The monomer-dimer equilibrium of stromal cell-derived factor-1 (CXCL 12) is altered by pH, phosphate, sulfate, and heparin

Abstract

Chemokines, like stromal cell-derived factor-1 (SDF1/CXCL12), are small secreted proteins that signal cells to migrate. Because SDF1 and its receptor CXCR4 play important roles in embryonic development, cancer metastasis, and HIV/AIDS, this chemokine signaling system is the subject of intense study. However, it is not known whether the monomeric or dimeric structure of SDF1 is responsible for signaling in vivo. Previous structural studies portrayed the SDF1 structure as either strictly monomeric in solution or dimeric when crystallized. Here, we report two-dimensional NMR, pulsed-field gradient diffusion and fluorescence polarization measurements at various SDF1 concentrations, solution conditions, and pH. These results demonstrate that SDF1 can form a dimeric structure in solution, but only at nonacidic pH when stabilizing counterions are present. Thus, while the previous NMR structural studies were performed under acidic conditions that strongly promote the monomeric state, crystallographic studies used nonacidic buffer conditions that included divalent anions shown here to promote dimerization. This pH-sensitive aggregation behavior is explained by a dense cluster of positively charged residues at the SDF1 dimer interface that includes a histidine side chain at its center. A heparin disaccharide shifts the SDF1 monomer-dimer equilibrium in the same manner as other stabilizing anions, suggesting that glycosaminoglycan binding may be coupled to SDF1 dimerization in vivo.

Figure 1.

SDF1α exists in a monomer–dimer equilibrium. (A) 2D ¹⁵N-¹H HSQC spectrum of SDF1α in 20 mM sodium phosphate at pH 6.0, with residue assignments indicated. (B) A subset of signals in the 2D ¹⁵N-¹H HSQC spectrum shift as the protein concentration is varied from 1.2 mM (green) to 10 μM (red). (C) Combined ¹H/¹⁵N chemical shift perturbations for each SDF1α residue upon dilution from 1.2 mM to 10 μM. No values are plotted for residues 2, 10, 32, and 53 (prolines) or Leu 26 (not observed). (D) Residues with the largest ¹H/¹⁵N chemical shift perturbations (>0.4) are highlighted in red on the surface of one monomer from the dimeric crystal structure (PDB code 1QG7). (E) Residues participating in the SDF1αintermonomer interface are highlighted on the structure as in D.

Figure 2.

SDF1α dimer formation requires phosphate or sulfate. (A) Differences in translational self-diffusion coefficients measured for SDF1α (250 μM) at pH 7.4 suggest that SDF1α oligomerization is favored in the presence of 20 mM phosphate relative to 20 mM MES. Peak intensities from a series of 1D ¹H spectra acquired in phosphate (♦, upper curve) or MES (○, lower curve) buffer using a longitudinal encode-decode pulsed field gradient diffusion pulse scheme (diffusion delay Δ = 80 msec; gradient pulse Δ = 5 msec) were analyzed by nonlinear fitting to Equation 1 (solid lines) to obtain values for the self-diffusion coefficient, D_s (Altieri et al. 1995). Every fifth data point has been enlarged for clarity. (B) Variations in D_s as a function of protein concentration were fit to a function describing a monomer–dimer equilibrium, revealing a ~10-fold change in the K_d for dimer dissociation depending on the presence (•) (K_d = 120 ± 80 μM) or absence (○) (K_d = 1000 ± 1000 μM) of phosphate. (C) Fluorescence polarization (FP) measurements over a range of SDF1α concentrations at pH 7.4 in the presence (▪) or absence (□) of 100 mM sodium phosphate were analyzed by nonlinear fitting to Equation 5. An equilibrium dissociation constant for the SDF1α dimer of 140 ± 19 μM was obtained in 100 mM sodium phosphate. In contrast, FP values for SDF1α measured in 100 mM HEPES at pH 7.4 vary only slightly, consistent with a dimer K_d > 10,000 μM.

Figure 3.

Titration of His 25 gives rise to the pH sensitivity of the SDF1α monomer–dimer equilibrium. Dimer dissociation K_d values were obtained by measuring fluorescence polarization (FP) as a function of wild type, H25R, and H25L SDF1α protein concentration and nonlinear fitting to Equation 5. (A) The K_d for SDF1α dimer dissociation rises from 261 ± 65 μM (♦) (100 mM HEPES [pH 7.4], 100 mM sodium sulfate) to 1.6 ± 0.4 mM (⋄) (100 mM MES [pH 5.5], 100 mM sodium sulfate), demonstrating that the monomer–dimer equilibrium is pH sensitive. Sulfate was used instead of phosphate because its ionization state is unchanged from pH 5.5–7.4. (B) Substitution of His 25 with Arg, which remains positively charged at pH 7.4, destabilizes the SDF1α dimer. In 100 mM sodium phosphate (pH 7.4), the K_d for H25R (□) is 1.5 ± 0.2 mM, compared to 140 ± 19 μM for wild-type SDF1α (▪). (C) Substitution of His 25 with Leu, an uncharged side chain, disrupts SDF1α dimerization only slightly, and eliminates the pH sensitivity. In 100 mM sodium sulfate the dimer K_d for H25L is 617 ± 170 μM at pH 7.4 (•) and 740 ± 160 μM at pH 5.5 (○). The K_d value for H25R in 100 mM sodium sulfate is 1.7 ± 0.5 mM at pH 7.4 (▾) and 2.0 ± 0.4 mM at pH 5.5 (▿).

Figure 4.

Electrostatic disruption and stabilization of the SDF1α dimer. Crystal structure of SDF1α with basic residues at the dimer interface highlighted (PDB 1QG7). The side chains of K24 (green), H25 (red), and K27 (blue) are shown. The dashed line highlights the dimer interface, and the close proximity (3.8 Å) of positively charged H25 and K27 side chains from different monomers.

Figure 5.

(Top) I-S heparin disaccharide stabilizes the SDF1α dimer. (Bottom) SDF1 self-association monitored by FP in the presence of 5 mM I-S heparin disaccharide (♦) revealed a dimer K_d of 172 ± 29 μM. In the absence of heparin (⋄), SDF1α is essentially monomeric in these buffer conditions (HEPES pH 7.4).

Figure 6.

Heparin binds preferentially to the SDF1α dimer. (A) ¹⁵N-¹H HSQC spectrum of 250 μM SDF1α in 25 mM MES (pH 6.8). (B) ¹⁵N-¹H HSQC spectrum of 250 μM SDF1α in 25 mM MES (pH 6.8) with 1 mM I-S heparin disaccharide. (C) ¹⁵N-¹H HSQC spectrum of 10 μM SDF1α in 25 mM MES (pH 6.8) with 1 mM I-S heparin disaccharide.