Endothelial Cell E- and P-Selectin and Vascular Cell Adhesion Molecule-1 Function as Signaling Receptors

Human umbilical vein endothelial cells (HUVEC) were isolated by collagenase treatment of the vessels as described elsewhere (Jaffe et al., 1973) and maintained in HUVEC medium (medium 199 buffered with 10 mM Hepes and supplemented with 20% newborn bovine serum [HyClone Laboratories Inc., Logan, UT], 50 μg/ml endothelial cell growth supplement [prepared from bovine brain, as described by Maciag et al., 1979], 90 μg/ml heparin [Sigma Chemical Co., St. Louis, MO], 50 U/ml penicillin, and 50 μg/ml streptomycin). For experiments, first-passage HUVEC were grown on gelatin-coated 22-mm glass coverslips placed on the bottom of 35-mm Petri dishes. The monolayers reached confluence within 2 d, as evidenced by prominent immunostaining for PECAM-1 along the cell borders (Muller et al., 1989; Albelda et al., 1990). The cells were used in assays at 4–8 d after plating.

Human PMN were isolated from acid-citrate-dextrose (ACD)-anticoagulated blood by Ficoll-Hypaque (Nycomed Pharma AS, Oslo, Norway) gradient centrifugation, 3% dextran (Pharmacia Biotech AB, Uppsala, Sweden) sedimentation, and hypotonic saline lysis of contaminating erythrocytes (Boyum, 1968). The entire procedure was conducted at 4°C. Purified PMN were suspended in PBS (Sigma Chemical Co.) containing 1% BSA (Sigma Chemical Co.) and held on ice (<3 h) until use.

Human monocytes were isolated from ACD-anticoagulated blood following the method of Peri et al. (1990), modified as follows. After erythrocyte sedimentation in 1.5% dextran, the white cell–rich plasma was washed once with PBS containing 13 mM sodium citrate and 0.5% BSA. The cell pellet was then resuspended in isotonic Percoll (Pharmacia Biotech AB) containing 13 mM sodium citrate and 0.5% BSA, pH 7.4. The density of the Percoll suspension was 1.0635 ± 0.0003 g/ml, as measured at 20°C by a DMA density meter (A. Paar, Graz, Austria), and Percoll osmotic value was 290 ± 2 mosM, as measured by a cryoscopic osmometer (Osmomat 030; Gonotec, Berlin, Germany). PBS containing 0.5% BSA was layered over Percoll and the cell suspension was centrifuged at 1,000 g for 20 min at 20°C. The cell ring formed at the PBS–Percoll interface was collected, washed twice, and then suspended in ice-cold PBS containing 1% BSA. The resulting cell suspension contained >90% (generally 95%) monocytes, as assessed by morphology and peroxidase-staining, and cell viability was >98%, as determined by trypan blue dye exclusion. Monocytes were then held on ice until use (<3 h).

In experiments involving a pretreatment of leukocytes with mAbs, before addition to HUVEC monolayers, PMN or monocytes were incubated on ice in the presence of mAb (20 μg/ml) for at least 1 h. The cells were then washed twice at room temperature with PBS containing 1% BSA, 5 mM glucose, and 0.7 mM CaCl₂, resuspended in the same medium, and immediately added to the HUVEC monolayers at a leukocyte/HUVEC ratio of 6–8:1.

Cytosolic-free calcium ([Ca²⁺]_i) was monitored in single HUVEC by using the fluorescent calcium probe Fura2-AM as described by Huang et al. (1993), with minor modifications. [Ca²⁺]_i measurements were performed on a digital fluorescence-imaging microscopy system built around a Zeiss Axiovert 135 (Oberkochen, Germany) as already described (Lorenzon et al., 1997). The excitation light at 340 and 380 nm was provided by a modified dual wavelength microfluorimeter (CAM-230; Jasco Intl. Co. Ltd., Tokyo, Japan) and the fluorescence images were collected by a low light level CCD camera (Hamamatsu Photonics, Hamamatsu, Japan). The camera output was fed into a digital image processor where video frames were digitized and integrated at real time. The image acquisition was performed at two frames/s. Determination of ratio and [Ca²⁺] calculation were performed pixel by pixel on pairs of corresponding 340 and 380 images according to Grynkiewicz et al. (1985). R_max and R_min were determined using ionomycin (40 μM) with calcium (30 mM) or a stoichiometric excess of EGTA, respectively. Thus, R_max represented the ratio of 340 nm/ 380 nm signal under saturating conditions of calcium, whereas R_min represented the same ratio in the absence of calcium. Values used for calibration included R_max = 650, R_min = 200, K_d = 224 nM, and f380_min/f380_max (ratio of fluorescence at high calcium to that at low calcium) = 2. Temporal plots were calculated from ratio images and each temporal plot indicated the mean value of variation of the fluorescent signal at the single cell level.

Endothelial monolayers grown for 3–5 d on gelatin-coated glass coverslips were loaded with 5 μM Fura2-AM (Sigma Chemical Co.) for 30 min. Cell loading was performed at room temperature to avoid compartmentalization of the dye. The monolayers were then washed twice and incubated in Hepes (10 mM) buffered medium 199 supplemented with 20% newborn bovine serum (M199-H/NBS) for an additional 20 min at room temperature to allow de-esterification of Fura2-AM. For experiments involving a pretreatment of the cells with Escherichia coli lipopolysaccharide (LPS), the monolayers were first incubated for 5 or 24 h at 37°C with 0.5 μg/ml LPS, and then loaded with Fura2, as described above. Tissue culture dishes were then mounted on the temperature-monitored stage of the Axiovert 135, and bathed with M199-H/NBS. To minimize spontaneous [Ca²⁺]_i spiking by HUVEC during the assay, the monolayers were maintained at 37°C for 30 min before starting [Ca²⁺]_i recordings. When indicated, the cells were further incubated 5 min in the presence of histamine (50 μM), washed three times with prewarmed M199-H/NBS medium, and then incubated in the same medium for another 5 min at 37°C. The homogeneous distribution of the fluorescent dye was carefully checked before starting [Ca²⁺]_i recordings. Agonists (mAbs or leukocytes) were pipetted onto the monolayers 2–3 min after starting the recordings, and HUVEC [Ca²⁺]_i was then monitored for 15 min unless otherwise indicated. As a control of variability among HUVEC, a maximal increase in HUVEC [Ca²⁺]_i (Jacob et al., 1988) was induced by the addition of 200 μM histamine by the end of each recording. By calculating the ratio between the integral value of the area covered by [Ca²⁺]_i spike(s) and the level of the spike induced by histamine, a numeric value was obtained for each cell. This value allowed to compare changes in HUVEC [Ca²⁺]_i obtained under different experimental conditions.

HUVEC for immunofluorescence assay were grown on gelatin-coated glass coverslips following the same procedure used to prepare the cells for [Ca²⁺]_i measurements. Since preliminary experiments indicated that endothelial stress fibers almost completely disappeared in long-term cultures, we performed these assays on monolayers grown for only 6–8 d after seeding. On the basis of preliminary experiments, care was also taken to change HUVEC medium every day. When indicated, E. coli LPS (0.5 μg/ml) was present in the cell medium during the last 5 or 24 h of incubation. Agonists (mAbs or leukocytes) were then added to the monolayers and the cells were further incubated at 37°C for 10 min. The monolayers were then fixed with freshly made 3% formaldehyde (Merck, Darmstad, Germany) in PBS for 5 min at room temperature and permeabilized with 0.5% Tx-100 (Merck) for 5 min at 0°C, as described by Dejana et al. (1987). After rinsing in PBS, coverslips to be examined for actin morphology were incubated in the presence of 0.2 μM fluorescein-labeled phallacidin (NBD-phallacidin; Molecular Probes Europe BV, Leiden, The Netherlands) in PBS for 20 min at room temperature, as suggested by the manufacturer. In contrast, cells to be examined for PECAM-1 or ICAM-1 immunostaining were incubated for 20 min at room temperature in the presence of fluorescein-labeled rabbit anti–mouse IgG, F(ab')₂ fragment. The slides were then mounted in glycerol/PBS and observations were carried out in a Zeiss Axiophot microscope equipped with epifluorescence. Fluorescence images were recorded on Fuji N1600 professional (Fuji Photo Film Co., Ltd., Tokyo, Japan) with a constant exposure of 40 s.

First-passage HUVEC were plated onto gelatin-coated 96-well tissue culture plates (Costar Cluster, Cambridge, MA) and grown to confluence in HUVEC medium. Histamine (50 μM) or LPS (0.5 μg/ml) or medium was then added to the monolayers, and the cells were further incubated at 37°C for the indicated time periods. The monolayers were then washed with a three-well volumes of medium and treated for 1 h at 4°C with mAbs to adherence molecules (5 μg/ml). For the detection of P-selectin, the cells were fixed with 1% PFA in PBS for 15 min at room temperature in the dark before exposure to the relevant mAb. The cell monolayers were then washed three times with PBS containing 1% BSA, incubated for 1 h at 4°C in the presence of alkaline phosphatase–labeled affinity-purified antibodies from goat to mouse IgG (Sigma Chemical Co.), and further washed three times with PBS containing 1% BSA. The reaction product was generated by the addition of para-nitrophenyl-phosphate (Sigma Chemical Co.) at the concentration of 1 mg/ml in 0.1 M glycine buffer, pH 10.4, containing 1 mM MgCl₂ as substrate for the enzyme. Quantification of reaction product was performed in a Multiscan 340 apparatus (Flow Laboratories Inc., Milan, Italy) at 405 nm.

First-passage HUVEC were plated onto gelatin-coated 96-well tissue culture plates (Costar Cluster) and grown to confluence in HUVEC medium. LPS (0.5 μg/ml) or medium was then added to the monolayers, and the cells were further incubated for 24 h at 37°C. Subsequently, the monolayers were washed with three-well volumes of medium and untreated monocytes, or monocytes that were pretreated for 60 min with mAb (20 μg/ml), were added to the monolayers (70 μl/well). Monocytes and HUVEC were then incubated for 15 min at 37°C. After incubation, the monolayers were washed with a two-well volume of PBS to remove non-adherent monocytes. A colorimetric assay was then used to detect the leukocytes adhering to the monolayers with tetramethyl benzidine (Sigma Chemical Co.) as the peroxidase substrate, as described previously in detail (Dobrina et al., 1991).

Anti–P-selectin/CD62P mAb CBL474 and anti–E-selectin/CD62E mAb CBL180 were from Cymbus Bioscience Ltd. (Southampton, UK). Anti– ICAM-1/CD54 mAb RR1/1 (Dustin et al., 1986) was obtained through the courtesy of R. Rothlein (Boehringer Ingelheim, Ridgefield, CT) and anti– ICAM-1/CD54 mAb MCA532 was obtained from Serotec Ltd. (Kidlington, Oxford, UK). Anti–VCAM-1/CD106 mAb 4B9 is a blocking mAb (Carlos et al., 1991). Anti–PECAM-1/CD31 mAb M89D3 inhibits leukocyte migration across HUVEC monolayers (Zocchi et al., 1996). mAb 60.3 recognizes a functional epitope on CD11/CD18 (Beatty et al. 1983). mAb CSLEX1 (Becton Dickinson, San Jose, CA) recognizes a functional epitope on sialyl Lewis x(sLe^x)/CD15s. mAb HP2/1 binds to a functional epitope on the CD49d (α chain) of leukocyte CD49d/CD29 (VLA-4) integrin and was a gift of Dr. F. Sanchez Madrid (Servicio de Immunologia, Hospital de la Princesa, Madrid, Spain; Sanchez Madrid et al., 1986). Preliminary observations confirmed that mAb M89D3 inhibited PMN and monocyte transendothelial migration and that all other mAbs used in this study inhibited PMN and/or monocyte adherence to HUVEC. Migration and adherence assays were performed as described previously (Dobrina et al., 1991; Zocchi et al., 1996).

To examine the potential involvement of adhesion proteins in endothelial stimulation, [Ca²⁺]_i was measured in Fura2-loaded HUVEC. After incubation of EC in the presence or absence of histamine or LPS, two agents known to induce expression of adhesion molecules on HUVEC (Schleimer and Rutledge, 1986; Lorant et al., 1991), EC were exposed to leukocyte adherence-blocking mAbs directed against endothelial P- and E-selectins, ICAM-1, VCAM-1, and PECAM-1. In HUVEC stimulated with agents such as histamine or thrombin, expression of P-selectin is known to peak by 10 min and return to baseline within 20–30 min (Hattori et al., 1989). In contrast, in LPS- or cytokine-activated HUVEC expression of E-selectin is known to peak by 4–5 h and to return to baseline within 8–12 h, while ICAM-1 and VCAM-1 peak at 5–10 h and decline by 48–72 h (Pober et al., 1986; Tedesco et al., 1997). Finally, PECAM-1 is strongly expressed at the intercellular junctions of unstimulated HUVEC (Muller et al., 1989). In preliminary experiments we confirmed this pattern of expression of adhesion molecules in HUVEC, by using mAb CBL474 to P-selectin, mAb CBL180 to E-selectin, mAb RR1/1 or mAb MCA532 to ICAM-1, mAb 4B9 to VCAM-1, and mAb M89D3 to PECAM-1 (Fig. 1). Thereafter, we performed [Ca²⁺]_i measurements after treatment of the cells for 5 min with histamine or for 5 h or 24 h with LPS. As shown in Fig. 2, the anti–P-selectin mAb induced an increase in [Ca²⁺]_i in the histamine-treated HUVEC, but not in untreated HUVEC. Similarly, the anti–E-selectin mAb and the anti– VCAM-1 mAb caused [Ca²⁺]_i changes in HUVEC treated with LPS for 5 or 24 h, but not in untreated HUVEC. Notably, the mAb-induced response, when present, was comparable in magnitude to the increase in [Ca²⁺]_i caused by the addition of 200 μM histamine. In contrast, no change in [Ca²⁺]_i was observed after the addition of the anti– ICAM-1 mAb RR1/1 to HUVEC treated with LPS for 5 or 24 h or after the addition of the anti–PECAM-1 mAb to unstimulated cells.

Further experiments with the anti–ICAM-1 mAb MCA532 confirmed the negative results obtained with RR1/1. The anti–PECAM-1 mAb also failed to activate histamine-treated HUVEC or LPS-treated HUVEC. Similarly, we were unable to induce [Ca²⁺]_i changes in HUVEC that were treated first with the anti–ICAM-1 mAb or anti–PECAM-1 mAb, and then washed and incubated with a goat anti–mouse IgG polyclonal antibody to induce cell surface clustering of ICAM-1 or PECAM-1 (results not shown).

Previous studies have shown that PMN adherence to EC elicits transient increases in endothelial [Ca²⁺]_i (Huang et al., 1993; Ziegelstein et al., 1994). Having established that stimulation of HUVEC with mAb directed against EC adhesion proteins was similarly effective in causing transient changes in the cell [Ca²⁺]_i, we investigated whether these molecules were actually involved in the transduction of signals generated upon PMN or monocyte adherence to HUVEC. PMN are known to adhere to HUVEC via sLe^x, CD11/CD18, and PECAM-1, counter-receptors for endothelial P- and E-selectins, ICAM-1 and PECAM-1, respectively. Hence, HUVEC were stimulated with histamine for 5 min to induce expression of P-selectin, or with LPS for 5 h to induce expression of E-selectin and ICAM-1. HUVEC [Ca²⁺]_i was then monitored after the addition of PMN to the monolayers. PMN were unable to induce [Ca²⁺]_i changes in untreated HUVEC, but actively stimulated [Ca²⁺]_i increases both in histamine- and in LPS-treated monolayers (Fig. 3). Recordings of HUVEC [Ca²⁺]_i were then repeated using PMN that had been preincubated in the presence of an adherence-blocking mAb directed against the sLe^x, CD18, or PECAM-1. As reported in Table I, pretreatment of PMN with the anti-sLe^x mAb (CSLEX1) led to a marked inhibition (>80%) of their capacity to induce [Ca²⁺]_i changes in both histamine- and LPS-treated HUVEC. In contrast, [Ca²⁺]_i changes in histamine- and LPS-treated HUVEC were not significantly inhibited by a pretreatment of PMN with an anti-CD18 mAb (60.3) or an anti–PECAM-1 mAb (M89D3).

The same experimental approach was then followed to investigate the effect of monocyte adhesion to HUVEC. Monocytes are known to share with PMN the counter-receptors for endothelial E-selectin, ICAM-1, and PECAM-1. In addition, monocytes bind to endothelial VCAM-1 via VLA-4. To induce the expression on HUVEC of E-selectin, ICAM-1, and VCAM-1, the monolayers were treated with LPS. E-selectin was assumed to be maximally expressed on HUVEC after 5 h of incubation with LPS, whereas expression of ICAM-1 and VCAM-1 was assumed to be induced after 24 h as well as 5 h of incubation with LPS. The cells were then loaded with Fura2, and [Ca²⁺]_i recordings were performed upon addition of monocytes to the monolayers. As with PMN, monocytes failed to induce [Ca²⁺]_i changes in untreated HUVEC. However, they actively stimulated [Ca²⁺]_i changes in HUVEC treated with LPS for 5 and 24 h (Fig. 4). Table II shows the inhibitory effect of preincubation of monocytes incubation with adherence-blocking mAb on their capacity to stimulate [Ca²⁺]_i changes in HUVEC monolayers. Pretreatment of monocytes with the anti-sLe^x mAb CSLEX1 alone, the anti-CD18 mAb 60.3 alone, the anti– VLA-4 mAb HP2/1 alone, the anti–PECAM-1 mAb M89D3 alone, or pretreatment of monocytes with combinations of these anti-CD18, anti-VLA-4, and anti-PECAM-1 mAbs, was ineffective in inhibiting monocyte-induced [Ca²⁺]_i changes in 5 h LPS-treated HUVEC. However, [Ca²⁺]_i changes were markedly reduced when monocytes were pretreated with a combination of anti-sLe^x mAb and anti– VLA-4 mAb. Since sLe^x and VLA-4 are ligands of endothelial selectins and VCAM-1, respectively, it appears that HUVEC treated with LPS for 5 h are stimulated by monocytes primarily via E-selectin and VCAM-1. In HUVEC treated with LPS for 24 h, treatment of monocytes with mAb to sLe^x alone or to CD18 alone did not inhibit monocyte-induced [Ca²⁺]_i changes. In contrast, pretreatment of monocytes with the anti–VLA-4 mAb HP2/1 alone was sufficient to inhibit strongly monocyte-induced [Ca²⁺]_i changes in HUVEC, indicating that in this setting the VLA-4/VCAM-1 interaction plays a major role in monocyte stimulation of EC.

Finally, we addressed the question whether the results obtained with mAb CSLEX1-treated PMN or with mAb CSLEX1- and/or mAb HP2/1-treated monocytes could be accounted for by some interference of these mAbs with leukocyte binding to HUVEC, thus preventing other leukocyte molecules from engaging HUVEC adhesion proteins. This possibility could not be excluded in our experiments with histamine-treated HUVEC, since Lorant et al. (1991) demonstrated that PMN adherence to thrombin-treated HUVEC was abolished by adherence-blocking mAb directed against P-selectin. In contrast, in a previous report we showed that PMN adherence to HUVEC treated with LPS for 4 h was only partially (by 50%) inhibited by the presence of an adherence-blocking mAb directed against E-selectin (Dobrina et al., 1991). Similarly, we showed that monocyte adherence to HUVEC was only minimally (∼15%) altered in the presence of anti–E-selectin mAb and/or anti–VLA-4 mAb (Carlos et al., 1991). These observations indicate that PMN and monocytes maintain their ability to adhere to HUVEC even in the presence of mAbs, which interfere with the sLe^x/E-selectin and/or the VLA-4/VCAM-1 pathways. As an additional control, we studied the adherence of monocytes to HUVEC treated with LPS for 24 h. As shown in Fig. 5, whereas pretreatment of monocytes with a combination of anti-CD18 and anti–VLA-4 mAbs significantly inhibited binding to the LPS-treated HUVEC, there was no inhibition when monocytes were pretreated with either mAb alone. Hence, it seems unlikely that the potent inhibition of monocyte-induced changes in HUVEC [Ca²⁺]_i by the anti-VLA-4 mAb HP2/1 was accounted for solely by its potential inhibitory effect on monocyte adherence.

Previous studies have shown that stimulation of EC with soluble agonists such as histamine and thrombin, which increase endothelial [Ca²⁺]_i, also induces rearrangements in endothelial actin morphology. Having established that an increase in HUVEC [Ca²⁺]_i results upon engagement of endothelial P-selectin, E-selectin, and VCAM-1, we sought evidence for a relationship between rearrangement of cytoskeletal actin filaments and endothelial stimulation via selectins or VCAM-1. Monolayers prepared identically to those used for [Ca²⁺]_i measurements were treated with E. coli LPS to induce expression of adhesion proteins. The cells were then treated with various mAbs to adhesion molecules, fixed, permeabilized, and then stained with fluorescein-labeled phallacidin. Fig. 6,A illustrates the F-actin distribution in a control monolayer treated with LPS for 5 h. Individual cells were clearly delineated by a rim of fluorescent staining. The central regions of the cells were devoid of prominent filament bundles. However, rare and randomly oriented actin fibers were present in some cells. When LPS-treated HUVEC were exposed to mAb CBL180 to E-selectin (Fig. 6,B), the rim of fluorescent staining at the cell periphery was less distinct, and the actin network was reorganized into thick cables mostly aligned parallel to the long axis of the cell (stress fibers). Similar results were obtained by treating the cells with mAb 4B9 to VCAM-1 (Fig. 6,C). Since not all of the cells responded similarly to the stimuli, we analyzed the percentage of stress fiber-positive and -negative cells under the various conditions. As shown in Table III, >50% of cells were positive for stress fibers after treatment of the monolayers with mAb CBL 180 to E-selectin or mAb 4B9 to VCAM-1. In contrast, we were unable to induce the appearance of stress fibers by incubating the cells with mAb RR1/1 or mAb MCA532 to ICAM-1, or with mAb M89D3 to PECAM-1. Similarly, there was no evidence of stress fiber formation when HUVEC monolayers that had not been pretreated with LPS were exposed to any of the above mAb to endothelial adhesion molecules or to the anti– P-selectin mAb CBL474. Finally, treatment of the cells with histamine to induce P-selectin expression caused formation of prominent stress fibers directly, thus preventing us from testing the effect of the anti–P-selectin mAb on activated HUVEC.

We next investigated whether the adherence of PMNs or monocytes to EC would induce changes in the organization of endothelial cytoskeleton. Therefore, leukocytes were added either to unstimulated HUVEC monolayers or to monolayers that had been pretreated with E. coli LPS to induce expression of adhesion molecules on their surface. Fig. 7 depicts the effect of PMN–HUVEC interaction on endothelial actin morphology. When PMN were added to unstimulated HUVEC, F-actin distribution on the periphery of EC remained unchanged and we could not detect the appearance of stress fibers (Fig. 7,A). In contrast, addition of PMN to monolayers that had been pretreated for 5 h with LPS caused the formation of prominent stress fibers in the large majority of the cells (Fig. 7,B). As shown in Table IV, pretreatment of PMN with the anti-sLe^x mAb CSLEX1 completely abolished the capacity of PMN to induce the formation of stress fibers in LPS-treated HUVEC, thus implicating endothelial selectins in EC stimulation. No inhibitory effect was observed with preincubation of PMN with the anti-CD18 mAb 60.3 or the anti–PECAM-1 mAb M89D3, suggesting that endothelial ICAM-1 and PECAM-1 were not involved in this response. Monocytes were used to explore the potential involvement of the endothelial VCAM-1 in leukocyte–EC interactions leading to formation of stress fibers. As in the experiments with PMN, monocytes were first added to unstimulated HUVEC. Under these conditions, we found no evidence of stress fiber formation in the monolayers (Fig. 8,A). To induce maximal expression of ICAM-1 and VCAM-1 on the endothelial surface, the monolayers were then incubated with LPS for 24 h. Under these conditions, the addition of monocytes actively stimulated a reorganization of endothelial actin filaments into stress fibers (Fig. 8,B). As shown in Table IV, formation of stress fibers was still induced by monocytes that had been pretreated with the adherence-blocking anti-sLe^x mAb CSLEX1, the anti-CD18 mAb 60.3, or the anti–PECAM-1 mAb M89D3. However, pretreatment of monocytes with the anti–VLA-4 mAb HP2/1 prevented changes endothelial actin morphology, thus confirming an involvement of VCAM-1 in EC stimulation by adhering monocytes.

Our results demonstrate a role in outside-in signaling for endothelial P-selectin, E-selectin, and VCAM-1 adhesion proteins. mAbs directed against these receptors induced a rise in endothelial [Ca²⁺]_i and rearrangements of the endothelial cytoskeleton. Moreover, adherence-blocking mAbs directed against the leukocyte counter-receptors of P-selectin, E-selectin, and VCAM-1 inhibited both endothelial [Ca²⁺]_i changes and cytoskeletal changes induced by the adhering PMNs or monocytes.

Yoshida et al. (1996) previously demonstrated a role for E-selectin in outside-in signaling. They observed that clustering of E-selectin on the endothelial surface caused its anchoring to the actin cytoskeleton. In their study, treatment of HUVEC with a mAb to a functional epitope in the extracellular domain was not alone sufficient to cause association of E-selectin with the cytoskeleton; cross-linking by a secondary polyclonal antibody was required. In contrast, by using another adherence-blocking mAb (CBL180) to E-selectin, we were able to induce [Ca²⁺]_i changes and formation of stress fibers in HUVEC without a secondary cross-linking antibody. These results suggest that the outside-in signaling function of E-selectin leading to [Ca²⁺]_i changes and formation of stress fibers is not linked to its association to the cytoskeleton. Moreover, our results establish the participation of the E-selectin epitope recognized by mAb CBL180 in outside-in signaling. Similarly, the leukocyte adherence-blocking mAbs to P-selectin (mAb CBL474) and VCAM-1 (mAb 4B9) were by themselves effective in triggering endothelial [Ca²⁺]_i changes and formation of stress fibers, indicating that they identify functional epitopes in outside-in signaling. It is important to note that virtually all leukocyte populations are endowed with adherence mechanisms recognizing E-selectin, P-selectin, or VCAM-1 (Carlos and Harlan, 1994), thus emphasizing the potential role of outside-in signaling by these receptors in regulating the leukocyte traffic across the endothelial barrier. It is also worth noting that the overlapping function of P- and E-selectin with respect to outside-in signaling is consistent with recent observations in gene-targeted animals with E- and/or P-selectin deficiency. Neutrophil extravasation into inflamed peritoneum was minimally altered in either the P- or the E-deficient mice, whereas it was markedly reduced in animals deficient in both selectins (Frenette et al., 1996; Munoz et al., 1997; Subramaniam et al., 1997).

Huang et al. (1993) were the first to demonstrate that leukocyte adherence to HUVEC induced a transient increase in endothelial [Ca²⁺]_i. Since [Ca²⁺]_i changes were induced by leukocytes equally well in the presence and in the absence of serum, these authors concluded that oxidants and/or proteases secreted by leukocytes were unlikely to be the effectors of endothelial stimulation. Rather, they noted that the changes in endothelial [Ca²⁺]_i coincided temporally with leukocyte transit across the endothelial monolayers, suggesting that endothelial stimulation was due in part to adhesive interactions of endothelial cells with diapedesing leukocytes. Recently, Muller et al. (1993) demonstrated that the homophilic interaction of leukocyte and endothelial PECAM-1 is crucial to the process of PMN and monocyte migration through intercellular junctions of endothelial cells. Therefore, in investigating the potential involvement of adhesion molecules in endothelial activation by leukocytes we expected that PECAM-1 would be involved. However, treatment of HUVEC with the anti–PECAM-1 mAb M89D3, which inhibits leukocyte migration across HUVEC monolayers (Zocchi et al., 1996), did not cause [Ca²⁺]_i changes nor induce the formation of stress fibers. In addition, endothelial [Ca²⁺]_i changes and the formation of stress fibers induced by the addition of PMN and monocytes to the EC monolayers were not inhibited by a pretreatment of the leukocytes with mAb M89D3. Since changes in surface expression of PECAM-1 were described in HUVEC exposed to various cytokines (Ferrero et al., 1996; Rival et al., 1996), we also tested the anti–PECAM-1 mAb on LPS-treated HUVEC, but again observed no effect. Similarly, negative results were obtained when clustering of PECAM-1 was induced by a secondary antibody. Nevertheless, we cannot exclude a potential contribution of PECAM-1 to signaling since: (1) the epitope recognized by M89D3, although functionally relevant in leukocyte–HUVEC adhesive interactions, might be distinct from the epitope(s) involved in outside-in signaling, and (2) the stimulation of the EC monolayers by PMN or monocytes pretreated with the anti–PECAM-1 mAb is likely dependent upon the interaction of the leukocytes with endothelial receptors other than PECAM-1 (e.g., selectins and/or VCAM-1), thereby obscuring any effect of the leukocyte pretreatment.

In contrast to PECAM-1, the inhibitory effect of leukocyte pretreatment with the anti-sLe^x or anti–VLA-4 mAb provides convincing evidence for the participation of their endothelial counter-structures in outside-in signaling. Notably, these treatments did not inhibit leukocyte adherence to HUVEC by the CD11/CD18–ICAM-1 adherence mechanism, nor leukocyte diapedesis. Regardless, with these mAbs leukocytes were unable to induce [Ca²⁺]_i changes or formation of stress fibers in the endothelial monolayers, thus excluding any major role in [Ca²⁺]_i signaling for PECAM-1 as well as for other molecules at the inter-endothelial junctions. A recent report by Gurubhagavatula et al. (1998) showed that an anti–PECAM-1 antibody (mAb 4G6) induced a slow but sustained increase in HUVEC [Ca²⁺]_i. Notably, the kinetics of the responses to mAb 4G6 were quite different from those induced by thrombin. In contrast, in all published reports to date, leukocyte adherence was shown to induce rapid, large changes in HUVEC [Ca²⁺]_i, comparable to those induced by agents such as histamine or thrombin. Hence, the study by Gurubhagavatula et al. (1998) supports our contention that endothelial molecules other than PECAM-1 function as signaling receptors for adhering leukocytes. Since anti-PECAM antibodies inhibit leukocyte migration across EC at the level of the junctions (Muller et al., 1993), our results suggest that changes in endothelial [Ca²⁺]_i and actin morphology are necessary, but not sufficient, to facilitate leukocyte diapedesis.

Four groups of investigators already investigated the involvement of ICAM-1, the endothelial ligand for CD11a/ CD18 and CD11b/CD18, in endothelial outside-in signaling. Durieur-Trautmann et al. (1994) reported that clustering of ICAM-1–induced tyrosine phosphorylation of cytoskeletal proteins potentially involved in the arrangement of inter-endothelial junctions, such as cortactin and src. More recently, the mechanisms leading to opening of endothelial adherence junctions during leukocyte diapedesis were investigated by Del Maschio et al. (1996) who did not observe changes in the junctional structures when ICAM-1 was clustered on the endothelial surface. Ziegelstein et al. (1994) and Pfau et al. (1995) examined the effect of anti-CD18 mAb, and anti–ICAM-1 mAb, respectively, on changes in endothelial [Ca²⁺]_i induced by the adherence of leukocytes. Ziegelstein et al. (1994) found that [Ca²⁺]_i changes induced by PMN adherence were partially (∼50%) inhibited by anti-CD18 mAb. Similarly, Pfau et al. (1995) observed that changes in HUVEC [Ca²⁺]_i caused by the adherence of natural killer lymphocytes were partially inhibited by an anti–ICAM-1 mAb. In the present study, we observed some degree of inhibition of the changes in HUVEC [Ca²⁺]_i induced by leukocytes by anti–ICAM-1 mAb only when PMN were added to histamine-stimulated HUVEC. However, the inhibition was weak and inconsistent, and did not achieve statistical significance. It should be noted that leukocytes treated with anti-sLe^x and/or anti–VLA-4 mAb were unable to induce endothelial [Ca²⁺]_i changes or stress fibers formation, even though CD11a/CD18 and CD11b/CD18 were free to interact with endothelial ICAM-1. From this result we suggest that the interaction of leukocyte CD11/CD18 with endothelial ICAM-1 acts primarily as a co-stimulus for endothelial [Ca²⁺]_i changes.

Formation of stress fibers was described in ECs undergoing mechanical stress (Sato and Ohshima, 1994; Cucina et al., 1995), or ECs exposed to soluble inflammatory agonists, such as histamine and thrombin (Rotrosen and Gallin, 1986; Kolodney and Wysolmerski, 1992). A novel finding in the present study is that PMN or monocyte adherence also induces rearrangements of endothelial actin filaments, leading to formation of stress fibers. Gross changes in endothelial morphology were described to occur during leukocyte adherence to and passage across the endothelial lining of blood vessels (Marchesi and Florey, 1960). More recently, Hixenbaugh et al. (1997) demonstrated that neutrophil adhesion to HUVEC induces endothelial myosin light chain phosphorylation and contraction of the monolayers. However, the mechanism(s) by which adhering leukocytes induce endothelial contraction have not been well characterized. Our evidence linking adhesion receptors, [Ca²⁺]_i changes, and formation of stress fibers contributes to our understanding of these mechanisms.

We acknowledge Dr. F. Ruzzier for helpful suggestions during the course of this work. We thank Dr. C. Vecile for supplying us with his software for analysis of our recordings and calculations. We also thank the Fondazione Callerio for hospitality.

This work was supported in part by grants provided from Italian Ministero della Università e della Ricerca Scientifica e Tecnologica (MURST 40% and 60%) to A. Dobrina and F. Tedesco and CNR (contract No. 97.01234.PF49) to F. Tedesco.

EC

endothelial cells

HUVEC

human umbilical vein endothelial cells

ICAM

PECAM, VCAM, intercellular, platelet/endothelial, vascular cell adhesion molecule

LPS

lipopolysaccharide

PMN

polymorphonuclear leukocyte

sLe^x

sialyl Lewis x