A novel CD34 glycoform acts as a P-selectin ligand and its interaction is dependent on O-glycans and tyrosine sulfation. (A) P-Ig was used to immunoprecipitate potential ligands from KG1a and CD34^pos-BM lysates, and the resultant proteins were analyzed by western blot for CD34 (QBend-10 and EP373Y mAb) or PSGL-1 (KPL-1-mAb). Note that CD34 immunoprecipitates were free from any PSGL-1 contamination (n = 3 independent experiments). (B) CD34 and PSGL-1 were immunoprecipitated from KG1a or CD34^pos-BM lysates and then analyzed by western blot for P-Ig binding (n = 3 independent experiments). (C) E-Ig was used to immunoprecipitate E-selLs from the KG1a lysate and resultant proteins were eluted with 30 mM EDTA. The eluate was then immunoprecipitated with CD34 mAbs (clones 4H11 and 581) prior to western blot analysis for CD34, E-Ig, and P-Ig (n = 3 independent experiments). (D) CD34 immunoprecipitates were prepared from CD34^pos-BM and KG1a lysates and spotted on glass slides to test for CHO-P binding using a Stamper-Woodruff assay. Adherent CHO-P cells were counted by light microscopy. The data are a representative experiment, and the error bars indicate the SEM of 7 fields per slide on 2 slides for each experiment (n = 3 independent experiments). (E) Adhesion bar graph representing results obtained for the blot-rolling assay using CHO-P cell rolling (red bars) at 0.25 dyne/cm² (rolling cells per field) over western blots of immunoprecipitated CD34, CD44, CD43, or PSGL-1 from KG1a cell lysates as in Figure 3D. As controls, CHO-E cells were incubated with EDTA (blue bars) or mock-transfected CHO cells (CHO-M; light green bars) were used. The adhesion bar graph is the average of 4 fields of view per experiment from n = 5 independent experiments, and data are reported as the mean ± SEM (error bars). (F) CD34 immunoprecipitates from KG1a lysates were treated with neuraminidase, OSGE, or PNGase F or no treatment followed by western blot for P-Ig binding. Note that CD34 (QBend-10) was used as an internal control (data not shown) as in Figure 6 (n = 3 independent experiments). (G) To inhibit sulfation, KG1a cells were treated with 150 mM sodium chlorate for 72 hours (+; left panel) while KG1a whole-cell lysates were treated with 5 U/mL arylsulfatase for 3 hours (+; right panel) as described in supplemental Materials and methods. Negative controls consisted of the buffers used for the treatment without the sodium chlorate (−; left panel) while KG1a whole-cell lysates were treated arylsulfatase (−; right panel). Following treatment, CD34 was immunoprecipitated from cell lysates, and the resulting proteins were analyzed by western blot for P-Ig binding (top). Note that both treatments abrogated PSGL-1 binding to the KPL-1-mAb from KG1a lysates (middle; KPL-1 is sensitive to the loss of sulfation on PSGL-1), whereas the treatments did not significantly affect CD34 protein levels as indicated by EP373Y-mAb staining (bottom) or PSGL-1 staining (data not shown). Blots are representative of n = 3 independent experiments. (H) Divalent metal ion dependency and binding of P-Ig to PSGL-1 and CD34 captured from KG1a cell lysate at 150 mM NaCl were measured by injecting P-Ig (67 nM) in the presence of 10 mM EDTA followed by consecutive injection of different concentrations of P-Ig (as indicated) in the presence of 2 mM CaCl₂. The experimental conditions are similar to those in Figure 4 with P-Ig injected for 170 seconds interrupted by a 60-second washing buffer step. KPL-1-mAb (11 000 RU) or MsIgG₁ isotype control (5000 RU) were immobilized to capture PSGL-1 (left panel). 563-mAb (6527 RU) or MsIgG₁ isotype control (5000 RU) were immobilized to capture CD34 (right panel) (n = 3). k_off-apparent and K_D were determined as described in Figures 3B and 4, respectively. The sensorgrams presented are corrected for the bulk refractive index and nonspecific interactions using the isotype controls.