In vivo, MHC class I-restricted injury of allogeneic tissue or cells infected by intracellular pathogens occurs in the absence of classical cytolytic effector mechanisms and Ab. Modulation of the target cell adhesion to matrix may be an additional mechanism used to injure vascular or epithelial cells in inflammation. We studied the mechanisms of human umbilical vein endothelial cell (EC) detachment from matrix-coated plastic following contact by concanamycin A-treated lymphocytes as an in vitro model of perforin-independent modulation of EC basement membrane adhesion. Human PBL were depleted of monocytes, stimulated, then added to an EC monolayer plated on either fibronectin or type I collagen matrices. Activated, but not resting, PBL induced progressive EC detachment from the underlying matrix. Injury of the EC monolayer required direct cell contact with the activated lymphocytes because no detachment was seen when the PBL were placed above a Transwell membrane. Moreover plasma membranes prepared from activated but not resting PBL induced EC detachment. Adherent EC stimulated with activated PBL did not show evidence of apoptosis using TUNEL and annexin V staining at time points before EC detachment was observed. Finally, neither the matrix metalloproteinase inhibitors o-phenanthroline and BB-94 nor aprotinin blocked EC detachment. However, activation of EC β1 integrin using mAb TS2/16 or Mg2+ decreased EC detachment. These data indicate that cell-cell contact between activated PBL and EC reduces adhesion of EC to the underlying matrix, at least in part by inducing changes in the affinity of the endothelial β1 integrin.
Cytotoxic T cells are thought to play a central role in the control of viral infections and in rejection of allografts (1, 2). CTL injure a target cell largely through one of three mechanisms: the perforin/granzyme, Fas ligand (FasL),3 or TNF pathways; however, CD8+ T cells may provide effective immunity in the absence of these classic cytolytic effector mechanisms. For example, in some cases CD8+ T cells control viral or intracellular bacterial infections in the absence of perforin- and FasL-dependent injury even in the absence of Ab or classic cytokines involved in delayed type hypersensitivity responses (3, 4). Moreover, allograft rejection in the absence of cell-mediated cytotoxicity has been observed in models using adoptive transfer of noncytolytic lymphocytes, and using perforin-deficient recipients to reject grafts deficient in Fas- and TNFR signaling (5, 6, 7).
These observations in vivo suggest a role for direct noncytolytic lymphocyte-mediated injury of target cells. A candidate mechanism to account for cell injury in the absence of cytolysis has been proposed previously by Russell and colleagues (8, 9, 10). They observed that activated lymphocytes induce loss of target cell adhesiveness to extracellular matrix independent of cytolytic damage. Although this form of injury was defined using fibroblast target cells, polarized epithelial and vascular cells may be most vulnerable to injury promoted by loss of matrix adhesion in vivo. In particular, loss of endothelial cell (EC) adhesion to the basement membrane may compromise vessel integrity and tissue perfusion. Focal loss of vascular EC is a feature of robust inflammatory infiltrates such as delayed-type hypersensitivity, antitumor responses, and allograft rejection (11, 12, 13). Shed microvessel EC displaying adhesion molecules characteristic of inflammation have been detected in the circulation (14).
EC adhesion to matrix proteins is mediated by transmembrane receptors of the integrin class that are anchored to the cytoskeleton through the focal adhesion complex. A tyrosine kinase, p125FAK, associated with the β subunit of the β1 family of integrin adhesion molecules is thought to play an important role in the assembly and maintenance of the focal adhesion complex and associated stress fibers (15). Cell contact between monocytes and EC stimulates p125FAK degradation and results in changes in EC architecture in vitro (16). This observation suggests that interaction between leukocytes and EC may influence EC adhesion to matrix.
In the present series of experiments we sought to develop a model to study the effect of lymphocytes on EC-matrix adhesion to determine whether EC adhesion was functionally altered by contact with lymphocytes. We observed that activated but not resting lymphocytes, or plasma membrane preparations of activated lymphocytes, stimulate EC release from matrix ligands. Detachment of the EC was partly inhibited by an Ab against the β1 integrin that fixes this integrin in the high-affinity state.
Materials and Methods
Cytokines and reagents
IL-4, TNF, and neutralizing pAb to TNF, lymphotoxin, and IL-12 were obtained from R&D Systems (Minneapolis, MN). mAb CH-11 to CD95 was obtained from Upstate Biotechnology (Lake Placid, NY), TS2/16 (anti-CD29) and OKT3 (anti-CD3) were from the American Type Culture Collection (Manassas, VA). Recombinant Fas-Ig was produced from stably transfected BW5147 cells (J. Elliott, unpublished data). PHA, concanamycin A (CMA), phenanthroline, MgCl2, aprotinin, and PMSF, were obtained from Sigma (St. Louis, MO). Annexin V-FITC was obtained from R&D Systems. BB-94 was a generous gift from D. Edwards (University of Calgary).
Human umbilical vein EC were isolated as described previously from single umbilical cords and cultured in M199 medium containing 20% FBS, EC growth factor, and antibiotics (17). The cells were expanded on gelatin substratum and then used in the experiments between passage 3 and 6. For the EC cytolysis assays, the EC were plated at 2 × 104 cells/well on a flat-bottom 96-well fibronectin (FN)-coated plate, and then were loaded with 111In essentially as described (18). For the EC detachment assays EC were labeled with [3H]thymidine at 4 μCi/ml during the logarithmic phase of growth. The EC were harvested using trypsin/EDTA, then replated at high density (1.8 × 105/well) on FN- or type I collagen-coated 24-well plates.
Human PBL were isolated from volunteer donors using leukapheresis, followed by centrifugation over a Ficoll gradient as described earlier (19). The resulting mononuclear cell population was then serially passed over uncoated and FN-coated plastic culture plates to remove mononuclear cells. Short-term lines of lymphocytes were generated by stimulating the PBL with 5 μg/ml PHA alone or in the presence of IL-4 (10 ng/ml) and neutralizing pAb to IL-12 (3 μg/ml) for two cycles. In other experiments, the PBL were suspended at 3 × 106 cells/ml, then stimulated with OKT3 (10 μg/ml) and sheep anti-mouse IgG (3 μg/ml) for 24 h before use. Except where indicated, the PBL were treated with 10 nM CMA for 4 h before use in the assays.
Subpopulations of PBL were isolated by negative selection using immunomagnetic beads. Briefly, anti-mouse IgG-coated beads were loaded with HB145 (anti-MHC class II), and either OKT4 or OKT8 as directed by the manufacturer (Immunotech, Westbrook, ME). PBL were incubated at 4°C with a mixture of beads for 1 h, then bound cells were removed as directed. The resulting CD8+- or CD4+-enriched T cell populations were studied by FACS analysis and determined to have <1% contaminating CD4+ or CD8+ T cells, CD19+ B cells, and CD16/56+ NK cells.
The target cells were prepared as described above, then unincorporated 111In was removed by one wash with medium, followed by a 1-h incubation and a further wash with medium. The effector lymphocytes were added in a final volume of 200 μl, then incubated at 37°C for 4 h. One replicate group of EC was treated with SDS at a final concentration of 0.1% to lyse the cells. Eighty microliters of cell-free supernatant was collected, and then the radioactivity was measured using a gamma counter. Each point represents the mean of four samples.
The target EC were labeled with [methyl-3H]thymidine, and then allowed to form a confluent monolayer on either FN or type I collagen matrix for 24 h. The cells were gently washed twice, then overlaid with the effector cell population in a final volume of 500 μl and cultured at 37°C for the indicated time. One group was treated with a 0.1% SDS final concentration to release all incorporated radioactivity. The loosely adherent cells were resuspended using the coculture supernatant, then 250 μl of the supernatant was added to liquid scintillation fluid and the radioactivity was determined using a beta counter. Each point represents the mean of four samples.
Specific release of label in the cytotoxicity or detachment experiments was calculated using the following formula: specific release (%) = [(experimental − spontaneous)/(total − spontaneous)].
Values less than zero are shown as zero. Spontaneous release was <20% of total counts.
Plasma membrane preparation
Resting or activated PBL were treated with 10 nM CMA for 4 h, and then plasma membranes were isolated as previously described (20). Briefly, PBL were washed with PBS, lysed in buffer (0.25 M sucrose, 0.1 M MgCl2, 10 mM Tris-HCl, 250 U/ml aprotinin, and 1 mM PMSF), and sonicated using three 5-s bursts at 40 W. The nuclei were pelleted by centrifugation at 1000 × g, and then the membrane fraction was isolated by centrifugation at 100,000 × g. The PBL membranes were resuspended in medium and used immediately in the assay.
To study the interaction between activated lymphocytes and EC in the absence of cytolytic EC injury, we first characterized the mechanisms of PHA-activated PBL-mediated EC lysis. PBL activated with PHA lyse EC as demonstrated previously (21, 22). Although lymphokine-activated killer cells generated in this way express both FasL and TNF, EC lysis is accomplished using the perforin/granzyme cytolytic pathway because pretreatment of the activated lymphocytes with CMA completely inhibits release of 111In from the target EC (Fig. 1) (23). CMA pretreatment of cytolytic cells is thought to inhibit cytolysis by altering the acidity of the lymphocyte cytolytic granules thereby promoting conformation changes in perforin or other granule proteins (23). This interpretation is supported by the observation that chelation of extracellular calcium using EGTA also completely blocks EC lysis and is consistent with earlier reports (data not shown) (24). To study EC detachment from matrix in the absence of perforin-mediated lysis, then, resting and PHA-activated PBL were pretreated with CMA in the subsequent experiments.
EC detachment from matrix
To examine lymphocyte-stimulated alterations in EC adhesion to matrix, we added CMA-treated resting or activated lymphocytes to a confluent monolayer of HUVEC. As shown in Fig. 2, interaction between the PHA-activated but not resting PBL and the EC monolayer promoted detachment of the EC from FN matrix. The extent of the EC detachment varied according to both the E:T cell ratio and the time of coculture, but was reproducibly and significantly increased after 16 h of coculture at an E:T cell ratio between 10 and 20:1. Detachment of EC from a type 1 collagen matrix and the complex matrix Matrigel was similarly promoted by contact with lectin-activated PBL (32.1 ± 2.7% vs 42.3 ± 6.2%, respectively; p = NS, in one of three experiments with similar results).
To determine whether both CD4+ and CD8+ subpopulations of T cells were able to promote EC detachment from matrix, we prepared highly enriched CD4+ and CD8+ lymphocytes by negative selection. Activated CD4+ and CD8+ PBL were both able to promote EC release from the matrix (Fig. 3 A). Moreover both populations appeared to be equally efficient to induce detachment.
Similarly, activated lymphocytes with a Th1- or Th2-like phenotype were tested to determine whether both lines of polarized lymphocytes induced EC detachment. In these experiments, we generated short-term lines of PBL stimulated with PHA alone (for Th1-like cells), or PHA in the presence of IL-4 and neutralizing Ab to IL-12 (to generate Th2-like cells). In agreement with previous work, activated lymphocytes stimulated in the presence of IL-4 and neutralizing Ab to IL-12 produced little IFN-γ compared with PBL stimulated with lectin alone, and lacked the ability to lyse EC targets in a lectin-directed 111In-release assay (data not shown) (25, 26). We observed that both lines stimulated EC detachment (Fig. 3 B). Taken together, these observations indicate that lymphocytes are able to promote EC detachment from matrix independent from cytolytic injury, and that this capability is broadly shared among different activated lymphocyte populations.
EC detachment requires direct contact with activated lymphocytes
To determine whether soluble products released following contact between the activated lymphocyte and EC could promote EC detachment, both lymphocytes and EC were cultured above a semipermeable membrane, and the detachment of radiolabeled EC in the lower chamber was assessed. As shown in Fig. 4, EC release was not stimulated by the activated PBL unless direct cell-cell contact occurred. To further confirm that soluble signals did not account for the effect of the activated lymphocytes, we tested the ability of activated-lymphocyte conditioned medium to stimulate EC detachment. Neither activated nor resting lymphocyte conditioned medium promoted EC detachment from FN (data not shown). These observations indicate that soluble mediators alone do not account for activated lymphocyte-stimulated EC detachment.
TNF, lymphotoxin, and FasL are cell surface molecules expressed by activated lymphocytes and are known to elicit pro-inflammatory and injury responses in some cell types. As shown in Fig. 5, inhibition of these signals failed to block EC detachment. However, at the concentrations used, apoptosis of the WEHI and Jurkat cell lines, used to test the TNF/LT or FasL blocking reagents, respectively, was inhibited by 96.4 ± 1.8% and 87.3 ± 3.8%, respectively. In contrast, we observed that plasma membrane preparations of activated, but not resting lymphocytes promoted EC release from matrix (Fig. 6). These data suggest that a cell surface ligand is presented to the EC by the activated lymphocyte and is sufficient to induce EC detachment.
EC are not apoptotic before detachment
Detachment from matrix is a late feature of apoptosis in the EC (27). To determine whether EC were undergoing apoptosis after contact by the activated lymphocytes but before detachment, we examined the adherent EC for evidence of phosphatidyl serine translocated to the outer leaflet of the cell membrane and genomic DNA degradation. Phosphatidyl serine displayed by the adherent EC was examined 4, 8, and 16 h after contact with CMA-treated activated lymphocytes by staining with FITC-conjugated annexin V and quantitated using flow microfluorometry. As shown in Fig. 7, a small fraction of EC stained with annexin V, but no increase was detected following stimulation with activated lymphocytes. To detect fragmentation of DNA, EC monolayers grown on FN-coated chamber slides were stimulated with lymphocytes at the same time points and then stained using the TUNEL technique. Small numbers (<5%) of the EC stained when stimulated with either resting or activated lymphocytes at any time point (data not shown). However, distinct staining was evident in control cultures treated with C6-ceramide in agreement with previous reports (28). These data indicate that the EC do not undergo apoptosis before detaching from their matrix ligands.
Matrix metalloproteinases (MMPs) do not mediate EC detachment
Because MMP transcription is increased in both EC and lymphocytes after activation, we considered the possibility that contact between the activated lymphocyte and EC might induce matrix-degrading MMPs to mediate EC detachment (29, 30). As shown in Fig. 8, EC detachment was not inhibited by either o-phenanthroline or BB-94, two broadly active MMP inhibitors. Supernatant from the coculture of activated lymphocytes and EC showed the presence of MMP2 and MMP9 proenzyme activity by gelatinase assay, but the active cleaved forms of the matrix MMPs were not detected. Taken together with the lack of detachment following EC culture with supernatant from the activated lymphocyte-EC coculture we conclude that matrix MMP activity does not account for lymphocyte-stimulated EC matrix release.
Activating mAb to CD29 stabilizes EC adhesion following contact by activated lymphocytes
Finally, we tested the hypothesis that contact with the activated lymphocyte influences EC-matrix adhesion through modulation of the affinity of the β1 integrin for its matrix ligand. We plated EC onto FN matrix in the presence of the anti-β1 integrin mAb, TS2/16, which has previously been shown to fix β1 integrin in the high affinity state for the matrix ligand (31, 32). Activated lymphocytes were then added to the coculture in the continued presence of TS2/16. As shown in Fig. 9, we observed that EC detachment was significantly reduced in the presence of this mAb. Similar inhibition of EC detachment was not observed with either irrelevant control mAb, or a binding control mAb directed against CD31.
The high-affinity state of β1 integrin is also favored by exposure to Mn2+ or Mg2+ (33). We found that prolonged exposure of EC to Mn2+ caused toxicity, but the addition of 1 mM Mg2+ to the culture medium reduced activated lymphocyte-stimulated EC detachment by 40.0 ± 4.1% vs control medium (mean ± SD, n = 3 experiments). These results support the interpretation that EC detachment stimulated by contact with activated lymphocytes involves a change in the EC β1 integrin matrix receptors to favor the low affinity conformation.
We have observed that contact between activated lymphocytes or plasma membranes derived from activated PBL, and EC in culture results in the detachment of the EC from its underlying matrix. Both CD4+ and CD8+ T cell subsets as well as noncytolytic T cell lines were competent to promote EC detachment following activation. Neither induction of apoptosis nor increased matrix MMP activity accounted for the phenomenon. However, EC detachment was inhibited if the β1 integrin matrix receptor was locked into the high-affinity conformation.
The simplest interpretation of these data is that EC detachment after contact by activated lymphocytes is mediated directly through changes in the affinity of the β1 integrin for the matrix ligand. Alternatively, dissociation of the transmembrane matrix receptor from the cytoskeletal anchoring proteins might relieve conformational constraints on the β1 integrin allowing more receptors to cycle back to the low-affinity state. Because isolated plasma membrane from activated lymphocytes is sufficient to promote EC detachment, we favor the idea that a contact-dependent signal is delivered to the EC following interaction between an unidentified cell surface ligand(s) on the lymphocyte and a receptor on the EC. This is consistent with the observation that monocyte adhesion to EC promotes EC rounding, a process associated with reduced matrix adhesion (16).
Inside-outside regulation of integrin ligand affinity has been described in several systems including platelet adhesion to fibrinogen through αIIbβ3, and leukocyte adhesion to ICAM-1 through αLβ2 or FN through α5β1 (34, 35, 36). In these systems the integrin assumes the high affinity conformation after the cell is stimulated by agents such as platelet-activating factor or chemokines. In contrast, β1 integrins, such as the FN receptor, may also exist in the active conformation or change to the high-affinity conformation after the ligand is engaged without a requirement for extracellular signals (37, 38, 39, 40).
Conformational change of β1 integrins from the high- to the low-affinity state is involved in morphogenesis, cell division/ras activation, and cell motility (41). For example, regional integrin-mediated attachment to the matrix at the trailing edge of the migrating cell is loosened in a process involving both decreased β1 integrin affinity and fracturing of the association between the integrin and the cytoskeleton (42, 43, 44). The calcium-dependent protease calpain has been implicated in regulation of integrin affinity through localized proteolysis of cytoskeletal associations (45, 46). However, we are unaware of examples of intercell contact-dependent signals regulating β1 integrin affinity.
Our observations confirm and extend earlier work in a murine system in which fibroblast target cells were observed to detach from matrix after contact with activated lymphocytes (8, 9, 10). In this model of injury, target cell detachment induced by cloned noncytolytic T cells required protein synthesis by both the lymphocyte and target cell, suggesting a newly synthesized effector molecule is placed on the surface following lymphocyte activation (10). Activated monocytes also injure EC in culture following direct cell contact, in some cases independent of reactive oxygen intermediates (47). Moreover, contact-dependent interaction between monocytes and EC has been shown to result in focal adhesion kinase degradation (16). Taken together with our results, these observations raise the possibility that activated leukocytes may be able to modulate matrix receptor affinity in a variety of different cell types.
EC detachment has been recognized as a feature of EC undergoing apoptosis (48). Recent work has emphasized the role of caspase 3 activation as a effector mechanism for the cytoskeletal remodeling involved in “rounding” that occurs before the apoptotic EC detaches (27). Several lines of evidence argue against endothelial apoptosis in our system. First, neither annexin V translocation to the outer leaflet of the cell membrane, nor genomic DNA nicking, early features of apoptosis, were detected in the adherent EC before matrix release. Second, lymphocyte-mediated injury of EC is dependent on perforin-dependent mechanisms that were excluded in our system by pretreatment of the effector cells with CMA, manipulation of the cells to a noncytolytic Th2-like phenotype, and by the demonstration that isolated lymphocyte plasma membrane preparations were able to elicit EC detachment. Finally, the EC detachment was significantly impaired by freezing the β1 integrin matrix receptor in the high affinity conformation. This is most consistent with a physiological, regulated mechanism controlling matrix adhesion rather than a prelethal event.
The extent of down-modulation of endothelial matrix avidity following contact with lymphocytes may occur over a wide range, encompassing the whole cell or in a strictly regional fashion depending perhaps on the duration of the intercellular interaction, and may be important in several settings. First, angiogenesis at sites of inflammation requires EC migration at the budding end of the newly forming capillary. This process requires that the endothelium remodel matrix attachments as the EC migrates outward. EC have been shown to express the high affinity IL-2 receptor at sites of inflammation and synthesize MMPs in response to lymphocyte-derived CD40 ligand, suggesting that contact between lymphocytes and the EC may condition the EC for angiogenesis (49, 50). Second, regional remodeling of the EC matrix attachments may play a permissive role in leukocyte transmigration similar to that identified for cytoskeletal remodeling and that proposed for lateral wall adhesion structure remodeling (51, 52). Finally, in the setting of high signal intensity of protracted duration endothelial detachment may be a mechanism of cell-mediated injury as proposed earlier (8).
We thank Doris Abley for her expert technical assistance, and Shannon Malowany for help in preparation of the manuscript.
This work was supported by grants from the Medical Research Council of Canada (to A.G.M.) and Juvenile Diabetes Foundation International (to J.F.E.). A.G.M. is the recipient of a Clinician-Investigator award from the Alberta Heritage Foundation for Medical Research.
Abbreviations used in this paper: FasL, Fas ligand; EC, endothelial cell(s); CMA, concanamycin A; FN, fibronectin; MMP, metalloproteinase; pAb, polyclonal antibody.