HECA-452–reactive CD44 functions as an E-selectin ligand. (A) KG1a membrane protein (10 μg) was resolved on a reducing 6% SDS-PAGE gel, blotted onto PVDF membrane, stained with HECA-452, and HECA-452 immunoblots blots were rendered transparent with 10% glycerol. CHO-E cells (2 × 10⁶/ml) were then perfused over these blots at a defined shear stress of 3.8 dynes/cm². Several HECA-452–stained bands from KG1a membrane protein–supported, E-selectin–dependent CHO-E cell rolling. (B) KG1a membrane proteins (10 μg) were treated with N-glycosidase F, separated on a reducing 6% SDS-PAGE gel, and immunostained with HECA-452. (C) Immunoprecipitated PSGL-1 was resolved on a reducing 6% SDS-PAGE gel and Western blotted with either HECA-452 (left) or anti–PSGL-1 antibody 4H10 (right). (lane 1) 10 μg of total KG1a membrane protein; (lane 2) immunoprecipitated PSGL-1 from 100 μg of KG1a membrane protein; (lane 3) 100 μg of total KG1a membrane protein, and (lane 4) immunoprecipitated PSGL-1, from 100 μg of KG1a membrane protein. Note that HECA-452–stained bands at 140 and 220 kD correspond to PSGL-1. (D) Isotype control or Hermes-1 immunoprecipitated CD44 from KG1a membrane protein (50 μg) was resolved on a reducing 9% SDS-PAGE gel and immunoblotted with HECA-452. Immunoprecipitated CD44 from KG1a membrane proteins (50 μg) treated with N-glycosidase F was also immunoblotted with HECA-452. Though CHO-E cell rolling frequencies are presented as the mean ± SD of E-selectin–mediated cell rolling at 3.8 dynes/cm² measured on the 100-kD isoform of CD44, no CHO-E rolling was observed along the entire length of the N-glycosidase F–treated lane.

Expression of HECA-452–reactive glycoproteins on human hematopoietic cells and E-selectin ligand activity of human hematopoietic cell lines. (A) Membrane preparations of human hematopoietic cell lines, KG1a (10 μg), HL60 (100 μg), RPMI-8402 (100 μg), and K562 (100 μg) were resolved on a reducing 6% SDS-PAGE gel and immunoblotted with HECA-452. (B) Parallel plate flow chamber analysis of CHO-E cell tethering and rolling on glutaraldehyde-fixed monolayers of hematopoietic cell lines (shear stress of 2.8 dynes/cm²). Twofold more CHO-E cell tethering and rolling was observed on KG1a than on HL60 cell monolayers, which was not affected by OSGE pretreatment. There was no CHO-E cell rolling on RPMI-8402 or K562 cell monolayers. Negative controls consisted of CHO-mock transfectants and CHO-E cells treated with anti–E-selectin Abs (10 μg/ml). Data are presented as mean ± SD CHO-E cell rolling per field × 5 fields, minimum of three experiments.

Human hematopoietic cell rolling on freshly isolated human BMEC. KG1a, HL60, RPMI-8402, and K562 were perfused over live IL-1α–treated primary BMEC cultures at 2.8 dynes/cm² in the parallel plate flow chamber, and cell rolling was observed and recorded for video analysis. Controls consisted of untreated BMEC and IL-1α–treated BMEC in the presence of anti–E-selectin mAb. Untreated BMEC showed no E-selectin ligand activity, and rolling on IL-1α–treated BMEC was eliminated by incubation with function-blocking anti–E-selectin mAb 68-5H11. Data represent mean ± SD cell rolling frequency per 100 × magnified field × 5 fields of view, minimum of three experiments. Note that KG1a cell rolling was 3.5-fold greater than that of HL60 cells.

Exhaustive immunoprecipitation of CD44 (Hermes-1) and blot rolling assay of residual E-selectin ligand activity. Hermes-1 immunoblot (A) or HECA-452 immunoblot ( B) of KG1a lysate (10 μg) subjected to three rounds of immunoprecipitation with Hermes-1 mAb. (Pre-Ippt) Total KG1a lysate 10 μg; (lane 1) first round Hermes-1 immunoprecipitate; (lane 2) second round Hermes-1 immunoprecipitate; (lane 3) third round Hermes-1 immunoprecipitate; (lane 4) residual lysate after three rounds of Hermes-1 immunoprecipitation. E-selectin ligand activity (CHO-E cell rolling) correlates with intensity of the Hermes-1 staining and of HECA-452 staining of 100-kD band, is reduced after each round of Hermes-1 immunoprecipitation (A), and was completely absent in the residual lysate material after the third round of immunoprecipitation (lane 4, A and B). CHO-E cell rolling was also eliminated over the 190-kD band by exhaustive Hermes-1 immunoprecipitation, whereas CHO-E cell rolling on 120-kD stained band was markedly reduced, and the 140-kD band (PSGL-1) retained full activity (B, lane 4).

HECA-452–reactive CD44 from freshly isolated normal human HPCs and from human leukemic blasts functions as an E-selectin ligand. (Panel A) HECA-452 staining of CD44 immunoprecipitated from (a) human BM mononuclear cells (10⁸ cells), (b) CD34−/lineage+ cells (10⁷ cells), (c) CD34+/lineage− cells (10⁷ cells), (d) CD34−/lineage+ cells (10⁸ cells). HECA-452 staining of CD44 and CHO-E cell rolling was observable only on the 100-kD CD44 immunoprecipitated from CD34+/lineage− cells (data are mean ± SD cell rolling/field on the 100-kD band). (B) Membrane proteins (50 μg) isolated from circulating blasts from an AML (M5) were resolved on a 9% SDS-PAGE gel and immunoblotted with HECA-452. Treatment of membrane proteins with N-glycosidase F markedly diminished HECA-452 staining. (C) Blot rolling assay results of AML (M5) membrane protein (50 μg) immunoprecipitated with isotype control or with Hermes-1 mAb, and of N-glycosidase F–treated Hermes-1 immunoprecipitates. Though data shown in parentheses is CHO-E cell rolling over the 100-kD band, no rolling was observed over the entire length of the lane corresponding to N-glycosidase F–treated protein. (D) Immunoprecipitated CD44 from membrane preparations (50 μg) of an AML (M0), AML (M1) and atypical CML (brc/abl⁻), and of human BMEC line (BMEC-1; 100 μg total protein), was separated on a 6% SDS-PAGE gel, immunostained with HECA-452, and evaluated for E-selectin ligand activity. E-selectin ligand activity correlates with intensity of HECA-452 staining of 100-kD band. (E) Western blot of immunoprecipitated CD44 from BMEC-1 stained with Hermes-1 mAb.

HECA-452–reactive CD44 is a more avid E-Selectin ligand than PSGL-1. Equivalent amounts (1 μg) of either immunoprecipitated CD44 or PSGL-1 were analyzed for E-selectin and P-selectin ligand activity in the parallel plate flow chamber. (A) E-selectin–mediated CHO-E cell rolling was observed at 2.8 dynes/cm² on KG1a CD44, but was significantly lower on KG1a PSGL-1 at 2.8 dynes/cm² (P < 0.001). N-glycosidase F- and α-L-fucosidase-treated KG1a CD44, and Vibrio cholerae neuraminidase treatment of KG1a membrane protein abrogated CHO-E cell rolling (P < 0.001), and no CHO-E cell rolling was observed on isotype control rat IgG– or mouse IgG–immunoprecipitated KG1a protein (data not shown). (B) CHO-P cell rolling was observed on KG1a PSGL-1 but not on KG1a CD44 (2.8 dyn/cm²). No rolling was observed on negative controls (CHO-Mock cells and CHO-P cells pretreated with function-blocking anti–P-selectin mAb AK-4 [10 μg/ml]).