Structural basis of CXCR4 sulfotyrosine recognition by the chemokine SDF-1/CXCL12

Abstract

Stem cell homing and breast cancer metastasis are orchestrated by the chemokine stromal cell-derived factor 1 (SDF-1) and its receptor CXCR4. Here, we report the nuclear magnetic resonance structure of a constitutively dimeric SDF-1 in complex with a CXCR4 fragment that contains three sulfotyrosine residues important for a high-affinity ligand-receptor interaction. CXCR4 bridged the SDF-1 dimer interface so that sulfotyrosines sTyr7 and sTyr12 of CXCR4 occupied positively charged clefts on opposing chemokine subunits. Dimeric SDF-1 induced intracellular Ca2+ mobilization but had no chemotactic activity; instead, it prevented native SDF-1-induced chemotaxis, suggesting that it acted as a potent partial agonist. Our work elucidates the structural basis for sulfotyrosine recognition in the chemokine-receptor interaction and suggests a strategy for CXCR4-targeted drug development.

Fig. 1

The NMR structure of disulfide-locked SDF1₂. (A) The amino acid sequence of SDF1₂ with the conserved intramolecular disulfide bonds (black lines) and the engineered intermolecular disulfide bonds (red lines) illustrated. (B) SDS-PAGE of SDF-1 and SDF1₂ treated with or without dithiothreitol (DTT). SDF-1 and SDF1₂ migrate near the monomeric molecular weight of 8 kD when treated with DTT. In contrast, whereas SDF1₂ migrates as a dimer, SDF-1 migrates as a monomer in the absence of DTT. (C) Translational diffusion measurements of SDF1₂ indicate that SDF1₂ is dimeric. Diffusion coefficients (Ds) of wild-type SDF-1 (black circles) in 20 mM sodium phosphate at pH 7.4 plotted against chemokine concentration (22). Nonlinear fitting of the Ds values of SDF-1 indicates a dimer dissociation Kd of 120 µM with a pure monomer Ds value of ~ 1.6 (×10⁻⁶ cm²s⁻¹) and a dimer value of ~1.0 (×10⁻⁶ cm²s⁻¹) [data from Veldkamp et al. (22)]. Ds values for 10, 50, and 150 µM SDF1₂ (red triangles) range from 1.08–1.09 (×10⁻⁶ cm²s⁻¹) consistent with those expected for SDF-1 in the dimeric state (data from this study). (D) Ensemble of 20 NMR solution structures of SDF1₂ (gray and tan) superimposed on the crystal structure of dimeric wild-type SDF-1 (blue, PDB ID 2J7Z) with an α-carbon RMSD of 1.2 Å for residues 9–66. Intermolecular Cys³⁶-Cys⁶⁵ disulfide bonds are shown in yellow. Flexible N-terminal residues of SDF-1 (–8) are omitted for clarity. Refinement statistics for the SDF1₂ structure ensemble are given in table S1.

Fig. 2

The N-terminus of CXCR4 binds to SDF1₂. (A) ¹⁵N-¹H HSQC spectra of 25 µM [U-¹⁵N]-SDF1₂ alone (black contours) and after the addition of 100 µM p38 peptide (green contours). (B) Combined ¹⁵N-¹H chemical shift perturbations plotted against SDF1₂ residue number. Secondary structure elements are indicated and regions involved in the dimer interface are highlighted in orange. Missing values correspond to proline residues (sequence positions , , , and 53) or amino acid residues not observed in the ¹⁵N-¹H HSQC spectra. (C) Chemical shift mapping on the SDF1₂ structure. Green surface highlighting corresponds to shift perturbations > 0.25 in (B).

Fig. 3

Structures of SDF1₂ dimers bound to the N-terminal domain of CXCR4. (A) N-terminal peptides corresponding to the first 38 amino acids of CXCR4 are illustrated. The sequence for p38 is identical to that of CXCR4 except for a Gly-Ser dipeptide on the N-terminus, which results from a cloning artifact, and a Cys²⁸ → Ala²⁸ mutation to prevent oxidative peptide dimer formation. The sulfated peptides are identical to p38 except for the inclusion of sulfotyrosine at position 21 for p38-sY₁ and at 7, 12, and 21 for p38-sY₃. (B) Representative intermolecular NOEs for the SDF1₂:p38-sY₁ complex. Strips from 3D F1-¹³C-fliltered/F3-¹³C-edited NOESY-HSQC spectra acquired from a complex containing [U-¹⁵N,¹³C]-SDF1₂ and unlabeled p38-sY₁ (left) and a complex containing [U-¹⁵N,¹³C]-p38-sY₁ and unlabeled SDF1₂ (right) contain equivalent NOEs between the methyl group of Val¹⁸ of SDF1₂ and sTyr^{21 1}H^δ of p38-sY₁. Ensembles of the 20 lowest energy conformers for the SDF1₂:p38 (C), SDF1₂:p38-sY₁ (D), and SDF1₂:p38-sY₃ (E) complexes. SDF1₂ is shown in gray and the CXCR4 N-termini are orange. Sulfotyrosine residues in N-termini of CXCR4 are shown in red.

Fig. 4

Recognition of sulfotyrosines by SDF1₂. (A) NMR structure of SDF1₂ bound to p38-sY₃. Individual subunits of the symmetric SDF1₂ dimer are shown in tan and white with symmetry-related p38-sY₃ peptides in blue and orange. Chemical shift perturbations greater than 0.25 ppm (Fig. 2C) are highlighted in green on the surface of SDF1₂. Flexible regions of SDF1₂ (residues –8) and p38-sY₃ (residues –38) are omitted for clarity. Sulfotyrosine side chains are shown in a ball-and-stick representation. In panels B–D, basic residues in SDF1₂ that pair with CXCR4 sulfotyrosines are shown in blue and SDF1₂ residues with NOEs to the sulfotyrosines are shown in green. (B) The sTyr⁷ residue of CXCR4 binds to SDF1₂ near Arg²⁰ and makes NOE contacts with Val²³. (C) The sTyr¹² residue of CXCR4 occupies a cleft bounded by residues Lys²⁷, Pro¹⁰, and Leu²⁹ of SDF1₂. (D) The sTyr²¹ residue of CXCR4 pairs with Arg⁴⁷ of SDF1₂ and makes NOE contacts with Val¹⁸ and Val⁴⁹.

Fig. 5

Amino acid substitutions in native SDF-1 corroborate the CXCR4 N-terminal binding site. One subunit of the SDF1₂ dimer and one p38-sY3 molecule from the SDF1₂:p38-sY₃ complex solved by NMR represent a model for the equivalent 1:1 complex. Front (A) and back (B) views of the SDF-1 surface are highlighted to indicate the location and functional impact of amino acid substitutions in the wild-type SDF-1 sequence. Substitutions at the sTyr¹²- and sTyr²¹-binding sites (red) showed increased EC₅₀ values for Ca²⁺ mobilization, whereas substitutions away from the CXCR4-binding site (cyan) showed no change in their EC₅₀ values. A binding site for sTyr⁷ is not defined in this model because sTyr⁷ binds to the opposing SDF-1 subunit in the SDF1₂:p38-sY3 structure.

Fig. 6

Dimeric SDF1₂ induces CXCR4-mediated Ca²⁺ mobilization but inhibits chemotaxis to wild-type SDF-1. (A) Ca²⁺ mobilization in THP-1 cells loaded with Fluo-3 indicates robust dose-dependant activation of CXCR4 by wild-type SDF-1 (●, EC₅₀ = 3.6 nM) and SDF1₂ (▲, EC₅₀ = 12.9 nM) (Data from Veldkamp et al. (53) and this study, respectively). (B) Wild-type SDF-1 induces the chemotaxis of THP-1 cells in a biphasic, concentration-dependent manner with a maximal migratory response at ~30 nM SDF-1. In contrast, SDF1₂ does not induce chemotaxis of THP-1 cells at any concentration from 1–1,000 nM. (C) Chemotaxis of THP-1 cells induced by 10 nM wild-type SDF-1 is inhibited by SDF1₂ (IC₅₀ ~ 4 nM). (D) Wild-type SDF-1 and the dimerization-impaired His²⁵ → Arg²⁵ variant [SDF1(H25R)] induce chemotaxis of THP-1 cells equally well at low concentrations (0.1–10 nM). SDF1(H25R) remains monomeric at higher concentrations than does wild-type SDF-1 and induces chemotaxis over a broader range of concentrations. (E) Monomeric SDF-1 generates the full range of cellular responses to CXCR4 activation, whereas dimeric SDF1 is a partial agonist of CXCR4 that fails to induce chemotaxis. Loss of migration could be a consequence of aberrant CXCR4 trafficking.