P-selectin Glycoprotein Ligand-1 Decameric Repeats Regulate Selectin-dependent Rolling under Flow Conditions*

P-selectin glycoprotein ligand-1 (PSGL-1) interacts with selectins to support leukocyte rolling along vascular wall. L- and P-selectin bind to N-terminal tyrosine sulfate residues and to core-2 O-glycans attached to Thr-57, whereas tyrosine sulfation is not required for E-selectin binding. PSGL-1 extracellular domain contains decameric repeats, which extend L- and P-selectin binding sites far above the plasma membrane. We hypothesized that decamers may play a role in regulating PSGL-1 interactions with selectins. Chinese hamster ovary cells expressing wild-type PSGL-1 or PSGL-1 molecules exhibiting deletion or substitution of decamers with the tandem repeats of platelet glycoprotein Ibα were compared in their ability to roll on selectins and to bind soluble L- or P-selectin. Deletion of decamers abrogated soluble L-selectin binding and cell rolling on L-selectin, whereas their substitution partially reversed these diminutions. P-selectin-dependent interactions with PSGL-1 were less affected by decamer deletion. Videomicroscopy analysis showed that decamers are required to stabilize L-selectin-dependent rolling. Importantly, adhesion assays performed on recombinant decamers demonstrated that they directly bind to E-selectin and promote slow rolling. Our results indicate that the role of decamers is to extend PSGL-1 N terminus far above the cell surface to support and stabilize leukocyte rolling on L- or P-selectin. In addition, they function as a cell adhesion receptor, which supports ∼80% of E-selectin-dependent rolling.

PSGL-1/GPIb␣ (PSGL/GP) was obtained by introducing a BamHI and a KpnI restriction site before and after the decameric repeats of PSGL-1 and by exchanging the 450-bp sequence encoding PSGL-1 decamers (Ala-118 to Thr-267) by the sequence of platelet GPIb␣ macroglycopeptide (C variant, Ser-312 to Thr-461) (24). Restriction sites were introduced in PSGL-1 sequence using the QuikChange site-directed mutagenesis kit (Stratagene, Amsterdam, The Netherlands). Sequences of the forward and reverse primers used to generate a BamHI restriction site before the first decameric repeat were: 5Ј-CCT GTC CAC GGA TGG ATC CGC TAT GGA GAT AC-3Ј and 5Ј-GTA TCT CCA TAG CGG ATC CAT CCG TGG ACA GG-3Ј. Primers used to generate a KpnI restriction site after the 15th decameric repeat were: 5Ј-CCA TGG AAC CTA CTG GTA CCA GAG GTC TGT TC-3Ј and 5Ј-GAA CAG ACC TCT GGT ACC CGT AGG TTC CAT GG-3Ј. The macroglycopeptide region of human platelet GPIb␣ was amplified by PCR, using genomic DNA extracted from bone marrow mononuclear cells as template with the following primers: forward 5Ј-GCT AGG ATC CTC ATG GTC CAC TGC TTC TCT AGA C-3Ј (containing a BamHI restriction site) and reverse 5Ј-CGA TGG TAC CGG TGG ATT CTA AGA GTG ATA CGG G-3Ј (containing a KpnI restriction site). The amplification was conducted for 30 cycles at 62°C using Pfu Turbo DNA polymerase (Stratagene). The 465-bp PCR product encoding the macroglycopeptide region of GPIb␣ was then digested with BamHI and KpnI and inserted in the restriction sites introduced by site-directed mutagenesis in PSGL-1 sequence.
PSGL-1 devoid of decameric repeats (PSGL/⌬DR) was obtained by amplifying the 351-bp 5Ј-sequence encoding the region located upstream of the decameric repeats with the forward primer: 5Ј-GCG GGA TCC AGC CAT GCC TCT GCA ACT CCT CCT-3Ј and the reverse primer 5Ј-GCT GGA TCC ACT TAC CTG CTG AAT CCG TGG ACA G-3Ј containing a splice donor site. The 408-bp 3Ј-sequence encoding PSGL-1 region located downstream of the decamers was amplified with the sense primer 5Ј-GCG GAA TTC CAG GTA CCA AAA GAG GTC TGT TC-3Ј containing a splice acceptor site and the antisense primer 5Ј-CGT TCT AGA GAG CTA AGG GAG GAA GCT GTG-3Ј. Both PCR products were inserted in pCR-Blunt vector (Invitrogen), then subcloned and ligated in pcDNA3 vector (Invitrogen). The donor and acceptor splice sites were separated by a 73-bp non-coding sequence.
Wild-type PSGL-1/ (WT PSGL/) chimera was generated by amplifying the sequence encoding the whole extracellular part of PSGL-1 (from Met-1 to Lys-308), as described (20). The cDNA sequence of PSGL/DR/ chimera, encoding decamers (from Ala-118 to Thr-267) coupled to IgM heavy chain , was constructed by deleting WT PSGL/ sequence encoding the propeptide and the sequence located upstream of the decameric repeats. A PvuII restriction site was introduced before PSGL-1 decamers by site-directed mutagenesis (Stratagene) using the following forward and reverse primers: 5Ј-GAA CCT GTC CAC GGA TCA GCT GGC TAT GGA GAT ACA GAC C-3Ј and 5Ј-GGT CTG TAT CTC CAT AGC CAG CTG ATC CGT GGA CAG GTT C-3Ј. Because a PvuII restriction site is present at the beginning of the propeptide sequence, the propeptide and the N-terminal sequence (from Leu-20 to Ala-117) were deleted by PvuII digestion of WT PSGL/. PSGL/ DR/ chimera cDNA sequence was constructed by ligating the PvuII-digested DNA fragments, allowing the fusion of PSGL-1 signal sequence to the decamer sequence. PSGL/⌬DR/ chimera was constructed by amplifying PSGL/⌬DR with the following primers: forward 5Ј-TCG CGA TAT CAA GCT TCT  CGA GCC ACC ATG CCT CTG CAA CTC CTC-3Ј and  reverse 5Ј-TAT AGA TAT CAT CGA TAC CTG AGA TGT GGT CTG GGG C-3Ј, the latter introducing a ClaI restriction site and a splice donor site allowing the fusion of the amplified sequence to IgM heavy chain.
Cells and Transfections-CHO dhFrϪ cells (ATCC number: CRL 9096) expressing FucT-VII and C2GlcNAcT-I (18) were stably transfected with cDNAs encoding wild-type or mutant PSGL-1 using TransIT reagent (Mirus, Madison, WI) according to the manufacturer's protocol. Transfected CHO cells were cultured in minimal essential medium ␣ containing 10% fetal bovine serum, hypoxanthine, and thymidine supplement (Invitrogen), 400 g/ml G418 and 200 g/ml Zeocin (Invitrogen). Individual clones expressing similar levels of the various forms of PSGL-1 and sLe x were isolated by limiting dilution and screened by flow cytometry analysis using KPL-1 and CSLEX-1 mAbs.
Immunophenotypic Analysis-PSGL-1, sLe x , and CLA expressions were assessed by incubating cells with saturating concentrations of appropriate unlabeled mAbs, followed by PEconjugated goat anti-mouse Ig. Cell staining with chimeric molecules was achieved by incubating chimeras with fluorescein isothiocyanate-conjugated rabbit anti-human IgM (20 g/ml). The concentrations of P-and L-selectin/ chimeras were measured by enzyme-linked immunosorbent assay (20). Chimeric proteins were suspended in RPMI 1640/1% fetal bovine serum, and binding specificity was assessed by its abrogation in the presence of 10 mM EDTA or anti-P-or -L-selectin blocking mAbs (WAPS 12.2 or LAM1-3). A total of 5000 cells was analyzed in each experiment. Flow cytometry was performed using a Cytomics TM FC 500 cytofluorometer (Beckman Coulter).
Epitopes present on chimeric proteins were identified by analyzing PSGL-1, sLe x , and CLA expressions by flow cytometry. 10-m polystyrene microspheres (2.5 ϫ 10 7 microspheres/ml, Polysciences, Eppelheim, Germany) were coated overnight at 4°C on a rotating wheel with goat anti-human IgM (Caltag), then blocked in phosphate-buffered saline/2% bovine serum albumin for 1 h and subsequently incubated with 2.5 g of chimeric protein in 50 l of phosphate-buffered saline (20). PSGL-1 epitopes were identified by incubating microspheres coated with captured chimeras with specific antibodies and PEconjugated goat anti-mouse Ig. Immunostaining was analyzed by flow cytometry.
Cell Rolling Assays-A laminar flow was generated in a parallel plate flow chamber (GlycoTech Corp., Rockville, MD) mounted on a glass coverslip (Polylabo SA, Plan-les-Ouates, Switzerland). Coverslips were coated with a confluent monolayer of transfected CHO cells or with recombinant P-selectin or L-selectin/ chimera (0.25 g) adsorbed on goat anti-human IgM heavy chain antibody (1 g in 50 l of 0.1 M borate buffer, pH 8.0, on an area of 38 mm 2 ) (20). Cells (0.5-1 ϫ 10 6 /ml in RPMI 1640/1% fetal bovine serum) were perfused through the chamber using a syringe pump (Harvard Apparatus, Indulab AG, Gams, Switzerland) under constant shear stress. Rolling cells were visualized using a phase contrast microscope (Leica Leitz DM IL, Renens, Switzerland) and a high resolution video camera (Sanyo CCD, Japan). Images were recorded on an S-VHS recorder (Panasonic MD830, Telecom Lausanne, Switzerland), and velocities were measured using a digital image analysis system (Mikado Software, GPL SA, Martigny, Switzerland) (23). Cell rolling interactions were analyzed from videotapes at 2-7 min of perfusion. Cell recruitment was counted in at least 24 microscopic fields of 0.25 mm 2 . Rolling interactions were analyzed when the interaction time was Ն1 s and when cell displacement during 20 s was Ն1 cell diameter. Rolling velocities illustrated in Fig. 4 were measured every 0.25 s by tracking individual cells in the flow direction within 0.25-mm 2 microscopic fields. Cell displacements illustrated in Fig. 5 were measured every 0.05 s, over 1-s observation periods. The median velocity of tracked cells was included between percentiles 40 and 60 of the velocity of cell populations illustrated in Fig. 4D. The S.D. value of the median velocity of each tracked cell, an indicator of the variation of cell rolling velocity, was used to calculate the mean Ϯ S.D. of each studied cell population. 156 -236 independent determinations of frame-by-frame displacements were measured for each tested condition. In each assay, CHO-WT and mutant PSGL-1 cells expressed similar levels of PSGL-1.
Statistical Analysis-Kruskal-Wallis nonparametric analysis of variance followed by Dunn's multiple comparisons test were used to determine statistical significance of difference between groups. p values Ͻ 0.05 were considered significant.
PSGL-1 Decamers Regulate L-and P-selectin-dependent Rolling-The involvement of decamers in regulating PSGL-1dependent rolling on P-or L-selectin was assessed under a constant shear stress of 1.5 dyn/cm 2 (Fig. 3). The ability of CHO cells expressing WT or mutant PSGL-1 to roll on L-or P-selectin was compared side by side (Fig. 3, A and B). Additional assays were performed in the reverse setting to examine L-or P-selectin-dependent rolling on WT or mutant PSGL-1 (Fig. 3,  C and D). These experiments mimic the interactions of flowing leukocytes with adherent leukocytes, activated platelets or inflamed endothelium.

JOURNAL OF BIOLOGICAL CHEMISTRY 28539
CHO transfectants (Fig. 3C). The mean number of rolling cells/min/mm 2 Ϯ S.E. on CHO-WT PSGL was 226 Ϯ 9. The deletion of decamers reduced rolling by 41% (137 Ϯ 13, p Ͻ 0.001), whereas substitution of decamers by GPIb␣ tandem repeats did not fully restore CHO cell rolling (153 Ϯ 13; 29% inhibition, p Ͻ 0.05). The strong decrease in cell recruitment observed in the absence of decamers was not due to a difference in PSGL-1 expression, as CHO-PSGL/⌬DR cell recruitment was compared with that of CHO-WT PSGL cells expressing the same level of cell surface PSGL-1.
Decamers Are Involved in Regulating Cell Rolling Velocity on L-and P-selectin-CHO-PSGL/⌬DR rolled much faster than CHO-WT PSGL cells on P-selectin (median: 9.  The distribution of travel distances illustrated in Fig. 5B was assessed by measuring cell displacements within successive 50-ms periods (156 -236 determinations). Rolling stability was correlated with shorter rolling distances (70% of 300.19-L cells rolled Ͻ4 m within 50-ms periods on CHO-WT PSGL). Longer rolling distances were observed in the absence of decamers. Substitution of decamers by GPIb␣ macroglycopeptide poorly stabilized cell displacements.
It should be noted that, because the lengths and structures of chimeric molecules (Fig. 7) and those of the extracellular domain of cell membrane PSGL-1 (Fig. 3) strongly differ, quantitative results obtained in Figs. 3 and 7 cannot be compared side by side (the length of cell membrane PSGL-1 is ϳ3-fold shorter than that of PSGL-1/ chimera bound to coverslip through a goat anti-human IgM antibody and PSGL-1/⌬DR expressed by CHO cells is ϳ5.5-fold shorter than PSGL-1/ ⌬DR/). The sole aim of experiments performed with chimeras was to identify PSGL-1 domains that contribute to support selectin-dependent rolling.

DISCUSSION
Selectins and PSGL-1 play a critical role in regulating leukocyte migration into inflammatory lesions (1). The involvement of core-2 O-glycans attached to PSGL-1 Thr-57 in regulating L-and P-selectin-dependent rolling has been clearly demonstrated (14,15,18,20,25). On the other hand, the role of decamers in regulating selectin-dependent rolling has not been previ-  ously examined. Data presented here lead to the following conclusions: 1) A primary function of decamers is to extend PSGL-1 N terminus away from the cell surface to present it to flowing leukocytes/activated platelets, inflamed endothelium/ adherent platelets, or leukocytes. They enhance the ability of PSGL-1 N terminus to support L-and P-selectin-dependent rolling. In addition, they play a critical role in stabilizing L-selectin-dependent rolling interactions. 2) Decamers directly interact with E-selectin and function as an adhesion receptor. In the absence of PSGL-1 N terminus, they support ϳ80% of E-selectin-dependent rolling. This suggests that they play a major role in mediating E-selectin-dependent rolling.
PSGL-1 decamers are rich in threonine residues and contain at least 50 potential O-glycosylation sites. Glycosylation of decamers contributes to almost two-thirds of PSGL-1 molecular weight. The extensive O-glycosylation (7) and high proline content of decamers confer a highly extended structure to PSGL-1, which was revealed by electron microscopy analysis (26). The rod-like structure of the mucin-like domain positions PSGL-1 N terminus far above cell surface and restricts the flexibility of the peptide backbone. This suggests that a major function of the mucin domain is to provide a rigid stalk to present PSGL-1 N terminus to flowing leukocytes/platelets, adherent platelets/leukocytes, or inflamed endothelium. This hypothesis was examined by comparing rolling interactions of CHO cells expressing WT PSGL-1, a PSGL-1 form devoid of decamers (PSGL/⌬DR) or a chimeric molecule in which decamers were replaced by the mucin domain of GPIb␣ (PSGL/GP) (24). CHO-PSGL/⌬DR cells suboptimally supported P-selectin-dependent rolling and very inefficiently L-selectin-dependent rolling (Fig. 3). Interestingly, CHO-PSGL/GP supported P-selectin-dependent rolling almost as efficiently as CHO-WT PSGL cells. This implies that a major function of decamers is to provide a stalk of sufficient length to extend PSGL-1 N terminus away from the leukocyte cell surface to support PSGL-1 interactions with P-selectin. On the other hand, CHO-PSGL/GP cells suboptimally supported L-selectin-mediated rolling, suggesting that PSGL-1 length is not the sole parameter that promotes L-selectin binding. Other biophysical mechanisms that regulate the formation of bonds between L-selectin and PSGL-1 may play a role.
Under hydrodynamic flow conditions, bonds form rapidly between selectins and their ligands and initiate cell tethering. In the flow direction, shear forces induce the dissociation of bonds at the back of the cell, while new bonds are formed at the front, resulting in cell rolling. A minimum level of fluid shear stress is required to sustain rolling interactions (27,28). Below the threshold level, fewer cells tether; they roll faster and begin to detach (28). Shear forces decrease L-selectin-PSGL-1 off-rates (catch bonds) between this threshold level and an optimal shear value, enabling the generation of long-lived tethers, which increases rolling stability and decreases rolling velocity (29,30). Above optimal shear stress, the increase in off-rates (slip bonds) accelerates rolling velocity and decreases rolling stability (30). Thus, L-selectin-mediated rolling is controlled by forcedependent alterations of bond lifetimes below and above a shear optimum. Protrusion of PSGL-1 N terminus into the bloodstream might be an important mechanism by which hydrodynamic flow could modulate PSGL-1 interactions. In the absence of decamers, shear forces generated on L-selectin-PSGL-1 N terminus could not be sufficient to promote efficient leukocyte rolling and to stabilize cell displacement. This may explain why 300.19-L cells rolled at high velocities, unstably traveling much longer distances on CHO-PSGL/ ⌬DR than on CHO-WT PSGL cells (Fig. 5). Substitution of PSGL-1 decamers by GPIb␣ was not sufficient to completely recover rolling stability. This may be explained by differences in the biophysical properties of WT PSGL and PSGL/ GP, in particular in their ability to form catch bonds promoting rolling stability.
By providing sufficient length and rigidity to PSGL-1, decamers play a critical role in stabilizing rolling interactions. Tandem repetition of decameric repeats has been an evolutionary mechanism used to change the biophysical properties of PSGL-1 by increasing the size of the mucin-like domain (21). A consequence is the exposure of PSGL-1 N terminus to shear forces, promoting its rapid binding to selectins and prompt initiation of leukocyte rolling at site of inflammation, without the need of exogenous modulators. The control of the lifetime of PSGL-1-selectin bonds by a threshold of fluid shear may also be important to prevent inappropriate leukocyte aggregation of flowing leukocytes exposed to low shear forces (31). A polymorphism in the number of decameric repeats has been observed in humans and other mammals (21,22). Variants with 16, 15, and 14 decameric repeats have been reported in the Caucasian population at frequencies of ϳ85, ϳ14, and ϳ1%, respectively. Interestingly, an association has been disclosed between expression of smaller alleles (B and C) and lower risk of developing cerebrovascular diseases (32). On the other hand, no correlation was observed with coronary heart disease or deep venous thrombosis (33)(34)(35).
Several parallels exist between PSGL-1 and GPIb␣. Both molecules exhibit a mucin-like domain that separates the ligand binding region from the cell membrane. PSGL-1 and GPIb␣ mucin-like regions are highly glycosylated and sialylated, rich in serine, threonine, and proline residues, and contain repeated tandem sequences with a variable number of repeats that account for polymorphism (21,22,24). PSGL-1, like GPIb␣, has N-terminal tyrosine sulfate residues that are important for ligand binding (15,18,36). Importantly, PSGL-1 and selectins, like GPIb␣ and its ligand the von Willebrand factor, require shear forces for binding. Furthermore, similarities are observed in the kinetics of PSGL-1 and GPIb␣ binding to their ligands. As described for PSGL-1 and selectins, catch-slip transitional bonds between GPIb␣ and von Willebrand factor regulate platelet adhesion (37,38). GPIb␣ interactions with von Willebrand factor mediate flow-enhanced platelet adhesion on subendothelium of disrupted arterial vessels, whereas platelets do not aggregate under low shear stress of sluggish flows. Interestingly, both PSGL-1 and GPIb␣ can interact with P-selectin (39). However, PSGL-1 and GPIb␣ differ by several aspects in their interactions with their ligands. Core-2 O-glycosylation and fucosylation of PSGL-1 and the presence of calcium is required to support PSGL-1 interactions with selectins, whereas GPIb␣ interaction with von Willebrand factor requires neither. Like GPIb␣ macroglycopeptide, decamers function as a rigid spacer extending the ligand-binding site far above the plasma membrane to support leukocyte rolling on L-or P-selectin. Importantly, we show here for the first time that decamers also function as an adhesion receptor by directly interacting with E-selectin and by cooperating with PSGL-1 N terminus in supporting E-selectin-dependent rolling (Fig. 7C).
The mucin domain can function as a rigid stalk that separates the ligand-binding site from the plasma membrane, as observed for GPIb␣, the human low density lipoprotein receptor (40) or fractalkine (41). However, it can also directly support adhesive interactions, as described for CD34, glycosylation-dependent cell adhesion molecule-1 (GlyCAM-1) or mucosal vascular addressin cell adhesion molecule-1 (MadCAM-1). These L-selectin ligands are highly fucosylated, sialylated, and sulfated (42). The sulfation of O-glycans allows the preferential interaction of these sialomucins with L-selectin (43). In contrast, the mucin-like domain of PSGL-1 is not sulfated. This might explain why L-and P-selectin do not interact with decamers, whereas they preferentially bind to N-terminal tyrosine sulfate residues and core-2 O-glycans linked to Thr-57. Because PSGL-1 sulfation is not required for E-selectin binding (19), E-selectin can interact with both the N-terminal selectin-binding site and with decamers (Fig. 7C). In the absence of PSGL-1 N terminus, E-selectin-dependent cell recruitment is decreased by only ϳ14% indicating that decamers play a major role in supporting E-selectin-dependent cell rolling. Interestingly, KPL-1 mAb binding to PSGL-1 N terminus partially inhibits E-selectin binding and slows down E-selectin-dependent rolling, whereas it strongly accelerates L-and P-selectin-dependent rolling and decreases cell recruitment. These results suggest that both PSGL-1 N terminus and decamers interact with E-selectin and promote slow rolling on E-selectin.
Taken together, data reported here show for the first time that PSGL-1 decamers play a crucial role in regulating leukocyte rolling on L-, P-, and E-selectin. They provide a sufficient length and an appropriate structure to PSGL-1 to interact with L-and P-selectin and to stabilize leukocyte rolling on L-selectin. Finally, PSGL-1 N terminus and decamers directly interact with E-selectin and promote slow rolling.