Distinct Sulfation Requirements of Selectins Disclosed Using Cells That Support Rolling Mediated by All Three Selectins under Shear Flow

l- and P-selectin are known to require sulfation in their ligand molecules. We investigated the significance of carbohydrate 6-sulfation and tyrosine sulfation in selectin-mediated cell adhesion. COS-7 cells were genetically engineered to express P-selectin glycoprotein ligand-1 (PSGL-1) or its mutant in various combinations with 6-O-sulfotransferase (6-Sul-T) and/or α1→3fucosyltransferase VII (Fuc-T VII). The cells transfected with PSGL-1, 6-Sul-T, and Fuc-T VII cDNAs supported rolling mediated by all three selectins and provided the best experimental system so far to estimate kinetic parameters in selectin-mediated cell adhesion for all three selectins using the identical rolling substrate and to compare the ligand specificity of each selectin. L-selectin-mediated rolling was drastically impaired if the cells lacked carbohydrate 6-sulfation elaborated by 6-Sul-T, but not affected when PSGL-1 was replaced with a mutant lacking three tyrosine residues at its NH2 terminus. L-selectin-mediated adhesion was also hardly affected by mocarhagin treatment of the cells, which cleaved a short peptide containing sulfated tyrosine residues from PSGL-1. In contrast, P-selectin-mediated rolling was abolished when PSGL-1 was either mutated or cleaved by mocarhagin at its NH2terminus, whereas the cells expressing PSGL-1 and Fuc-T VII but not 6-Sul-T showed only a modest decrease in P-selectin-mediated adhesion. These results indicate that L-selectin prefers carbohydrate 6-sulfation much more than tyrosine sulfation, whereas P-selectin favors tyrosine sulfation in the PSGL-1 molecule far more than carbohydrate 6-sulfation. E-selectin-mediated adhesion was sulfation-independent requiring only Fuc-T VII, and thus the three members of the selectin family have distinct requirements for ligand sulfation.

In addition to leukocyte-endothelial interactions, L-selectin is proposed to contribute to leukocyte-leukocyte interactions (31,32). In this context, recent reports have suggested that the ligand specificity of L-selectin is similar to that of P-selectin and that PSGL-1 also serves as a L-selectin ligand by utilizing both the carbohydrate determinants and the three tyrosine sulfate residues at the extreme NH 2 -terminal region of PSGL-1 (33)(34)(35)(36)(37)(38)(39)(40). Transfection of cells with cDNA for PSGL-1 together with that for fucosyltransferases is reported to confer binding activity for L-selectin, even without any exogenous introduction of 6-Sul-T cDNA (37,38,40,41). This implies that tyrosine sulfation may substitute for carbohydrate 6-sulfation in Lselectin-mediated cell adhesion. Another possibility is that the cells used in the cell adhesion assays expressed endogenous 6-Sul-T. The relative importance of tyrosine sulfation versus carbohydrate 6-sulfation of PSGL-1 in L-selectin-mediated cell adhesion thus remains to be fully determined. For instance, treatment of cells with chlorate, which was sometimes employed in earlier experiments to inhibit tyrosine sulfation of PSGL-1, also abolishes carbohydrate 6-sulfation.
In this study, we evaluated the relative contribution of carbohydrate 6-sulfation and the tyrosine sulfation of PSGL-1 on cell adhesion mediated by selectins using monolayer cell adhesion assays and rolling assays under shear flow. We employed two different approaches. One was to use cells expressing a mutated form of PSGL-1, indicated as the FFFE mutant of PSGL-1. In this mutant, the three tyrosine residues essential for the recognition by P-selectin were replaced by phenylalanine residues at the NH 2 -terminal region of PSGL-1 (24,37). This is known to nearly completely abrogate P-selectin-mediated cell adhesion. We then generated stable transfectant cells expressing either intact PSGL-1 or its FFFE-mutant, with various combinations of co-transfection with Fuc-T VII and/or 6-Sul-T cDNAs. The other approach was mocarhagin treatment of the cells. Mocarhagin is a metalloproteinase known to specifically remove the NH 2 -terminal 10 amino acids of the mature form of PSGL-1, which includes the three potential sulfated tyrosine residues, thus specifically inhibiting P-selectin-mediated cell adhesion (42)(43)(44).

EXPERIMENTAL PROCEDURES
Monoclonal Antibodies (mAbs) and Flow Cytometric Analyses of Transfectant Cells-The mAbs used in these studies were as follows: G72, recognizing anti-6-sulfo LacNAc, and G152, directed to sialyl 6-sulfo Lewis X (both murine IgM), were prepared as described previously (20). CSLEX-1, a classical anti-sialyl Lewis X antibody (murine IgM), was obtained from American Type Culture Collection (Manassas, VA). The anti-PSGL-1 antibodies, PL1, PL2 (both murine IgG 1 from Immunotech), and KPL1 (murine IgG 1 , BD PharMingen, Mountain View, CA), and an anti-CD34 antibody, clone QBEnd 10 (murine IgG 1 , Seikagaku Corp., Tokyo), were commercially obtained. Flow cytometric analysis was performed using FACScan (BD PharMingen). The cells were harvested at a semiconfluent stage and stained with the respective, purified monoclonal antibody at 1 g/ml or culture supernatant at a dilution of 1:10 at 4°C for 30 min. The cells were then washed three times with phosphate-buffered saline containing 0.5% bovine serum albumin and stained with a 1:200 dilution of fluorescein isothiocyanateconjugated goat anti-mouse Ig (Silenus Laboratories) at 4°C for 30 min.
Non-static Monolayer Cell Adhesion Assays-In vitro non-static monolayer cell adhesion assays were performed as described previously (20,45). Briefly, parental COS-7 cells or transfectant clones were cultured in monolayer in 24-well plates. After the addition of BCECF-AMlabeled selectin transfectant of cultured murine B lymphoma cells (300.19/L, 300.19/P, or 300.19/E, 8 ϫ 10 5 cells/0.5 ml/well) (46), the plate was incubated with horizontal rotation at 100 rpm for 20 min at 25°C. Non-adherent cells were washed out three times with phosphatebuffered saline containing Ca 2ϩ and Mg 2ϩ , the cells were lysed with 0.5% Nonidet P-40, and the attached cells were counted by measuring fluorescence using an Arvo 1420 multilabel counter (Wallac, Gaithersburg, MD).
Adhesion Assay under Shear Flow-Rolling assays for COS-7 transfectant cells were performed using the flow chamber of Lawrence et al. Mocarhagin Digestion-The cobra venom metalloproteinase, mocarhagin, was purified as described previously (42)(43)(44). The PSGL-1/7,6 cells were cultured in a monolayer in 24-well plates or on chamber slides, and then washed twice with DMEM supplemented with 1% FBS. The cells were incubated for 30min at 25°C with mild shaking with or without mocarhagin at 12.5 g/ml in DMEM supplemented with 1% FBS. The cells were then washed two times with 500 l of DMEM supplemented with 10% FBS and used for cell adhesion assays. The change in cell surface antigens after mocarhagin treatment was examined by flow cytometry.

Influence of 6-Sul-T and Fuc-T VII Transfection on Cell Adhesion Mediated by Selectins-
The 300.19/L cells adhered only to the 7,6/COS-7 cells and did not adhere to the 7/COS-7, 6/COS-7, and parental COS-7 cells in monolayer cell adhesion assays ( Fig. 2A). The same specificity was noted in the rolling of 300.19/L cells on the transfectant cells under shear flow (Fig.  2B). Significant rolling of 300.19/L cells was observed only on 7,6/COS-7, but not on 7/COS-7 cells (Fig. 2B). Maximum rolling of 300.19/L cells on the 7,6/COS-7 cells was observed at a shear stress of 1.4 dyne/cm 2 or below, and at this shear stress, the average number of 300.19/L cells rolling on the 7,6/COS-7 cells was 1125 Ϯ 145/mm 2 , while that for the 7/COS-7 cells was only 22 Ϯ 9/mm 2 . The rolling velocity of the 300.19/L cells on the 7,6/COS-7 cells was estimated to be 124 m/s (Table I).
On the other hand, 300.19 cells expressing E-selectin (300.19/E) adhered to the 7/COS-7 and 7,6/COS-7 cells in both the non-static monolayer adherent assay and in terms of rolling under shear flow, indicating that the introduction of Fuc-T VII is sufficient for E-selectin-mediated interaction. Maximum rolling of the 300.19/E cells on the 7,6/COS-7 and 7/COS-7 cells was observed at 1.1 dyne/cm 2 or below, and the average number and rolling velocity of 300.19/E cells on 7,6/COS-7 cells at this shear stress were 1111 Ϯ 35/mm 2 and 49 m/s, respectively, while those for 7/COS-7 cells were 972 Ϯ 54/mm 2 and 43 m/s. The 300.19 cells expressing P-selectin (300.19/P) did not interact with these cells until we further introduced cDNA encoding PSGL-1 (see below).
Expression of PSGL-1 and FFFE-PSGL-1 Mutant on Stable Transfectant Cells-We next introduced cDNAs for PSGL-1 and its FFFE-PSGL-1 mutant to the 7/COS-7 or 7,6/COS-7 cells and generated stable transfectants, namely PSGL-1/7, PSGL-1/7,6, FFFE/7, and FFFE/7,6 cells, respectively. Fig. 3 shows flow cytometric analyses for expression of carbohydrate determinants and PSGL-1 or FFFE-PSGL-1 mutant protein on these cell lines. The KPL1 antibody is known to strictly recognize a peptide epitope that includes the sulfated tyrosine residues at the extreme NH 2 -terminal region of PSGL-1 (37). The antibody strongly labeled the PSGL-1/7 and PSGL-1/7,6 cells, while FFFE/7 and FFFE/7,6 cells lacked the KPL1 reactivity (Fig. 3), indicating, as expected, that the sulfated tyrosine residues were not expressed on these cells. The PL2 antibody, an anti-PSGL-1 mAb, recognizes a peptide epitope in the middle region of the molecule common to intact PSGL-1 and its FFFE-mutant (25,49). The reactivity of the FFFE/7 or FFFE/7,6 cells to PL2 mAb was essentially identical to that of the PSGL-1/7 or PSGL-1/7,6 cells (Fig. 3). The expression level of carbohydrate determinants on the cells transfected with FFFE-PSGL-1 cDNA was almost the same as that on the cells transfected with intact PSGL-1 cDNA (Fig. 3). Influence of the Tyrosine Residues at the NH 2 -terminal Region of PSGL-1 on the Adhesion Mediated by Selectins-When these transfectant cells were tested in the cell adhesion assays, the 300.19/L cells adhered to the cells transfected with both Fuc-T VII and 6-Sul-T, regardless of the presence or absence of the NH 2 -terminal tyrosine residues (Fig. 4, A and B). A clear peak of maximum rolling of 300.19/L cells on the PSGL-1/7,6 or FFFE/7,6 cells was observed at around 2.0 dyne/cm 2 . At this shear stress, the average number and velocity of 300.19/L cells rolling on the PSGL-1/7,6 cells were 2474 Ϯ 410/mm 2 and 66 m/s, while those for FFFE/7,6 cells were 2595 Ϯ 388/mm 2 and 93 m/s, respectively (see Table I for rolling velocities). These results indicate that the introduction of PSGL-1 or its FFFE mutant on the cells increased the number of rolling cells about 2.2ϳ2.3 times and significantly increased the resistance against the hydrodynamic shear stress. These results, however, also indicate that the increase was not directly related to the tyrosine residues at the NH 2 terminus of the PSGL-1 molecule, since introduction of the FFFE mutant of PSGL-1 had essentially the same effect as that of intact PSGL-1.
Interestingly, a small increase was observed in the number of 300.19/L cells rolling on the PSGL-1/7 cells compared with that on FFFE/7 cells, which was statistically significant (Fig.  4B). The average number of the 300.19/L cells rolling on FFFE/7 cells was only 12 Ϯ 10/mm 2 , against 135 Ϯ 53/mm 2 for PSGL-1/7 cells at the shear stress of 1.4 dyne/cm 2 , which supported the maximum rolling of the 300.19/L cells on the PSGL-1/7 cells. This would imply that sulfation of tyrosine residues at the NH 2 -terminal region of PSGL-1 molecule had some facilitative effect on L-selectin-mediated cell adhesion, but it was still much less than the effect of 6-sulfation of the GlcNAc moiety in the carbohydrate portion of the ligand molecules.
These results were in a clear contrast to P-selectin-or Eselectin-mediated adhesion of these cells. The 300.19/P cells adhered only to the PSGL-1 expressing cells (PSGL-1/7 and PSGL-1/7,6), but not to the cells expressing its FFFE mutant (Fig. 4, A and B), indicating the specific role of the sulfated tyrosine residues near the NH 2 -terminal region in P-selectinmediated cell adhesion as reported previously (22)(23)(24)(25)(26)(27)(28)(29)(30). Maximum rolling of the 300.19/P cells on PSGL-1/7 or PSGL-1/7,6 cells was observed at the shear stress of 1.1 dyne/cm 2 or below, and the average number and velocity of the rolling 300.19/P cells were 1104 Ϯ 52/mm 2 and 49 m/s for the PSGL-1/7 cells (at 0.6 dyne/cm 2 ), while they were 1181 Ϯ 125/mm 2 and 43 m/s for the PSGL-1/7,6 cells (at 1.1 dyne/cm 2 ). The 300.19/E cells adhered to all the transfectants expressing Fuc-T VII cDNA shown in Fig. 4, A and B. For these cells, maximum rolling was always observed around 1.1 dyne/cm 2 , and at this shear stress the number and velocity of rolling of the 300.19/E cells were in the range of 2241 ϳ 2860/mm 2 and 28 ϳ 43 m/s. The PSGL-1/7,6 cells were noted to support rolling mediated by all three selectins, and this enabled a direct comparison of the kinetic parameters for the three selectins on the same rolling substrate. The optimal hydrodynamic shear stress of rolling on the PSGL-1/7,6 cells was much higher for L-selectin (2.0 dyne/cm 2 ) than that for P-and E-selectin, which was around 1.1 dyne/cm 2 ( Fig. 4 and Table I). Rolling velocities on the PSGL-1/7,6 cells were highest for L-selectin (66 m/s), followed by P-selectin (43 m/s), and lowest for E-selectin (31 m/s), and the difference was statistically significant in every combination ( Table I).
Influence of Mocarhagin Treatment on the Adhesion Mediated by Selectins-The importance of sulfated tyrosine residues in the NH 2 -terminal region of PSGL-1 in selectin-mediated cell adhesion was also tested by treatment with mocarhagin. Specific cleavage of the NH 2 -terminal decapeptide from PSGL-1 by mocarhagin (44) was assessed by flow cytometric analysis of enzyme-or mock-treated PSGL-1/7,6 cells using mAbs KPL1 and PL1 (Fig. 5A). The binding of KPL1, recognizing the NH 2terminal epitope that includes the three tyrosine-sulfated residues in human PSGL-1 (37), was markedly reduced by mocarhagin treatment, while the binding of PL1, which is known to recognize an epitope further downstream (25,37,49), remained unaltered (Fig. 5A). The reactivity of the mAbs G152 and CSLEX-1 was also essentially unaltered, indicating that the limited hydrolysis induced by mocarhagin treatment had negligible effect on the expression of the carbohydrate determinants defined by these antibodies (Fig. 5A).
Results of cell adhesion assays before and after mocarhagin treatment clearly indicated that it strongly affected P-selectinmediated cell adhesion, but was much less effective on L-and E-selectin-mediated adhesion in both monolayer cell adhesion assays (Fig. 5B) and rolling assays (Fig. 5C). In monolayer cell adhesion assays (Fig. 5B), the number of adherent 300.19/P cells decreased to less than 30% of that of non-treated cells, while more than 70% of the adhesion of 300.19/L cells was retained after mocarhagin treatment. The 300.19/E cells adhered to the PSGL-1/7,6 cells at a similar level after treatment with mocarhagin. The results of rolling assays (Fig. 5C) were even more clear in that mocarhagin treatment almost completely abolished P-selectin-mediated adhesion, but had virtually no effect on L-and E-selectin-mediated adhesion in terms of the number of rolling cells. The velocity and optimum shear stress for rolling of the 300.19/L and 300.19/E cells also showed no significant change after mocarhagin treatment (Fig. 5C).
Influence of the Expression of CD34 on the Adhesion Mediated by Selectins-CD34 is a sialomucin that is regarded as one of the major core proteins for the L-selectin ligands on HEVs (50,51). The 7/COS-7 or 7,6/COS-7 cells were transfected with cDNA for CD34, and the resulting CD34/7 or CD34/7,6 cells were examined for the expression of CD34 protein and carbohydrate determinants by flow cytometry (Fig. 6A). Both CD34/7 cells and CD34/7,6 cells expressed the CD34 molecule at almost equivalent levels. Only CD34/7,6 cells, but not CD34/7 cells, significantly expressed sialyl 6-sulfo Lewis X. Non-sulfated sialyl Lewis X determinant defined by CSLEX-1 was expressed on both CD34/7 cells and CD34/7,6 cells.
The 300.19/L cells adhered to and rolled on the CD34/7,6 cells, but not on the CD34/7 cells. Significant adhesion and rolling of the 300.19/E cells was observed on both CD34/7 and CD34/7,6 cells, while 300.19/P cells interacted with none of these cells (Fig. 6, B and C). Introduction of CD34 resulted in a marginal increase in the number of cells involved in L-and E-selectin-mediated adhesion or rolling. This, however, was much less prominent than the effect exerted by the introduction of PSGL-1. Similar findings were reported previously for CD34 in L-selectin-mediated cell adhesion (41). The optimum shear stress for rolling showed no significant difference before ( Fig. 2B) or after (Fig. 6C) introduction of CD34 to the cells. The velocity of the 300.19/L cells rolling on the CD34/7,6 cells was 156 Ϯ 47 m/s at 1.4 dyne/cm 2 , while the rolling velocity of the 300.19/E cells on the CD34/7 or CD34/7,6 cells was 43 Ϯ 16 or 44 Ϯ 17 m/s at 1.1 dyne/cm 2 , respectively. DISCUSSION We have previously shown that the enzymatic action of both Fuc-T VII and 6-Sul-T is required for the optimum expression of L-selectin ligands using monolayer cell adhesion assays (20). In this study we have expanded our analysis to include all three members of the selectin family. Furthermore, we have examined the effect of specific core proteins in addition to the introduction of Fuc-T VII and 6-Sul-T, mainly using rolling assays under shear flow. The results obtained in this study clearly indicate that each member of the selectin family cell adhesion molecules has a distinct ligand binding specificity of its own.
The simultaneous expression of 6-Sul-T and Fuc-T VII was essential for the synthesis of ligand for L-selectin, while introduction of core proteins such as PSGL-1 or CD34 augmented this interaction. Negligible or very weak rolling of 300.19/L cells was observed on the cells expressing Fuc-T VII but lacking 6-Sul-T, including the COS-7/7, PSGL-1/7, FFFE/7, and CD34/7 cells. Far more significant rolling was always observed on the cells expressing both 6-Sul-T and Fuc-T VII, such as the 7,6/COS-7, PSGL-1/7,6, FFFE/7,6, and CD34/7,6 cells. In contrast, P-selectin strictly required the presence of PSGL-1 in addition to the action of Fuc-T VII for ligand binding, and the presence of 6-Sul-T did not seem to be essential. All the transfectant cells lacking expression of PSGL-1 failed to support P-selectin-mediated rolling without exception. The inability of cells expressing the FFFE mutant of PSGL-1 to support rolling of P-selectin-expressing cells is in line with the suggested importance of sulfation of the tyrosine residues at the NH 2 -terminal region of PSGL-1 (23,24,37,44). Although many investigators have reported a similarity in the ligand specificity of Pand L-selectins in that they both required sulfation for their ligands (33)(34)(35)(36)(37)(38)(39)(40), the present results clearly establish that the sulfation required by L-selectin is carbohydrate 6-sulfation, whereas that required by P-selectin is sulfation of tyrosine residue(s) at the NH 2 terminus of the PSGL-1 molecule. In contrast, the introduction of Fuc-T VII was necessary and sufficient for the synthesis of functional ligands for E-selectin. P-selectin is known to require the NH 2 -terminal region of PSGL-1 having two distinct posttranslational modifications, sulfation of tyrosines clustered at residues Tyr-5, -7, and -10 of the mature protein and attachment of a core2 O-glycan capped with sialyl Lewis X-related carbohydrate determinants at Thr-16 (23,24,26,27,37,38,44). It was unexpected that sulfated tyrosine residues in PSGL-1 turned out to be much less effective in promoting L-selectin-mediated adhesion than carbohydrate 6-sulfation, since many preceding reports have stressed the importance of the molecule in L-selectin-mediated adhesion. This led to the assumption that L-selectin ligands carried by PSGL-1 may be the preferential L-selectin ligand in leukocyte-leukocyte interactions (37)(38)(39)(40)(41). Our results also suggest some augmentative effect of PSGL-1 on L-selectin-mediated cell adhesion. Introduction of PSGL-1 increased the number of rolling cells about 2.2-fold and significantly increased the resistance against shear stress in L-selectin-mediated adhesion. However, this did not appear to be due to a specific effect of sulfation of tyrosine residues in the PSGL-1 molecule. Both cells expressing either the FFFE mutant of PSGl-1 or cells expressing PSGL-1 and treated with mocarhagin to remove the NH 2 -terminal decapeptide that contains the sulfated tyrosine residues retained a similar facilitative effect for L-selectinmediated adhesion. Moreover, a similar enhancing effect was also noted for E-selectin-mediated rolling on introduction of either intact PSGL-1 or FFFE-PSGL-1. Most probably these quantitative enhancing effects were largely due to the so-called density effect. The significance of PSGL-1 in selectin-mediated adhesion is not restricted to its sulfated tyrosine residues, since it also supports high-density presentation of carbohydrate determinants preferentially at the termini of microvilli (52). The only evidence supporting a more specific role for tyrosine sulfation in L-selectin-mediated adhesion was a small but significant increase of rolling on the cells transfected with both PSGL-1 and Fuc-T VII cDNAs (the PSGL-1/7 cells) compared with the cells transfected with the FFFE mutant PSGL-1 and Fuc-T VII cDNAs (the FFFE/7 cells). This would account for some of the earlier observations that reported a role for PSGL-1 in L-selectin-mediated cell adhesion and would imply that sulfation of tyrosine residues of PSGL-1 may partially substitute the role of carbohydrate 6-sulfation when the carbohydrate determinants lack GlcNAc 6-sulfation. Its effect, however, was far less (estimated to be ϳ5%) than the effect exerted by the introduction of 6-Sul-T. In this regard, a significant delay in P-selectin-mediated leukocyte migration to the peritoneum and skin has been observed in the PSGL-1-deficient mouse, but so far no evidence for any serious impairments of L-selectin-mediated cell adhesion has been detected in these mice (53,54).
It should be noted that the earlier studies that suggested a significant role for PSGL-1 in L-selectin-mediated rolling were performed before the elucidation of the essential role of carbohydrate 6-sulfation in L-selectin-mediated cell adhesion and performed without introducing the 6-Sul-T cDNA in the tested cells. Under such conditions, the level of endogenous 6-Sul-T, which will be different for each experimental system, would strongly affect the experimental outcome. Similar problems seem to be involved in the characterization of the kinetic parameters for L-selectin-mediated rolling. In some investigations, the optimum shear stress for L-selectin-mediated rolling fell within the same range as that for E-or P-selectin-mediated rolling (55), while significantly higher optimum shear stresses for L-selectin-mediated rolling were obtained in other experimental systems (56 -58). Our results indicate the optimum shear stress for L-selectin-mediated rolling on the PSGL-1/7,6 or FFFE/7,6 cells to be significantly higher (1.5-2.5 dyne/cm 2 ) than that for E-and P-selectin-mediated rolling (0.5-1.5 dyne/ cm 2 ) on the same cells. L-selectin, but not E-or P-selectin, is known to require a minimum level of hydrodynamic shear stress to initiate rolling (56,59). A similar requirement for hydrodynamic shear flow above a threshold level around 1.0 dyne/cm 2 was clearly observed for L-selectin-mediated rolling on the PSGL-1/7,6 and FFFE/7,6 cells. However, our results also indicate that a higher optimum shear stress was observed with PSGL-1/7,6 cells as substrate, while much lower optimum fluid shears were obtained with PSGL-1/7 or 7,6/COS-7 cells. This finding is compatible with the assumption that the higher shift of optimal hydrodynamic shear is mainly due to the carbohydrate 6-sulfation as well as the increased density of carbohydrate determinants. In this regard, it has been reported that modification of the carbohydrate portion of L-selectin ligand results in a drastic decrease in the shear threshold of L-selectin-mediated rolling (60). The optimal shear stress was independent of tyrosine sulfation, since in every experiment the FFFE-PSGL-1 cells showed essentially the same shear stress dependence as the cells expressing intact PSGL-1. A shift in the optimum shear stress was noted even for P-selectinmediated rolling on the PSGL-1/7,6 cells compared with the PSGL-1/7 cells, which would, although less prominent than that observed in L-selectin-mediated rolling, further support a role for carbohydrate 6-sulfation in the resistance against hydrodynamic shear stress. The velocity of L-selectin-mediated rolling is known to be significantly higher than the rolling velocity for P-or E-selectin-mediated adhesion (38,55,60,61). The velocities of 147 Ϯ 114 m/s for L-selectin and 19 Ϯ 33 m/s for E-selectin reported from in vivo experiments (62) are also in good agreement with the values reported in the current study. It is pertinent that the PSGL-1/7,6 cells employed in this study represent a universal reagent in that they supported optimum adhesion and rolling mediated by all three selectins. We believe this is the first example where the kinetic parameters for selectin-mediated cell adhesion have been compared for all three selectins using an identical rolling substrate.