An Analysis of Trafficking Receptors Shows that CD44 and P-Selectin Glycoprotein Ligand-1 Collectively Control the Migration of Activated Human T-Cells

Abstract

Selectins guide the traffic of activated T-cells through the blood stream by mediating their tethering and rolling onto inflamed endothelium, in this way acting as beacons to help navigate them to sites of inflammation. Here, we present a comprehensive analysis of E-selectin ligands expressed on activated human T-cells. We identified several novel glycoproteins that function as E-selectin ligands. Specifically, we compared the role of P-selectin glycoprotein ligand-1 (PSGL-1) and CD43, known E-selectin ligands, to CD44, a ligand that has not previously been characterized as an E-selectin ligand on activated human T-cells. We showed that CD44 acts as a functional E-selectin ligand when expressed on both CD4⁺ and CD8⁺ T-cells. Moreover, the CD44 protein carries a binding epitope identifying it as hematopoietic cell E- and/or L-selectin ligand (HCELL). Furthermore, by knocking down these ligands individually or together in primary activated human T-cells, we demonstrated that CD44/HCELL, and not CD43, cooperates with PSGL-1 as a major E-selectin ligand. Additionally, we demonstrated the relevance of our findings to chronic autoimmune disease, by showing that CD44/HCELL and PSGL-1, but not CD43, from T-cells isolated from psoriasis patients, bind E-selectin.

Keywords: CD44; E-selectin; P-selectin glycoprotein ligand-1 (CD162); cell adhesion; cell migration; hematopoietic cell E- and/or L-selectin ligand; human activated T-cells; psoriasis.

Figure 1

CD44/HCELL isolated from activated human T-cells serves as an E-selectin ligand. (A) sLe^x expression was assayed by the reactivity of activated (black) and naïve (dark-gray) T-cells to HECA-452 (left panel) or CD15s (middle panel) monoclonal antibodies. The isotype controls were included as negative controls (light-gray). E-Ig binding to activated (black) versus naïve (dark-gray) human T-cells (right panel) is shown in the right panel. To assay for non-specific interactions, 10-mM EDTA (light-gray) was added to the assay buffer. These are representative flow cytometric histograms of n = 3 independent experiments. (B) CD44 was immuno-purified from naïve and activated T-cells (right panel) and from activated human CD4⁺ and CD8⁺ T-cells (left panel) purified using autoMACS (Figure S1 in Supplementary Material). The samples were prepared for Western blot analysis and stained with either CD44 or E-Ig (or secondary alone controls; Figure S4 in Supplementary Material). E-Ig specifically binds to CD44 immuno-purified from both activated CD4⁺ and CD8⁺ T-cells but not naïve T-cells. These are representative blots of n = 3 independent experiments. (C) CD44 was immuno-purified from activated T-cells that were either untreated or treated with PNGaseF to remove N-glycans (upper panels), sialidase to remove sialic acid (middle panels), or OSGE to remove O-glycans (lower panels) and prepared for Western blot analysis for CD44 expression or for binding to E-Ig. Note that the data suggest that the interaction between E-selectin and CD44 is N-glycan and sialic acid dependent but that it is not affected by the removal of O-glycans. The efficiency of OSGE enzymatic treatment was confirmed by its ability to abolish the E-selectin binding capacity of P-selectin glycoprotein ligand-1 (PSGL-1) after the removal of O-glycan. These are representative blots of n = 3 independent experiments.

Figure 2

CD44s, not CD44 variant (CD44v), is an E-selectin ligand on activated human T-cells. (A) To analyze the expression of CD44v on the surface of naive (left panel; CD25⁻CCR7⁺CD45RA⁺) and activated (right panel) T-cells at different time points after activation (24 h, 48 h, and Day 5), the cells were incubated with monoclonal antibodies against specific CD44v isoforms (black line) or isotype controls (gray line) and analyzed by flow cytometry. The activity of the CD44v antibodies were confirmed using breast cancer cell lines (Figure S5 in Supplementary Material). (B) CD44v6 (left panel) and CD44s (right panel) were consecutively immuo-purified from cell lysates prepared at different time points (24 h, 48 h, and Day 5), subjected to Western blot analysis and stained with CD44v6 and CD44v10 (left panel) and CD44s (right panel). The protein immuno-purified by the anti-CD44v6 also stained positive for CD44v10 (left panel), which indicates that activated T-cells express a form of CD44 that comprises both exon 6 and exon 10. (C) Western blot analysis of immuno-purified CD44s and CD44v6 illustrates that E-selectin interacts with CD44s but not CD44v6/10. This interaction was time dependent initially detected at 48 h and increasing significantly at Day 5. Immuno-purification for CD44 (s and v) was performed from cell lysates that were normalized for cell number at each time point (24 h, 48 h, and Day 5). Data are representative of n = 3 independent experiments.

Figure 3

CD44/HCELL is a functional E-selectin ligand on activated human T-cells. (A) Blot rolling assays were performed on immuno-purified CD44/HCELL and P-selectin glycoprotein ligand-1 (PSGL-1). Initially, immuno-purified protein was resolved by SDS-PAGE. Then, CHO-E cells were allowed to roll over immuno-purified proteins in the presence of 2-mM Ca²⁺ at 0.5 dyn cm⁻². After cell perfusion, the numbers of rolling cells per square millimeter in five distinct fields of view were counted and the average is presented (black bars). The specificity of CHO-E binding to membrane glycoproteins was assessed by adding 5-mM EDTA to the buffer containing CHO-E cells (gray bars). The x-axis reflects the mean of rolling cells per field per minute from one representative experiment of multiple independent membrane preparations (n = 3). The mean velocity ± SEM of rolling cells was determined and presented here on top of the bars. (B) BIAcore assay was performed to determine the ability of native CD44/HCELL and PSGL-1 to bind E-Ig. To determine the ability of captured protein to bind E-Ig, we injected 100-nM of E-Ig at 20 μl min⁻¹ over the native CD44/HCELL or PSGL-1 (upper panels) in the presence of 5-mM EDTA to control for specificity or 1-mM Ca²⁺ to study the binding. To determine the kinetics of the CD44/HCELL interaction with E-Ig and compare it to that with PSGL-1 (lower panels), monoclonal antibodies or their isotype controls were immobilized on the chip, and then the ligands were captured. Afterward, a series of 10 E-Ig concentrations (0.456–466.7 nM) were injected at 30 μl min⁻¹ over CD44/HCELL or PSGL-1 in 150-mM NaCl running buffer. The equilibrium dissociation constant (K_D) was derived from fitting the binding isotherm using the steady-state model and the maximum response unit (RUmax) of E-Ig binding just before the washing step. After the last E-Ig injection, we determined the apparent dissociation rate constant (K_off-apparent) by fitting the stable phase obtained during the buffer wash. Data shown here are the mean ± SEM of n = 3, after the correction against the bulk refractive index and subtraction of the non-specific binding of the isotype control.

Figure 4

CD44/HCELL collaborates with P-selectin glycoprotein ligand-1 (PSGL-1) to confer the optimal rolling of activated human T-cells over E-selectin expressing monolayers. Small interfering RNA (siRNA) knockdown technologies were used to effectively decrease the expression of various E-selectin ligands on activated human T-cells. To evaluate the effect of bromelain treatment on the transfection efficiency, the cells were treated with bromelain and then transfected with scrambled control siRNA, CD44 siRNA, or PSGL-1 siRNA. After 24 h, live cells were collected to evaluate transfection using Western blot analysis. (A) The knockdown of CD44 (left panel), but not PSGL-1 (right panel), was enhanced by bromelain treatment prior to the siRNA transfection. (B) The specificity of gene silencing was demonstrated by Western blot analysis. The removal of one ligand did not affect the expression of other ligands. (C) Transfected cells were perfused over CHO-E cell monolayers for 30 s at 1 dyn cm⁻² wall shear stress, and then detachment assays were performed by a stepwise increment of the shear stress every 30 s. The average number of rolling cells at the end each shear stress is depicted. A significant difference was observed between scrambled control cells and all other knockdown transfections except CD43 knockdown. No major differences were observed between cells treated with CD44, PSGL-1, double (CD44 and PSGL-1), or triple (CD44, PSGL-1, and CD43) siRNAs. Data are mean ± SEM; n = 3. p-Values comparing the different groups versus scrambled control siRNA are indicated: *p < 0.05, **p < 0.01, and ***p < 0.001.

Figure 5

CD44/HCELL on T-cells from psoriatic patients binds E-selectin. (A) Expression of sialofycosylated glycans on T-cells isolated from psoriatic patients. After isolation of CD3⁺ cells from psoriatic patients (representing mixed subsets as illustrated in Table S2 in Supplementary Material), we performed flow cytometry analysis to confirm the expression of sialofycosylated glycans using HECA-452 antibody. (B) CD44, P-selectin glycoprotein ligand-1 (PSGL-1), and CD43 were immuno-purified from T-cells obtained from three psoriatic patients and then samples were prepared for Western blot analysis and blotted for each ligand (upper panels) and for E-Ig (lower panels). Note that CD44/HCELL and PSGL-1 exhibit strong binding to E-Ig, while CD43 binding was very weak.