers, a relatively long first loop (the N-loop), 3 antiparallel β-strands separated by short loops, and a C-terminal α-helix. The 8 kDa CCR5 chemokines RANTES, MIP-1α and MIP-1β, like many other chemokines, were found to form a symmetric homodimer in solution and larger aggregates at pH > 4 [7]. The aggregation of chemokines can be abolished by specific mutations such as the E66S mutation for RANTES, but the chemokines remain dimeric [11].

While the structures of many chemokines have been solved, the crystal structure of CXCR4 is the only high-resolution structure available for a chemokine receptor to date [12]. The CXCR4 structure reveals the classical seven-helix bundle motif but has significant structural divergence in both the helical core and loop domains from structures reported for Rhodopsin and for other class-A GPCRs that are triggered by non-peptide ligands.

In the absence of a high-resolution structure, site-directed mutagenesis as well as biochemical and immunological techniques have been used to characterize CCR5’s interactions with its respective ligands. Functional studies of different chemokines have led to the general conclusion that the N-terminus of these proteins is necessary for triggering receptor signaling activity, while core domains are the primary contributors to binding to the receptors [13–15]. In the case of CCR5, the N-terminus (Nt-CCR5) and the second extracellular loop (ECL2) of CCR5 have been identified as most important for binding to CC-chemokines and HIV-1 gp120 [16–19]. The major binding determinant is within a region rich in tyrosine and acidic amino acids in Nt-CCR5 [20, 21] and sulfation of Y10 and Y14 is essential for CCR5 activity [22–24].

Although structure determination of intact GPCRs by NMR remains an extremely challenging goal, complexes of chemokines with receptor-derived peptides offer an attractive option for studying chemokine-receptor interactions and for obtaining structural information on discrete regions of the receptor/ligand complexes. Binding site residues were mapped by following changes in NMR chemical shifts upon addition of peptides corresponding to the N-terminal regions of the CXCR1, CXCR3, CCR3 and CX3CR1 receptors to their respective chemokine partners [25–30]. In these studies, the affinity of the N-terminal receptor peptides to the chemokines was found to be in the 20–100 µM range [25, 26, 28].

Currently there are only two high-resolution structures of chemokines bound to receptor peptides. An NMR structure has been reported for an N-terminal CXCR1 peptide analog in complex with interleukin-8 [26]. The peptide used in this study contained unsulfated tyrosine although the receptor N-terminus contains a signal for tyrosine sulfation at Y27 (as analyzed by sulfinator [31]). More recently the NMR structure of a 38-residue trisulfotyrosine containing peptide bound to SDF-1 dimer was reported [32]. Both studies provide insight into specific residue to residue interactions and into the spatial details of the binding interface between the chemokines and their receptors. Such information is extremely valuable in mechanistic studies and in designing receptor antagonists [33].

Despite the prominence of the N-terminal region of CCR5, the extracellular loops of this receptor have also been implicated in ligand binding [34]. However, the binding of chemokines to peptides representing the individual ECLs seems to be much weaker than their affinities to the receptor's N-terminus [35]. This weak binding has prevented a complete mapping of the chemokines' binding site for their respective receptors. It is possible to increase the affinity of a domain of a multi-topic receptor for its interacting ligand by combining it with other regions of the receptor-interacting interface. This approach was used in the development of a soluble receptor analog consisting of the N-terminus and the third extracellular loop of CCR3 displayed on the B1 domain of protein G [36]. Remarkably, although the N-terminal peptide had an affinity of 80 µM and the ECL3 peptide didn’t bind to eotaxin, the chimeric peptide called CROSS bound with an affinity of 4 µM. Similar results have been reported for CCR2 peptides [37]. In a more recent study, a completely synthetic, multi-topic mimetic (ECD1–4) of the ectodomains of the corticotropin releasing factor receptor type 1 (CRF-1) was shown to bind to sauvagine with high affinity, whereas the N-terminus of CRF-1 or the multi-topic mimetic without the N-terminus (ECD2–4) did not bind to this natural agonist of CRF-1 [38]. Very recently Eichler and coworkers designed a CXCR4 mimetic that contained the three extra-cellular loops of this receptor with the native disulfide bond connecting ECL1 and ECL2 [39, 40]. The CXCR4 mimetic was shown to bind gp120 in the presence of soluble CD4, sCD4, and inhibited HIV-1 infection at 10 µM concentration [39]. These studies show that combination of discontinuous regions of the extracellular surface of a GPCR can result in increased affinity and thereby enable mapping and structural investigations on domains that cannot be studied using peptides representing single domains.

In the present investigation we created a panel of peptides by linking a few extracellular regions of CCR5, enabling us to study the contributions of Nt-CCR5, ECL1, ECL2 and ECL3 to ligand binding and to map the RANTES residues interacting with each of these regions. Specifically, we biosynthesized a soluble CCR5 ectodomain composed of unsulfated Nt-CCR5, ECL1 and ECL2 and have synthesized chimeric peptides containing disulfated-Nt-CCR5 conjugates with ECL1 and ECL3 and a conjugate of ECL1 and ECL2. These peptides, together with Nt-CCR5 as a control, were used to map the interactions of the CCR5 ECLs with RANTES. To our knowledge this is the first example of NMR studies investigating the interactions of multi-topic receptor surface mimetics with their natural ligands. This approach is general and, in principle, can be applied to NMR studies aimed at investigating the involvement of the extracellular loops of other chemokine receptors and GPCRs in binding to their ligands.

Results

Interactions of RANTES with Nt-ECL1-ECL2

A recombinant construct (Nt-ECL1-ECL2) containing the unsulfated N-terminal segment (residues 1–34) as well as ECL1 (residues 89–101) and ECL2 (residues 168–194) of CCR5, connected by flexible linkers was designed to facilitate NMR studies of RANTES interactions with ECL1 and ECL2. Cys101 in ECL1 and Cys178 in ECL2 were retained to enable the formation of the native disulfide bond between ECL1 and ECL2. C20 in the N-terminal segment was mutated to alanine to prevent formation of homodimers and alternative disulfide bonds involving C101 and C178. Due to expression in E. colithe tyrosine residues at the Nt-CCR5 segment were not sulfated.

To resolve the aggregation problem of RANTES, the E66S mutant [11] was used in our studies. Residues at the dimerization interface were previously found to manifest changes in chemical shifts resulting in the appearance of two populations of cross-peaks for the interface residues [35]. The appearance of two sets of cross-peaks indicated that the exchange between the monomer and dimer was slow in the NMR time scale. ¹H-¹⁵N-HSQC measurements at different chemokine concentrations indicated that, whereas E66S-RANTES was almost completely dimeric at 200 µM, a significant amount of monomer (~60%) coexisted with dimer at 12.5 µM [8, 35]. As shown in Fig. 1, the ratio of monomer/dimer peak intensities, I_M/I_Dof uniformly labeled (U-) ¹⁵N-RANTES decreased from 0.45 at 310 K to 0.13 at 293 K. Since both chemokines and chemokine receptors form homo and heterodimers [4] and since dimer is almost the exclusive form of RANTES at neutral pH, there is an interest in understanding the binding of CCR5’s extra cellular regions to the dimeric form of RANTES. Moreover, the homologous CXCR4 receptor forms a homodimer in the crystal structure [12].

Fig. 1. ¹H-¹⁵N-HSQC of RANTES.

Segments of ¹H-¹⁵N-HSQC spectra of 50 µM RANTES at pH 4.8 and 310 K (8 scans) or 293 K (16 scans) (left and right, respectively). Peaks are labeled with residue name and number. Superscript M and D stand for monomer and dimer, respectively. At 293 K only dimer peaks are observable. Spectra measured on a Bruker AVIII800 NMR spectrometer equipped with a cryoprobe.

At 293 K and pH 4.8, most of the RANTES molecules are dimers and the monomeric form almost disappears. Titration of U-¹⁵N-RANTES with Nt-ECL1-ECL2 resulted in gradual changes in chemical shifts of dimeric RANTES indicating fast exchange. Since the Nt-ECL1-ECL2 molecule has limited solubility and precipitated upon addition into RANTES, only half the concentration of RANTES could be obtained. Nevertheless, the titration experiment allowed mapping the binding interface for Nt-CCR5, ECL1 and ECL2 of CCR5 on RANTES. Significant changes in chemical shift (ΔCS > 0.02 ppm) were observed for a large number of residues (>20). The RANTES residues affected by Nt-ECL1-ECL2 binding are Y3-S5 and T7-T8 in the N-terminal segment, F12-I15 and L19 in the N-loop, K25, F27-Y28 and T30-S31 in the β1 strand, N36 in the 30's loop, V39 and R44-Q48 in the β2-β3 hairpin and the 40's loop and K55, W57 and M67-S68 in the 50's loop and C-terminal helix (Fig. 2A). A ribbon representation of the three dimensional structure of RANTES is shown in Fig. 2B. As can be seen in Fig. 2C, two extensive surfaces on opposing sides of RANTES close to the dimer interface were affected by Nt-ECL1-ECL2 binding.

Fig. 2. RANTES binding to Nt-ECL1-ECL2.

(A) Maximal U-¹⁵N-RANTES chemical shifts changes observed upon titration with Nt-ECL1-ECL2 at 293 K and pH 4.8 as a function of RANTES residue number. Sequence and secondary structure of RANTES are shown on bottom. Changes to sequence are highlighted in bold. A threshold of chemical shift changes of 0.02 ppm is shown in dotted line. (B) Ribbon representation of RANTES (PDB: 1RTO) depicting one subunit in magenta and the other in yellow. (C) Mapping RANTES residues involved in Nt-ECL1-ECL2 binding. Diagram of RANTES (PDB: 1RTO) showing in blue residues affected upon Nt-ECL1-ECL2 binding. One monomer is depicted in space filling representation and the other as a semi-transparent surface, which allows a view of the ribbon diagram and stick representation of the binding residues. Proline residues, which do not appear in ¹H-¹⁵N-HSQC spectra, are dotted.

The Nt-CCR5 binding surface on monomeric and dimeric RANTES

In order to assess the contribution of each of the CCR5 extracellular domains composing Nt-ECL1-ECL2 to RANTES binding, we first studied the binding of Nt-CCR5 to RANTES. We used an Nt-CCR5 peptide (residues 1–27) disulfated at Y10 and Y14 since sulfation at these positions was found to be important for RANTES binding. Unlike previously published results [35], Nt-CCR5 binding was observed also to the dimeric form of RANTES as shown in Figs. 3A, 3C and 3D. The binding of this peptide to both monomeric and dimeric RANTES was compared by carrying out the experiment at different temperature and pH conditions. To investigate Nt-CCR5 binding to RANTES monomers, titration experiments were conducted at 310 K and pH 4.8 using RANTES concentrations of 100 and 10 µM (data not shown). At this temperature and pH conditions, the approximate I_M/I_D was 0.2 (Fig. 3A) and 1.57 at 100 and 10 µM concentrations, respectively. Gradual changes in chemical shifts were observed for both monomeric and dimeric RANTES (Fig. 3A; eg C50^M and C50^D) indicating that Nt-CCR5 binds both the monomeric and dimeric forms of RANTES. As the titration progressed and the concentration of Nt-CCR5 increased, the intensity of the dimer peaks diminished while the intensity of the monomer-associated peaks increased. By the end of the titration, the ratio I_M/I_D was approximately 2.5 for cross-peaks that did not change their chemical shift upon Nt-CCR5 binding and therefore their intensity was not affected by the exchange rate. This observation suggests that Nt-CCR5 binding weakens the tendency of RANTES to form dimers. The change in monomer-dimer population upon Nt-CCR5 binding was also evident from the trajectories of some of the peaks, which were not always linear, as expected for a simple one-site binding event, but curved, implying a more complex mechanism of binding as illustrated in Fig. 3B. To obtain a rough estimate of the K_D we had to assume a constant concentration of monomeric and dimeric RANTES (although, as mentioned above, this assumption is not correct). A rough K_D estimate of 16 and 100 µM for the monomeric and dimeric form, respectively, was obtained (Fig. 4). The K_D value for monomeric RANTES/Nt-CCR5 is one order of magnitude larger than that previously determined using the same assumption [35]. Examination of the previously published results (Fig. 3b in [35]) reveals that the Nt-CCR5 concentration at which 50% saturation of RANTES was achieved (this concentration is equal to the K_D for a simple one site binding equilibrium) is in the 50 µM range (according to the published results) and not in the 1 µM range as reported for the K_D in that paper.

Fig. 3. RANTES binding to Nt-CCR5.

(A) Sections of overlaid ¹H-¹⁵N-HSQC spectra of U-¹⁵N-RANTES titrated with doubly sulfated Nt-CCR5 at pH 4.8 and 310 K (left panel) and 293 K (right panel). U-¹⁵N-RANTES at a concentration 80–100 µM alone (blue) and with increasing amount of Nt-CCR5: 1:1 (red), 1:2 (green) and 1:8 or 1:7 (left and right, respectively; purple) at 310 K and pH 4.8. Dimer population (labeled D) decreases, while monomer population (labeled M) increases as Nt-CCR5 is titrated into the sample. Experiments carried out in 100 mM d₄-acetate buffer, 5% D₂O, 0.005% thimerosal. Spectra measured on a Bruker AVIII800 NMR spectrometer equipped with a cryoprobe. (B) A diagram describing the system’s equilibria. (C) Maximal dimeric (yellow) and monomeric (blue) U-¹⁵N-RANTES chemical shifts changes observed upon titration with 8 times molar ratio (100 and 800 µM RANTES and doubly sulfated Nt-CCR5, respectively) at 310 K, as a function of RANTES residue number. The cutoff is set to 0.055. (D) Mapping RANTES residues involved in Nt-CCR5 binding. RANTES (PDB: 1RTO) showing the residues that underwent a chemical shift change larger than the 0.055 ppm cutoff upon binding doubly sulfated Nt-CCR5. In red, residues affected in both monomers and dimers; in blue and yellow residues affected only in monomeric or dimeric RANTES, respectively. One RANTES molecule is depicted in space filling diagram and the other in ribbon and semi-transparent surface representation. (E) Ribbon diagram of RANTES (PDB: 1RTO) colored as in (D) showing residues involved in Nt-CCR5 binding in the monomeric form (blue sticks).

Fig. 4. Determination of RANTES/Nt-CCR5 dissociation constant.

Plots of observed changes in chemical shifts of RANTES as a function of added Nt-CCR5 concentration. (A) Dimeric RANTES. Experiment was conducted at 293 K and pH 7.0. (B) Monomeric RANTES. Experiment was conducted at 310 K and pH 4.8. The curves were fit globally to the observed changes in chemical shifts of selected residues as a function of added peptide concentration according to equation 3 (Experimental Procedures) using the OriginLab software.

As shown in Fig. 3C and 3D, the RANTES surface affected by Nt-CCR5 binding was mapped to residues T7, F12-A16, L19, R21, H23, K25, T30, R47 and C50 for both the monomer and the dimer forms. For clarity, the secondary structure of the RANTES dimer is presented in Fig. 3E and residues that underwent changes in chemical shift upon Nt-ECL1-ECL2 binding are given in a stick representation. Residues Y3, S5-D6 in the N-terminus of RANTES dimers, were also affected upon Nt-CCR5 binding. The N-terminus of RANTES is involved in dimerization and binding of Nt-CCR5 may cause destabilization of the dimer interface. Peaks originating from residues C10 (at the dimer interface), S31-G32, A38, F41, T43, R47, V49, A51, and V58 changed their chemical shift upon binding Nt-CCR5 only for RANTES monomers. Residues F41, A51 and V58 (Fig. 3E), affected only upon monomer binding the peptide, are found in the core of the chemokine. This, together with the differences between monomeric and dimeric RANTES in binding Nt-CCR5, might imply that the monomeric RANTES is more malleable and its conformation is more affected by the binding of Nt-CCR5 in comparison with the dimeric RANTES that is more rigid.

When the residues in the RANTES dimer that exhibited the largest changes in chemical shift upon Nt-CCR5 binding are highlighted on the structure of RANTES (Fig. 3D left panel), an almost continuous large surface, which contains four positively charged residues, namely R47, K45, H23 and K25, is revealed on one face of RANTES while on the opposite face there are fewer residues which changed their chemical shift, and these do not form an extensive continuous surface. The surface area affected by Nt-CCR5 binding is very extensive and includes numerous residues in the β1 and β4 strands as well as in loops connecting the β-strands. We hypothesize that the extensive surface containing the four positively charged residues is involved in actual contact with the negatively charged Nt-CCR5 peptide, while the few changes in chemical shifts observed on the opposite face of RANTES (Fig. 3D right panel) are a result of some conformational changes, possibly due to movements of the β-strands.

To further characterize the binding of Nt-CCR5 to dimeric RANTES we repeated the titration experiment at a lower temperature and at pH 7.0, conditions in which RANTES is almost exclusively dimeric. At 293 K, the residues which were affected upon binding Nt-CCR5 are the same as at 310 K (data not shown). The preference towards peptide-binding monomers was evident at 293 K as well, and monomer peaks began to appear during the titration as Nt-CCR5 concentration in the sample increased. Initially I_M/I_D was less than 0.1 and by the end of the titration it increased to 0.5 (Fig. 3A).

In order to transfer the RANTES assignments made at pH 4.8 to pH 7.0, we performed a series of ¹H-¹⁵N-HSQC experiments of ¹⁵N-RANTES at increasing pH values. In most cases the transfer of the assignment from pH 4.8 to 7.0 was straightforward since most RANTES resonances did not experience large changes in chemical shifts. This suggests that RANTES does not undergo any major conformational changes at higher pH. The main difference in Nt-CCR5 binding between pH 7.0 and 4.8 is located at the N-terminus of RANTES, for which residues Y3-S5, affected upon peptide binding at pH 4.8, are not affected at pH 7.0.

The strong preference towards the dimeric form of RANTES at 293 K and pH 7.0 allows a rough estimate of the dissociation constant (as detailed in the Experimental Procedures) for the RANTES/Nt-CCR5 complex assuming, as discussed above, that the concentration of RANTES dimers does not change during the experiment. The K_D value extracted from this experiment is 0.084 ± 0.012 mM (Fig. 4A). This estimated value can serve as an upper limit for the actual dissociation constants at both pH 7.0 and pH 4.8. Unlike the results of Grzesiek and coworkers [35] we observed Nt-CCR5 binding to dimeric RANTES. However, this binding is about five-fold weaker than the binding to monomeric RANTES (Fig. 4B).

Interactions of RANTES with ECL1

As shown in Figs. 5A and 5B, residues Y27, F28, S31, R44, N46, Q48, K55, M67 and S68 were affected by Nt-ECL1-ECL2 binding while Nt-CCR5 binding caused considerably smaller changes in the chemical shifts of these residues (below the set threshold value). This suggests that these residues are involved in ECL1 and/or ECL2 binding. The contribution of ECL1 to RANTES binding was further investigated using a synthetic chimeric peptide (Nt-ECL1). In this chimera, a doubly sulfated Nt-CCR5 segment is connected to ECL1 via linker residues and a non-native disulfide bond connecting the C-terminal end of the first peptide with the N-terminal strand of the following peptide (Table 1). The length of the linker was designed to approximate the ~10 Å distance between the C-terminal end of Nt-CCR5 and the N-terminus of ECL1 in a CCR5 model calculated on the basis of the CXCR4 structure [41], while allowing some flexibility for optimal binding. The results of ¹H-¹⁵N-HSQC titration experiments of RANTES with Nt-ECL1 (data not shown) revealed the same RANTES surfaces affected by binding of this chimeric peptide as found in the titration experiments of RANTES with Nt-CCR5 under the same conditions (pH 4.8 and 293 K), suggesting ECL1 does not contribute significantly to RANTES binding. Therefore, the continuous surface, including P9, on RANTES (Fig. 5B, right panel) discovered upon Nt-ECL1-ECL2 binding but not Nt-CCR5 binding, is likely to be involved in ECL2 binding.

Fig. 5. The contribution of ECL2 to RANTES binding.

(A) Comparison of maximal U-¹⁵N-RANTES chemical shift changes observed upon titration with Nt-CCR5 (green bars) an Nt-ECL1-ECL2 (black bars) at 293 K and pH 4.8 as a function of RANTES residue number. (B) Mapping RANTES residues involved in Nt-ECL1-ECL2 binding. Space filling diagram of RANTES (PDB: 1RTO) showing in green residues affected upon Nt-CCR5 as well as Nt-ECL1-ECL2 binding, in magenta residues which underwent a chemical shift change upon binding Nt-ECL1-ECL2 and do not bind Nt-CCR5 (presumably involved in binding ECL1 and ECL2) and in yellow residues affected upon Nt-CCR5 binding but not upon Nt-ECL1-ECL2 binding (presumably involved in interaction with sulfated tyrosine). Proline residues are dotted.

Table 1.

CCR5 peptides ^a

Nt-CCR5 Residues 1–27	MDYQVSSPIY(SO3)DINY(SO3)YTSEPAQKINVKQ
ECL2 Residues 167–188	TRSQKEGLHYTSSSHFPYSQYQ
Nt-ECL1-ECL2 Residues 1–34, 89–101 (ECL1) and 168–194 (ECL2)	MDMDYQVSSPIYDINYYTSEPAQKINVKQIAARLLPGGGSGYAAAQWDFGNTMCHGGGNGRSQKEGLHYTCSSHFPYSQYQFWKNFQ
Nt-ECL1 Residues 1–27 and 88–102 (ECL1)	MDYQVSSPIY(SO3)DINY(SO3)YTSEPAQKINVKQGSGC // CGSGSHYAAAQWDFGNTMSQGSGS
Nt-ECL3 Residues 1–27 and 259–274 (ECL3)	MDYQVSSPIY(SO3)DINY(SO3)YTSEPCQKINVKQ // GSTFQEFFGLNNCSSSNRG
ECL1-ECL2 Residues 91–102 (ECL1) and 167–188 (ECL2)	AAQWDFGNTMCQGSDDGSTRSQKEGLHYTCSSHFPYSQYQ

^aAll cysteine residues (in bold) are involved in disulfide bonds. Underlined residues correspond to CCR5 domains. // symbolizes end of peptide.

Interactions of RANTES with ECL3

The contribution of ECL3 to RANTES binding was measured using the Nt-ECL3 peptide. This construct includes ECL3 linked to the doubly-sulfated N-terminal segment of CCR5 by the native disulfide bond connecting these segments in CCR5 (Table 1). This alleviates the need to use a flexible linker between the two segments, which would be ~20 Å long, and better mimics the mutual orientation of Nt-CCR5 and ECL3 [41]. A ¹H-¹⁵N-HSQC titration experiment of Nt-ECL3 into RANTES was conducted at pH 4.8 and 293 K. In comparison with the results obtained with Nt-CCR5 under the same conditions, no new RANTES residues were found to be affected by Nt-ECL3 titration (data not shown). However, many of the RANTES cross-peaks disappeared at increasing concentrations of the added peptide. This might indicate a change in binding kinetics for these residues and an exchange rate that is medium rather than fast on the scale of the chemical shift difference between bound and free forms. This prevented us from extracting a K_D value from the experiment.

Interactions of RANTES with ECL1-ECL2

ECL1 and ECL2 contribution to RANTES binding was further investigated using a chemically synthesized ECL1-ECL2 peptide, in which the sequences of CCR5’s ECL1 and ECL2 were connected by a flexible linker (Table 1). In addition, a disulfide bond corresponding to the native bond between ECL1 and ECL2 in the CCR5 molecule was formed in the ECL1-ECL2 peptide. A series of ¹H-¹⁵N-HSQC spectra measured for increasing ECL1-ECL2/RANTES concentration ratios was measured at pH 4.8 and 293 K. Despite the fact that the ECL1-ECL2 peptide lacks the N-terminal segment of CCR5, significant changes in RANTES chemical shifts (ΔCS > 0.03 ppm) were observed upon addition of large quantities of the peptide (50 µM RANTES and 2 mM ECL1-ECL2 at end of titration). The RANTES residues affected by ECL1-ECL2 binding are Y3-S4, T8, F12-A13, L19, H23, K25, Y28, T30-S31, K33, S35-N36, F41, R44-N46, Q48, K55, W57 and S66-S68 (Fig. 6A). As with Nt-ECL1-ECL2, two extensive surfaces on opposing sides of RANTES were affected by ECL1-ECL2 binding (Fig. 6B). A K_D of 2.3 ± 0.3 mM was determined for the ECL1-ECL2/RANTES complex (Fig. 6C). The surface shown in Fig. 6B (left panel) partially overlaps the segment implicated in Nt-CCR5 binding (Fig. 3D left panel) while the surface shown in Fig. 6B (right panel) matches the surface affected by Nt-ECL1-ECL2 and that was implicated in ECL2 binding (Fig. 5B right panel). The excellent agreement between the results obtained using Nt-ECL1-ECL2 and the chemically synthesized ECL1-ECL2 supports our conclusion regarding the involvement of ECL2 in RANTES binding.

Fig. 6. RANTES binding to an ECL1-ECL2 peptide.

(A) Maximal U-¹⁵N-RANTES chemical shifts changes observed upon titration with ECL1-ECL2 at 293 K and pH 4.8 as a function of RANTES residue number. (B) Mapping RANTES residues involved in ECL1-ECL2 binding. Diagram of RANTES (PDB: 1RTO) showing in violet residues affected upon ECL1-ECL2 binding. One monomer is depicted in space filling representation and the other as a semi-transparent surface, which allows a view of the ribbon diagram and stick representation of the binding residues. Proline residues, which do not appear in ¹H-¹⁵N-HSQC spectra, are dotted. (C) Determination of RANTES/ECL1-ECL2 dissociation constant at 293 K and pH 4.8. The curves were fit globally to the observed changes in chemical shifts as a function of added peptide concentration according to equation 3 (Experimental Procedures) using the OriginLab software. The orange curve shows CSD data for residue T8 obtained during titration with an ECL2 peptide. As can be appreciated from this curve, the change in chemical shift does not reach saturation.

A control experiment using an ECL2 peptide (Table 1) alone was also conducted. In comparison to the ECL1-ECL2 titration, the chemical shift changes upon ECL2 binding were much smaller, although similar RANTES residues underwent changes in chemical shifts upon addition of the two peptides. Despite the use of a larger peptide excess (up to 2.9 mM, 64 molar excess relative to RANTES), saturation was not achieved with the ECL2 peptide (Fig. 6C, orange), and therefore a K_D value could not be extracted. However, it is clear that the binding affinity of ECL1-ECL2 is stronger than that of ECL2 possibly because of conformational stabilization of ECL2 induced by ECL1.

Discussion

Elucidation of the binding surfaces in the RANTES-CCR5 complex

In this study we evaluated the use of chimeric peptides to simulate the extracellular surface of CCR5 and enable mapping of the interactions of this GPCR with its chemokine RANTES. We have shown that conjugation of the high affinity N-terminal segment of CCR5 to extracellular loop sequences of this chemokine receptor and the conjugation of two low affinity-extracellular loop peptides increased the overall affinity enough to use NMR for identifying residues affected by the binding of these CCR5 peptides. Two methods were evaluated for production of the chimeric molecules, each having advantages and limitations. The first involved bacterial expression of a soluble CCR5 surface mimetic containing the N-terminal segment connected by a flexible linker to the first loop, which is further linked by a flexible linker to the second extracellular loop. The advantage of this method was the simplicity of the protein production, little limitation on the length of the chimera and the homogeneity of the product obtained. Its disadvantage is that bacterial expression does not duplicate post-translational modifications, such as sulfation, that occur in eukaryotic proteins. The second method involved chemical synthesis of two extracellular regions connected by a flexible linker and/or a disulfide bond. This method, while allowing the simulation of eukaryotic post translational modifications is limited to some degree by the size of the chimera that can be obtained. Both methods should be widely applicable to the study of interactions of chemokines with other receptors.

The use of a panel of the conjugated peptides enabled identification of RANTES residues affected by peptide binding. Most of these residues form almost continuous surfaces on RANTES. As mentioned above, by taking into account electrostatic complementarity, and surface continuity we identify those surfaces interacting with each of the CCR5 extra-cellular regions. Our study has almost doubled the area of RANTES implicated in CCR5 binding in comparison to that determined by NMR using only a disulfated Nt-CCR5 peptide [35]. A suggested schematic model for RANTES binding to CCR5 is presented in Fig. 7. Extensive binding surfaces are located on opposing sides of RANTES. While one face is involved in binding the highly negatively charged Nt-CCR5, the opposite face is assumed to interact mostly with ECL2. ECL1 and ECL3, when conjugated to the higher affinity Nt-CCR5 and ECL2 regions, do not exhibit any significant influence on additional RANTES residues upon binding although they slightly increase the binding affinity. This is not surprising when the CXCR4 crystal structure and the model of CCR5 based on this structure are examined [12]. Both in the structure and the model, ECL1 and ECL3 are located outside the ligand binding site. These loops are connected by disulfide bonds to ECL2 and Nt-CCR5, respectively, which surround a ligand binding pocket formed by the trans-membrane helices. These same disulfide bonds were formed in our ECL1-ECL2 and Nt-ECL3 peptide surrogates. Given the likely proximity of ECL1 and ECL3 to ECL2 and Nt-CCR5, respectively, in the receptor, it is reasonable that they influence or stabilize the native conformation of these interactive domains. This might explain the lack of a direct interaction of ECL1 or ECL3 with RANTES residues accompanied by a modest increase in binding affinity in their peptide conjugates. Nt-CCR5 binding affects similar residues of monomeric and dimeric RANTES. Moreover, using a RANTES molecule in which the six N-terminal residues were deleted and which forms predominantly monomers, similar surfaces were observed to be involved in Nt-CCR5 and ECL2 binding (unpublished results). Thus, we conclude that monomeric and dimeric RANTES interact similarly with the extracellular regions of the CCR5 receptor (Nt-CCR5 and the ECLs). However, it is likely that dimerization of RANTES interferes with the interactions of the RANTES N-terminal residues with the binding pocket of CCR5, formed by the trans-membrane helices as is depicted in Fig. 7. This Fig. shows a schematic model of the complex, similar to the one suggested previously by Permentier and coworkers [42]. This model is now supported by NMR mapping of the extensive surfaces involved in Nt-CCR5 and ECL2 binding and the conclusions drawn in this study regarding the absence of any direct interactions between ECL1 and ECL3 and RANTES.

Fig. 7. Schematic representation of RANTES binding CCR5.

RANTES (orange) bound to CCR5 (red). The N-terminus segment of CCR5 is bound to one side of RANTES (depicted as left, behind), while ECL2 binds on an opposite face of RANTES (depicted as right, in front).

The surfaces of RANTES affected by Nt-CCR5 and ECL2 binding form almost a closed ring above the short N-terminal segment consisting of residues S1-T7. Nt-CCR5 interacts with a surface of ~800 Å and ECL2 interacts with a somewhat smaller face. Nt-CCR5 and the long ECL2 have been already implicated as being very important for chemokine and gp120 binding. Given the heptahelical bundle of the CCR5 (GPCR) and the lengths of both the N-terminus (27 residues) and the ECL2 (22 residues), the distances that must be spanned to interact with the opposing faces of the ellipsoidal RANTES can be readily accommodated. The CXCR4 crystal structures found with small molecule and cyclic peptide antagonists reveal a wide binding pocket formed primarily by the trans-membrane domains with significant contribution from ECL2. The N-terminal segment of the receptor was missing in the structure due to its flexibility. The antagonistic cyclic peptide, resembling V3 of X4 viruses in its high content of positively charged residues, was found to form a β-hairpin structure, similar to that found for the V3, and to fill the CXCR4 binding pocket, while creating an intermolecular β-sheet. Given these results, it is likely that the trans-membrane helices of CCR5 contribute to RANTES and gp120 binding. The pocket formed by them could accommodate the gp120 V3 region or the N-terminal segment of chemokines that have been implicated in receptor binding. The negatively charged residues E288 and E283 in the CXCR4 structure and CCR5 model, respectively, could interact with R315 in the GPGR segment at the tip of the V3 or the positively charged N-terminus of chemokines. Our NMR mapping results are in agreement with recent mutagenesis data that showed that ECL2 and the trans-membrane helices are major contributors to chemokine and gp120 binding, while the contribution of ECL1 and ECL3 is rather minor [41]. The role of ECL2 in chemokine binding was also highlighted by the crystal structure of the CXCR4 receptor in complex with an antagonistic peptide, where one of the β-strands of the peptide formed an intermolecular β-sheet with residues at the C-terminal segment of the CXCR4 ECL2 segment [12].

Comparison with gp120

The RANTES face that is most affected by Nt-CCR5 binding contains almost a continuous surface of three positively charged residues, namely K25, K45 and R47. As mentioned above, it is very likely that some or all of these residues interact with the negatively charged residues of Nt-CCR5 which has 5–7 negatively charged residues, depending whether Tyr3 and Tyr15 are sulfated in addition to Tys10 and Tys14. This surface on RANTES is reminiscent of a highly positively charged surface formed by the C4 region and the base of the V3 of HIV-1 gp120, which contains seven positively charged residues, namely R298, R304, K305, R328, R419, K421 and R440. A docking model of the Nt-CCR5 peptide to gp120 suggests that three of these positively charged residues interact with the sulfated tyrosines at positions 10 and 14 of Nt-CCR5 [44].

We previously suggested that the V3 β-hairpin structure mimics the β-hairpin structure of the 40's loop of CC-chemokines, especially that the gp120 triad ³⁰⁷IHI³⁰⁹ mimics the triad ⁴⁰VFV⁴² and ⁴⁰IFL⁴² in RANTES and MIP-1α, respectively [45]. Examination of the RANTES surface affected by ECL2 binding (K25, F28, V41, R44, N46 and Q48) and the V3 surface formed upon binding the 447-52D antibody reveals interesting similarity when the V3 β-hairpin is rotated by 180° relative to the orientation of the chemokines 40’s β-hairpin. This manipulation reveals a surface homologous in its amino-acid composition, which extends the previously identified homology between the V3 and the 40's loop. The surfaces include two positively charged residues at the top (R304 and K305 in V3, K25 and R44 in RANTES and K44 and R45 in MIP-1α and MIP-1β), a central hydrophobic surface (I307, F317 and I309 in V3, V41 and F28 in RANTES and L42 and F28 in MI-1α) and polar residues on the left side (S306 and H308 in V3, Fig. 7A, N46 and Q46 in RANTES and S46 and Q48 in MIP-1α, Fig. 8 B–D). The 40's loop of RANTES underwent significant changes in chemical shift upon binding ECL2 (Fig. 6A). The affected residues include the tip of the loop (R44 and K45), as well as N46 and Q48 in the N-terminal strand of the 40's loop β-hairpin. The similarity is especially striking when MIP-1α is compared to the V3 (Fig. 8A and 8C). When the location of the Cα and Cβ of the V3 residues R304, K305, I307, I309 and F317 are compared to F28, I40, L42, K44 and R45 of MIP-1α, an RMSD of 2.2 Å is found. R304 and K305 were suggested to be involved in Nt-CCR5 binding [44] and, in this study, K44 and R45 seem to participate in the RANTES surface forming the sulfated Nt-CCR5 binding site. Thus, it seems that V3 mimics a more complex surface of RANTES that, in addition to five 40's-loop residues, includes two residues from the β1 strand.

Fig. 8. Surface homology between HIV-1 gp120-V3 and chemokines.

Space filling diagram of (A) the V3 loop of HIV-1 gp120 (PDB: 2ESX); (B) RANTES (PDB: 1RTO); (C) MIP-1α (PDB: 1B53); and (D) MIP-1β (PDB: 1JE4) showing aromatic residues in green, hydrophilic residues in magenta, positively charged residues in blue, hydrophobic residues in cyan and proline residues in yellow.

Lusso and co-worker designed RANTES peptides that inhibit R5 HIV-1 infection at 0.1–0.7 µM concentration. The essential components for high potency were the C11-P18 segment found to be affected by Nt-CCR5 binding and the K25-Y29 segment [46] discovered in this study to be involved in ECL2 binding. F28A mutation was previously found by the same group to reduce the potency of a RANTES peptide by 80% [47]. These biological results are in agreement with our NMR results that show that K25 and F28 are involved in binding of RANTES to CCR5. Although there is no direct experimental evidence for the involvement of the crown of V3 in ECL2 binding, this has been suggested by Dragic and coworkers [48]. This suggestion gets further support from the recent crystal structure of CXCR4 in complex with a highly positively-charged antagonistic peptide resembling gp120 V3 in some sequence and conformation features. It has been suggested by Stevens and coworkers [12] that the gp120 V3 region is inserted into the CXCR4 binding pocket, similar to the antagonistic peptide. Further experimental data on pairwise interactions between RANTES and CCR5 and between gp120 and CCR5 are required to corroborate these suggestions.

Experimental Procedures

Synthesis of Sulfated Nt-CCR5

The Nt-CCR5 peptide, MDYQVSSPIY(SO₃)DINY(SO₃)Y-TSEPAQKINVKQ, sulfated at Y10 and Y14, was synthesized as described in Schnur et al. [44]. The sulfated peptide was >95% homogeneous on HPLC and had the expected molecular weight as judged by ESI-MS in the positive and negative ion modes. The yields of the peptides were about 20–30%. Both HPLC and NMR analysis showed that the sulfotyrosine moieties were stable over time under the conditions of the NMR experiments (data not shown). Analogs of the above, with spacers and cysteine residues, used in the preparation of chimeric peptides (see below) were synthesized using similar methods.

Synthesis of Chimeric Peptide Surrogates of the Extracellular Surface of CCR5

Peptides serving as surrogates of the extracellular surface of CCR5 were synthesized either directly using solid-phase peptide synthesis or using guided heterodisulfide formation. The peptides synthesized joined either Nt-CCR5 and ECL1 [¹MDMDYQVSSPIY(SO₃)DIN-Y(SO₃)YTSEPAQKINVKQ²⁷-GSGC--CGSGS-⁸⁸HYAAAQWDFGNTMSQ¹⁰²-GSGS (C101A)], Nt-CCR5 and ECL3 [DYQVSSPIY(SO₃)DINY(SO₃)YTSEPCQKINVKQ + GSTFQEFFGLNNCSSSNRG] or ECL1 and ECL2. All component peptides used in heterodisulfide formation were assembled using automated solid-phase peptide synthesis with Fmoc protection and HOBT/HBTU coupling. The yields of the tyrosine sulfate containing peptides were similar to those previously reported (10–20% overall yield of >95% homogeneous product). The yields of the ECL1 and ECL3 peptides containing cysteine for heterodisulfide formation were 20–30%. All peptide intermediates were characterized by electrospray ionization mass spectrometry (ESI-MS).

Synthesis of Cyclic ECL1-ECL2

The linear 40-residue AAQWDFGNTMCQGSDDGSTRSQKEGLHYTCSSHFPYSQYQ peptide was synthesized using standard ABS-133A chemistry with an extended coupling cycle. 20 single coupling cycles were followed by 19 double coupling during sequence assembly. Capping of unreacted amino groups with acetic anhydride was used at every step. The peptide was cleaved for 1.5 hours in 93.5% trifluoroacetic acid (TFA), 2.5% water, 2.5% ethane dithiol, 1.5% trimethyl silane at room temperature. The resin was removed, the recovered peptide solution was concentrated using a rotary evaporator, and the peptide was precipitated with cold ether. Yield by resin weight gain was 72%. 20 mg of the crude peptide was dissolved in 0.4 mL acetonitrile (0.1%TFA) and 1.6 mL water (0.1%TFA) by sonication, after filtration the filtrate was loaded onto preparative HPLC using gradient 10–40% acetonitrile (0.1%TFA) over 80 min at temperature +46 °C. 160 mg of the crude peptide was purified via 8 injections to yield 18.3 mg (11%) of 85–90% pure linear ECL1-ECL2 peptide with the correct molecular weight calc. 4519.84, found: 4520.7. Linear ECL1-ECL2 (10 mg) was dissolved in 10 mL 50% acetonitrile (0.1%TFA) and the solution was added to an ammonium acetate buffer (200 mL, pH 8, 0.1M) containing 7 mL acetonitrile and 20 mg gluthathione (oxidized form). The resulting solution was stirred at room temperature for 16 hours. After acidification with trifluoroacetic acid to pH 4 the entire solution volume was loaded (pumped) onto a preparative HPLC column (C₁₈Deltapak). After loading a gradient from 10–40% acetonitrile (0.1%TFA) over 80 min was applied to isolate the cyclic peptide. Purity of recovered cyclic ECL1-ECL2 (5.7 mg) was 94%, MS Calc: 4517.82; found: 4518.90. The 2.07 Da mass-difference compared with the linear peptide measured at the same time confirmed the cyclic product.

Synthesis of hetero-dimeric Nt-ECL1 and Nt-ECL3

The formation of the heterodisulfides of the Nt-CCR5 and either ECL1 or ECL3 were carried out by the same strategy. In the first step the peptide (ECL1 or ECL3) corresponding to the loop was activated with Ellman's reagent [5,5'-dithiobis-(2-nitrobenzoic acid); DTNB], the activated cysteine containing peptide was purified by HPLC characterized by MS and then added to the desired Cys containing Nt-CCR5 peptide and heterodisulfide formation was allowed to occur at pH 8.5 in a guanidine buffer (6 N). Using this strategy Nt-ECL1 was obtained in >98% purity in ~50% yield (8.2 mg); MS calc. 6064.56, MS found 6063.8. Similarly the Nt-ECL3 chimera was formed by activating an ECL3 peptide (GS-TFQEFFGLNNCSSSNR-G) with DTNB, isolating the activated intermediate and adding it to DYQVSSPIY(SO₃)DINY(SO₃)YT-SEPCQKINVKQ in guanidinine buffer pH 8.5. After HPLC purification 12.8 mg (~50% yield) of >95% pure heterodisulfide as the ammonium salt was obtained MS calculated 4855.24; found 4856.34

Nt-ECL1-ECL2 Expression

The 87-residue recombinant and soluble CCR5 polypeptide, Nt-ECL1-ECL2 (¹MDMDYQVSSPIYDINYYTSEPAQKINVKQIAARLLP³⁴-GGGSG-⁸⁹YAAAQWDFGNTMC¹⁰¹-HGGGNG-¹⁶⁸RSQKEGLHYTCSSHFPYSQYQFWKNFQ¹⁹⁴), containing the unsulfated N-terminal segment (residues 1–34) as well as ECL1 (residues 89–101) and ECL2 (residues 168–194) of CCR5 was expressed in Escherichia coli. The different CCR5 segments were connected by flexible linkers (GGGSG and HGGGNG). Since the N-terminal residue cysteine-20 is substituted with alanine, the Nt-ECL1-ECL2 construct contains two cysteine residues. In the context of full CCR5 C101 from ECL1 forms a disulfide bond with C173 from ECL2 and C20 from Nt-CCR5 forms a disulfide bond with a cysteine residue at ECL3. However, to enable the high concentration of Nt-ECL1-ECL2 necessary for the NMR experiments, without concern of homodimer formation via disulfide bonds, C20 at the N-terminal segment was mutated to alanine. Nt-ECL1-ECL2 was expressed in BL21(DE3) cells, which were grown in LB medium supplemented with 25 µg/ml kanamycin at 310 K. At OD₆₀₀ = 0.7 the cells were induced by 1 mM IPTG for 4 hours, followed by harvesting by centrifugation at 5000 rpm for 30 min at 277 K. The cell pellet was resuspended in lysis buffer (50 mM Tris-HCl, 50 mM NaCl, 1 mM TCEP, 0.5 mM EDTA, 5% glycerol, Lysonase Bioprocessing reagent (Novagen), pH 8.0), incubated on the rotator for 15 min at room temperature and sonicated. The inclusion bodies were repeatedly washed with wash buffer (50 mM Tris-HCl, 50 mM NaCl, 1 mM TCEP, 0.5 mM EDTA, 5% glycerol and 1% Triton X-100, pH 8.0) and centrifuged at 8000×g for 15 min at 277 K. The pellet was solubilized in urea buffer (50 mM Tris-HCl, 200 mM NaCl, 10 mM TCEP, 2 mM EDTA, 4M Urea, pH 8.0), briefly sonicated, incubated on the rotator for 2 hours at room temperature and centrifuged at 25000×g for 15 minutes. The denatured peptide was purified by ion exchange chromatography using a Source 15Q column and following gel filtration using a Superdex 200 column (GE Healthcare Life Sciences). The peptide containing fractions were pooled and analyzed by SDS-PAGE.

The purified Nt-ECL1-ECL2 was concentrated to a minimum of 5 mg/ml by Amicon Ultra concentrator (Millipore) in the presence of 4.5% N-Lauroylsarcosine and then refolded by dialysis, once against 100 volumes of 0.06% N-Lauroylsarcosine, 10 mM K-phosphate, pH 8.0 overnight at 277 K and three times against 100 volumes of 10 mM K-phosphate, pH 8.0 for a total of 12 hours at 277 K. After dialysis, the refolded protein was lyophilized and stored at −80 °C. ESI-MS and amino-acid analysis were used to validate the peptide.

RANTES Expression

Wild type human RANTES (SPYSSDTTPCCFAYIARPLPRAHIK EYFYTSGKCSNPAVVFVTRKNRQVCANPEKKWVREYINSLEMS) was expressed from a Novagen pET32LIC expression vector, which allows production of the desired protein with a thioredoxin fusion tag (kindly provided by the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH). E66 was mutated to serine to reduce oligomerization of RANTES (named hereafter: RANTES). The original enterokinase cleavage site between RANTES and thioredoxin was replaced by a TEV protease cleavage site, which ultimately resulted in a Gly residue addition at the amine terminus of the truncated RANTES. The mutated TEV-RANTES was transformed into Rosetta-gami B DE3 pLysS cells (Novagen) which were subsequently grown in M9 medium containing ¹⁵NH₄Cl and/or ¹³C-glucose (for ¹⁵N- and/or ¹³C-labeling) and 100 µg/ml ampicillin. Cells were induced at A₅₉₅ = 0.7 by adding IPTG to 0.2 mM, and the temperature was reduced to 293 K. After over-night incubation, the cells were harvested by centrifugation. The cell pellet was resuspended in lysis buffer (500 mM NaCl, 20 mM Tris pH 7.5) containing 10 mM imidazole, sonicated for 2.5 minutes, twice and then centrifuged at 12,500 rpm for 40 minutes, twice. The supernatant was loaded onto a Ni-NTA column, and eluted with lysis buffer containing 200 mM imidazole. Cleavage of the thioredoxin fusion tag was carried out with TEV at a 1:10 mg ratio over-night at 303 K. The cleaved protein was loaded on a heparin column and eluted using a NaCl gradient. The final purification step employed a Superdex 75 column. These purification steps resulted in a protein yield of ~13 mg/liter with >95% purity. Identity of RANTES was verified by MALDI-TOF.

NMR Sample Preparation

All samples of ¹⁵N-RANTES were dissolved in a 100 mM aqueous solution of D₄-acetate buffer, pH 4.8, containing 95% H₂O, 5% D₂O and 0.005% thimerosal. The samples used for sequential assignment of ¹⁵N-¹³C-RANTES contained 50 µM protein. To keep RANTES dilution to a minimum in titration experiments, a concentrated solution of the peptide was prepared in D₆-DMSO and added stepwise (single microliters) into a sample of 0.1 – 0.01 mM ¹⁵N-RANTES. The DMSO concentration at the end of titration did not exceed 10%, at which, no chemical shift perturbations are expected. RANTES concentrations were determined using UV absorbance measurements at 280 nm, using an extinction coefficient of 11710 M⁻¹cm⁻¹.

NMR Measurements

All NMR spectra were measured on a Bruker AVIII800 spectrometer equipped with a 5mm TCI cryoprobe except for the CBCACONH experiment, which was measured on a Bruker DMX500 spectrometer equipped with a TXI cryoprobe. Water suppression was achieved using the WATERGATE (WATER suppression by GrAdient-Tailored Excitation) sequence [49], 3-9-19 pulse sequence with gradients [50] or excitation sculpting with gradients [51]. The three dimensional HNCA and CBCACONH experiments used for sequential assignment were measured at 310 K. Typical carrier positions used in all experiments were 118 ppm for ¹⁵N, 46 ppm for ¹³C, 177 ppm ¹³CO and 4.75 ppm for ¹H. Data were processed and analyzed using Topspin (Bruker-De), NMRPipe [52] and NMRView [53, 54] software.

Evaluation of the Dissociation Constants

The dependence of the changes in chemical shifts on the concentration of the added peptide provides a convenient indirect measure of the complex concentration and therefore can be used to determine its dissociation constant. The weighted change in chemical shift, $\mathrm{\Delta }CS=\sqrt{{\left(\mathrm{\Delta }C{S}_H\right)}^2+{(\mathrm{\Delta }C{S}_N/5)}^2}$ relative to the chemical shift difference between the free and bound conformations, ΔCS₀serves as a measure of the complex concentration, [P]_bound, which is calculated using equation (1) $(1)\frac{\mathrm{\Delta }CS}{\mathrm{\Delta }C{S}_0}=\frac{{\left[P\right]}_{\mathit{\text{bound}}}}{{\left[P\right]}_{\mathit{\text{total}}}}$. Five residues were chosen for the analysis because their ¹H-¹⁵N-HSQC peaks were well resolved throughout the titration course, allowing accurate determination of ΔCS for each titration point. The changes in chemical shift of those residues were plotted as a function of the added peptide concentration and assuming a 1:1 simple binding model, $(2)K_D=\frac{\left[P\right]·\left[L\right]}{\left[PL\right]}$, were fit globally to the equation:

$$(3)\frac{\mathrm{\Delta }CS}{\mathrm{\Delta }C{S}_0}=\left(\right(\frac{K_D}{\left[P\right]}+1+\frac{\left[L\right]}{\left[P\right]})-\sqrt{{(\frac{K_D}{\left[P\right]}+1+\frac{\left[L\right]}{\left[P\right]})}^2-4\frac{\left[L\right]}{\left[P\right]}})/2$$ (3)

where [P] and [L] represent the concentrations of ^7–68RANTES(E66S) and the CCR5 peptides, respectively, and ΔCS₀ and K_D are varied and fitted by OriginLab.

Abbreviations

CCR5

CC chemokine receptor 5

CS

chemical shift

CSD

chemical shift difference

CXCR4

CXC chemokine receptor 4

ECL1, ECL2 and ECL3

the three extracellular loops of CCR5

K_D

equilibrium dissociation constant

Nt-CCR5

N-terminal segment of CCR5

Tys

sulfotyrosine

V3

the third variable loop of gp120, the envelope glycoprotein on HIV-1