Regulation of PSGL-1 interactions with L-selectin, P-selectin, and E-selectin: role of human fucosyltransferase-IV and -VII.

P-selectin glycoprotein ligand-1 (PSGL-1) interactions with selectins regulate leukocyte migration in inflammatory lesions. In mice, selectin ligand activity regulating leukocyte recruitment and lymphocyte homing into lymph nodes results from the sum of unequal contributions of fucosyltransferase (FucT)-IV and FucT-VII, with FucT-VII playing a predominant role. Here we have examined the role of human FucT-IV and -VII in conferring L-selectin, P-selectin, and E-selectin binding activities to PSGL-1. Lewis x (Le(x)) carbohydrate was generated at the CHO(dhfr)(-) cell surface by FucT-IV expression, whereas sialyl Le(x) (sLe(x)) was synthesized by FucT-VII. Both human FucT-IV and -VII had the ability to generate carbohydrate ligands that support L-selectin-, P-selectin-, and E-selectin-dependent rolling on PSGL-1, with FucT-VII playing a major role. Cooperation was observed between FucT-IV and -VII in recruiting L-, P-, or E-selectin-expressing cells on PSGL-1 and in regulating cell rolling velocity and stability. Additional rolling adhesion assays were performed to assess the role of Thr-57-linked core-2 O-glycans in supporting L-selectin-, P-selectin-, and E-selectin-dependent rolling on PSGL-1. These studies confirmed that core-2 O-glycans attached to Thr-57 play a critical role in supporting L- and P-selectin-dependent rolling and revealed that additional binding sites support >75% of E-selectin-mediated rolling. The observations presented here indicate that human FucT-IV and -VII both contribute and cooperate in regulating L-selectin-, P-selectin-, and E-selectin-dependent rolling on PSGL-1, with FucT-VII playing a predominant role in conferring selectin binding activity to PSGL-1.

Leukocyte ␣1,3-fucosyltransferases (FucT) synthesize Le x , sLe x , and cutaneous lymphocyte antigen (CLA), which are essential to confer an activity to selectin ligands (38 -42). FucT-IV and -VII are constitutively expressed in neutrophils, whereas lymphocytes express FucT-VII after cell activation (43)(44)(45). Human FucT-VII is involved in the synthesis of sLe x and CLA, whereas human FucT-IV generates Le x and VIM-2 carbohydrate determinants; the physiological role of human FucT-IV in synthesizing sLe x is uncertain (38, 40, 41, 46 -49). Biochemical studies showed that FucT-IV fucosylates all Nacetyllactosamine units in neutral polylactosamines and internal lactosamine units in ␣2,3-sialylated polylactosamines, whereas FucT-VII transfers fucose only to distal ␣2,3-sialylated lactosamine units to generate the sLe x terminus (50). Studies performed in FucT-VII-deficient mice revealed that FucT-VII confers P-selectin and E-selectin binding activity to PSGL-1 as well as L-selectin binding activity to ligands expressed on high endothelial venules of peripheral lymph nodes (51)(52)(53). FucT-VII-deficient mice exhibit a defect in leukocyte migration in inflammatory lesions and lymphocyte homing into peripheral lymph nodes (51,52). In contrast, FucT-IV null mice presents only subtle defects in leukocyte adhesion with an increase in rolling velocities along inflamed vessels (52,53). Unlike FucT-VII null mice, mice deficient in both FucT-IV and -VII exhibit an extreme leukocytosis with leukocytes virtually devoid of E-selectin and P-selectin ligand activity and an almost abrogated lymphocyte homing in peripheral lymph nodes (52). These observations indicated that both FucT-IV and -VII contribute to regulate leukocyte recruitment and lymphocyte homing in mouse (52). More recently, observations made with leukocytes obtained from FucT-VII-deficient patients suggested that human FucT-IV may play a role in generating Pand E-selectin ligands (49,54). Thus, despite a near complete absence of sLe x synthesis, neutrophils from these patients display normal rolling on P-selectin and E-selectin (49). However, a compensatory increase in FucT-IV activity was observed in human neutrophils deficient in FucT-VII that may have rescued their rolling capacity on P-selectin and E-selectin (49).
Because differences in carbohydrate synthesis produced by human and mouse glycosyltransferases have been observed (39,(55)(56)(57), we compared human FucT-IV and -VII in generating selectin ligand activity on human PSGL-1, and in particular we examined the role of FucT-IV in regulating L-selectin interactions with its major ligand PSGL-1 (25). In addition, as the complimentary acceptor site specificities of FucT-IV and -VII suggest that they may cooperate in generating selectin ligands, we examined whether human FucT-IV and -VII collaborate in regulating PSGL-1 interactions with L-, P-, or Eselectin. These points had not been examined in studies previously devoted to the relative role of FucT-IV or -VII in generating E-selectin or P-selectin counter-receptor activities (18, 26, 43, 46, 47, 49, 58 -61). Of note, previous observations showed that CHO dhfr Ϫ cells and CHO Pro-5 cells (41,42,46,62) and leukocytic cell lines (43) may differ in their capacity to synthesize selectin ligands. This property may limit the physiological relevance of assays performed with cell lines. By taking into account this limitation, we chose to study the specific involvement of human FucT-IV and the cooperation of human FucT-IV and -VII in conferring selectin binding activity to PSGL-1 in CHO dhfr Ϫ cells, because these cells, when cotransfected with C2GnT, FucT-IV, or -VII and PSGL-1 cDNAs, have the ability to generate ligands for the three selectins (26,46,58,63). In addition, results presented here with CHO dhfr Ϫ cells expressing FucT-IV are close to those reported with human FucT-VII-deficient neutrophils (49). Our observations show that the expression of human FucT-IV or/and -VII in CHO dhfr Ϫ cells generate selectin binding activity on PSGL-1 and that FucT-IV and -VII unequally contribute but cooperate in supporting L-, P-, and E-selectin-mediated rolling on PSGL-1. Additional experiments have been performed to examine the role of core-2 O-glycans attached to Thr-57 in regulating E-selectin-dependent rolling on PSGL-1. Results indicate that Thr-57-core-2 O-linked glycans are not the sole E-selectin-binding sites; additional binding sites support Ͼ75% of E-selectin-dependent rolling present on PSGL-1.
Lowe (Howard Hughes Institutes, Ann Harbor, MI), and core-2 ␤1,6-Nacetylglucosaminyltransferase transferase (C2GnT) was from M. Fukuda (the Burnham Institute, La Jolla Cancer Research Center, San Diego). The cDNA sequence encoding the internal ribosome entry site (IRES) of the encephalomyocarditis virus sequence was a gift from P. Aebischer (Swiss Federal Institute of Technology, Ecublens, Switzerland) (26). The pZeoSV and the pcDNA3.1 vectors were from Invitrogen. The sequence encoding the IRES was inserted in the multiple cloning site of the pZeoSV vector (pIRES ZeoSV vector). C2GnT and FucT-VII cDNAs were then subcloned in the pIRES ZeoSV vector to permit the translation of C2GnT and FucT-VII or FucT-IV cDNA sequences from one messenger RNA, as described (26). This vector allows the expression of the target gene (C2GnT) and of the selection marker (sLe x versus Le x expression associated to FucT-VII versus FucT-IV activity) from the same promoter so that virtually all transfected cells expressing the selection marker also express the gene of interest (65)(66)(67). FucT-IV cDNA sequence was also inserted into the pcDNA3.1hygromycin vector (Invitrogen) and used to transfect CHO dhfr Ϫ cells co-expressing C2GnT and both FucT-IV and -VII cDNAs.
Cells and Transfections-Neutrophils were isolated from normal heparinized blood samples by Ficoll-Paque centrifugation, dextran sedimentation, and erythrocyte hypotonic lysis (14). CHO dhfr Ϫ cells (ATCC number CRL 9096) were stably transfected with pCDNA3.1 vector expressing the cDNA sequences of wild-type PSGL-1, PSGL-1/ chimera, or PSGL-1T57A/ mutant. When indicated, CHO dhfr Ϫ cells were co-transfected with the following: 1) FucT-VII cDNA sequence subcloned in pZeoSV vector (Invitrogen); 2) FucT-IV cDNA sequence subcloned in pCDNA3.1 vector (Invitrogen); or 3) the pIRES ZeoSV expression vector that allows the simultaneous translation of C2GnT and FucT-VII or FucT-IV sequences (26). Transfections were performed using Gene Porter (Core BioServices) or Trans-IT LT1 (Mirus Corp., Madison, WI). CHO dhfr Ϫ were cultured in minimum Eagle's ␣-medium containing ribonucleosides, deoxyribonucleosides, 10% fetal calf serum (FCS), and 1% penicillin/streptomycin. Transfectants were selected in medium containing 400 g/ml G418 (Invitrogen) and, when required, 200 g/ml Zeocin (Invitrogen) and 650 g/ml hygromycin (Calbiochem-Novabiochem). Individual clones co-expressing high levels of PSGL-1 and FucT-IV or -VII-dependent expression of Le x , sLe x , or CLA were isolated by limiting dilution and identified by immunophenotypic analysis using PL2, CSLEX-1, HECA-452, and 80H6 mAbs. CHO dhfr Ϫ cells that were selected for adhesion studies expressed similar levels of PSGL-1, sLe x , CLA, and/or Le x . The expression levels of the various PSGL-1 glycoforms of Le x and sLe x determinants were measured (68) on CHO cell transfectants, mentioned in Figs. 1-5, with the DAKO QIFIKIT® (Dako, Glostrup, Denmark). The anti-PSGL-1 mAb PS5 was used to assess PSGL-1 expression, the 80H6 mAb to measure Le x expression, and CSLEX-1 mAb to assess sLe x expression. 443 The levels of C2GnT mRNAs were compared in stable transfectants by semi-quantitative reverse transcription-PCR. Total RNA was extracted from each CHO transfectant (5 ϫ 10 6 cells) using Trizol LS Reagent (Invitrogen). After treatment with DNase I, 100 ng of RNA was amplified for 30 min at 50°C (30 cycles) using the Superscript One-Step RT-PCR with Platinum® Taq System (Invitrogen). C2GnT cDNA was detected with the sense primer GGCAGTGCCTACTTCGTGGTC and the antisense primer ATGCTCATCCAAACACTGGATGGCAAA. The number of PCR cycles was titered below the plateau phase of amplification to reflect the relative starting concentrations of analyzed mRNAs. Cycles were performed in a Primus 96 Plus Thermal Cycler (MWG Biotec, Germany) at 94°C for 30 s, 56°C for 45 s, and 72°C for 50 s. The final extension was performed at 72°C for 10 min. ␤-Actin was detected using the sense primer GAGACCTTCAACACCCC and the antisense primer GTGGTGGTGAAGCTGTAGCC. PCR products were separated by agarose gel electrophoresis containing ethidium bromide. Similar levels of ␤-actin and of C2GnT mRNAs were observed in CHO-C2F4F7, CHO-C2F4PSGL-1, CHO-C2F7PSGL-1, and CHO-C2F4F7PSGL-1 cells (not illustrated). P-selectin, E-selectin, and L-selectin cDNA sequences were subcloned in pcDNA3.1 vector (Invitrogen) and used to transfect K-562 cells, CHO dhfr Ϫ cells, or pre-B murine 300.19 cells by lipofection. K-562-P-selectin cells (K-562-P cells), CHO-P-selectin cells (CHO-P cells), K-562-E-selectin cells (K-562-E cells), and 300.19-L-selectin cells (300.19-L cells) were selected in medium containing 0.4 mg/ml G418 (geneticin, Invitrogen) and isolated by limiting dilution. K-562-P, K-562-E, and 300.19-L transfectants expressing high levels of selectins were cultured in RPMI 1640 medium containing 10% FCS, 1% penicillin/streptomycin, and 400 g/ml G418. 300.19-L cells used to perform the experiments described here expressed L-selectin levels similar to those of human neutrophils; K-562-E cells expressed E-selectin levels that were similar to those of human umbilical vein endothelial cells exposed for 6 h to 50 units/ml tumor necrosis factor-␣ (not illustrated).
Immunophenotypic Analysis-Expression of PSGL-1, sLe x , CLA, Le x on transfected CHO dhfr Ϫ cells, and of L-, P-, or E-selectin on 300.19, K-562, or CHO dhfr Ϫ cells was detected by flow cytometry using saturating concentrations of appropriate FITC-or PE-conjugated mAbs or unlabeled mAbs (26). When required, PE-conjugated goat anti-mouse Ig (Dako), FITC-labeled goat anti-mouse IgM (Jackson ImmunoResearch Laboratories, West Grove, PA), or rabbit anti-rat IgM were used as secondary antibodies (Rockland Immunochemicals, Gilberts, PA). Cells were incubated on ice for 30 min with appropriate antibodies, washed, fixed in 1% paraformaldehyde, and analyzed on an Epics XL-MCL cytofluorimeter (Beckman Instruments, Nyon, Switzerland). A total of 5000 cells was analyzed in each experiment.
L-, P-, E-selectin/ binding to CHO dhfr Ϫ transfectants was assessed by incubating CHO cells, for 30 min on ice, with chimeric proteins used at optimal concentrations for immunostaining, in RPMI 1640 supplemented with 1% FCS. Cell surface binding of chimera was revealed using a polyclonal PE-conjugated goat anti-human IgM heavy chain antibody (10 g/ml; Dako) (14,26). The specificity of L-, P-, or Eselectin/ chimera binding to PSGL-1 was indicated by the abrogation of binding in the presence of 5 mM EDTA or 100 g/ml of appropriate anti-L-, -P-, or -E-selectin blocking mAb (LAM1-3, WAPS 12.2, and H18/7). 5000 cells were analyzed in each experiment on an Epics XL-MCL cytofluorimeter (Beckman Instruments).
The ability of the various fucosylated forms of PSGL-1/ chimera to inhibit P-selectin/ binding to sLe x /BSA (Calbiochem-Novabiochem) was studied by ELISA. 96-Well ELISA microtiter plates (96-well ELISA plate, Costar, Corning, NY) were coated overnight at 4°C with sLe x coupled to bovine serum albumin (sLe x /BSA, 0.15 g/well in sodium carbonate/bicarbonate buffer, pH 9.4) and blocked with PBS containing BSA (10 mg/ml). P-selectin/ chimera (2 g/ml), precomplexed with biotinylated goat anti-human IgM (0.5 g/ml), horseradish peroxidaseconjugated streptavidin (0.3 g/ml), and serial diluted concentrations of PSGL-1/ chimera were co-incubated in sLe x /BSA-coated ELISA microtiter wells, for 4 h, at room temperature. After extensive washing with PBS/Tween 20 (0.05%), P-selectin binding to sLe x /BSA was revealed by the addition of 0.67 mg/ml o-phenylenediamine dihydrochloride (Sigma) and 0.16‰ H 2 O 2 . The chromogenic reaction was stopped by the addition of 50 l of 3 M H 2 SO 4 /well. Absorbance was read at 490 nm. IC 50 values of PSGL-1/ glycoforms on P-selectin/ chimera binding to sLe x /BSA were calculated using the GraphPad Prism 3.0 software (GraphPad Software Inc., San Diego).
The involvement of core-2 O-glycans linked to Thr-57 in the presentation of Le x , sLe x , and CLA was examined by comparing the expression levels of Le x , sLe x , and CLA on PSGL-1/, C2F7PSGL-1/, and C2F7PSGL-1T57A/ chimera adsorbed on 10.0-m polystyrene microspheres (2.5 ϫ 10 7 microspheres/ml, Polysciences Inc., Warrington, PA) coated with goat anti-human IgM (Caltag). Microspheres (15 l/sample) were washed twice with 0.1 M borate buffer, pH 8.5, and incubated overnight with 100 l of goat anti-human IgM (100 g/ml, Caltag) with end-to-end rotation. Microspheres were then washed, blocked for 1 h at 20°C with 0.1 M borate buffer, pH 8.5, containing 2% BSA, and incubated for 1 h at 4°C with 50 l of PSGL-1/ chimera (50 g/ml). The expression of Le x , sLe x , and CLA on PSGL-1 chimera was detected with the anti-sLe x mAb CSLEX-1 or the anti-CLA mAb HECA-452 and PE-labeled secondary antibodies (Dako). Microspheres were then washed and fixed with 1% paraformaldehyde in PBS. PE fluorescence was assessed on a Cytomics FC 500 cytofluorimeter (Beckman Instruments) by analyzing the fluorescence of 10,000 microspheres.
Cell Adhesion Assays-Adhesion assays were performed as described previously (26,63). A laminar flow was generated in a parallel plate flow chamber (GlycoTech Corp., Rockville, MD) mounted on a glass coverslip covered with a confluent monolayer of transfected CHO dhfr Ϫ cells or coated with L-selectin/ chimera (1.0 g diluted in 50 l of 0.1 M borate buffer, pH 8.5), or recombinant E-selectin or P-selectin (0.25 g in 50 l of borate buffer, R&D Systems, Minneapolis, MN), or wild-type PSGL-1/ versus PSGL-1T57A/ chimera (0.75 g in 50 l of borate buffer) adsorbed for 1 h at 37°C on coverslips coated with rabbit anti-human IgM antibody (Dako, 1 g in 50 l of 0.1 M carbonate buffer, pH 9.4, with a 75-mm 2 surface). Neutrophils, 300.19-L cells, or K-562-P cells (0.5 ϫ 10 6 /ml in RPMI 1640, 1% FCS) were perfused through the chamber using a syringe pump (Harvard Apparatus, Indulab AG, Gams, Switzerland) for 6 min, at room temperature, under a constant shear stress. Leukocyte interactions with CHO dhfr Ϫ cells were recorded during the entire period of cell perfusion with a phase contrast microscope (Leica Leitz DM IL, Renens, Switzerland), a high resolution Sony CCD-IRIS video camera (Japan), and an S-VHS recorder (Panasonic MD 830, Telecom, Lausanne, Switzerland). Leukocyte interactions with transfected CHO dhfr Ϫ cells or chimeric molecules were analyzed by manually tracking the motion of individual cells for 2-3 min. Leukocyte rolling interactions with CHO dhfr Ϫcells were considered for the analysis when the interaction time was Ͼ2.0 s and when the distance traveled by leukocytes during observation periods (20 s) was Ͼ1 cell diameter. Rolling velocities were determined with a digital image analysis system (Mikado software, GPIL SA, Martigny, Switzerland) and a Power-Macintosh 8600/200 workstation equipped with a Scion LG-3 board (Scion, Frederick, MD) (63). Cell interactions were analyzed from videotapes at 2-5 min of perfusion. Experiments were performed in triplicate under constant shear stress. Cell recruitment was determined in 32-64 microscopic fields (0.08 mm 2 ) for each tested condition. Rolling velocities illustrated in Figs. 3c, 5d, 6b, and 8b were measured in the direction of flow by tracking individual cells every 0.25 s, for 1-20 s, using digitized images of 0.7-mm 2 microscopic fields. 194 -894 independent determinations of cell rolling velocity were measured for each tested condition. Frame-by-frame velocity data obtained by tracking cells every 0.032 s, within 0.7-mm 2 microscopic fields, are illustrated in Fig. 4a and were used to assess the mean velocity Ϯ S.D. of each tracked cell over 2.7-s observation periods. The mean velocity of frame-by-frame tracked cells was included between percentile 40 and 60 of the velocity of each cell population illustrated in Fig. 3c. The S.D. value of the mean velocity of each tracked cell was then used to calculate the mean S.D. of cell rolling velocities of each cell population. The mean S.D. was used as an indicator of the variation of cell rolling velocity. 93-131 independent determinations of frame-by-frame velocity were measured for each tested condition. The specificity of L-, P-, and E-selectin-dependent cell interactions was indicated by the complete inhibition (Ͼ95%) of cell rolling by LAM 1-3, WAPS 12.2, and H18/7 mAbs. Isotype-matched mAbs were used as controls. Cell recruitment was not supported by mock-transfected CHO dhfr Ϫ cells. Adhesion assays performed at various shear stress (0.5-4.0 dynes/cm 2 ) indicated that 300.19-L and K-562-P cell recruitment on PSGL-1 was optimal at 1.5-2.0 dynes/cm 2 (data not shown). Experiments analyzing neutrophil rolling on CHO dhfr Ϫ cells expressing PSGL-1 were performed in the presence of the adhesionblocking anti-CD18 mAb TS1/18 (Endogen, Woburn, MA) to prevent CD18-dependent neutrophil arrest on CHO dhfr Ϫ cells. 300.19-L, CHO-P, K-562-P, and K-562-E cells did not arrest on CHO transfectants.
CHO-C2F4PSGL-1, CHO-C2F7PSGL-1, and CHO-C2F4F7PSGL-1 used to perform adhesion studies slightly differed in Le x , sLe x , and PSGL-1 expression levels (Fig. 1). The impact of the expression levels of sLe x on the ability of CHO-C2F7PSGL-1 to roll on P-, L-, or E-selectin was examined by performing rolling adhesion assays on recombinant P-selectin (0. 25  Statistical Analysis-Analysis of variance and the Bonferroni multiple comparison test or the Kruskal-Wallis non-parametric analysis of variance test was used to assess statistical significance of differences between groups. The non-parametric Mann-Whitney test was used to compare the medians of two unpaired groups. p values Ͻ 0.05 were considered as significant.
Because immunophenotypic analysis (Fig. 1)  These results indicate that FucT-IV and -VII cooperate in regulating L-and P-selectin-dependent cell rolling velocity, slowest rolling velocities, and maximal cell recruitment being observed when PSGL-1 O-glycans were proximally and distally fucosylated. Rolling velocities were significantly slower on CHO-C2F7PSGL-1 cells than on CHO-C2F4PSGL-1 cells (p Ͻ 0.001), suggesting that the generation of sLe x and CLA by FucT-VII on PSGL-1 is more efficient than Le x and CLA synthesis by FucT-IV in supporting L-and P-selectin-dependent cell rolling.
Stabilization of L-selectin-dependent Rolling on PSGL-1; Role of FucT-IV and -VII-The stability of rolling velocities on PSGL-1, under a constant shear stress of 1.5 dynes/cm 2 , was examined by measuring distances traveled by tracked cells within successive video frames (0.032 s) for 2.7 s in the flow direction. The mean velocity of tracked cells was included between percentile 40 and 60 of velocity curves illustrated in Fig. 3c. Each increase in velocity of tracked cells is represented in Fig. 4a by a peak and each decrease  The distribution of travel distances illustrated in Fig. 4b was assessed by measuring cell displacements within successive video frames (100 -128 determinations for each cell category). Data obtained for each cell category were pooled and illustrated in Fig. 4b. Cell displacements observed among compared cell categories were significantly different (p Ͻ 0.0001). Cell displacements were strongly affected by the pattern of PSGL-1 fucosylation. A higher percentage of 300.19-L cells rolled Ͼ3.0 m, within a video frame, on CHO-C2F4F7 and on CHO-C2F4PSGL-1 (76.0 and 59% respectively, number of determinations 119 and 100) than on CHO-C2F7PSGL-1 or on CHO-C2F4F7PSGL-1 (28 and 18%, respectively, n ϭ 128 and 100; Fig. 4b).
Regulation of E-selectin-dependent Rolling on CHO Transfectants; Role of FucT-IV, -VII, C2GnT, and PSGL-1-The role of PSGL-1, C2GnT, FucT-IV, and -VII in supporting K-562-E cell and neutrophil rolling on CHO dhfr Ϫ cell monolayers was studied under a constant shear stress of 1.5 dynes/cm 2 (Fig.  5a). CHO dhfr Ϫ transfectants expressed similar levels of Le x and/or sLe x and/or CLA and/or PSGL-1. In contrast to observations made with neutrophils (Fig. 5c) indicate the following: 1) that both FucT-IV and -VII generate carbohydrate determinants that efficiently support E-selectindependent rolling; 2) that core-2 O-glycans attached to PSGL-1 are not required to present Le x and sLe x carbohydrates and to support cell rolling on CHO transfectants; and 3) that the co-expression of both FucT-IV and -VII in CHO dhfr Ϫ-PSGL-1 transfectants does not increase cell recruitment over that observed on cells expressing only one fucosyltransferase.
As E-selectin is physiologically expressed by adherent endothelial cells, we examined if similar results were observed when flowing CHO transfectants rolled on recombinant Eselectin. CHO transfectants expressing FucT-IV and/or -VII were similarly recruited on E-selectin, and PSGL-1 expression was not required to support E-selectin-dependent rolling three experiments). These results are in contrast with observations made with neutrophils, 300.19-L cells, and K-562-P cells that showed slower rolling velocities on PSGL-1 glycoforms fucosylated by FucT-VII (Fig. 3).
Rolling velocities of CHO-P cells were significantly faster on C2F7PSGL-1T57A/ than on C2F7PSGL-1/ chimera (median rolling velocity on C2F7PSGL-1T57A/: 36.6 m/s, range 0.  Fig. 8). The lack of 300.19 cell recruitment on C2F7PSGL-1T57A/ chimera confirms that core-2 O-glycans attached to Thr-57 are essential to support L-selectin-dependent rolling and indicate that they support also ϳ80% of P-selectin-mediated rolling. In contrast, they may play a lesser role in mediating E-selectin-dependent rolling, cell recruitment being only weakly decreased (by ϳ24%) on C2F7PSGL-1T57A/ chimeras.
Location of sLe x , CLA, and Le x Epitopes on PSGL-1; Role of Thr-57-Because experiments illustrated in Fig. 8 showed that an important part of E-selectin-mediated rolling does not depend on core-2 O-glycans linked to Thr-57, we examined if sLe x /CLA residues are expressed on C2F7PSGL-1T57A/ chimera. In keeping with the results of adhesion assays illustrated in Fig. 8, sLe x residues, identified by CSLEX-1 mAb, and CLA, recognized by HECA-452 mAb, were detected on both C2F7PSGL-1/ and C2F7PSGL-1T57A/ chimera indicating that O-glycans attached to Thr-57 are not the sole lactosamine chains decorated by sLe x and CLA (Fig. 9). DISCUSSION Core-2-O-glycans attached to Thr-57 and terminated by sLe x determinants play a critical role in supporting L-selectin and P-selectin interactions with PSGL-1 (20,26,27,(71)(72)(73). Although the role of human FucT-VII and of C2GnT in supporting leukocyte rolling was well characterized, the involvement of human FucT-IV in regulating PSGL-1 interactions with L-selectin and the potential cooperation of human FucT-IV and -VII in generating selectin ligand activity on PSGL-1 had not been examined previously. Results presented here indicate the following: 1) that both human FucT-IV and -VII confer to PSGL-1 the ability to support L-selectin-, P-selectin-, and E-selectin-dependent rolling; 2) that the contribution of these two human fucosyltransferases is unequal, FucT-VII playing a predominant role in regulating L-selectin-, P-selectin-, and E-selectin-dependent rolling on PSGL-1; and 3) in addition, we show that core-2 O-glycans attached to Thr-57 are not the sole E-selectin-binding sites on PSGL-1, additional binding site supporting Ͼ75% of E-selectindependent rolling interactions with PSGL-1.
FucT-IV controls the synthesis of Le x residues by mediating the proximal or internal ␣1,3-fucosylation of polylactosamines linked to PSGL-1, whereas FucT-VII preferentially mediates the distal fucosylation of mono-or polylactosamines (38,50). The importance of FucT-IV and FucT-VII in generating Le x and sLe x determinants was indicated by observations made with cell lines or transfectants expressing FucT-IV or -VII cDNAs (18, 26, 40, 42, 43, 45-48, 57, 58, 62, 74) and with leukocytes isolated from mice deficient in ␣1,3-fucosyltransferases (18,(51)(52)(53). Our observations confirm that the transfection of CHO cells with human FucT-IV cDNA is sufficient to induce Le x expression, whereas CHO cells expressing FucT-VII generate sLe x (26, 38, 40 -42, 46, 47, 57, 58, 62, 71, 75). CHO dhfr Ϫ cells used in this study strongly expressed CLA and sLe x when they were transfected with FucT-VII cDNA, whereas they synthesized only low levels of sLe x when they expressed FucT-IV (Fig.  1). The generation of sLe x by CHO cells expressing FucT-IV has been observed previously (41,46,62) in CHO dhfr Ϫ cells but not in CHO Pro-5 cells. The biochemical basis of sLe x synthesis of CHO dhfr Ϫ cells still remains unexplained. In the present study, the lack of the inhibitory effect of V. cholerae neuraminidase on the ability of CHO-C2F4PSGL-1 cell to roll on L-, E-, and P-selectin indicates that sLe x expression is not required to support CHO DHFR Ϫ-C2F4PSGL-1 cell rolling on L-and P-selectin. In contrast, the exposure of CHO-C2F7PSGL-1 cells to neuraminidase strongly decreased cell recruitment on selectins underlining the critical role played by sLe x in regulating C2F7PSGL-1 interactions with selectins. A similar lack of requirement in sLe x for mediating C2F4PSGL-1 interactions with selectins was observed in adhesion assays performed with human neutrophils deficient in FucT-VII that do not synthesize sLe x and efficiently interact with P-and E-selectin (49).
Functional studies performed with chimeric selectins showed that the co-expression of PSGL-1 with human FucT-IV or -VII is required to support L-and P-selectin binding (Fig. 2), whereas FucT-IV or -VII expression was sufficient to generate E-selectin ligand activity on CHO dhfr Ϫ cells. The PSGL-1-independent reactivity of E-selectin most likely results from Eselectin binding to E-selectin ligand-1 (ESL-1) or other potential ligand(s) to which both FucT-IV and -VII confer a functional activity (47). L-selectin/ and P-selectin/ significantly bound to CHO dhfr Ϫ cells co-expressing human FucT-IV, C2GnT, and PSGL-1 (Fig. 2). However, the reactivity of Lselectin/ and P-selectin/ chimera with CHO-C2F4PSGL-1 cells was weaker than with CHO-C2F7PSGL-1 or CHOC2F4F7PSGL-1, indicating that the distal fucosylation of PSGL-1 by FucT-VII may generate binding sites that bind Land P-selectin more efficiently than the proximal and/or internal fucosylation promoted by FucT-IV (Fig. 2).
In keeping with results obtained with chimeric selectins (Fig.  2), we observed a contribution of both FucT-IV and -VII in regulating L-selectin-and P-selectin-dependent rolling on PSGL-1 (Fig. 3). The crucial role played by PSGL-1 was confirmed by the strong increase in cell recruitment observed in its presence (Fig. 3, a and b). The side by side comparison of L-selectin-and P-selectin-dependent rolling on PSGL-1 (Fig.  3a) and of PSGL-1-dependent rolling on L-selectin and P-selectin (Fig. 3b) indicated that FucT-IV and -VII unequally contribute to support L-and P-selectin-dependent rolling. Indeed, Land P-selectin-dependent cell recruitment was lower when C2PSGL-1 was fucosylated in CHO cells by FucT-IV, suggesting a predominant role for FucT-VII (Fig. 3). These adhesion studies also indicated that FucT-IV and -VII cooperate in sup- porting cell rolling on PSGL-1, cell recruitment being maximal when PSGL-1 is synthesized in cells that co-express both fucosyltransferases (Fig. 3, a and b). The critical involvement of FucT-VII in controlling leukocyte trafficking and in synthesizing selectin ligands was established in FucT-VII-deficient mice that exhibit an impaired E-selectin-and P-selectin-dependent leukocyte extravasation in inflammatory lesions and lymphocyte homing in peripheral lymph nodes (51,52) and more recently in a model of atherogenesis in apolipoprotein EϪ/Ϫ mice deficient in FucT-IV or -VII (76). A minor role was attributed to mouse FucT-IV in synthesizing P-selectin and E-selectin ligands, mice deficient in FucT-VII exhibiting only a weak residual leukocyte rolling along postcapillary venules and migration in inflamed tissues (52,53) and a subtle reduction of atherosclerotic lesions (76). Results presented here show that human FucT-IV can generate L-and P-selectin binding activities on PSGL-1. Further studies will be required, however, to establish the physiological role of FucT-IV in human leukocytes. Our results parallel observations made with leukocytes isolated from individuals presenting a missense mutation G329A of the FucT-VII gene that abolishes FucT-VII enzyme activity (54). Neutrophils isolated from these patients do not significantly express sLe x but conserve a significant E-selectin and an unchanged P-selectin ligand activity that was attributed to FucT-IV (49). However, the detection of a compensatory overexpression of FucT-IV mRNA (Ͼ100-fold increase) in human neutrophils deficient in FucT-VII suggested that it may have rescued their rolling capacity (49). In contrast to patients with leukocyte adhesion deficiency II, who present a large defect in fucosylation secondary to a disabled GDP-fucose transporter (77)(78)(79)(80), patients deficient in FucT-VII do not suffer from recurrent severe infections (79). These clinical observations suggested that the overexpressed FucT-IV activity may have compensated for the lack of FucT-VII activity (54).
Rolling velocities were significantly slower when PSGL-1 was fucosylated by FucT-VII instead of FucT-IV (Fig. 3c). In addition, L-selectin-dependent rolling was more stable on PSGL-1 glycoforms fucosylated by FucT-VII than by FucT-IV (Fig. 4). These results further indicate that FucT-VII serves as the main human ␣1,3-fucosyltransferase that regulates leukocyte rolling on PSGL-1. However, at a lesser extent than FucT-VII, FucT-IV may also participate in the regulation of leukocyte rolling on PSGL-1. Most interestingly, the analyses of cell recruitment and of rolling velocities on PSGL-1, illustrated in Fig. 3, show that human FucT-IV and -VII cooperate in regulating L-and P-selectin-dependent rolling. These results are in agreement with observations made in fucosyltransferase-deficient mice, which showed that leukocyte E-and P-selectin ligand activities result from the sum of unequal contributions provided by FucT-IV and -VII (52). Our results are, however, in contrast to the results of binding studies performed with PSGL-1 amino-terminal peptides expressing core-2 O-glycans containing sialylated and polyfucosylated polylactosamine, which exhibit a decreased binding affinity to P-selectin (73). Distinct requirements between peptides and the whole extracellular domain of PSGL-1 for binding to P-selectin as well as heterogeneity in fucosylated PSGL-1 molecules produced by CHO dhfr Ϫ cells might explain these results.
In agreement with previous studies, CHO dhfr Ϫ cells expressing human FucT-IV or -VII generated E-selectin ligands (26,42,46,47,58) that supported E-selectin-dependent rolling (Fig. 5). Of note, CHO dhfr Ϫ cells and CHO Pro-5 cells expressing human FucT-IV were shown previously (46) to differ in their ability to synthesize cell surface E-selectin binding activity. Human FucT-IV generated E-selectin ligands at the surface of CHO dhfr Ϫ cells but not on CHO Pro-5 cells (41,42,62). These observations and studies performed with leukocytic cell lines (43) indicated that the glycosylation phenotype of cell lines contributes to the ability of FucT-IV to synthesize E-selectin ligands. This may limit the physiological relevance of assays performed with cell lines. However, our observations made in CHO dhfr Ϫ cells are consistent with observations made with leukocytes obtained from patients with FucT-VII deficiency that exhibit a conserved leukocyte rolling on E-selectin despite a lack of FucT-VII activity (49). C2GnT co-expression with FucT-IV or -VII did not increase E-selectin-mediated rolling suggesting that E-selectin mainly interacts with N-glycosylated ligands modified with sLe x or Le x determinants, such as ESL-1, present on CHO cells expressing FucT-IV and/or -VII (34,47).
As several reports indicated the involvement of PSGL-1 in regulating leukocyte rolling on E-selectin (21-23), we examined the importance of PSGL-1 fucosylation by analyzing E-selectindependent rolling on PSGL-1/ chimera. As observed for Lselectin and P-selectin, FucT-VII played a predominant role in regulating E-selectin-dependent recruitment and rolling velocities on PSGL-1 (Fig. 6). As illustrated in Fig. 6a, FucT-IV and -VII cooperate in recruiting E-selectin-expressing cells on PSGL-1. These observations indicate that FucT-IV and -VII do not contribute equally in conferring E-selectin ligand activity to PSGL-1, and, as illustrated in Fig. 5; however, they do not differ in their ability to generate E-selectin ligand activities on CHO dhfr Ϫ cells.
Recombinant soluble PSGL-1 is a potent inhibitor of leukocyte migration that reduces leukocyte adhesion and migration into inflamed tissues and that exhibits broad anti-inflammatory properties (69,70,(81)(82)(83)(84)(85). Because fucosylation may affect the inhibitory activity of soluble PSGL-1, we compared the anti-P-selectin activities of PSGL-1/ glycoforms in a P-selectin/ binding assay to sLe x /BSA. The inhibitory activity of PSGL-1/ chimera was higher when the chimera was fucosylated by FucT-VII or both FucT-IV and -VII than by FucT-IV (Fig. 7). These results show that FucT-VII is more efficient than FucT-IV in conferring an inhibitory activity to recombinant PSGL-1 and suggest that maximal anti-P-selectin activity is obtained when the recombinant molecule is produced in CHO dhfr Ϫ cells co-expressing both fucosyltransferases.
The importance of core-2 O-glycans linked to Thr-57 in supporting L-selectin-, P-selectin-, and in particular E-selectinmediated rolling was examined on isolated C2F7PSGL-1/ and C2F7PSGL-1T57A/ chimeras. Consistent with previous studies (26,27), L-selectin-dependent rolling was abrogated in the absence of Thr-57. P-selectin-mediated rolling was strongly decreased on C2F7PSGL-1T57A/, but some residual rolling remained, suggesting that P-selectin may react with additional binding site(s) distinct from the major P-selectin-binding site located at the amino-terminal end of PSGL-1 (Fig. 8a) (20,28,71). Most importantly, E-selectin-dependent rolling was reduced by only 24% on PSGL-1T57A/ chimera indicating that additional E-selectin-binding sites, distinct from core-2 O-glycans linked to Thr-57, are expressed on wild-type PSGL-1. Because the expression of CLA and sLe x was correlated to the capacity of E-selectin to react with PSGL-1 (31,33,86,87), we examined whether PSGL-1s express these carbohydrate determinants at sites distinct from O-glycans attached to Thr-57. Flow cytometry analysis using HECA-452 and CSLEX-1 mAbs confirmed the presence of CLA and sLe x determinants on C2F7PSGL-1T57A/ chimera (Fig. 9).
In conclusion, the data presented here indicate that human FucT-IV and -VII have both the ability to confer to PSGL-1 ligand activity for L-selectin, P-selectin, and E-selectin. However, FucT-VII plays a predominant role. Most importantly, E-selectin-mediated rolling on PSGL-1 was only partially dependent on core-2 O-glycans attached to Thr-57, additional binding sites expressing CLA and sLe x mediating Ͼ75% of E-selectin-dependent rolling. These data enlighten the role played by human FucT-IV and -VII in regulating L-, P-, and E-selectin-mediated interactions with PSGL-1, a key initial event of the inflammatory response (1,38).