Effect of extracellular pH on selectin adhesion: theory and experiment

Abstract

Selectins mediate circulatory leukocyte trafficking to sites of inflammation and trauma, and the extracellular microenvironments at these sites often become acidic. In this study, we investigated the influence of slightly acidic pH on the binding dynamics of selectins (P-, L-, and E-selectin) to P-selectin glycoprotein ligand-1 (PSGL-1) via computational modeling (molecular dynamics) and experimental rolling assays under shear in vitro. The P-selectin/PSGL-1 binding is strengthened at acidic pH, as evidenced by the formation of a new hydrogen bond (seen computationally) and the observed decrease in the rolling velocities of model cells. In the case of L-selectin/PSGL-1 binding dynamics, the binding strength and frequency increase at acidic pH, as indicated by the greater cell-rolling flux of neutrophils and slower rolling velocities of L-selectin-coated microspheres, respectively. The cell flux is most likely due to an increased population of L-selectin in the high-affinity conformation as pH decreases, whereas the velocities are due to increased L-selectin/PSGL-1 contacts. In contrast to P- and L-selectin, the E-selectin/PSGL-1 binding does not exhibit significant changes at acidic pH levels, as shown both experimentally and computationally.

Figure 1

(A) Equilibrated PSGL-1 (magenta) and sLe^X (orange) structures bound to P-selectin (gray) at physiologically neutral pH. (B) Equilibrated P-selectin/PSGL-1 structure at physiologically acidic pH, where the H114/Y7 contact forms a hydrogen bond. All hydrogens except for those on the imidazoles of histidines are omitted for clarity, and the yellow sphere is the calcium ion. All figures available online in color.

Figure 2

(A) Representative SMD force versus time graphs depicting the dissociation of PSGL-1 from P-selectin in physiologically neutral (red) and acidic (blue) conditions. Time point 1 represents the rupture of all amino acid contacts, except for coordination bonding, in the acidic case. Time point 2 represents a long force plateau in the neutral case, where several sliding and rebinding events occur. Time point 3 represents the complete rupture of all bonds, leading to the dissociation of PSGL-1 from P-selectin in both conditions. (B) Representative P-selectin/PSGL-1 structure in neutral conditions for time periods < ∼6 ns, where R85 forms a hydrogen bond with sLe^X, and Y5 forms a hydrogen bond with the backbone structure of P-selectin. (C) Representative structure in neutral conditions for a time period > ∼10 ns, where Y5 has no contacts and Y7 forms a contact with the backbone structure of P-selectin. (D) Representative structure in acidic conditions for time periods < ∼6 ns, where Y5 has two hydrogen bonds with K8 and the backbone structure of P-selectin, Y7 forms a hydrogen bond to H114, and Y10 is hydrogen bound to R85. (E) Representative structure in acidic conditions for time periods > ∼10 ns, where Y7 is still hydrogen bound to H114 and forms a hydrogen bond with the P-selectin backbone structure, and R85 is now hydrogen bound to sLe^X.

Figure 3

(A) 10⁶ cells/ml of KG1a cells were perfused through MRE tubing functionalized with 2 μg/ml of recombinant human P-selectin/Fc at a wall shear stress of 7 dyn/cm². (B and C) Protein-A-coated microspheres (10⁶ beads/ml) functionalized with 14.47 μg/mg particles of recombinant human PSGL-1/Fc (B), or with 0.017 μg/mg particles of sialyl Lewis-x-PAA-biotin (C) were perfused through MRE tubing functionalized with 10 μg/ml recombinant human P-selectin/Fc at a wall shear stress of 1 dyn/cm². Experiments were recorded for 1 min each at three randomly selected locations along the tube (unpaired t-test was performed; errors are mean ± SE; ^∗∗∗ p < 0.0001; n = 3).

Figure 4

(A) Equilibrated L-selectin (blue) at physiologically neutral pH, where N138 is hydrogen bound to the backbone structure of L-selectin. (B) Equilibrated L-selectin structure at physiologically acidic pH, where N138 is no longer close enough to form a hydrogen bond. All hydrogens except for those on the imidazoles of histidines are omitted for clarity, and the yellow sphere is the calcium ion.

Figure 5

(A) Representative SMD force versus time graphs depicting the dissociation of PSGL-1 from L-selectin in physiologically neutral (red) and acidic (blue) conditions. Time interval 1 represents a long force plateau in the neutral case, where several sliding and rebinding events occur. Time point 2 represents the complete rupture of all bonds, leading to the dissociation of PSGL-1 from L-selectin in both conditions. (B and C) Equilibrated PSGL-1 (magenta) and sLe^X (orange) structures bound to L-selectin (blue) at physiologically neutral pH. (D and E) Equilibrated PSGL-1/L-selectin structure at physiologically acidic pH, where Y7 forms contacts with the L-selectin backbone and L13 contacts H110 via a stacking interaction.

Figure 6

(A and B) Human neutrophils (10⁶ cells/ml) (A) or protein-A-coated microspheres functionalized with 14.47 μg/mg particles of recombinant human L-selectin/Fc (B) were perfused through MRE tubing functionalized with 10 μg/ml of recombinant human PSGL-1/Fc at a wall shear stress of 2 dyn/cm². (C) Human neutrophils (10⁶ cells/ml) were perfused in buffer of indicated pH through MRE tubing functionalized with 10 μg/ml recombinant human PSGL-1/Fc, and the number of rolling cells through the cross-sectional area of the tube was recorded for varying shear stresses (unpaired t-test; errors are mean ± SE; ^∗∗ p < 0.0012, ^∗∗∗ p < 0.0001; n = 3).

Figure 7

(A) Equilibrated E-selectin (green) at physiologically neutral pH. (B) Equilibrated E-selectin structure at physiologically acidic pH. All hydrogens except for those on the imidazoles of histidines are omitted for clarity, and the yellow sphere is the calcium ion.

Figure 8

(A and B) KG1a cells (10⁶ cells/ml) (A) or protein-A-coated microspheres functionalized with 14.47 μg/mg particles of recombinant human E-Sel/Fc (B) were perfused through MRE tubing functionalized with 2 μg/ml of recombinant human E-selectin/Fc at a wall shear stress of 7 dyn/cm² (unpaired t-test; errors are mean ± SE; ^∗∗∗ p < 0.0001; n = 3).