Endothelial Chemokines Destabilize L-selectin-mediated Lymphocyte Rolling without Inducing Selectin Shedding* 210

Chemokines presented on specialized endothelial surfaces rapidly up-regulate leukocyte integrin avidity and firm arrest through Gi-protein signaling. Here we describe a novel, G-protein-independent, down-regulatory activity of apical endothelial chemokines in destabilizing L-selectin-mediated leukocyte rolling. Unexpectedly, this anti-adhesive chemokine suppression of rolling does not involve L-selectin shedding. Destabilization of rolling is induced only by immobilized chemokines juxtaposed to L-selectin ligands and is an energy-dependent process. Chemokines are found to interfere with a subsecond stabilization of selectin tethers necessary for persistent rolling. This is a first indication that endothelial chemokines can attenuate in situ L-selectin adhesion to endothelial ligands at subsecond contacts. This negative feedback mechanism may underlie the jerky nature of rolling mediated by L-selectin in vivo.

endothelial ligands, and thereby stabilize secondary leukocyte adhesion, arrest, and subsequent extravasation (3). Whether chemokine signals transduced to a rolling leukocyte can also modulate the adhesive properties of its selectin or selectin ligands has not been demonstrated. Soluble chemoattractants have been shown, on the other hand, to trigger L-selectin shedding by cell surface endoproteolysis, implicating bloodborne chemokines as potential down-regulators of selectin-mediated rolling (4,5). Selectin rolling is a highly dynamic process that depends on subsecond coupling of tethers successively formed and broken at the cell front and trailing edge under disruptive shear forces (6,7). L-selectin rolling adhesions can be mediated by single tethers preferentially formed at microvillar surface projections where L-selectin is preferentially localized (8,9). Notably, L-selectin-mediated leukocyte rolling in vivo is extremely fast and jerky in nature (10), even at endothelial sites expressing high levels of L-selectin ligands such as the peripheral lymph node high endothelial venules (HEV) (10,11). Stabilization of L-selectin-mediated rolling, characterized by smooth rather than jerky motion, is critically dependent on the number of bonds simultaneously formed at each microvillar contact site (12). We therefore speculated that the jerky nature of L-selectin-mediated rolling in various in vivo settings might be caused by reduced L-selectin adhesiveness on leukocytes interacting with endothelium-displayed L-selectin ligands. In the present in vitro study, we found that several key chemokines, shown to be displayed on endothelial surfaces in vivo, are capable of strongly destabilizing the rolling activity of Lselectin in different types of leukocytes. Notably, this suppression of rolling was mediated by immobilized rather than soluble chemokines. Leukocyte capture to ligand, although normal, allowed the rapid encounter of immobilized chemokines codisplayed with L-selectin ligands, thereby eliciting the in situ reduction of L-selectin tether avidity to ligand. Surprisingly, destabilization of rolling was not the result of proteolytic Lselectin shedding. Furthermore, chemokine-mediated suppression of selectin rolling, although dependent on metabolic energy, did not involve intracellular signaling through the chemokine receptor. This is a first demonstration that endothelial chemokines may regulate selectin-mediated leukocyte rolling through a nonproteolytic G i -protein-independent process, prior to and independent of their triggering of integrin adhesiveness.

Cells
Human peripheral blood lymphocytes (PBL; obtained from healthy donors) were isolated from citrate-anticoagulated whole blood by dextran sedimentation and density separation over Ficoll-Hypaque. The mononuclear cells thus obtained were washed and further purified on nylon wool and plastic adherence as described (22). The resulting purified PBL consisted of more than 90% CD3 ϩ T lymphocytes and were cultured in lipopolysaccharide-free RPMI, 10% FCS for 15-18 h before use. Peripheral blood neutrophils were isolated from anticoagulated blood after dextran sedimentation and density separation over Ficoll-Hypaque (23). Murine B lymphocytes were derived from fresh splenocytes by positive immunoselection with mAb B220 followed by magnetic cell sorting purification, as described (24). The murine pre-B 300.19 cell line, stably expressing either native human L-selectin or tail-truncated L-selectin, was a generous gift from Dr. G. S. Kansas (Northwestern University, Chicago, IL) (25). Clones were maintained in RPMI 1640, supplemented with antibiotics, 10% FCS, 2 mM glutamine, and 0.1 M 2-mercaptoethanol. The human umbilical vein endothelial cell-derived line, ECV-304 (LS12), stably transfected with ␣-1,3-fucosyltransferase and N-acetylglucosamine 6-O-sulfotransferase and expressing functional sulfated L-selectin ligands (26) was a kind gift from Dr. R. Kannagi (Aichi Cancer Center, Nagoya, Japan). Cells were maintained in RPMI 1640, 10% FCS, 2 mM glutamine, and antibiotics.

Immunofluorescence Flow Cytometry
Indirect immunofluorescence was performed on washed cells that were suspended in PBS supplemented with 5% FCS and 5 mM EDTA. Cells were incubated at 4°C either with 10 g/ml L-selectin mAb DREG-200 or with pre-immune mouse IgG. Cells were washed, stained with fluorescein isothiocyanate-conjugated goat anti-mouse Ig, resuspended in PBS supplemented with 0.05% sodium azide, and immediately analyzed in a FACScan flow cytometer (BD PharMingen, Erembodegem, Belgium). To assess protein kinase C-induced L-selectin shedding, PBL or neutrophils were suspended in cell binding medium (see below) in the presence of protease inhibitors or with control carrier solution for 15 min at 25°C as described previously (20,21). Leukocytes were then treated with PMA (100 ng/ml, 2-10 min, 25°C) and immediately incubated at 4°C with 10 g/ml DREG-200, followed by staining with secondary mAb, as described above.

Preparation of Ligand-coated Substrates
Aliquots of GlyCAM-1, PNAd, or PSGL-1 were diluted in coating medium (PBS, supplemented with 20 mM bicarbonate, pH 8.5) and adsorbed onto polystyrene plates as described previously (27). Washed substrates were adsorbed with 0.1-4 g/ml amount of either intact or heat-inactivated chemokines for 3 h at 4°C. The anti-L-selectin mAb DREG-200 was mixed with either intact or heat-inactivated chemokines in the presence of 2 g/ml HSA and coated onto polystyrene plates overnight at 4°C. Neutralite avidin was diluted in PBS, 40 mM bicarbonate, pH 9.0, and adsorbed onto a polystyrene plate overnight at 4°C, followed by HSA blocking at 4°C. An equimolar mixture of biotinlabeled PSGL-1-derived selectin-binding peptide (2 ϫ 10 Ϫ2 nM) and either biotin-labeled SDF-1␣ or an inactive biotin-labeled control PSGL-1 peptide was diluted in cell binding medium (see below) and adsorbed for 4 h at 4°C on the avidin-coated plate. Substrates coated with avidin complexed with inactive biotin-labeled glycopeptides lacked any adhesive activity to all L-selectin-expressing leukocytes tested.

Laminar Flow Assays
Cell Tethering and Rolling Measurements-The polystyrene plate, on which purified ligand was adsorbed, was assembled in a parallel plate laminar flow chamber as described previously (28). Various leukocyte populations were washed in H/H medium (Hanks' balanced salt solution, 10 mM HEPES, pH 7.4, supplemented with 2 mg/ml bovine serum albumin) containing 5 mM EDTA, resuspended in cell binding medium (H/H medium supplemented with 2 mM CaCl 2 ) at 2 ϫ 10 6 cells/ml, and perfused at room temperature through the flow chamber at a rate generating wall shear stress of 0.1 dyn/cm 2 , as described (27). Once reaching the upstream side of the test adhesive substrate, the flow rate was elevated to generate a shear stress of 0.75, 1, or 1.75 dyn/cm 2 , and all cellular interactions were visualized at two different fields of view (each one 0.17 mm 2 in area) using a 10ϫ objective of an inverted phase contrast microscope (Diaphot 300, Nikon Inc., Tokyo, Japan). An imaging system was used for analysis of instantaneous velocities of leukocytes, WSCAN-Array-3 (Galai, Migdal-Ha'emek, Israel) as described previously (27). Accumulation of rolling leukocytes on the test fields was determined by computerized cell motion tracking. Adhesive interactions of transiently tethered cells were also manually analyzed as described (27). The frequency of rolling cells was defined as the number of cells out of the cell flux that initiated persistent rolling on the adhesive substrate lasting at least 3 s after initial tethering. Transient tethers were cells that attached only briefly to the substrate (Ͻ 0.2 s). Frequency of each category of tethers was expressed in % units; 1% unit measured at 1.75 dyn/cm 2 corresponded to tethering rate of 5.25 ϫ 10 Ϫ3 event ϫ cell Ϫ1 mm Ϫ1 s Ϫ1 .
To block GPCRs on target leukocytes, cells suspended in binding medium were preincubated for 45 min at 37°C with 0.5 g/ml soluble chemokines and perfused into the chamber. For blocking G i -protein signaling, PBL were cultured for 15 h at 37°C in culture medium alone or in the presence of 100 ng/ml PTX. For blocking JAK/STAT pathway stimulation by chemokines, lymphocytes were preincubated for 2 h at 37°C with 150 M JAK inhibitor, tyrphostin AG490, or with control Me 2 SO solution (0.1%, v/v). To inhibit metabolic energy without interfering with intact L-selectin rolling activity (29), lymphocytes were pretreated with 0.05% sodium azide for 2 min at room temperature, and perfused unwashed into the chamber. To adsorb SDF-1␣ on an endothelial monolayer, SDF-1␣ (100 ng/ml) was overlaid for 5 min on an L-selectin ligand-expressing ECV-304 cell monolayer assembled on the lower plate of the flow chamber. Unbound chemokine was removed by extensive washing. Overlaid chemokine remained bound to the monolayer throughout the assay, as confirmed in repetitive experiments after extensive washings.
Dissociation Kinetics of Individual Transient Tethers and Successive Rolling Tethers-Microkinetics of individual cells exhibiting jerky rolling on medium density GlyCAM-1 was analyzed on digitized video segments using the WSCAN-Array-3 software as described previously (27). Individual cell displacement analysis at 0.02-s intervals monitored changes in instantaneous cell velocities in the flow direction, depicted as successive transient pauses (27). The natural log of the number of all pauses or tethers formed by a given number of perfused leukocytes that remained bound after a given duration was plotted as a function of time, yielding tether dissociation curves with slopes representing Ϫk off .

Immobilized Chemokines Suppress Rolling Activity of L-selectin on Peripheral Blood
Lymphocytes-To study the effect of chemokine encounter by rolling lymphocytes on the dynamic stability of their selectin contact, the ability of L-selectin-expressing leukocytes to tether to and roll on surfaces bearing both purified L-selectin ligands and chemokines was investigated in a parallel plate flow chamber. Human PBL were perfused on a substrate coated with PNAd or GlyCAM-1, pro-totypic lymph node HEV-derived L-selectin ligands (15,16). At physiological shear stresses, freely flowing PBL rapidly tethered to and established continuous rolling on PNAd or Gly-CAM-1 coated alone (data not shown) or co-adsorbed with a nonfunctional secondary lymphoid tissue chemokine (SLC, CCL21) (Fig. 1A). However, PBL perfused over substrates containing identical densities of PNAd or GlyCAM-1 each co-adsorbed with functional SLC failed to accumulate on the ligands (Fig. 1A, and videos A and B available as supplemental information in the on-line version of this article). The distribution and intrinsic functional activity of GlyCAM-1 were unaffected by SLC adsorption, because human neutrophils, which lack CCR7 and do not respond to SLC (30), rolled normally on GlyCAM-1, regardless of SLC presence (data not shown). Indeed, pre-exposure of lymphocytes to soluble SLC at levels saturating its GPCR, CCR7, completely eliminated the suppressive effect of immobilized SLC on L-selectin-mediated lymphocyte rolling (Fig. 1A, and video C available as supplemental information in the on-line version of this article) without altering intrinsic L-selectin activity (Fig. 2).
Notably, short exposure of lymphocytes to saturating levels of soluble SLC or to SDF-1␣ (CXCL12), a chemokine ligand of the CXCR4 GPCR, expressed by most circulating PBL (31), did not alter their ability to tether to and roll on substrates containing GlyCAM-1 (Fig. 2), although this treatment triggered robust G i -protein-dependent ERK activation (data not shown). Importantly, the ability of SLC to destabilize lymphocyte rolling on a fixed GlyCAM-1 density decreased with reduced chemokine coating density (Fig. 2). Low density SLC, in contrast to high density chemokine, had a dual effect on L-selectin-mediated rolling on GlyCAM-1 co-immobilized with the chemokine; it partially reduced the fraction of tethered lymphocytes capable of rolling on GlyCAM-1 and increased the velocity of this rolling fraction. Thus, the rapid suppressive activity of SLC on L-selectin-mediated rolling adhesion depends on both the mode and density of chemokine presented to lymphocytes in juxtaposition to the L-selectin ligand to which they are tethered.
Chemokine Destabilization of L-selectin Rolling Does Not Involve Selectin Shedding or G i -protein Signaling-L-selectin rolling velocity can be affected by proteolytic shedding of the selectin by a cell surface protease (21). Interestingly, although soluble chemoattractants induce L-selectin shedding in myeloid cells (5), soluble SLC or SDF-1␣ failed to induce L-selectin shedding in PBL (Fig. 3A). Furthermore, the suppressive activity of SLC on rolling of T cells pretreated with the potent protease inhibitor Ro31-9790, confirmed to significantly block PMA-induced L-selectin shedding in lymphocytes (Fig. 3A, right panel), could not be rescued by the inhibitor (Fig. 3B) or by another hydroxamic acid-based L-selectin sheddase inhibitor, KD-IX-73-4 (data not shown). The protease inhibitor also did not augment L-selectin-mediated rolling of PBL on substrate coated with GlyCAM-1 alone (Fig. 3B), suggesting that spontaneous L-selectin shedding in lymphocytes does not take place during rolling on purified ligand, as reported previously for neutrophils (21,32). Indeed, L-selectin lacking a protease recognition site has been reported to retain wild type activity when expressed in a lymphocyte line (33). L-selectin suppression was also induced by immobilized SDF-1␣, and, as with SLC, suppression could not be rescued by inhibition of L-selectin shedding (Fig. 3B). Thus, chemokine destabilization of Lselectin-mediated rolling did not involve in situ L-selectin shedding on tethered lymphocytes.
Immobilized chemokines signal to target lymphocytes within subseconds of contact by binding their seven spanner receptors and activating their associated heterotrimeric G i -proteins (2). Surprisingly, however, PTX inactivation of these G i -proteins on PBL, which did not interfere with intrinsic L-selectin adhesive activity (Fig. 3C), also had no effect on chemokine suppression of L-selectin rolling (Figs. 3C and 4A). Consistent with the G-protein independence of this process, an SDF-1␣ mutant, P2G, with retained affinity to the SDF-1␣ receptor, CXCR4, but lost signaling capacity (17,34) and GPCR internalization activity (data not shown), fully reproduced the suppressive activity of native SDF-1␣ on intact or PTX-treated lymphocytes (Fig.  4A). Nevertheless, destabilization of L-selectin-mediated rolling by chemokine required metabolic energy, because lymphocyte pretreatment with low levels of sodium azide could rescue chemokine-induced suppression of L-selectin rolling, whereas it did not affect spontaneous L-selectin rolling (Fig. 4A). The suppressive activity of chemokines on L-selectin-mediated rolling appeared GPCR-specific, because rescue of rolling suppression by immobilized SLC could be achieved only by pre-exposure of lymphocytes to saturating levels of soluble SLC or to a second CCR7 ligand, ELC (CCL19) but not to SDF-1␣ (Fig. 3C). Taken together, these results suggest that chemokine destabilization of L-selectin-mediated rolling, although receptor-medi- For SLC pretreatment, lymphocytes were preincubated with the chemokine (0.5 g/ml) in binding medium for 45 min. Data points represent the means Ϯ range of flow experiments performed at a shear stress of 1.75 dyn/cm 2 taken in two fields of view. Mean rolling velocities Ϯ S.E. are indicated near respective accumulation plots. Representative segments of video frames taken at t ϭ 5 s, which depict lymphocytes interacting with GlyCAM-1 under the three experimental conditions mentioned above, are shown at the right panels. Blue images depict continuously rolling lymphocytes; green images depict transiently tethered lymphocytes. The experiment shown is representative of six independent experiments using different donor lymphocytes. ated and energy-dependent, did not involve G i -protein signaling by the chemokine receptor and was not the result of L-selectin shedding.
Leukocyte tethering to immobilized mAbs is inefficient under flow and is not followed by rolling adhesions, but serves as a sensitive measure of antigen density or availability on the surface of the tethered leukocyte at subsecond contact sites (9,17). To investigate whether chemokine suppression of L-selectin rolling involves interference with local surface density of L-selectin at these tether sites, SDF-1␣ was co-immobilized with an anti-L-selectin mAb and its effect on lymphocyte tethering to the mAb was determined. Consistent with such interference, immobilized SDF-1␣ and SLC (data not shown), but not their soluble counterparts, strongly suppressed the ability of T lymphocytes to tether to anti-L-selectin mAb (Fig. 4B). In contrast, SDF-1␣ dramatically augmented lymphocyte tethering to an immobilized ␣ 4 integrin-specific mAb (Fig. 4B), consistent with its ability to induce in situ VLA-4 clustering at lymphocyte-substrate contact zones (17). Similar to the effects of SDF-1␣ on suppression of L-selectin rolling on authentic ligand (Fig. 4A), intact signaling capacity through the GPCR was not essential for the chemokine to suppress L-selectin binding to mAb, because suppression was insensitive to PTX pretreatment and could be induced by the non signaling SDF-1␣ mutant, P2G (Fig. 4B). Thus, lymphocyte binding to both authentic carbohydrate ligands and to an L-selectin-binding mAb was similarly sensitive to suppression by immobilized chemokines.
To further understand the specific requirement for immobilized chemokines in destabilization of L-selectin rolling, anti-CXCR4 mAb was immobilized alone or with SDF-1␣ (Fig. 4C). Specificity of the assay was confirmed by the ability of SDF-1␣, but not of SLC, to augment CXCR4-dependent lymphocyte adhesion to immobilized anti-CXCR4 mAb (Fig. 4C). Notably and consistent with its inability to destabilize L-selectin-mediated rolling (Fig. 2), soluble SDF-1␣ did not augment or suppress CXCR4-dependent mAb adhesion in this assay (Fig. 4C). On the other hand, and consistent with its suppressive effects on L-selectin rolling and L-selectin mAb binding (Fig. 4, A and  B), P2G could significantly augment CXCR4-dependent PBL adhesion in this assay (Fig. 4C). SDF-1␣-augmented PBL adhesion to anti-CXCR4 mAb was also PTX-insensitive (Fig. 4C) and azide-sensitive (data not shown), reminiscent of SDF-1␣ suppression of L-selectin rolling (Fig. 4A). Thus, the ability of SDF-1␣ to destabilize L-selectin rolling of PBL correlated with ability to augment CXCR4-dependent binding of these PBL to immobilized CXCR4-binding mAb. Assuming that this binding depends on locally elevated densities of CXCR4 on the PBL surface at the site of immobilized CXCR4 ligand, SDF-1␣, it appears that local clustering of CXCR4, induced by immobilized, but not by soluble SDF-1␣, underlies the ability of SDF-1␣ to suppress L-selectin-mediated adhesion at subsecond contacts.

Chemokine Suppression of L-selectin Rolling Is Shared among Different Leukocytes and Requires GPCR Recognition-
The destabilizing effects of chemokines on L-selectin-dependent rolling were not restricted to human PBL. The ability of murine B lymphocytes to establish rolling following tethering to GlyCAM-1 was similarly abolished in the presence of immobilized SDF-1␣ or B cell attracting chemokine 1 (BCA-1, CXCL13), both potent B cell chemokines (Fig. 5, A and B). As in PBL, soluble chemokines did not suppress L-selectin rolling in these cells, but could selectively rescue suppression of rolling by immobilized counterparts (Fig. 5B and data not shown). The suppressive effects of chemokines on L-selectin rolling were also not restricted to endothelial L-selectin ligands or to lymphoid cells, as they could be also observed with neutrophils interacting with a nonendothelial L-selectin ligand, PSGL-1 (Fig.  5C). Interleukin 8 (IL-8, CXCL8), a key chemokine implicated in neutrophil recruitment to inflamed endothelial sites (35), strongly suppressed neutrophil rolling on PSGL-1 (Fig. 5C). PSGL-1 immobilized with inactive chemokines or with SLC supported efficient rolling of neutrophils (Fig. 5C), consistent with low expression levels of the SLC GPCR, CCR7, on these leukocytes (data not shown). SLC strongly suppressed, however, L-selectin-mediated rolling of PBL on PSGL-1 (data not shown). Thus, the correct GPCR on a target leukocyte is required for its ligand to suppress L-selectin-mediated rolling.
Interestingly, the L-selectin rolling activity of PSGL-1 did not depend on its native mucin scaffold, because short PSGL-1-derived peptides, corresponding to the NH 2 -terminal selectin-binding region of PSGL-1, supported efficient L-selectinmediated lymphocyte rolling when immobilized through a biotin spacer on an avidin-coated substrate (Fig. 6A), compa- rable with that of native PSGL-1 (Fig. 6B). Notably, when SDF-1␣ was co-adsorbed with the biotinylated PSGL-1 peptide on an avidin-coated substrate through its non GPCR-binding COOH terminus, it could efficiently suppress L-selectin-dependent PBL rolling on the PSGL-1 peptide, through its functionally exposed NH 2 terminus (Fig. 6A, NЈ-ter). In contrast, SDF-1␣, coupled to the avidin-coated substrate via its NH 2terminal CXCR4 binding site, lost its ability to destabilize L-selectin-mediated PBL rolling (Fig. 6A). Thus, a functionally intact GPCR binding domain is required for immobilized SDF-1␣ to suppress L-selectin-mediated rolling, whereas the mode of chemokine presentation, either direct, by a plastic surface (Fig. 6B) or by an avidin scaffold (Fig. 6A), does not affect the capacity of the chemokine to suppress L-selectin rolling.
Chemokine Suppresses Stable Rolling but Not Transient Tethers-To destabilize L-selectin rolling, chemokines must be properly displayed on endothelial surfaces. We next tested the ability of cell surface determinants to display SDF-1␣ in a conformation capable of suppressing L-selectin-mediated rolling. In light of low ligand expression on endothelial-derived cell lines, we made use of a human umbilical vein endothelial cell-derived line, ECV-304, stably transfected with ␣-1,3-fucosyltransferase and N-acetylglucosamine 6-O-sulfotransferase (26), two key regulators of L-selectin ligand biosynthesis on lymph node HEV and inflamed endothelia (36,37). A murine pre-B lymphocytic cell line expressing CXCR4 but lacking endogenous L-selectin and transfected with human L-selectin cDNA could establish persistent rolling on these ligand-expressing ECV-304 cells under physiological shear flow (Fig.  7A). As observed with purified proteins, SDF-1␣ immobilized on the ECV-304 cell surface strongly suppressed L-selectinmediated rolling of the L-selectin transfected B cells, without interfering with initial capture (Fig. 7, A and inset). SDF-1␣ also strongly destabilized L-selectin rolling of these cells on purified GlyCAM-1 (Fig. 9A). In contrast to the pre-B transfectants, PBL captured only transiently to the same ligand-expressing ECV-304 monolayers and failed to establish rolling (Fig. 7A, inset). Strikingly, transient PBL tethers, although L-selectin-dependent, were completely resistant to SDF-1␣ suppression (Fig. 7A, inset). This result suggested that chemokines do not interfere with transient tethers mediated by weak L-selectin-ligand interactions. To further delineate this observation, we next tested how the suppressive effect of a chemo- kine varies with the density of L-selectin ligand on the substrate. Reminiscent of the inability of SDF-1␣ to suppress transient PBL tethering (Fig. 7A, inset), immobilized SLC ef-fectively suppressed L-selectin rolling of PBL observed on high or medium physiological GlyCAM-1 densities, but could not suppress either the formation or the duration of transient FIG. 4. Immobilized but not soluble SDF-1␣ reduces effective L-selectin density on tethered lymphocytes at local adhesive contacts while increasing CXCR4 density. A, effect of signaling capacity and context of presentation of SDF-1␣ on the frequency and type of tethers formed by PBL interacting with PNAd at a shear stress of 1.75 dyn/ cm 2 . PNAd was coated at 100 ng/ml, and either SDF-1␣ or the nonsignaling SDF-1␣ derivative, P2G, were coimmobilized with it. Where indicated, PBL were pretreated with either soluble SDF-1␣ (0.5 g/ml) or 0.05% azide, as described under "Experimental Procedures." B, frequency and type of tethers mediated by PBL perfused over substrates coated with L-selectin-specific mAb DREG-200 (at 0.1 g/ml) in the presence of inactive (Ϫ) or active SDF-1␣ or P2G. Tethers were measured at 1 dyn/cm 2 . For comparison, effect of SDF-1␣ on frequency of PBL tethers formed with surface-bound ␣ 4 integrin-specific mAb HP1/2 is shown. C, frequency and type of tethers mediated by PBL perfused at 0.75 dyn/cm 2 over surface-bound anti-CXCR4 mAb (0.1 g/ml) coated with inactive (Ϫ) or active SDF-1␣ or the P2G derivative (each at 2 g/ml). Note that co-immobilized SLC does not induce any adhesive activity in this setting. The scheme depicts a CXCR4-bearing microvillus tethered to a substrate coated with anti-CXCR4 mAb and SDF-1␣. The experiments shown in A-C are each representative of three. L-selectin-mediated PBL tethers to low density GlyCAM-1 (Fig. 7B). Taken together, these findings suggest that chemokine binding to a tethered leukocyte selectively suppresses L-selectin-mediated rolling adhesions without interfering with transient L-selectin tethers.
Chemokine Suppresses a Subset of High Avidity L-selectin Tethers Independently of Cytoskeletal Association of the Selectin-To gain further insights into the kinetics of GPCR-mediated suppression, the motion of individual T lymphocytes interacting with medium density GlyCAM-1 and SLC (Fig. 7B) was next compared by computerized image analysis (27). The duration of L-selectin tethers comprising leukocyte rolling motions was shown to be progressively shorter at reduced bond numbers within each tether (12). Thus, tether duration can serve as a sensitive reporter of effective L-selectin avidity at the contact zone. Upon initial tethering to the ligand-coated surface, captured lymphocytes continued to roll on the ligand through engagement of closely spaced successive tethers (Fig.  6A, right panel). Over 80% of these tethers dissociated from the ligand with a first order dissociation rate corresponding to a t1 ⁄2 of 19.5 ms (Fig. 8A, left panel, open circles). The remaining tethers lasted significantly longer, with a t1 ⁄2 of 43 ms (Fig. 8A,  left panel, closed circles). As reported earlier, lymphocytes interacting with GlyCAM-1 in the presence of immobilized SLC could not establish rolling (Fig. 8B, right panel) and engaged with the L-selectin ligand exclusively through short-lived tethers (Fig. 8B, left panel; t1 ⁄2 of 17.8 ms). PBL pretreated with soluble SLC engaged with GlyCAM-1 co-immobilized with SLC with microkinetics nearly identical to that of intact PBL interacting with GlyCAM-1 in the absence of SLC (Fig. 8, A and C,  left panels). Thus, persistent PBL rolling on GlyCAM-1 is mediated by a subset of prolonged tethers with a t1 ⁄2 of 45 Ϯ 2 ms, which make up about one fifth of all L-selectin-mediated tethers (Fig. 8, A and C). Chemokine suppression of rolling adhesions in this system is therefore associated with the ability to suppress this subset of nascent tethers by reducing their lifetime to a t1 ⁄2 shorter than 20 ms.
The cytoplasmic domain of L-selectin has been shown to regulate leukocyte rolling under shear flow through stabilizing association of L-selectin with the cell actin cytoskeleton (25,38). An L-selectin mutant lacking the carboxyl-terminal 11 residues of the cytoplasmic domain, with deficient cytoskeletal association, expressed on pre-B cells, can establish weak but persistent rolling on high density L-selectin ligands (39). Notably, SDF-1␣ could fully destabilize the rolling adhesions mediated by this tail-truncated L-selectin mutant on high density GlyCAM-1, despite of the very fast rolling mediated by this mutant (Fig. 9, A and B). Thus, chemokine destabilization of  6. GPCR binding domain is necessary for chemokine-suppressed L-selectin-mediated rolling. A, rolling of PBL on a PSGL-1derived biotinylated peptide (0.1 g/ml) co-immobilized with equimolar concentrations of control biotinylated peptide (Ϫ) or with biotinylated SDF-1␣ (ϩ). PSGL-1 peptides were labeled with a single biotin at their COOH terminus and were co-immobilized together with the biotinylated SDF-1␣ to substrates pre-coated with avidin (see "Experimental Procedures"). NЈ-ter, SDF-1␣ modified in its COOH terminus with biotin was anchored on avidin leaving its NH 2 CXCR4 binding domain functionally exposed. C'-ter, SDF-1␣ modified in its NH 2 terminus with biotin, and anchored on avidin leaving its COOH terminus functionally exposed. Where indicated, PBL were pretreated with 0. L-selectin-mediated rolling does not require an intact cytoplasmic domain of L-selectin, suggesting that the suppressive signals exerted by chemokine do not require L-selectin association with the cytoskeleton.

DISCUSSION
Selectin-mediated tethering and rolling are prerequisite for leukocytes to survey the endothelial lining for proadhesive and promigratory signals, primarily apical chemokines (3,10). Rolling leukocytes must integrate chemokine signals within subsecond contacts along the direction of flow (17,40,41). Here we suggest that apically displayed endothelial chemokines may not merely transmit integrin-activating signals to rolling leukocytes, but in fact directly modulate the rolling process itself, through an in situ reduction of selectin tether stability. Thus, rolling adhesions that allow a captured leukocyte to sample the endothelium for specific chemokines are subjected to a negative feedback mechanism by these very chemokines. Rather than being discrete and sequential events (42), reversible selectin interactions and chemokine receptor occupancy events appear to simultaneously operate at particular adhesive zones bearing immobilized chemokines juxtaposed to L-selectin ligands. As a result, selectin-mediated rolling, which has been predicted to increase encounter of endothelium-displayed chemokines (43), is in fact attenuated by this encounter. Attenuation of rolling is predicted to be more robust at sites of leukocyte interaction with high densities of L-selectin ligand, probably at endothelial regions within lymph node HEV enriched with L-selectin ligands (11). In addition, because the local density of chemokine on endothelial surfaces is heterogeneous (35), this attenuation mechanism may result in multiple dynamic outcomes. In regions of low chemokine density, L-selectin-mediated rolling is expected to be accelerated (Fig. 2), whereas L-selectin-mediated rolling on specific regions expressing high density chemokine is expected to be strongly suppressed (Fig. 2). This would cause a rolling leukocyte to detach from such sites, while allowing it to jerk and rebind ligand at an adjacent downstream sites. Furthermore, chemokine distribution on individual endothelial cells is nonuniform, as chemokines can be found in clusters on endothelial microvilli (35). These domains could be preferential sites of chemokine destabilization of L-selectin rolling. The jerky nature of L-selectin rolling is not controlled solely by chemokines. Anti-adhesive glycoproteins like CD43 (44), topological heterogeneity of both leukocyte and endothelial surfaces (45), as well as intrinsic properties of L-selectin bonds (46,47) can each contribute to the jerky nature of Lselectin-mediated rolling of leukocytes along various blood vessels. The existence of such multiple mechanisms for attenuating L-selectin rolling suggests that the jerky nature of L-selectin-mediated rolling is of major physiological significance. One possible outcome of such suppression of rolling could be to attenuate direct integrin activation by L-selectin, a process that depends on L-selectin ligation by ligand and bypasses chemokine regulation of leukocyte arrest on integrin ligands (48 -50).
Spontaneous and chemoattractant-induced proteolytic shedding of L-selectin was traditionally proposed as a major negative feedback mechanism of L-selectin-mediated leukocyte rolling (21,51). However, the G i -protein-independent chemokine destabilization of L-selectin rolling studied here did not involve L-selectin shedding, previously shown to involve activation of GPCR signaling (4,51). The insensitivity of chemokine suppression of rolling to PTX blockage of G i signaling, demonstrated here, also rules out the possibility that chemokines suppress L-selectin rolling through G i -protein-dependent phosphorylation of the L-selectin cytoplasmic tail (52). Indeed, even lymphocytes expressing tail-truncated L-selectin were sensitive to chemokine suppression of rolling. Similar to our finding of an accelerated L-selectin rolling induced by low level chemokine (Fig. 2), Campbell and co-authors (40) reported 2-fold faster L-selectin-dependent rolling of murine lymphocytes on PNAd co-immobilized with chemokines. The study attributed the accelerated rolling to chemokine blockage of L-selectin binding carbohydrates on the substrate. Our evidence that chemokines suppress L-selectin binding to mAb, an L-selectinbinding protein that lacks any selectin-binding carbohydrates, rules out the possibility of L-selectin ligand masking by chemokine. Furthermore, our finding that chemokines fail to suppress L-selectin adhesion to low density ligand (Fig. 7B) also rules out a direct blockage of ligand activity by chemokine. Instead, our data strongly suggest that chemokines induce rapid redistribution of both chemokine receptors and L-selectin at adhesive contact sites, possibly through extracellular or membranal associations of their receptors with juxtaposed Lselectin molecules. Recent electron microscopic analysis of the chemokine receptors for SDF-1␣ and RANTES (regulated on activation normal T cell expressed and secreted), CXCR4 and CCR5, respectively, in PBL has demonstrated that these GPCRs localize to lymphocyte microvilli (53), where L-selectin as well as ␣ 4 integrins are preferentially co-expressed (8,9,54,55). These observations suggest that these chemokine receptors and probably other GPCRs of endothelial chemokines, including CCR7, CXCR5 and CXCR1/2, receptors for SLC/ELC, BCA-1, and IL-8 or Gro␣ (CXCL1), respectively, may also be found on leukocyte microvilli. We propose a model whereby, upon binding immobilized chemokines juxtaposed to endothelial L-selectin ligands, the leukocyte-based GPCRs may cluster in proximity to L-selectin molecules, sterically hindering Lselectin clustering with high density ligand (Fig. 4C, inset). Such selectin/ligand clusters appear essential to stabilize a newly formed tether at subsecond endothelial contacts (12, 56 -58).
We have previously demonstrated that PBL GPCR occupancy by immobilized, but not by soluble chemokines, induces rapid ␣ 4 integrin avidity at adhesive contacts containing chemokine and integrin ligand (17). The present study suggests the reverse activity of GPCR occupancy by immobilized chemokines, i.e. interference with L-selectin avidity to ligand. The ability of chemokine to suppress L-selectin rolling is closely correlated with the ability of immobilized chemokine to augment binding of its GPCR to a GPCR-binding mAb on a countersurface (Fig. 4C). This suggests that the local density of the leukocyte GPCR at sites containing immobilized chemokine FIG. 8. Effect of immobilized chemokine on the microkinetics of L-selectin-mediated tethering and rolling of PBL on GlyCAM-1. A, the duration of instantaneous velocity drops (pauses) of 5-20 PBL continuously rolling on medium density GlyCAM-1 (100 sites/m 2 coated with inactive SLC) at a shear stress of 1.75 dyn/cm 2 was determined and used to calculate the dissociation kinetics of all pauses. Data points that fit a first order dissociation curve are connected by straight lines, with slopes equal to Ϫk off . The fraction of the total tethers that dissociated with the indicated k off values are shown. r, coefficient of correlation. Right panel, instantaneous velocities of a representative lymphocyte interacting with the GlyCAM-1 substrate. B, motion analysis of lymphocytes tethered to GlyCAM-1 (100 sites/m 2 ) coated with active SLC (4 g/ml). The k off value was determined as in A. A velocity pattern of a representative tethered lymphocyte detaching from the substrate shortly after initial tethering is depicted in the right panel. C, motion analysis and k off values of PBL pretreated with soluble SLC and allowed to tether to and roll on a substrate coated with GlyCAM-1 co-immobilized with SLC, identically as in B. Right panel, instantaneous velocities of a representative SLC-treated lymphocyte interacting with the substrate. Background Lselectin-independent tethering to the substrates was determined in the presence of EDTA. Values represent the mean of two independent experiments with a total of 70 transient events analyzed for each experimental group. must be rendered high to effectively destabilize selectin avidity to respective ligands. Notably, to suppress selectin rolling, immobilized chemokines do not need to activate G i -protein signaling or trigger GPCR internalization. We considered that chemokine-induced GPCR clustering could activate JAK/STAT signaling cascades independent of G i -protein signaling (59). However, blockage of JAK activation with the specific inhibitor tyrphostin AG490 did not block chemokine suppression of Lselectin rolling in PBL (data not shown). Notably, suppression of L-selectin rolling appeared to require specialized GPCRassociated machinery, because engagement of a non-GPCR cytokine receptor, IL-2 receptor, by high level IL-2 co-immobilized with L-selectin ligand had no effect on L-selectinmediated rolling of PBL. Nevertheless, suppression of L-selectin-mediated rolling by chemokine-occupied GPCRs required metabolic energy, consistent with an involvement of active cellular machinery in chemokine suppression of L-selectin function.
The new notion that chemokines can reciprocally modulate selectin and integrin adhesiveness predicts that a proper balance and coordinated timing of these opposite chemokine activities at leukocyte/endothelial contacts must be maintained to allow optimal conversion of selectin-mediated rolling into integrin-mediated leukocyte arrest. Chemokine recognition by a rolling leukocyte poses a regulatory hurdle in the propagation of multistep L-selectin initiated adhesive cascades, not previously realized. This novel activity of chemokines and its unique G-protein-independent mode of action should be considered when using specific chemokine antagonists to attenuate pathological L-selectin-mediated leukocyte recruitment at various target tissues.