Lymphocytes migrate from the blood across endothelial cells to reach foreign substances sequestered in peripheral lymphoid organs and inflammatory sites. To study intracellular signaling in endothelial cells during lymphocyte migration, we used murine endothelial cell lines that promote lymphocyte migration and constitutively express VCAM-1. The maximum rate of resting splenic lymphocyte migration across monolayers of the endothelial cells occurred at 0–24 h. This migration was inhibited by anti-VCAM-1 or anti-α4 integrin, suggesting that VCAM-1 adhesion was required for migration. To determine whether signals within the endothelial cells were required for migration, irreversible inhibitors of signal transduction molecules were used to pretreat the endothelial cell lines. Inhibitors of NADPH oxidase activity (diphenyleneiodonium and apocynin) blocked migration >65% without affecting adhesion. Because NADPH oxidase catalyzes the production of reactive oxygen species (ROS), we examined whether ROS were required for migration. Scavengers of ROS inhibited migration without affecting adhesion. Furthermore, VCAM-1 ligand binding stimulated NADPH oxidase-dependent production of ROS by the endothelial cells lines and primary endothelial cell cultures. Finally, VCAM-1 ligand binding induced an apocynin-inhibitable actin restructuring in the endothelial cell lines at the location of the lymphocyte or anti-VCAM-1-coated bead, suggesting that an NADPH oxidase-dependent endothelial cell shape change was required for lymphocyte migration. In summary, VCAM-1 signaled the activation of endothelial cell NADPH oxidase, which was required for lymphocyte migration. This suggests that endothelial cells are not only a scaffold for lymphocyte adhesion, but play an active role in promoting lymphocyte migration.
Lymphocyte recirculation is crucial for the regulation of an effective immune response, because lymphocytes must reach foreign substances sequestered in peripheral lymphoid organs and inflammatory sites. The first step in this process is the adhesion of circulating lymphocytes to receptors on high endothelial venule (HEV)3 cells in lymph nodes or cytokine-activated endothelial cells in inflammatory sites. Adhesion is mediated by several receptors. Selectins bind to addressins on the endothelial cells, and integrins bind to members of the Ig superfamily of molecules, such as ICAM-1 and VCAM-1 (1, 2). The combination of these receptors is thought to determine the specificity of lymphocyte migration into tissues (2). Subsequent to adhesion, lymphocytes migrate between adjacent endothelial cells and into the tissue. Leukocyte migration across some endothelial cells uses adhesion molecules such as the homophilic adhesion molecule PECAM-1 (platelet endothelial cell adhesion molecule-1) (3). It has also been suggested that lymphocytes may migrate on the adhesion molecule VCAM-1 on endothelial cells, because lymphocytes migrate on VCAM-1/Fc on a solid support (4).
Despite significant advances in our understanding of adhesion events, much less is known about the mechanism(s) for migration, especially regarding the role of the endothelial cell in this process. A few studies suggest that intracellular signaling pathways in endothelial cells are required for endothelial cell promotion of lymphocyte migration. For example, neutrophil migration across HUVECs requires activation of endothelial cell calcium/calmodulin-dependent myosin light chain kinase (5), but the receptor(s) that mediates this signal is not known. It has also recently been shown that lymphocyte adhesion to ICAM-1 on brain endothelial cells activates the small GTP binding protein, Rho, and this Rho activity is required for lymphocyte migration across the endothelial cells (6). A signaling pathway has also been reported for VCAM-1. Ab cross-linking of VCAM-1 on the surface of HUVEC induces a calcium flux (7). However, it is not known whether this VCAM-1-mediated signaling is important for lymphocyte migration.
Signaling via adhesion molecules may be important for changes in cell shape during transendothelial migration. It has been reported that lymphocyte migration across rat HEV cell layers is an active process, requiring cytoskeletal structural changes, whereas the lymphocyte-HEV cell binding events are passive (8, 9). Furthermore, inhibition of actin filament polymerization by cytochalasin D and inhibition of tubulin elongation by colchicine block spleen lymphocyte migration across monolayers of a rat lymph node HEV cell line (10). It is well established that cytoskeletal proteins, such as actin, vinculin, and tubulin, regulate cell shape, suggesting that these proteins may regulate cell shape changes during lymphocyte migration across endothelial cells. It is not known whether changes in endothelial cell shape are activated by lymphocyte binding to endothelial cell adhesion molecules.
To examine whether VCAM-1 ligand binding activates changes in endothelial cell shape and whether these signals are involved in the regulation of lymphocyte migration, we used continuously cultured endothelial cell lines that were previously developed in our laboratory (11). These endothelial cell lines, mHEVa and mHEVc cells, were derived from BALB/c mouse axillary and cervical lymph nodes, respectively, by spontaneous immortilization. They exhibit contact inhibition and do not form transformed foci. These endothelial cell lines bind resting lymphocytes and then promote the migration of the lymphocytes under the endothelial cell monolayer (11). During lymphocyte migration across the mHEV cell lines, the endothelial cells change shape by retracting their membrane at the site of the lymphocyte, the lymphocyte migrates between adjacent endothelial cells within a few minutes, and the endothelial cells reform their cell-cell junction (11). The mechanism for lymphocyte adhesion to the mHEV cell lines is by lymphocyte α4 integrin binding to VCAM-1 on the mHEV cells and not to any other known adhesion molecule (12). The expression of VCAM-1 by the mHEV cell lines is constitutive (11), in contrast to other endothelial cells that require cytokine activation of VCAM-1 expression. Importantly, this constitutive VCAM-1 expression by the mHEV cell lines enables the examination of intracellular signals transduced by VCAM-1 without the complications of signals for activation of VCAM-1 expression. Thus, these cell lines were used as a simplified model to examine lymphocyte interactions with VCAM-1 on endothelial cells. In this report we demonstrate that adhesion to VCAM-1 on these endothelial cell lines induced endothelial cell NADPH oxidase activity, which was required for lymphocyte migration and endothelial cell actin-mediated cell shape changes. In addition, the VCAM-1-mediated activation of NADPH oxidase was not required for the initial lymphocyte adhesion to the endothelial cells.
Materials and Methods
Male BALB/c mice, 4–6 wk old (Harlan Industries, Indianapolis, IN), were the source of resting splenic lymphocytes.
The following inhibitors were obtained from Biomol (Plymouth Meeting, PA): wortmannin, diphenyleneiodonium chloride (DPI), herbimycin A, fluphenazine, phenoxybenzamine, methoxsalin, and troleandomycin. Other inhibitors included apocynin and allopurinol (Acros Organics, Pittsburgh, PA); NG-methyl-l-arginine (l-NMMA), and N5-(1-iminoethyl)-l-ornithine (Molecular Probes, Eugene, OR); and superoxide dismutase and catalase (Sigma, St. Louis, MO.) Purified rat anti-mouse VCAM-1 (clone MVCAM.A), rat IgG2a, rat anti-mouse α5 integrin (clone MFR5), FITC-conjugated rat anti-mouse CD45, and mouse anti-human VCAM-1 (clone 51-10C9) were obtained from PharMingen (San Diego, CA). Rat anti-mouse VLA-4 (clone PS-2) and rat anti-mouse CD44 (clone KM201) were obtained from BioDesign International (Kennebunk, ME). Mouse anti-human PECAM-1 (clone JC/70A) was purchased from Dako (Carpinteria, CA). Biotin-conjugated goat anti-rat Ig and biotin-conjugated goat anti-mouse IgG1 were obtained from Southern Biotechnology Associates (Birmingham, AL). FITC- and tetramethylrhodamine isothiocyanate (TRITC)-phalloidin were purchased from Sigma. Dihydrorhodamine 123 (DHR) was obtained from Molecular Probes. Streptavidin-conjugated 10.4-μm beads were purchased from Bangs Laboratories (Fishers, IN).
Cells and culture medium
Two endothelial cell lines, mHEVa and mHEVc, were previously derived from BALB/c mice axillary or cervical lymph nodes (11). Spleen cells were isolated as previously described (11). B78H1 cells (derived from B16 mouse melanoma) were a gift from Dr. Lloyd Graff (Department of Biochemistry, University of Illinois, Chicago, IL). The cells were incubated at 37°C and 6% CO2 in culture medium consisting of RPMI 1640 (Fisher Scientific, Cincinnati, OH) supplemented with 20% FCS, 2 mM glutamine (Sigma), 1 mM HEPES (Sigma), 10 mM sodium bicarbonate (Sigma), 100 U/ml penicillin (Fisher Scientific), 100 μg/ml streptomycin (Fisher), and 50 μg/ml gentamicin (Life Technologies, Grand Island, NY). For phenol red-free culture medium, phenol red-free RPMI 1640 (Life Technologies) was substituted for RPMI 1640 in the above culture medium. HUVECs were obtained from Clonetics (Walkersville, MD) and grown in endothelial growth medium (Clonetics) plus 5% FCS. The HUVECs were used at passage 4 or 5.
Lymphocyte migration assay
mHEVa cells or mHEVc cells were plated and grown to confluence in the upper chamber of Transwells with 12-μm pores (Costar Cambridge, MA). Inhibitors were added to both the upper and lower chambers of the Transwells as described in Results. After pretreatment with the irreversible inhibitors, the mHEV monolayers were washed by placing the Transwell in a new plate with 1.2 ml of fresh medium in the bottom chamber and 500 μl of fresh medium in the top chamber. After five washes, the last wash was added to a set of untreated wells to determine whether the washes were sufficient to remove effective concentrations of the inhibitor from the medium. This wash had no effect on migration (data not shown). Splenic lymphocytes (4 × 106 optimal dose) and 4–8 × 106 splenic RBC were added to the upper chamber on top of the mHEV monolayer. The cells were incubated at 37°C under static (nonflow) conditions. RBC served as a control for confluence of the monolayer. The RBC are smaller than the lymphocytes and do not migrate. Therefore, on the rare occasion that RBC were in the bottom chamber, the monolayer was not confluent, and the Transwell was discarded. Lymphocytes were collected from the bottom chamber and counted at the indicated times. Asynchronous lymphocyte migration occurs up to 48 h, whereas migration by a particular lymphocyte occurs in minutes (11). The effects of treatments on migration are only compared within an experiment, as the number of migrated lymphocytes varies among experiments but varies little among triplicate determinations within an experiment. Each experiment was performed at least twice. The inhibitors at the concentrations used had no affect on lymphocyte viability or mHEV cell viability, as determined by trypan blue exclusion (data not shown).
Lymphocyte adhesion assay
As previously described (11), the mHEV cell lines and the control cell line B78H1 were grown to confluence in 96-well, tissue culture-treated plates (Corning, Corning, NY). RBCs in spleen cell preparations were lysed by hypotonic shock with water and washed with culture medium. RBCs were not lysed by hypotonic shock with the ammonium chloride method (13), because it increased nonspecific binding in the adhesion assay. The splenic lymphocytes were then labeled with calcein acetoxymethyl ester (AM) (1 μM) for 15 min at 37°C and washed twice with PBS. Calcein-acetoxymethyl ester (AM) is a vital dye that is membrane permeable but becomes membrane impermeable and fluorescent when cleaved by intracellular esterases. Lymphocytes (1 × 106) were added to each well, then were then gently centrifuged onto the monolayers at 500 rpm without the brake. Centrifugation is not required for the adhesion, but it initiates simultaneous interactions between lymphocyte and cell monolayers. The cells were then incubated for 30 min at 37°C. To remove the nonadherent cells, the plates were gently vortexed twice for 2 s on a Genie Vortex I (Fisher) at setting 5. The medium containing the nonbound cells was then removed, and 100 μl of PBS supplemented with 0.2 mM CaCl2 and 0.1 mM MgCl2 was added. The plates were read on a microplate fluorometer (Cambridge Technology International, Watertown, MA).
Reactive oxygen species
mHEV cells or HUVECs were grown to confluence on 35-mm petri dishes constructed with bottoms of 22-mm round glass coverslips. In experiments with HUVECs, the endothelial cells were stimulated overnight with 1 ng/ml TNF-α to induce VCAM-1 expression. As expected, the TNF-α-treated HUVECs expressed VCAM-1 as determined by immunofluorescent labeling of HUVECs and fluorescent microscopy (data not shown). The mHEV cells and HUVECs were preloaded with DHR (1 or 1.5 μM) in phenol red-free culture medium 15 min before stimulation of the endothelial cells. DHR fluoresces when oxidized (14, 15). The mHEV cells were incubated with 5 × 106 lymphocytes, with anti-VCAM-1-coated 10.4-μm beads, or with anti-CD44-coated 10.4 μm beads as controls. To Ab-coat beads, 40 μl of streptavidin-conjugated 10.4 μm beads (Bangs Laboratories) were labeled with 6 μg of biotin-conjugated goat anti-rat Ig in 75 μl of PBS with gentle rocking for 1 h at 4°C and then washed. These beads were incubated with 8 μg of rat anti-mouse VCAM-1 or a rat control Ab (anti-mouse CD44) in 40 μl of PBS for 30 min at 4°C with gentle rocking and then washed. For experiments with the HUVECs, 40 μl of streptavidin-conjugated 10.4-μm beads (Bangs Laboratories) were incubated with 12 μg of biotin-conjugated goat anti-mouse IgG1 in 75 μl of PBS with gentle rocking for 1 h at 4o C. These beads were washed and incubated with 8 μg of mouse anti-human VCAM-1 or a control Ab (mouse anti-human PECAM-1) in 40 μl of PBS for 30 min at 4o C with gentle rocking and then washed. HUVECs, preloaded with DHR, were incubated with the anti-human VCAM-1-coated beads. Rhodamine 123 fluorescence was examined as previously described (14) at 0–40 min using a confocal microscope equipped with a ×100 objective (Leica TCS 4D microscope/SCANware system, Heidelberg, Germany) equipped with an Omnichrome krypton-argon laser (Chino, CA). After each time point, the field of cells was changed to avoid laser-induced rhodamine 123 fluorescence in the next image.
Mouse HEV cells were grown to confluence on Permanox eight-well chamber slides (Corning). Lymphocytes were added to the wells and incubated 0, 5, 15, or 30 min at 37°C, and the nonbound cells were removed. The bound cells were fixed with 3% paraformaldehyde/2% sucrose and then permeabilized for 5 min on ice with 0.1% Triton X-100. Lymphocyte and mHEV cell actin was incubated with 0.1 μg/ml TRITC-phalloidin for 40 min on ice and washed. Lymphocytes were labeled with 2.5 μg/ml FITC-conjugated rat anti-mouse CD45. Fluorescence was examined by confocal microscopy.
To examine anti-VCAM-1-coated bead induction of actin coalescence, the mHEV cells were incubated with 5 × 106 anti-VCAM-1-coated beads for 30 min at 37°C. The nonbound beads were removed, and the mHEV cells were fixed, permeabilized, and labeled with FITC-phalloidin as described above. Fluorescence was examined by confocal microscopy. The bead location was examined using reflected light and the confocal microscope.
Data were analyzed by ANOVA followed by a multiple comparisons test (SigmaStat, Jandel Scientific, San Ramon, CA). The specific statistical tests are indicated in the figure legends.
Lymphocyte adhesion is required for lymphocyte migration across two endothelial cell lines
Lymphocytes bind to endothelial cells and then migrate between adjacent endothelial cells. We have reported that lymphocytes bind and migrate across monolayers of the endothelial cell lines, mHEVa and mHEVc (11). We have also previously determined by Ab inhibition studies that adhesion of resting lymphocytes to these endothelial cell lines is dependent on lymphocyte α4 integrin binding to VCAM-1 on the mHEV cells (12). Therefore, we determined whether adhesion via α4 integrin to VCAM-1 was required for lymphocyte migration. Mouse HEV cells were grown to confluence in the upper chamber of Transwells, and resting spleen cells were added. The upper chamber was treated every 4 h with either 4 μg of anti-α4 integrin, 4 μg of anti-VCAM-1, 4 μg of an isotype control Ab, or no Ab (Fig. 1). In nontreated controls, 2–10% of the lymphocytes migrated in 24 h, which is similar to other studies with endothelial cell lines or cytokine-activated microvascular endothelial cells (16, 17, 18, 19). Furthermore, because about half of the lymphocytes bind to the mHEV cells (11), this results in about 4–25% of the bound cells migrating. Anti-α4 integrin and anti-VCAM-1 significantly blocked lymphocyte migration across the mHEVa and mHEVc cells, whereas the isotype control Ab had no effect on migration (Fig. 1). This indicates that adhesion was required for migration across these cell lines.
It is possible that lymphocytes also migrate on VCAM-1 to cross the mHEV cell monolayer, because lymphocytes migrate on VCAM-1/Fc on a solid support (4). However, we could not allow adhesion and then treat the cells with the anti-α4 integrin or anti-VCAM-1 Abs to determine whether these molecules are also involved in migration, because these Abs reversed lymphocyte adhesion to the mHEV cell lines (data not shown). Furthermore, CD44, which mediates migration of activated lymphocytes into inflammatory sites as well as across cytokine-activated endothelial cells in vitro (20, 21), did not inhibit resting lymphocyte migration across the mHEV cell lines, as determined using an anti-CD44 blocking Ab (clone IM7; data not shown) (11). It is also possible that lymphocytes migrate between the endothelial cells on fibronectin. Inhibition of one fibronectin receptor VLA-5 (α5β1 integrin) with blocking anti-α5 integrin (clone MFR5) had no effect on lymphocyte migration (data not shown). This could not be tested further with CS-1 peptides that block integrin binding to fibronectin, because these peptides also block binding to VCAM-1 (22). Furthermore, RGD-containing peptides, which block binding to fibronectin, reversed endothelial cell attachment to the solid support, causing the monolayer to dissociate from Transwells or 96-well plates (data not shown). In summary, resting lymphocyte α4 integrin binding to VCAM-1 on the mHEV cells was required for lymphocyte migration.
Lymphocyte migration across endothelial cell lines requires endothelial cell NADPH oxidase-catalyzed production of ROS
The role of VCAM-1 as a receptor for α4 integrin during lymphocyte adhesion has been clearly established (1, 2). However, it is not known whether ligand binding to VCAM-1 initiates intracellular signals in the endothelial cell that are required for endothelial cell promotion of lymphocyte migration. Therefore, our approach was to determine whether inhibitors of signal transduction molecules blocked lymphocyte migration and then determine whether VCAM-1 activated these signal transduction molecules in endothelial cells. Because VCAM-1 is a member of the Ig gene superfamily and other members of this superfamily, such as the Ig receptor and the TCR, use tyrosine kinases to initiate signal transduction, we first examined endothelial cell tyrosine kinase involvement in lymphocyte migration. To determine whether tyrosine kinase activity in the endothelial cells was required for lymphocyte migration, an irreversible inhibitor of tyrosine kinase (herbimycin A) was preincubated with the mHEV cells, the cells were washed, and then lymphocyte migration across monolayers of the mHEV cells was examined. This allowed the examination of the effect of the inhibitor on one cell type during coculture of the lymphocytes with the mHEV cells. Lymphocyte migration was not affected when mHEV cells were pretreated with the irreversible tyrosine kinase inhibitor herbimycin A (1–5 μM; Fig. 2) (23, 24). We then examined whether calmodulin was required for lymphocyte migration. It has been reported that Ab cross-linking of VCAM-1 initiates an endothelial cell calcium flux, but it is not known whether this calcium flux is required for lymphocyte migration. Furthermore, HUVEC calcium/calmodulin-dependent myosin light chain kinase is required for neutrophil migration (5), but the receptor(s) that mediates this signal is not known. Therefore, we examined whether irreversible inhibitors of calmodulin blocked endothelial cell promotion of lymphocyte migration. Pretreatment of the endothelial cells with the irreversible calmodulin inhibitors phenoxybenzamine (10–50 μM) (25) or fluphenazine (25–50 μM) (26) did not inhibit lymphocyte migration (Fig. 2). Pretreatment of the endothelial cells with the irreversible phosphatidylinositol 3-kinase (PI3-K) inhibitor wortmannin (0.5–1 μM) (27, 28) also did not block lymphocyte migration (Fig. 2). In contrast, pretreatment of lymphocytes with these inhibitors significantly decreased lymphocyte migration across the mHEV cells (Fig. 2), but did not affect lymphocyte adhesion to the mHEV cell lines (data not shown), suggesting that, subsequent to the initial cell adhesion event, these signaling molecules in lymphocytes were required for migration. The inhibition of lymphocyte function during migration also indicates that the inhibitors were active. Although the inhibitors had no effect on endothelial cells for promotion of lymphocyte migration, the inhibitors were used at concentrations reported to modulate other endothelial cell functions (29, 30, 31, 32). Furthermore, at these concentrations, the inhibitors had no effect on lymphocyte or mHEV cell viability as determined by trypan blue exclusion (data not shown). Thus, these results are consistent with a requirement for activation of lymphocyte, but not endothelial cell, tyrosine kinase, calmodulin, and PI3-K during lymphocyte migration across the endothelial cell lines.
Recently, ROS such as NO and superoxide have been recognized as potential signal transduction molecules (33). Furthermore, endothelial cells have been reported to produce ROS (34, 35) for the modulation of smooth muscle cell shape (36). Therefore, we determined whether pretreatment of endothelial cells with an irreversible inhibitor of enzymes that catalyze the production of ROS could block lymphocyte migration and thus potentially modulate the endothelial cell shape changes observed during lymphocyte migration (11). The mHEV cells or lymphocytes were pretreated for 30 min with DPI (1 or 5 μM). DPI inhibits both NO synthase (NOS) and NADPH oxidase (37, 38), which catalyze the production of NO and superoxide, respectively. The cells were washed five times, and migration was examined. The last wash was added to nontreated cells and had no effect on migration (data not shown), indicating that effective concentrations of the inhibitor had been removed by the washes. DPI pretreatment of the mHEV cells exhibited dose-dependent inhibition of lymphocyte migration across the mHEVa cells and mHEVc cells (Fig. 3). In contrast, pretreatment of lymphocytes with DPI had no effect on lymphocyte migration across the mHEV cells (data not shown). Because VCAM-1 is required for migration, it was determined whether DPI affected VCAM-1 expression. Pretreatment of mHEV cells with DPI for 30 min did not affect VCAM-1 expression 24 h later as determined by immunofluorescent labeling and flow cytometry (98 ± 1% of the mHEVa cells expressed VCAM-1 with a mean fluorescence intensity of 318 ± 33; 80 ± 4% of the mHEVc cells expressed VCAM-1 with a mean fluorescence intensity of 53 ± 8). To determine which enzyme was important for migration, reversible inhibitors that were specific for NOS and NADPH oxidase were used. Coincubation with reversible inhibitors was used, because DPI had no effect on the lymphocytes. The arginine analogues l-NMMA (5–1000 μM) and N5-(1-iminoethyl)-l-ornithine (0.5–100 μM), which inhibit NOS (39), did not reduce lymphocyte migration (data not shown). Apocynin, which inhibits NADPH oxidase (40), exhibited dose-dependent (0.5–4 mM) inhibition of lymphocyte migration across the mHEVa and mHEVc cells (Fig. 3), suggesting that endothelial cell NADPH oxidase activity is required for lymphocyte migration. Apocynin had no effect on mHEV cell expression of VCAM-1 after 24 h as determined by immunofluorescent labeling and flow cytometry (95 ± 5% of the mHEVa cells expressed VCAM-1 with a mean fluorescence intensity of 271 ± 70; 75 ± 12% of the mHEVc cells expressed VCAM-1 with a mean fluorescence intensity of 46 ± 9). Inhibitors of other enzymes (xanthine oxidase and cytochrome P450) that catalyze the production of ROS were also examined. These inhibitors were used at concentrations reported to modulate endothelial cell functions other than promotion of lymphocyte migration (41, 42, 43, 44, 45). Inhibitors of xanthine oxidase (100–300 μM allopurinol) (46, 47) and cytochrome P450 (0.03–0.3 μg/ml methoxsalen or 10–20 μM troleandomycin) (48, 49, 50) did not affect lymphocyte migration or cell viability (data not shown).
To further support the hypothesis that NADPH oxidase is important for a signaling cascade within mHEV cells, scavengers of ROS were used. Scavenging of superoxide with superoxide dismutase or scavenging of its metabolite hydrogen peroxide with catalase inhibited lymphocyte migration across the mHEVa cells and mHEVc cells (Fig. 3). Therefore, NADPH oxidase production of ROS by mHEV cells was required for lymphocyte migration.
Lymphocyte adhesion stimulated the production of ROS by endothelial cells
To demonstrate whether ROS were generated by lymphocyte binding to mHEV cells, the production of low levels of intracellular ROS in the mHEV cells was examined using the reactive oxygen-sensitive fluorescent indicator DHR and confocal microscopy. The mHEV cell monolayers were preloaded with DHR for 15 min. Lymphocytes were added in the presence and the absence of the NADPH oxidase inhibitor apocynin. At 5, 10, 20, 30, and 40 min, rhodamine 123 fluorescence was examined by confocal microscopy. Lymphocytes stimulated detectable accumulation of endothelial cell rhodamine 123 fluorescence at 20–30 min after addition of lymphocytes compared with nonstimulated endothelial cells incubated for the same length of time (Fig. 4). Furthermore, apocynin inhibited this generation of rhodamine 123 fluorescence (Fig. 4), indicating that mHEV cell NADPH oxidase was required for lymphocyte-stimulated production of ROS. Therefore, lymphocyte adhesion stimulated endothelial cell production of ROS.
Lymphocytes bind to the endothelial cell lines via the adhesion molecule VCAM-1. Therefore, to determine whether lymphocyte adhesion to VCAM-1 was required for the production of ROS, VCAM-1 was blocked with soluble anti-VCAM-1, and lymphocyte stimulation of rhodamine 123 fluorescence in mHEV cell monolayers was examined. Anti-VCAM-1 inhibited lymphocyte stimulation of rhodamine 123 fluorescence in the endothelial cells (Fig. 4). As controls, rhodamine 123 fluorescence of anti-VCAM-1-treated mHEV cell monolayers was examined without the addition of lymphocytes. Soluble anti-VCAM-1 did not stimulate accumulation of rhodamine 123 fluorescence compared with that in nonstimulated endothelial cells (Fig. 4). Therefore, VCAM-1-mediated adhesion was required for lymphocyte stimulation of endothelial cell NADPH oxidase-dependent synthesis of ROS.
Cross-linking VCAM-1 stimulates the production of ROS by endothelial cells
VCAM-1 as well as fibronectin are constitutively expressed on the surface of the mHEV cell lines (11). To determine whether VCAM-1 signaling could induce the activation of NADPH oxidase, anti-VCAM-1-coated 10.4-μm diameter beads were used to cross-link VCAM-1 on the endothelial cell surface. The mHEV cell monolayers were preloaded for 15 min with DHR. Anti-VCAM-1-coated beads or anti-CD44-coated control beads were added to the mHEV cell monolayers, and rhodamine 123 fluorescence was examined by confocal microscopy at 5, 15, 30, and 60 min. The beads bound to the endothelial cells and were not phagocytosed by the endothelial cells (data not shown). Cross-linking of VCAM-1 stimulated detectable accumulation of rhodamine 123 fluorescence by 15–30 min compared with that in nonstimulated endothelial cells (Fig. 5). Anti-CD44-coated control beads did not stimulate rhodamine 123 fluorescence (Fig. 5). The anti-VCAM-1 bead-stimulated fluorescence was inhibited by the addition of the NADPH oxidase inhibitor, apocynin (Fig. 5), indicating that signals initiated by VCAM-1 stimulated NADPH oxidase activity in the mHEV cells for the production of ROS.
Primary endothelial cell cultures were used to confirm that VCAM-1 activates endothelial cell NADPH oxidase. Confluent monolayers of HUVECs were incubated with 1 ng/ml TNF-α overnight to induce VCAM-1 expression. The TNF-α-treated HUVECs expressed VCAM-1, as determined by immunofluorescent labeling and fluorescent microscopy (data not shown). Cross-linking VCAM-1 on TNF-α-treated HUVECs with anti-human VCAM-1-coated beads stimulated accumulation of rhodamine 123 fluorescence by 30 min (Fig. 6) as examined by confocal microscopy at 5, 15, 30, and 45 min. The anti-VCAM-1 bead-stimulated fluorescence was blocked by the NADPH oxidase inhibitor, apocynin (Fig. 6), indicating that VCAM-1 cross-linking activates endothelial cell NADPH oxidase. Anti-PECAM-1-coated control beads did not stimulate rhodamine 123 fluorescence (Fig. 6). In summary, VCAM-1 cross-linking activated endothelial cell NADPH oxidase.
Lymphocytes and anti-VCAM-1-coated beads stimulate an NADPH oxidase-dependent coalescence of actin in the endothelial cells at the site of contact
We have shown by time lapse confocal microscopy that during lymphocyte migration across mHEV cells, the mHEV cell changes shape by retracting its membrane, the lymphocyte migrates between adjacent mHEV cells within a few minutes, and the mHEV cells reform their cell-cell junctions (11). Cell shape changes are known to involve rearrangement of the cytoskeletal structure of the cell (51). Furthermore, membrane ruffling as well as NADPH oxidase activation have been shown to be regulated by the G protein Rac-1 (52, 53). Therefore, we hypothesized that VCAM-1 activates NADPH oxidase to produce low concentrations of localized ROS that modulate actin structure and therefore mHEV cell shape. To examine this, it was determined whether cross-linking VCAM-1 activated changes in mHEV cell actin structure and whether these changes could be blocked by the NADPH oxidase inhibitor apocynin. Lymphocytes or anti-VCAM-1-coated beads were added to mHEV monolayers and allowed to adhere for 5 min. The cells were fixed with paraformaldehyde and permeabilized. The actin in these cells was then labeled with TRITC-phalloidin and examined by confocal microscopy. The mHEV cell actin locally coalesced around the bound bead or lymphocyte (Fig. 7), whereas the actin structure in the center of the mHEV cells did not change (Fig. 7). The induction of actin coalescence was inhibited by apocynin or cytochalasin D (Fig. 7). Therefore, adhesion to VCAM-1 caused changes in the endothelial cell cytoskeleton, and these changes were dependent on NADPH oxidase activity. In summary, the data indicate that lymphocyte binding to VCAM-1 on the endothelial cell lines stimulated endothelial cell NADPH oxidase for the generation of ROS. This NADPH oxidase activity was required for lymphocyte migration and for endothelial cell shape changes that occur during the migration of lymphocytes.
An essential component for adequate immune regulation is lymphocyte recirculation, which ensures the exposure of effector cells to Ag, selective distribution of effector cells, and control of proliferation and differentiation by cell-cell interactions. Information about the mechanisms by which endothelial cells control lymphocyte recirculation will provide a basis for proposing interventions in migration of lymphomas and metastatic cancer cells. Although several cell surface receptors for lymphocyte adhesion to endothelial cells have been identified, the mechanisms for lymphocyte migration across endothelial cells are not understood. Binding of adhesion molecules on lymphocytes to adhesion molecules on endothelial cells is required for lymphocyte transendothelial cell migration (54). After adhesion, lymphocytes migrate between adjacent endothelial cells. Little is known regarding the function of adhesion molecules on endothelial cells during lymphocyte transmigration. Using a simplified endothelial cell model system that constitutively expresses VCAM-1, thus excluding complications from signals for activation of VCAM-1 expression, we report here that VCAM-1 on endothelial cells activates endothelial cell NADPH oxidase to catalyze the production of ROS. Furthermore, this activity is important for endothelial cell shape changes and lymphocyte transendothelial migration. Anti-VCAM-1-coated beads also stimulate NADPH oxidase activity in primary cultures of TNF-α-treated HUVECs.
In the past, NADPH oxidase has been primarily examined in leukocytes such as neutrophils and macrophages, where it functions to produce large amounts of superoxide for the destruction of pathogens (55). More recently, the production of low subtoxic concentrations of oxygen metabolites has been recognized as a potential cell signaling mechanism (33). For example, superoxide activates mitogenic signaling in Ras-transformed fibroblasts (56), mediates chemotactic responses by fibronectin- or platelet-derived growth factor-stimulated fibroblasts (57, 58), and inhibits phosphatases in phorbol ester-treated macrophages (59). Oxygen metabolites also play a role in signaling in endothelial cells. For example, in endothelial cells, exogenous superoxide stimulates a transient calcium release (60, 61). Furthermore, it has been shown that endothelial cells have NADPH oxidase components that produce low levels of ROS (35, 62, 63, 64). ROS also mediate signals required for induction of endothelial cell gene expression of VCAM-1 and M-CSF (62, 63). However, it is unlikely that ROS modulated VCAM-1 expression in the mHEV cell lines during lymphocyte migration, because mHEV cells constitutively express VCAM-1, and the migration of a lymphocyte across a particular mHEV cell occurs within minutes (11).
Oxygen metabolites have also been shown to modulate actin structure. For example, catalase scavenging of H2O2 blocks FCS-stimulated mesangial cell contraction and production of ROS (65). Furthermore, addition of oxygen to posthypoxic cultured aortic endothelial cells induces reorganization of actin into ruffles, and this actin change is blocked by overexpression of superoxide dismutase, which scavenges superoxide (66).
It is interesting to discuss potential mechanisms for ROS modulation of endothelial cell actin structure. ROS have been shown to inhibit phosphatases and activate metalloproteinases, both of which regulate cell shape and/or cell attachment. H2O2 can directly oxidize phosphatases (67) by oxidizing cysteine residues in the phosphatase catalytic site (68, 69). Phosphatase activity in neutrophils and Fao hepatoma cells is inhibited by exogenous H2O2 or activation of NADPH oxidase (67, 70). Phosphatases regulate many cell functions, including cytoskeletal structure, cell adhesion to extracellular matrix, and cell-cell junctions in fibroblasts, neutrophils, platelets, and endothelial cells (71, 72, 73, 74, 75, 76, 77, 78, 79). For example, cell constriction and reorganization of actin and microtubules in HUVECs are induced by inhibitors of phosphatases PTP1 and PTP2A (80), and cell-cell junction separation occurs when phosphatase PTP1B activity is blocked (81). In addition, the phosphatase inhibitor pervanadate initially induces an increase in endothelial cell phosphotyrosine labeling at cell junctions and an increase in levels of the cytoskeletal proteins vinculin, actin, and plakoglobulin, whereas a prolonged incubation with pervanadate induces dissociation of cell-cell junctions (71). Inhibition of protein phosphatases also decreases endothelial cell barrier function, increases protein phosphorylation, and induces localization of actin at the endothelial cell periphery (82). Therefore, the endothelial cell actin coalescence at the site of VCAM-1 binding may be mediated by reactive oxygen inhibition of phosphatases.
Another potential target of ROS is matrix metalloproteinases (MMPs), which degrade extracellular matrix (ECM), thus altering cell shape (83). MMP degradation of ECM modulates the shape of endothelial cells and endothelial cell growth, because the concentration of ECM regulates capillary endothelial cell growth (84) and endothelial cell spreading (85). Furthermore, T cell adhesion to VCAM-1 on the rat microvascular endothelial cell line RFC or T cell adhesion to recombinant VCAM-1 induces T cell MMP2 mRNA and MMP2 enzyme activity (86, 87). Furthermore, lymphocyte migration across these cells is inhibited by the MMP2 inhibitor TIMP2 (tissue inhibitor of metalloproteinase-2) (86, 87). However, the mechanism for activation of the MMPs during lymphocyte migration across endothelial cells is not known. Latent MMP2 is activated by low concentrations of H2O2 (4 μM), whereas higher concentrations of H2O2 (50 μM) inactivate MMP2 (83). We report that production of H2O2 by endothelial cell lines was required for lymphocyte migration across these endothelial cells that express ECM on their cell surface (11). Therefore, low localized production of H2O2 by endothelial cells may activate local MMP activity to degrade endothelial cell ECM at cell junctions. The endothelial cells would then retract at that site and allow the lymphocyte to migrate beneath the endothelial cells. Future studies will focus on whether ROS inhibit phosphatases and/or activate MMPs for the migration of lymphocytes across endothelial cells.
In summary, VCAM-1 is not simply a scaffold for lymphocyte adhesion, but activates endothelial cell functions that regulate lymphocyte migration. Specifically, VCAM-1 mediates outside-in signaling, and this signaling is via endothelial cell NADPH oxidase activity.
We thank Drs. Leslie Myatt and Lawrence Sherman for critical review of the manuscript.
This work was supported by National Institutes of Health Grants AI34585 and AI40640.
Abbreviations used in this paper: HEV, high endothelial venule; PECAM-1, platelet endothelial cell adhesion molecule-1; TRITC, tetramethylrhodamine isothiocyanate; DHR, dihydrorhodamine 123; DPI, diphenyleneiodonium; ECM, extracellular matrix; l-NMMA, NG-methyl-l-arginine; MMP, matrix metalloproteinases; NOS, NO synthase; ROS, reactive oxygen species; PI3-K, phosphatidylinositol 3-kinase; PTP, protein tyrosine phosphatase.