Important Contributions of P-Selectin Glycoprotein Ligand-1-Mediated Secondary Capture to Human Monocyte Adhesion to P-Selectin, E-Selectin, and TNF-α-Activated Endothelium Under Flow In Vitro¹

Monocyte adhesion to the vascular endothelial lining and subsequent diapedesis are crucial events that occur during chronic inflammation, immune-mediated reactions, and atherosclerosis (1, 2, 3, 4). Previous in vitro studies have examined monocyte interactions with cytokine-activated endothelium and determined the adhesion molecules that mediate adhesion under defined flow conditions. Under defined laminar flow conditions, in vitro monocyte initial attachment, rolling, arrest, and transmigration across cytokine-activated HUVEC involved sequential and overlapping adhesion pathways, including the selectins (CD62L, CD62P), VCAM-1, VLA-4, β₂-integrin, ICAM-1, and platelet endothelial cell adhesion molecule-1 (PECAM-1, CD31) (5, 6).

Recent studies have identified leukocyte structures that present carbohydrate ligands for recognition by E- and P-selectins (reviewed in 7 . A glycoprotein described initially as a ligand for P-selectin, termed PSGL-1³ (8), was cloned and demonstrated subsequently to interact with both E- and P-selectin in some in vitro assays. PSGL-1 is a mucin-like homodimer consisting of two disulfide-linked subunits with an apparent molecular mass of 120 kDa each. PSGL-1 undergoes posttranslational modifications, including cleavage by paired basic converting enzyme, sulfation on one or more of the three tyrosines located in the anionic N-terminal region, and significant glycosylation, depending upon the enzymatic activities of different leukocyte cell lineages. PSGL-1 is expressed by essentially all blood leukocytes, including neutrophils, monocytes, and lymphocytes, and is glycosylated extensively with N-linked glycans and closely spaced O-linked glycans, a portion of which are modified with sialyl Lewis^x epitopes (9) and the HECA452 Ag (10, 11). In assays performed under flow, PSGL-1 has been reported to mediate adhesion of human neutrophils (12, 13) and T lymphocytes (13, 14, 15) to P-selectin and to be required for optimal attachment to E-selectin (16). L-selectin is expressed on neutrophils, monocytes, and most lymphocytes, and also has been proposed to present carbohydrate ligands to E- and P-selectin. Similar to PSGL-1, L-selectin carries sialyl Lewis^x-type structures and is localized on the tips of microvilli (17). However, more recent studies in vitro suggested that L-selectin does not serve as a ligand for E- or P-selectin during leukocyte adhesion to endothelium; rather, it is more likely that L-selectin serves to mediate leukocyte/leukocyte (interleukocyte) adhesion, which can greatly amplify leukocyte adhesion to and accumulation on endothelium in vitro (18).

Recent studies report that L-selectin participates in interleukocyte adhesion under flow, which promotes rapid and efficient adhesion of neutrophils (18, 19) and γ/δ T cells (20) to cytokine-activated endothelium. L-selectin on newly arriving leukocytes interacts with multiple ligands on leukocytes already stably adherent on the HUVEC surface, and thus mediates formation of strings of rolling cells (18) that greatly enhance leukocyte accumulation under defined laminar flow conditions. The ability of adherent leukocytes to form strings on endothelium or isolated adhesion molecules has been noted earlier for both neutrophils (21) and monocytes (5, 6). Subsequently, PSGL-1 has been shown to serve as one of the ligands for L-selectin to mediate neutrophil/neutrophil interactions (22), with blocking mAb inhibiting 60% of these interactions. In addition, several less well-defined leukocyte structures have been described to act as ligands for L-selectin, and thus amplify leukocyte capture on various substrates (18, 22). While the role of L-selectin in monocyte/monocyte (18) and monocyte/HUVEC interactions (5, 6) under flow has been explored, the role of PSGL-1 has yet to be examined directly for monocyte/monocyte and monocyte/endothelial adhesive (e.g., E- and P-selectin) interactions under flow conditions.

In this study, we have examined in detail the adhesive interactions of human blood monocytes with TNF-α-activated human endothelial cell monolayers, as well as the inducible endothelial cell adhesion molecules E-selectin and P-selectin, under conditions likely to simulate blood flow in microvessels in vivo. The experiments using mAb KPL1, which recognizes a sulfated N-terminal epitope (9, 13), demonstrate that PSGL-1 is the predominant monocyte ligand for P-selectin, but not E-selectin, and that PSGL-1, in conjunction with L-selectin, mediates crucial adhesive interactions for both monocyte/monocyte and monocyte/HUVEC adhesion, suggesting that targeting these adhesion pathways may provide beneficial therapies in chronic vascular disease settings.

Human rTNF-α was obtained from Genentech (S. San Francisco, CA) and was free of detectable endotoxin, as reported previously (6). A concentration of 20 ng/ml for 4 to 6 h gave optimal expression of VCAM-1, E-selectin, P-selectin, and ICAM-1 on HUVEC (6). HBSS with and without Ca²⁺ and Mg²⁺, M199, α-media, and HEPES were obtained from BioWhittaker (Walkersville, MD). FBS was obtained from HyClone (Urem, UT). All other chemicals were of the highest grade available from Baker Chemical (Phillipsburg, NJ). All buffers that came in contact with monocytes were purchased commercially, and all subsequent solutions were from sterile disposable plasticware to minimize endotoxin contamination.

A function-blocking murine mAb, KPL1 (IgG1, used as purified IgG at 20 μg/ml), recognizes the tyrosine-sulfated motif at the N-terminal portion of human PSGL-1, as detailed (13). Murine mAb directed to L-selectin (LAM1-4 (blocking), LAM1-14 (nonblocking), both IgG1) (23) were purified IgG (used at 10 μg/ml). Function-blocking mAb HEL3/2 (IgG1) (24) recognizes human E-selectin and was used as IgG (10 μg/ml). Function-blocking mAb HPDG2/3 (IgG1) recognizes P-selectin (8) and was used as IgG (10 μg/ml). Control mAb W6/32 (IgG2a) recognizes human class I Ag and is expressed at high levels on HUVEC and leukocytes, and does not alter leukocyte adhesion to cytokine-activated endothelium (6).

HUVEC were isolated from two to five umbilical cord veins, pooled, and established as primary cultures in M199 containing 20% FBS (5). Primary HUVEC cultures were passed serially (6) at a 1:3 split ratio and maintained in M199 containing 10% FBS, endothelial cell growth factor (50 μg/ml; Biomedical Technologies, Stoughton, MA), porcine intestinal heparin (100 μg/ml; Sigma, St. Louis, MO), and antibiotics. For use in the flow apparatus, HUVEC (passage 1) were plated at 80% confluence on 25-mm circular glass coverslips (No. 1 thickness; Thomas Scientific, Swedesboro, NJ) previously precoated overnight with human fibronectin (2 μg/cm²). HUVEC were allowed to reach confluence and were used in experiments within 24 and 72 h. CHO cells stably transfected with P-selectin (CHO-P) or E-selectin (CHO-E) have been reported previously (8). CHO-E and CHO-P cells were cultured in α-medium with 10% dialyzed FBS (HyClone) and plated on glass coverslips, as detailed above for HUVEC monolayers, and used when they achieved ∼90% confluence, as detailed earlier (25). Purified recombinant human E-selectin and P-selectin molecules have been described (8, 26) and were adsorbed to 25-mm-diameter glass coverslips using a stock solution of 10 μg/ml in 10 mM sodium bicarbonate, pH 9.5 (9, 27). Just before use in flow experiments, the selectin-coated coverslips were incubated at ambient temperature with 1% Tween-20 in DPBS containing Ca²⁺ and Mg²⁺ for 1 min to reduce nonspecific cell adhesion (21, 27).

Human monocytes were isolated from platelet pheresis residues by centrifugation on density gradients (LSM; Organon Teknika, Durham, NC), followed by counterflow centrifugation elutriation (28). Monocyte suspensions were >91% pure with 6 to 8% lymphocyte, <2% granulocyte, and essentially no platelet contamination, as determined by light scatter (FACScan; Becton Dickinson, Mountain View, CA) and cell surface Ag analysis with mAb directed to CD14, CD41, and CD61, and P-selectin (5, 6). L-selectin surface expression was not significantly changed following elutriation, and CD18 expression increased minimally in 10 separate preparations that were analyzed (5). Thus, monocytes isolated by this technique had minimal alterations in surface adhesion molecules that are sensitive markers for activation, implying that these cells were minimally perturbed. Isolated monocytes were resuspended in warm perfusion buffer (DPBS containing 0.75 mM Ca²⁺, 0.75 mM Mg²⁺, and 0.2% HSA) and were used immediately in flow studies. All buffers and reagents contained <10 pg/ml of endotoxin (E-toxate kit; Sigma).

The parallel plate flow chamber used in this study has been described in detail (5). Defined levels of flow are applied to the HUVEC monolayer by drawing perfusion media through the channel via a syringe pump (model 44; Harvard Apparatus, Natick, MA). Saturating levels of each mAb were used for all incubations and also were added in the perfusion buffer. The entire time period of monocyte perfusion was videotaped, and accumulation was determined from videotape at 8 to 10 min of perfusion by enumerating adherent cells in six random fields. Monocytes were considered to be adherent after 20 s of stable contact with the monolayer (6). Formation of strings of adherent monocytes and accumulation on HUVEC monolayers or isolated E-selectin and P-selectin coverslips was assessed during the first 4 min of monocyte perfusion using a ×20 objective to visualize the entire field of view for contact between incoming monocytes and the adhesive substrates. We used the same strategy as Alon et al. (18) to identify primary and secondary adhesion (or tethers) on substrates. Strings of adherent monocytes at the end of 4 min were identified first, and then the tape was replayed frame by frame to identify and track incoming monocytes that either attached to substrate surface without previous interactions (primary adhesions) with adherent monocytes, or monocytes that attached to HUVEC via interactions with previously adherent monocytes (secondary adhesions). Primary and secondary adhesions reported in Figure 6 were determined in 15-s time periods over the first few minutes of perfusion to determine the kinetics, or over a 4-min time period, as described for Figure 3. All fields of observation were located as far upstream toward the inlet port as possible to enable the viewers to track the history of the incoming monocyte to allow precise identification of primary and secondary cell adhesions. Typically, two or three blinded observers viewed the tape to determine primary and secondary adhesion, and these data were pooled to give the mean and SDs.

Monocyte-adhesive substrates were prepared using a modification of a previous report (22). Circular 25-mm glass coverslips were immersed in 100% ethanol and air dried, and a 1-cm-diameter region delineated with marker was washed three times with DPBS. After washing, 40 μl of a monocyte suspension (10⁷ cells/ml in perfusion buffer) was placed on the washed area and incubated for 7 min at ambient room temperature. The region of adherent cells was immersed in freshly prepared 0.25% paraformaldehyde in DPBS without Ca²⁺ and Mg²⁺ (5 min) at room temperature, washed extensively with perfusion buffer, and blocked (10 min) with 1% HSA-DPBS without Ca/Mg²⁺ containing 0.2% sodium azide. Coverslips were used immediately in flow studies.

Data were analyzed by one-way analysis of variance, followed by Tukey analyses of multiple comparisons for multiple mAb treatments in Figures 1, 2, and 4 (Minitab for Windows, release 10; Minitab, State College, PA). Significance was set to 0.05.

To evaluate the role of PSGL-1 in monocyte attachment to E-selectin or P-selectin under flow at 1.8 dynes/cm², function-blocking studies were conducted using monolayers of CHO cells stably transfected with full-length human E-selectin or P-selectin (CHO-E and CHO-P, respectively). Murine mAb, KPL1, has been shown previously to recognize the N-terminal tyrosine sulfation motif of mature PSGL-1, which is expressed on human circulating blood leukocytes (13). As shown in Figure 1 (bottom panel), monocytes attached and rolled on CHO-P cell monolayers, and pretreatment of monocytes with anti-PSGL-1 mAb (KPL1) reduced significantly monocyte interactions (50% inhibition) as compared with control anti-HLA class I (mAb W6/32). Blockade of CHO-P cells with anti-P-selectin mAb (HPDG2/3) totally blocked monocyte interactions. Anti-L-selectin mAb (LAM1–4) also reduced adhesion, and if combined with anti-PSGL-1, reduced adhesion by 92%. Blocking studies performed with CHO-E monolayers revealed that KPL1 treatment of monocytes also reduced adhesion (45% inhibition) as compared with control mAb to class I (W6/32). mAb LAM1–4 inhibited interactions by 82%, and the combination of KPL1 and LAM1–4 mAb blocked most of the interactions (91% inhibition). As a control, anti-E-selectin mAb (HEL3/2) totally inhibited all monocyte interactions. These data suggest that PSGL-1 can serve as a ligand for both E- and P-selectin, and are quite distinct from our recent findings that KPL1 mAb totally blocked all neutrophil and peripheral blood T cell adhesion to CHO-P, but had no effect on interactions with CHO-E monolayers (13).

Leukocyte/leukocyte interactions, which occur via L-selectin on the incoming leukocyte binding to L-selectin ligands on already adherent leukocytes, can lead to leukocyte adhesion independent of any contact with the purified adhesion molecules E-selectin and P-selectin (18, 20, 29). To distinguish between the two possibilities, we attempted to determine the contributions of L-selectin and PSGL-1 to initial monocyte/E-selectin or P-selectin interactions (primary adhesions, as originally described by Alon et al. (18)) or monocyte/monocyte interactions (secondary adhesions). However, because of the varied topology of the CHO cell monolayers, we were not able to confidently track monocyte trajectories across the monolayer surface or to clearly distinguish primary and secondary adhesions. As an alternative strategy, purified recombinant E-selectin or P-selectin molecules were adsorbed on glass coverslips to overcome the topology issue. Using this strategy, monocyte interaction with P-selectin on coverslips is ablated with anti-PSGL-1 mAb (Fig. 2, bottom panel). The specificity is demonstrated by the ability of anti-P-selectin mAb to totally block monocyte interactions, while control anti-class I (mAb W6/32) is without effect. This result demonstrates that the epitope recognized by mAb KPL1 on PSGL-1 is essential for binding to P-selectin under flow conditions. Interestingly, mAb KPL1 also reduced monocyte interactions with E-selectin by 62% as compared with W6/32 mAb control (Fig. 2, top panel). Careful analysis of the videotapes from these studies revealed that KPL1 mAb inhibited monocyte/monocyte interactions (i.e., secondary capture), but not primary adhesion to E-selectin (Fig. 3). In contrast, we found no evidence in any experiment for monocyte binding to P-selectin in the presence of KPL1 mAb. Given published reports (18, 22), we conclude that PSGL-1 expressed on monocytes already adherent to E-selectin acts as a ligand for incoming freely flowing monocytes (also see below section).

Figure 4 shows that monocyte adhesion to media-treated HUVEC monolayers is negligible, whereas adhesion to 6-h TNF-α-activated HUVEC monolayers is increased by 12- to 15-fold, which is in line with our previous report (6). Pretreatment of monocytes with mAb KPL1 dramatically reduced adhesion (72% inhibition), as assessed after 10 min of leukocyte perfusion (Fig. 4). This level of inhibition is significantly more than one would expect if PSGL-1 interacted solely with P-selectin expressed on TNF-α-activated HUVEC, because F(ab′)₂ preparations of anti-P-selectin HPDG2/3 mAb block only 30% of monocyte adhesion (data not shown and (6)). Given the above data that L-selectin and PSGL-1 can mediate monocyte/monocyte interleukocyte adhesion on both E- and P-selectin, and recent reports demonstrating PSGL-1 interacts with both L-selectin and P-selectin (13, 22), we compared the effect of anti-L-selectin mAb alone and in combination with anti-PSGL-1 mAb. LAM1-4, a function-blocking anti-L-selectin mAb, reduced monocyte adhesion (75% inhibition) to level similar to anti-PSGL-1 (Fig. 4). Control nonblocking anti-L-selectin mAb LAM1-14 or control anti-class I (mAb W6/32), which is expressed on both monocytes and HUVEC, did not alter adhesion. The combination of mAb KPL1 and LAM1-4 blocked better than either alone (p < 0.015; n = 4 paired experiments), and reduced monocyte adhesion by 86%, as compared with the combination of nonblocking control mAb (W6/32 + LAM1-14). Based on the results above showing that monocytes form interleukocyte adhesions very well on both P- and E-selectin, we assessed the role of PSGL-1 and L-selectin in this process of monocyte accumulation on monolayers of E- and P-selectin and cytokine-activated HUVEC.

After 4 min of monocyte perfusion across 6-h TNF-α-activated HUVEC, examination of monocyte interactions with the apical endothelial cell surface revealed that freely flowing monocytes often collided with and momentarily bound to already adherent monocytes, and then subsequently traveled downstream and bound to other already adherent monocytes, or stably arrested on the endothelial surface. This resulted in the formation of multiple linear strings of rolling and adherent monocytes (data not shown). The exact same process has been detailed for neutrophils and γβ T cells on a variety of substrates (unpublished observations, F. W. Luscinskas; and (18, 19, 20, 22)), and we have reported this for purified human monocytes interacting with IL-4- or TNF-α-activated HUVEC monolayers (5, 6). At later time points (8 min of perfusion), the linear strings of adherent monocytes increased in size (length and width) and evolved to larger foci of adherent/rolling monocytes that could be clearly distinguished using a ×40 phase objective (Fig. 5). The shape of these foci is broader after 8 min with smaller satellite strings, identified with arrows, branching out from the initial string (Fig. 5, A and B, arrows identify satellite strings). In Figure 5, the fluid flow is from right to left. Preincubation of monocytes with control nonblocking L-selectin mAb (LAM1-14) had no effect (Fig. 5,B). In contrast, anti-PSGL-1 mAb (Fig. 5,C) or anti-L-selectin mAb (not shown) blocked most string formation (arrows identify single, adherent, and/or transmigrated monocytes). When monocytes were preincubated with both function-blocking mAb (LAM1-4 and KPL1), no monocyte strings were observed (Fig. 5 D; n = 4 separate experiments). Interestingly, the pattern of monocyte accumulation after blockade of both L-selectin and PSGL-1 was essentially random when compared with accumulation with control mAb.

Further analyses of the kinetics of monocyte adhesion were performed for both E- and P-selectin monolayers and activated HUVEC monolayers. As shown in Figure 6, the secondary kinetics for monocyte accumulation on P-selectin (Fig. 6,A) and E-selectin (Fig. 6,B) monolayers fit best to quadratic functions (y = 0.0267x² + 0.674x − 21.4, R² = 0.975; y = 0.011x² + 1.053x − 22.8, R² = 0.9753; y = accumulation, x = time), while the primary interactions fit a linear function. The kinetics of secondary monocyte accumulation on activated HUVEC monolayers also fit best to a quadratic function (y = 0.002x² + 0.237x − 6.007, R² = 0.9942), whereas primary accumulation best fit to a linear function (Fig. 6,C). The effects of blocking L-selectin (Fig. 6,D) or PSGL-1 (Fig. 6,E) also are shown and demonstrate that blockade of either adhesion molecule dramatically reduces accumulation, consistent with data in Figure 4. The most significant effect was that KPL1 or LAM1–4 alone prevented most monocyte/monocyte adhesion as compared with control mAb. These data demonstrate that monocyte-mediated capture of incoming freely flowing monocytes contributes significantly to the accumulation of monocytes on activated HUVEC monolayers.

To further understand the molecular interactions underlying monocyte capture under flow at 2 dynes/cm², experiments were performed with mAb to PSGL-1 and L-selectin using monolayers of purified adherent monocytes. Monolayers of adherent monocytes were prepared and fixed with 0.25% paraformaldehyde to preserve their round, unactivated morphology and maintain their adhesion to the glass surface, and to prevent loss or redistribution of L-selectin and PSGL-1 (30, 31, 32). Perfusion of freshly isolated monocytes across monolayers of fixed monocytes led to a steady flux of rolling or transiently attached monocytes that were enumerated easily. Few, if any, monocytes adhered longer than 2 s, which is quite different from monocyte behavior on purified E- and P-selectin or activated HUVEC reported above. After 3 min of perfusion of media-treated monocytes, a new cohort of monocytes pretreated with and containing mAb to class I (W6/32 mAb), followed by cells treated with nonblocking L-selectin mAb (LAM1-14, 10 μg/ml), was added to the reservoir. None of these treatments affected monocyte/monocyte rolling interactions (Fig. 7, top panel). As shown in the middle panel, after 3 min of perfusion of media-treated monocytes, a cohort of monocytes treated with PSGL-1 (20 μg/ml of KPL1) was added to the buffer reservoir. Anti-PSGL-1 mAb consistently and totally blocked monocyte rolling and transient attachments. After a 3-min washout with media alone (no monocytes) to remove unbound mAb, a new cohort of W6/32 mAb-treated monocytes was perfused. As can be seen, monocyte rolling interactions were essentially ablated by mAb to PSGL-1, but not control mAb (Fig. 7, compare top to middle panels). Moreover, monolayers treated with mAb to PSGL-1 did not support monocyte rolling (5% of control), even after media washout and perfusion with control mAb-treated monocytes. We next pretreated monocytes with anti-PSGL-1 mAb and then prepared fixed monolayers to confirm the effects of PSGL-1 blockade. Although several experiments were conducted, studies performed using this strategy did not yield consistent interpretable results.

Next, mAb to L-selectin (10 μg/ml, blocking LAM1-4) were tested in an identical fashion on fixed monocyte monolayers. Anti-L-selectin mAb blocked all monocyte rolling interactions (Fig. 7, bottom panel). After a short washout period with media alone (no monocytes) to remove unbound mAb, a fresh cohort of W6/32 mAb-treated monocytes did resume rolling interactions. Taken together, these results demonstrate that the predominant molecular mechanisms underlying monocyte capture by adherent monocytes are L-selectin on incoming monocytes recognizing the sulfated N-terminal motif of PSGL-1 on adherent monocytes.

An in vitro flow model and mAb to L-selectin in combination with mAb to PSGL-1 were used to demonstrate that L-selectin on flowing monocytes binds to PSGL-1 on already adherent monocytes, and that this mechanism constitutes the predominant adhesion pathway for monocyte initial attachment, adhesion, and accumulation on P- or E-selectin monolayers, and TNF-α-activated HUVEC monolayers in vitro. mAb to PSGL-1 ablates monocyte-adhesive interactions with P-selectin and partially blocks adhesion to E-selectin. The partial inhibition on E-selectin monolayers is due to inhibition of monocyte/monocyte interactions, not monocyte/E-selectin interactions. This result demonstrates that the epitope recognized by KPL1 mAb, which maps to the sulfated N-terminal of PSGL-1 (9, 13), is the binding site for P-selectin, but not for E-selectin. We conclude that human monocytes bind to E-selectin either by utilizing another site(s) on PSGL-1, or through other surface-expressed ligands. Remarkably, blockade of either PSGL-1 or L-selectin alone consistently blocked most adhesion (70 and 76%) to TNF-α-activated endothelial monolayers, and the combination of mAb was greater than either alone (86%). Based on current literature, no other leukocyte type exhibits this level of dependence on the L-selectin and PSGL-1 molecules for adhesion to activated HUVEC.

A similar pattern for mAb inhibition emerges from our studies with monocyte monolayers. Injection of blocking mAb to PSGL-1 ablated monocyte interactions within 2 min. Moreover, after washout with media, monolayers treated with anti-PSGL-1 mAb did not regain the ability to support rolling of control mAb-treated monocytes (<5% of control), demonstrating that PSGL-1 is the predominant L-selectin ligand on adherent monocytes (Fig. 7, middle panel). The complementary study with mAb to L-selectin and the above results with PSGL-1 mAb reveal that L-selectin on incoming monocytes is the predominant molecule that recognizes PSGL-1 on adherent monocytes. As noted in Results, we were unable to perform selective pretreatment of monolayers with mAb to distinguish by a second approach whether PSGL-1 on the monocyte monolayers is the only ligand for L-selectin-mediated rolling. Because blockade with either anti-PSGL-1 or L-selectin mAb totally blocked monocyte/monocyte rolling interactions (Fig. 7) and more than 90% of interactions on E- and P-selectin substrates (Fig. 3), other ligands may not be required in this in vitro model.

Previous reports also suggest an important role for PSGL-1 in leukocyte adhesion to endothelium in vivo (33, 34) or endothelial cell adhesion molecules under flow conditions in vitro (13, 22, 35, 36, 37). In vivo studies have found that pretreatment of human neutrophils or the HL60 cell line with blocking mAb (mAb PL1) significantly reduced the leukocyte rolling flux (80% inhibition) and increased leukocyte rolling velocities in rat mesentery venules (34). A second study reported that rat mAb to murine PSGL-1 (mAb 2PH1) blocked 80% of leukocyte rolling in venules of mouse cremaster muscle, and also blocked (80% at 2 h and 60% at 4 h) neutrophil accumulation in a murine model of peritonitis (33). In in vitro studies with neutrophils and eosinophils, PSGL-1 is the predominant ligand for P-selectin and L-selectin, but not for E-selectin (20, 22, 35). Certain T cell subsets also utilize PSGL-1 for recognition of P-selectin under flow (20, 35); however, they also appear to utilize PSGL-1, as well as other sialylated and fucosylated structures, to bind L-selectin under flow (38, 39).

In contrast to PSGL-1, numerous in vivo and in vitro studies have identified a role for L-selectin in adhesion of monocytes (5, 6, 40, 41), as well as neutrophils (7, 42, 43), eosinophils (36, 44, 45), and γδ T cells (20) to activated endothelium, but not for PBL (39) or CD4⁺ T cell (46) to endothelium. The current results support the existing paradigm of L-selectin on incoming cells interacting with PSGL-1 expressed on already adherent monocytes (secondary adhesion), thus mediating enhanced monocyte accumulation. Because the combination of anti-L-selectin and anti-PSGL-1 was slightly greater than either alone, one also can conclude that L-selectin recognizes an inducible endothelial ligand(s) and/or other leukocyte ligands, in addition to PSGL-1. Previous studies have also offered indirect evidence for the existence of an inducible endothelial cell ligand for L-selectin using a variety of models (40, 41, 47).

Using unfractionated peripheral blood mononuclear leukocytes, monocyte binding and accumulation on immobilized E-selectin or VCAM-1 were shown to entail monocyte/monocyte interleukocyte interactions, which were blocked completely by fucoidin and mAb to L-selectin (DREG56) (18). The current study confirms and extends these findings. First, elutriation-purified human monocytes, blocking mAb to PSGL-1, alone and in combinations with mAb to L-selectin were used, and the analyses fully documented both primary and secondary adhesion kinetics and the molecular mechanisms utilized for rolling and accumulating on purified E- and P-selectin as well as TNF-α-activated HUVEC monolayers. The requirement for PSGL-1 (and L-selectin) for both initial attachment and accumulation (Figs. 5 and 6) is substantially greater for monocytes as compared with neutrophils and eosinophils (22, 36, 44, 45). The kinetics of monocyte primary and secondary adhesion in our in vitro flow model is fit best to quadratic and linear functions, respectively. Secondary kinetics for monocyte accumulation determined in this study and for neutrophil accumulation on P-selectin described previously (22) is fit best to a quadratic function, whereas the primary adhesion kinetics for either leukocyte type fits best to a linear function (see Fig. 6). The analysis of E-selectin and TNF-α-activated HUVEC monolayers indicates that monocyte primary and secondary kinetics of accumulation are best fit to quadratic and linear functions, respectively. We note, however, that the kinetics of monocyte secondary accumulation on all substrates also could be fit to a linear function, but the R² was less favorable. Because of the rapid accumulation on these substrates, the primary and secondary adhesion cannot be determined confidently after 4 min.

It is of interest to note that a recent report on human monocyte interactions with IL-1β-activated HUVEC or L cells stably expressing E-selectin (L-ELAM) under flow at 2 dynes/cm² did not observe leukocyte string formation (primary adhesion only, no secondary adhesions) (48). In this study, monocytes accumulated at a constant linear rate, and anti-L-selectin treatment of monocytes (DREG56 mAb) did not alter the pattern of monocyte accumulation on IL-1β-treated HUVEC or L-ELAM. The basis for this discrepancy is not known, but one explanation may reside in the different dimension of the actual flow channel used in these studies or other differences in the design characteristics of the flow chambers.

The importance of secondary capture of leukocytes in vivo remains to be determined. A recent report (49) indicates that leukocyte capture by already adherent or rolling leukocytes occurred in vivo, but was not an important determinant for leukocyte adhesion to endothelium of venules in the mouse cremaster muscle. Although in this specific model a small effect for leukocyte-dependent capture was observed, one cannot discount the role of this mechanism in vivo without further study in other animal models of acute or chronic inflammation, especially in response to strong physiologic stimuli (endotoxin, TNF-α, immune reactions, wounds) or in small microvessels or postcapillary venules, in which such interactions would be favored. Moreover, the ease with which monocytes form secondary adhesions, leading to more efficient capture on HUVEC and endothelial cell adhesion molecules, suggests that monocytes may use secondary capture to overcome their numerical disadvantage in blood, as compared with neutrophils and lymphocytes, and thus preferentially accumulate at sites of inflammation or chronic immune reactions and atherosclerosis. We speculate that the secondary interactions described in this study for monocytes could be important in a setting of a chronic stimulus, such as enhanced adhesion and intimal accumulation in fatty streaks in atherosclerosis-prone vessels, in which endothelial leukocyte ligands may be limiting.

We thank William J. Atkinson and Kay Case for providing cultured HUVEC and Dr. Thomas F. Tedder, Duke University Medical Center, for providing purified mAb LAM1-4 and LAM1-14. We also thank Drs. Jennifer R. Allport and other members of Vascular Research Division for helpful discussions and critical review of this manuscript.