Vcam-1 is a cs1 peptide-inhibitable adhesion molecule expressed by lymph node high endothelium

Lymphocyte recirculation between the blood and lymphoid organs ensures effective immunosurveillance of the body, enabling lymphocytes to encounter antigen within the lymphoid microenvironment. The migration of lymphocytes from the blood into lymph nodes is mediated by the interaction of adhesion molecules expressed by lymphocytes (homing receptors), with their complementary ligands (vascular addressins) expressed by specialised high endothelial cells (HEC; Michl et al., 1991). Studies of neutrophil interactions with endothelial cells (EC) have demonstrated that extravasation is regulated by a sequence of adhesion events. The selectin family of adhesion molecules mediates the initial low-affinity reversible binding of neutrophils to EC and integrins mediate the subsequent stabilisation of adhesion and transendothelial migration (for review see Zimmerman et al., 1992). A number of molecules have been implicated in lymphocyte homing to distinct regions of the immune system, including L-selectin and the integrins, α₄β₇ and LFA-1; however, their precise roles in lymphocyte extravasation remain to be determined.

We have developed an in vitro model using cultured HEC, which allows the initial binding and subsequent migration of lymphocytes across high endothelium to be studied. In culture, HEC retain some HEV-associated properties including the synthesis of GlyCAM-1 (Ager, unpublished), a 50 kDa sulphated glycoprotein that binds to L-selectin (Lasky et al., 1992), the expression of phenotypic markers and vascular addressins (Ise et al., 1988; Chin et al., 1990), and the adhesion of large numbers of unactivated lymphocytes in comparison with non-specialised endothelial cells (Ager and Mistry, 1988). Using these cells we have shown that CS1 peptide, which represents the major cell adhesion domain in the type III connecting segment of fibronectin (Humphries et al., 1987), blocks the binding of lymphocytes to the surface of HEC (Ager and Humphries, 1990), and monoclonal antibodies against the α₄ and β₁ integrin subunits block the adhesion of human lymphocytes to rat HEC (Szekanecz et al., 1992). Together these observations suggest that lymphocytes employ α₄β₁integrin (VLA-4) to recognise a CS1 peptideinhibitable ligand expressed by HEC. Previous studies have shown that treatment of HEC with interferon-γ (IFN-γ) increases the number of lymphocytes that bind by up to 5fold. Interestingly, CS1 peptide blocks the adhesion of lymphocytes to both IFN-γ and untreated HEC equally (May and Ager, 1992), which suggests that the CS1 peptide-inhibitable ligand is upregulated by IFN-γ.

The CS1 peptide receptor, VLA-4, is known to bind to at least one other ligand apart from fibronectin. This ligand is VCAM-1, an inducible adhesion molecule expressed by cultured umbilical vein EC after pretreatment with TNFα, IL-1β or LPS, but not IFN-γ (Rice and Bevilacqua, 1989; Osborn et al., 1989; Carlos et al., 1990). VCAM-1 expression in vivo has been found on endothelial cells associated with a variety of immune-mediated inflammatory lesions, whereas expression in normal tissues (Rice et al., 1991; Cybulsky and Gimbrone, 1991; van Dinther-Janssen et al., 1991) and untreated cultured EC (Carlos et al., 1990; Rice et al., 1990) is minimal. In view of the association of VCAM-1 with immune-mediated inflammation and our demonstration of a role for VLA-4 in lymphocyte-HEV interactions we have determined the potential role of VCAM-1 in this interaction and its relationship to the CS1 peptide-inhibitable ligand that we have previously demonstrated.

Inbred AO and (AO×DA)F₁ hybrid rats bred and maintained in the Animal Unit, The Medical School, University of Manchester, and LOU rats bred and maintained under specific pathogen-free conditions at the National Institute for Medical Research, London, were used in this study.

Individual strains of HEC were isolated from cervical LN of single (AO×DA)F₁ hybrid rats and cultured as previously described (Ager, 1987). Confluent primary cultures of HEC were serially subcultured using 0.1% trypsin/0.025% EDTA in phosphate buffered saline (PBS) and plated at 50% of confluent density. For the experiments described in this study, 3 different HEC strains between passage numbers 12 and 36 were used. Previous studies have shown that the interactions between lymphocytes and cultured HEC are independent of passage number (Ager and Mistry, 1988).

HEC were pretreated with cytokines as previously described (May and Ager, 1992). The cytokines used in this study were rat rIFN-γ (generously provided by Dr Peter Van de Meide, Primate Centre TNO, Rijswijke, The Netherlands), human rTNFα (Genentech Inc., CA) and rat rIL-1β (generously provided by Dr Alan Shaw, Glaxo Institute for Molecular Biology, Geneva, Switzerland). Cytokines were diluted to the appropriate concentrations in RPMI 1640 plus 20% FCS and incubated on subconfluent HEC for 72 hours at 37°C in 5% CO₂/95% air.

The monoclonal antibodies (mAbs) used in this study were: HP2/1 (mouse anti-human α₄ integrin subunit purchased from Immunotech, Marseilles, France), which cross-reacts on rat α₄ (Yednock et al., 1992); Protein A-purifed 5F10 (mouse anti-rat VCAM-1) from a fusion using spleen cells of mice immunized with CHO cells stably transfected with 7 domain rat VCAM-1 cDNA (Lobb, unpublished) and generously provided by Dr P. Chisholm, Biogen; WT3 (mouse anti-rat CD18), generously provided by Dr M. Miyasaka, Tokyo Metropolitan Institute of Medical Science, Japan) and MRC-OX1 (mouse anti-rat leucocyte common antigen) purchased from Serotec Ltd., Bicester, UK.

Protein A- and affinity-purified rabbit immunolgobulin against rsVCAM-1 (generously provided by Dr C. Benjamin, Biogen) and normal rabbit immunoglobulin G were also used. The secondary antibodies used when staining HEC for FACS analysis (described below) were FITC-conjugated swine anti-rabbit Ig and FITC-conjugated F(ab)₂ fragments of rabbit anti-mouse Ig (both purchased from DAKO Ltd., High Wycombe, UK).

CS1, CS1-C and CS1-scr were synthesized and purified as described previously (Humphries et al., 1986). The sequences of the peptides are as follows: CS1, DELPQLVTLPHPNLHG-PEILDVPST; CS1-C, VTLPHPNLHGPEIL (Mould et al., 1990); CS1-scr, DELPQLVTLPHPNLHGPPVTSELID (Komoriya et al., 1991). Previous studies have shown that CS1-C and CS1-scr have no effect on VLA-4 integrin-dependent adhesion of cells to fibronectin (Komoriya et al., 1991); these peptides were therefore used to control for specific effects of CS1 peptide in this study.

HEC were plated at 5×10³/well in 96-well cluster trays (6 mm diameter; Nunc, Life Technologies) and pretreated in triplicate with IFN-γ (200 units/ml; 72 h). Total lymphocyte adhesion to either untreated or IFN-γ-treated confluent layers of HEC was studied using [³H]leucine-labelled lymphocytes as described previously (Ager and Humphries, 1990). Briefly, lymphocytes syngeneic to cultured HEC were collected from cervical, brachial and axillary LN, pooled and radiolabelled for 60 minutes in leucine-free MEM plus 5% dialysed FCS containing 10 μCi/ml of [³H]leucine ([4,5-³H]leucine, 60 Ci/mmol, Amersham International, UK). Lymphocytes were resuspended in assay medium (HEPES-buffered RPMI 1640 plus 1% FCS) and incubated on HEC for 60 minutes at 37°C (10⁷ cells/ml; 0.1 ml/well). The wells were then washed five times with Dulbecco’s-PBS plus 1% FCS. HEC and adherent lymphocytes were checked by phase-contrast microscopy before being solubilised in 0.1 ml 1 M NH₄OH. Samples were suspended in Optiphase “Hisafe II” (LBK, UK) liquid scintillant and counted on a Beckman LS1801 β-counter. HEC-associated radioactivity was expressed as a percentage of the total plated to give the percentage of lymphocyte adhesion. Results are mean ± s.d. (n=3).

During the adhesion assay one well each of untreated and IFN-γ-treated HEC was incubated in RPMI 20% FCS alone and harvested for endothelial cell counts. Cells were detached using trypsin/EDTA as for HEC subculture, resuspended in Isoton II (Coulter Electronics Ltd., Luton, UK) and counted using a Coulter counter.

The effects of peptides and monoclonal antibodies (mAb) on lymphocyte adhesion to untreated and IFN-γ-treated HEC were measured using the adhesion assay. Peptides were dissolved in assay medium at twice the final concentration and 50 μl was added to triplicate wells. An equal volume of lymphocytes (2×10⁷ cells/ml) was added and the assay was continued as described above. To study the effects of mAbs, lymphocytes were resuspended to 10⁷ cells/ml in mAb and incubated for 30 minutes at 4°C, after which they were warmed to 37°C for 15 minutes. Lymphocytes were then added to the HEC layer in the presence of mAb and the assay continued as described above.

Triplicate wells of 96-well cluster trays were coated with CS1-IgG conjugate, the control CS1-scr-IgG conjugate (1/40 dilution each; Komoriya et al., 1991) or 20 μg/ml rsVCAM-1 (Lobb et al., 1991; generously provided by Dr R. Lobb, Biogen) in Dulbecco’s-PBS for 1 hour at room temperature. Non-specific binding sites were blocked for 2 hours using 10 mg/ml heat inactivated (85°C; 10 minutes) BSA in PBS and the wells washed twice in assay medium. [³H]leucine-labelled lymphocytes at a final concentration of 1 ×10⁷/ml (50 μl/well), in the presence or absence of mAbs or peptides, were incubated for 30 minutes at 37°C. Non-adherent cells were removed by aspiration and the wells washed twice with 0.1 ml Dulbecco’s-PBS containing 1% FCS. Adherent cells were examined by phase-contrast microscopy before being processed for β-scintillation counting, as described above. Bound radioactivity was expressed as a percentage of the total plated to give the percent lymphocyte adhesion. Results are mean ± s.d. (n=3).

HEC plated at 10⁵/ml in 6-well cluster trays (30 mm diameter; Nunc) were pretreated with cytokines as described above. After removal of the cytokines the cells were detached using 0.025% EDTA in PBS in the presence or absence of 0.1% trypsin. HEC were then washed twice in Dulbecco’s-PBS plus 1% FCS, resuspended in 50 μl of primary antibody/antiserum diluted in RPMI 1640 plus 1% FCS and incubated at 4°C for 30 minutes. After washing three times, cells were stained with 50 μl of RPMI 1640 containing the appropriate secondary antibody plus 15% normal rat serum, and incubated a further 30 minutes at 4°C. HEC were then washed a further three times, fixed with 1% formaldehyde in Dulbecco’s-PBS and analyzed by flow cytometry (Becton Dickinson FACScan) using Consort30 software (Becton Dickinson, Cowley, UK).

Lymph nodes, Peyer’s patches and spleens from untreated rats were frozen in isopentane at −70°C and stored in liquid nitrogen. Cultured HEC were plated at 10⁴/well in 8-well glass slides (Lab-Tek Tissue Culture Chamber Slide, ICN Flow) and grown to confluence over 3 days. Cryostat sections (5 μm) and HEC were fixed at −20°C for 10 minutes (dried acetone for tissues and methanol for HEC), air dried and stored at −70°C for up to 4 weeks. All antibodies were diluted in PBS containing 0.1% BSA and incubated on sections at room temperature in a humidified atmosphere. Tissues and cells were incubated with the primary antibodies (30 μg/ml rabbit anti-rsVCAM-1 or rabbit IgG; 10 μg/ml 5F10 or 10 μg/ml anti-rat CD4; W3/25; Serotec, UK) for 60 minutes, washed 3 times in PBS (5 min) and incubated for 30 minutes with a 1:50 dilution of either peroxidase-conjugated goat anti-rabbit immunoglobulin or rabbit anti-mouse immunoglobulin (Dako, Ltd). Tissues and cells were washed 3 times with 0.05 M Trisbuffered saline, incubated for 10 minutes in 0.5 mg/ml diaminobenzidine plus 0.002% H₂O₂ peroxidase and counterstained with haematoxylin.

A previous study showed that the adhesion of human lymphocytes to rat HEC is inhibited by antibodies to α₄ and β₁ integrin subunits (Szekanecz et al., 1992). In order to confirm that VLA-4 is also used by rat lymphocytes and to study further the effect of IFN-γ on this adhesive pathway, the adhesion of rat lymphocytes to either untreated or IFN-γ-treated (200 units/ml for 72 h) HEC was measured in the presence of either mAb HP2/1 (anti-human α₄ integrin; Sanchez-Madrid et al., 1986) or CS1 peptide.

Fig. 1A shows that CS1 peptide caused a dose-dependent inhibition of lymphocyte adhesion to untreated and IFN-γ-treated HEC, whereas, CS1-C (a truncated control peptide) had no effect. Inhibition of adhesion to untreated and IFN-γ-treated HEC was detectable with 11 μM (30 μg/ml) CS1 and at 370 μM CS1, a dose that gives maxi - mal inhibition (Ager and Humphries, 1990), adhesion to untreated HEC was reduced from 16.1 ± 1.7 in the presence of CS1-C to 8.6 ± 1.1. At this dose of CS1, adhesion to IFN-γ-treated HEC was reduced from 62.3 ± 4.0 in the presence of CS1-C to 36.1 ± 2.8. These reductions represent inhibitions of 47% and 42%, respectively, showing that CS1 peptide inhibits adhesion to untreated and IFN-γ-treated HEC equally.

Inclusion of HP2/1 in the adhesion assay also caused dose-dependent inhibition of lymphocyte adhesion to untreated and IFN-γ-treated HEC (Fig. 1B). The isotype matched control mAb, MRC-OX1, had no effect when compared with levels of adhesion in the absence of antibodies (data not shown). Inhibition of adhesion was detectable with 7 μg/ml HP2/1 and at 60 μg/ml, the highest dose tested, HP2/1 reduced adhesion to untreated HEC from 21.4 ± 2.1 to 8.2 ± 2.1. Adhesion to IFN-γ-treated HEC was reduced from 72.3 ± 1.9 to 28.2 ± 4.7 at this dose of HP2/1. These reductions represent inhibitions of 62% and 61%, respectively, showing that, at this dose, HP2/1 inhibits adhesion to untreated and IFN-γ-treated HEC equally.

VCAM-1 has been identified as an alternative ligand for VLA-4, and recognition of fibronectin and VCAM-1 is thought to occur via distinct sites on VLA-4. We therefore determined which of these two ligands is blocked by HP2/1 in the rat.

Rat lymphocytes bound to immobilized CS1 peptide and, in individual experiments, the level of adhesion varied between 13 and 30%. Adhesion to the control peptide, CS1-scr, was low in comparison at <5%. HP2/1 inhibited adhesion to CS1 in a dose-dependent manner (Fig. 2A). Inhibition was detectable with 1 μg/ml HP2/1 and at >20 μg/ml adhesion was reduced to <5%, the level of adhesion to control peptide. The control antibody, MRC-OX1, had little effect on lymphocyte adhesion to either CS1 or CS1-scr. When compared with MRC-OX1, 20 μg/ml HP2/1 reduced adhesion from 14.1 ± 1.0 to 2.4 ± 0.8, to give inhibition of 83%.

Rat lymphocytes bound equally well to immobilized rsVCAM-1 and the level of adhesion in individual experiments ranged from 6 to 27%. Fig. 2B shows that HP2/1 also inhibited adhesion to VCAM-1. Inhibition was dosedependent and detectable at 2 μg/ml. In separate experiments, 20 μg/ml HP2/1 reduced adhesion to the level of that in BSA-coated wells (data not shown). MRC-OX1 had little effect on lymphocyte adhesion. HP2/1 (at 20 μg/ml) reduced adhesion from 9.3 ± 0.1 in the presence of MRC-OX1 to 2.7 ± 0.5 to give inhibition of 71%.

To test whether CS1 can inhibit binding to VCAM-1, lymphocyte adhesion to immobilized rsVCAM-1 was measured in the presence of either CS1 or the control peptide, CS1-C. CS1 caused a dose-dependent inhibition of lymphocyte adhesion to VCAM-1 reducing adhesion to the level in BSA-coated wells whereas CS1-C had no effect. Inhibition was detectable with 4 μM CS1 and it saturated at 110 μM and above (Fig. 3). At saturating concentrations of peptide, lymphocyte adhesion was reduced from 31.1 ± 1.2 to 5.7 ± 1.9 to give inhibition of 82%.

The results described in the previous section suggest that VCAM-1 is a CS1 peptide inhibitable ligand in this model. In order to confirm this, the role of VCAM-1 in lymphocyte adhesion to either untreated or IFN-γ-treated HEC was measured in the presence of mAb 5F10 (anti-rat VCAM-1) or the control antibody MRC-OX1.

As shown in Fig. 4, 5F10 inhibited adhesion to untreated and IFN-γ-treated HEC in a dose-dependent manner, whereas MRC-OX1 had no effect. Inhibition of adhesion to IFN-γ-treated HEC was detectable with 0.2 μg/ml 5F10 but the higher dose of 7 μg/ml was required to inhibit adhesion to untreated HEC. However, at 60 μg/ml, the highest concentration tested, 5F10 inhibited adhesion to untreated and IFN-γ-treated HEC equally, with inhibitions of 48% and 49%, respectively.

The effect of 5F10 was slightly less than that of HP2/1, which gave ∼60% inhibition. However, when the two anti-bodies were tested in the same experiment, the effects of HP2/1 and 5F10 on lymphocyte adhesion to untreated and IFN-γ-treated HEC were similar. When tested together, the combined effects of the antibodies were no greater than their individual effects. Inclusion of mAb WT3, which blocks the function of the β₂ integrin subunit (CD18), had no effect on lymphocyte adhesion to either untreated or IFN-γ-treated HEC (data not shown).

The results described above suggest that a CS1 peptide-inhibitable ligand expressed by HEC is VCAM-1. VCAM-1 expression by HEC was therefore studied directly by staining the cells with a rabbit antiserum against rsVCAM-1, which cross-reacts on rat tissues, and analysing this by flow cytometry. As shown in Fig. 5, the majority of HEC (86%) expressed VCAM-1 under basal conditions. In eight separate experiments the level of VCAM-1 expression ranged from 59 to 98% with a mean of 81%. Since VCAM-1 expression by non-specialised endothelial cells in culture is reported to be low, we determined whether the high basal expression by cultured HEC was due to endotoxin present in the culture media. Endotoxin levels in all media and sera used throughout this study were <25 pg/ml (ICN Flow). Consistent with this low level of endotoxin, pretreatment of cultured HEC for 24 hours with up to 10 μg/ml polymyxin B to inhibit endogenous endotoxin had no effect on the level of VCAM-1 staining. For example, in one experiment HEC were 98% positive with a mean fluorescence intensity (MFI) of 49; the number of positive cells and fluorescence intensity both remained the same following polymyxin B treatment. Inclusion of endotoxin in the form of LPS had a marginal effect on VCAM-1 expression. Incubation with 50 μg/ml LPS for 24 hours increased VCAM-1 expression slightly to a MFI of 51 and this effect was completely inhibited by 10 μg/ml polymyxin B (MFI: 48). The basal expression of VCAM-1 by cultured HEC is therefore not simply due to endotoxin in the culture media.

A previous study has shown that pretreatment of HEC for 72 hours with IFN-γ, TNFα (200 units/ml each) or 10⁻⁹ M IL-1 β causes maximal increased lymphocyte adhesion, with IFN-γ having the greatest effect (May and Ager, 1992). We therefore studied the effects of these cytokines on VCAM-1 expression to determine whether there was a correlation between VCAM-1 levels and lymphocyte adhesion. After pretreatment with IFN-γ, TNFα and IL-1β the percentage of VCAM-1 positive cells increased from 86% to 99, 97 and 93%, respectively. Analysis of the mean fluorescence intensities (MFI) also showed increased expression after cytokine pretreament with the effects ranked IFN-γ > TNFα ≥ IL-1β. In the representative experiment shown in Fig. 5, the mean fluorescence intensity of IFN-γ-treated HEC was increased by >2-fold to 71, compared with 33 for untreated cells. Both TNFα and IL-1β increased VCAM-1 expression 1.5-fold up to MFI of 50. As found previously for ICAM-1 (May and Ager, 1992), the level of VCAM-1 detected on HEC was unaffected by trypsin used to detach the cells for FACS analysis (data not shown).

Of the cytokines tested, IFN-γ caused the greatest increase in VCAM-1 expression, consequently, the effects of IFN-γ were examined in more detail. VCAM-1 expression increased in a dose-dependent manner with maximal increases at 100 units/ml and above (Table 1A). The number of positive cells increased slightly from 86% to 99% but the mean fluorescence intensity increased >2-fold from 38 to 84. At concentrations above 100 units/ml the mean fluorescence intensity decreased slightly but the percentage of positive HEC remained at maximal levels (99%). Time course analysis showed that both mean fluorescence intensity and the percentage of positive HEC increased in a time-dependent manner (Table 1B). Increases in the percentage of VCAM-1 positive HEC were detectable after 24 hours and continued to increase up to 120 hours. Increased mean fluorescence, however, was not detectable until 72 hours, after which it continued to increase up to 120 hours.

Following removal of IFN-γ, lymphocyte adhesion and VCAM-1 expression returned to basal levels over similar time courses in 3 separate experiments. The percentage of lymphocytes that bound to cultured HEC varied between experiments, which reflects the limited number of binding sites on the surface of HEC rather than the binding of a minor lymphocyte population, as we have shown previously (Hourihan et al., 1993). However, IFN-γ consistently caused a 2-to 3-fold increase in lymphocyte adhesion to HEC, which decreased to basal levels 8 hours after removal of the cytokine (Fig. 6). The half-life of recovery was approximately 180 minutes. VCAM-1 expression by HEC also decreased in a time-dependent manner. After 8 hours, mean fluorescence intensity had decreased from 390 to 220, which, compared with untreated HEC (MFI:150), represented a decrease of 71%. The recovery of VCAM-1 expression was slightly delayed in comparison with that of lymphocyte adhesion, since lymphocyte adhesion had returned to basal levels at this time point. VCAM-1 expression returned to basal levels 24 hours after removal of IFN-γ.

The majority of cultured HEC were found to express VCAM-1 in the absence of cytokines and contaminating endotoxin in the culture media. This suggests that VCAM-1 may be constitutively expressed by HEV in vivo. We therefore determined the distribution of VCAM-1 in lymphoid organs of the rat. The staining pattern seen using mAb 5F10 was indistinguishable from that using rabbit anti-rsVCAM-1 antiserum in all tissues tested, demonstrating that the rabbit antisera cross-reacts on rat tissues. The results described below were obtained using 30 μg/ml rabbit anti-VCAM-1. Rabbit IgG at equivalent concentrations, which was used as control, did not stain any of the tissues tested. Some cells in the lymph node medulla and red pulp of the spleen stained non-specifically, corresponding with cells expressing endogenous peroxidase.

High endothelial venules (HEV) in lymph nodes were identified using the following criteria: their location in the paracortical or T-cell area as delineated using anti-CD4 antibody, the cuboidal morphology of lining endothelial cells and the presence of infiltrating lymphocytes in the vessel wall. All HEV in cervical lymph nodes of untreated rats stained using the rabbit anti-VCAM-1 antibody. The staining extended throughout the vessel wall and was not obviously restricted to the luminal surface (Fig. 7). Two other cell types stained using this antibody; a distinct population of dendritic cells in the T-cell area and, as described previously in other species, follicular dendritic cells in the germinal centres (Rice et al., 1991). Non-specialised blood vessels in the medullary region of the node did not stain with the antibody, neither did lymphocytes. The levels of VCAM-1 expression by HEV were similar in animals bred under conventional and specific pathogen-free conditions, indicating that endogenous VCAM-1 expression is not directly regulated by environmental stimuli. The staining pattern of other nodes was similar to that of cervical lymph nodes. In the spleen the only cells to stain were dendritic cells in the T-cell areas of the white pulp. Central arterioles and other non-specialised blood vessels in the marginal zone were not stained.

To compare the levels of VCAM-1 expression by lymph node HEV and cultured HEC the latter were analysed by indirect immunoperoxidase staining. As shown in Fig. 7, the intensity of staining of cultured HEC and lymph node HEV was similar. There was a visible increase in staining intensity following treatment of cultured HEC with 200 units/ml IFN-γ for 72 hours. To determine whether VCAM-1 expression by lymph node HEV is also upregulated by IFN-γ lymph nodes were maintained in organ culture for 70 hours in the presence and absence of 500 units/ml IFN-γ. The characteristic morphology of HEV and the distinct architecture of the lymph node were not maintained in organ culture. However, VCAM-1 staining was seen following organ culture and the level of staining was visibly higher in IFN-γ-treated tissues. Although it was not possible to identify which cells stained in these tissues, VCAM-1 expression was localised to the paracortical areas of the node as delineated using an anti-rat CD4 monoclonal antibody.

VLA-4 and other α₄ subunit-containing integrins have been implicated in the migration of lymphocytes to lymphoid organs and sites of inflammation. A homing receptor for mucosal HEV was originally identified as α₄ associated with a novel β chain, β_p (Holzmann and Weissman, 1989), which is now known as β₇ (Ruegg et al., 1992). Studies using large vessel endothelial cells in culture identified an important role for VLA-4 (α₄β₁)-mediated recognition of VCAM-1 expressed on cytokine-activated EC (Elices et al., 1990). These results suggested that the VLA-4/VCAM-1 pathway may be used by lymphocytes during migration to inflammatory sites rather than to lymphoid organs. The high level of VCAM-1 found on EC at sites of inflammation (Rice et al., 1990) supports this proposal. However, our identification of a CS1 peptide-inhibitable ligand expressed by cultured HEC suggested that VLA-4 may regulate the physiological migration of lymphocytes to normal lymph nodes of the rat, as well as to sites of inflammation. Studies using a xenogeneic model have shown that the adhesion of human lymphocytes to rat HEC is blocked by antibodies to α₄ and β₁ subunits, which supports this proposal (Szekanecz et al., 1992). In this study we have confirmed a role for an α₄ integrin in mediating lymphocyte adhesion to HEC using a syngeneic model and identifed VCAM-1 as a CS1 peptide-inhibitable ligand expressed by HEC.

Previous studies using antibodies have suggested that VLA-4 can recognise its two known ligands, fibronectin and VCAM-1, independently. Antibodies believed to recognise separate epitopes on VLA-4 have been characterised according to their ability to inhibit cell adhesion. Anti-epitope A mAbs inhibit adhesion to a 40 kDa CS1-containing fragment of fibronectin named FN-40, but not to VCAM-1. In contrast, antibodies that map to epitope B block adhesion to both FN-40 and VCAM-1 and anti-epitope C mAbs are without effect (Pulido et al., 1991). Further evidence that VLA-4 recognises its ligands independently is that soluble FN-40 does not block adhesion to VCAM-1 (Elices et al., 1990). However, the results presented here clearly show that soluble CS1 peptide can completely inhibit the adhesion of rat lymphocytes to immobilized VCAM-1. The discrepancies between our results and those already published warrant further consideration. The previously reported lack of effect of soluble FN-40 on cell adhesion to VCAM-1 may be due to the low dose tested (100 μg/ml; ∼2.5 μM), which is below the effective threshold concentration of 3 μM reported here. The effective concentration range of CS1 in this study (3-100 μM) is within previously reported dose ranges for inhibition of lymphocyte adhesion to HEC (Ager and Humphries, 1990) and melanoma cells to fibronectin (Komoriya et al., 1991) by CS1 peptide. There are two published reports demonstrating differential recognition of FN-40 and VCAM-1 by VLA-4; however, the results are conflicting. In the first, mAb HP1/3 blocked adhesion of Ramos cells to VCAM-1-transfected COS cells but not to immobilised FN-40 (Elices et al., 1990); however, in the second report, HP1/3 (and other mAbs to epitope A) blocked adhesion to FN-40 but not to VCAM-1-transfected COS cells (Pulido et al., 1991). These conflicting results warrant further investigation of the differential recognition of VCAM-1 and FN-40 by VLA-4.

The functional domains of VCAM-1 are currently being mapped by other workers and results using mAbs and VCAM-1/ICAM-1 chimeric constructs have identified important roles for the first Ig domain as well as the alternatively spliced, fourth Ig domain (Osborn et al., 1992; Vonderheide and Springer, 1992). A previous study identified the minimal sequence in CS1 that supports VLA-4-dependent melanoma cell adhesion to be the tripeptide, leucine-aspartic acid-valine (LDV; Komoriya et al., 1991). Sequence alignment of predicted amino acid sequences of rat fibronectin and the 7-domain form of rat VCAM-1 cDNAs identifies an homologous LDV-containing sequence in VCAM-1, which is located in the seventh Ig domain. The potential role of this LDV-containing domain in VCAM-1 in cell adhesion has not yet been addressed. Our studies do not indicate whether CS1 peptide directly competes with VCAM-1 because of sequence homology or whether it binds directly to VLA-4 and alters its ability to interact with VCAM-1 via steric hindrance or a conformational change in the integrin. Further work is therefore required to determine the precise mechanism of CS1 peptide inhibition of VCAM-1 recognition that we report here.

These results show that the binding of lymphocytes to peripheral lymph node HEC is mediated, in part, by VCAM-1 recognition. Anti-VCAM-1 antibodies blocked adhesion to cultured HEC and VCAM-1 was constitutively expressed by these cells both in vitro and in vivo. In addition, the expression of VCAM-1 was increased by cytokines and closely correlated with the time course and magnitude of increases in lymphocyte adhesion induced by cytokines that we have previously reported (May and Ager, 1992). VCAM-1 was upregulated by IL1-β and TNF-α, as reported previously using umbilical vein EC; however, a novel finding from our study was that IFN-γ was the most potent inducer of VCAM-1 expression. Previous reports have shown that IFN-γ does not induce VCAM-1 on HUVEC (Carlos et al., 1990). This may be a property of large vessel EC, since IFN-γ has been shown to induce VLA-4-dependent adhesion to rat microvascu-lar EC (Issekutz and Wykretowicz, 1991). The endogenous expression of VCAM-1 by HEV support the concept that these specialised blood vessels constitutively express several differentiated properties that together regulate lymphocyte homing. In addition to expressing ligands for l-selectin (Lasky et al., 1992), that mediate the initial binding of lymphocytes, HEV also express ligand(s) for integrin-mediated interactions that have been shown to regulate the transendothelial migration of lymphocytes in this (Hourihan et al., 1993) and other models of leucocyte-endothelial interaction (Zimmerman et al., 1992).

Thus far VCAM-1 has only been detected on HEV in inflamed lymph nodes of humans and sheep (Rice et al., 1991; Mackay et al., 1992). The endogenous expression of VCAM-1 by rat HEV may be a species difference. Alternatively, since several different splice variants exist (Cybulsky and Gimbrone, 1991), it is possible that the variant expressed by HEV is not recognised by all the antibodies currently available. It will be important to determine whether VCAM-1 expressed by rat lymph node HEV plays any role in lymphocyte migration from the blood using in vivo models of homing. Although an anti-α₄ subunit antibody has been shown to inhibit lymphocyte migration to rat peripheral lymph nodes the role of VCAM-1 was not determined (Issekutz, 1991). Antibodies are not currently available to distinguish between the use of α₄β₁ and α₄β₇ by rat lymphocytes and therefore we cannot definitively conclude that VLA-4 is the only integrin which recognises VCAM-1 on cultured HEC. Previous studies have shown that antibodies to α₄ and β₁ integrin subunits equally inhibit the binding of human peripheral blood lymphocytes to rat HEC (Szekanecz et al., 1992), which suggests a dominant role for VLA-4 in this model. Lymphomas that express both integrins have been shown to use α₄β₁ rather than α₄β₇ to bind to VCAM-1 (Chan et al., 1992). However, α₄β₇-expressing lymphomas can bind to VCAM-1 following phorbol ester stimulation (Ruegg et al., 1992). The role of α₄β₇ on unactivated, recirculating lymphocytes in mediating VCAM-1 recognition therefore remains to be determined.

Our results do not indicate the relative contribution of VCAM-1 and fibronectin to α₄ integrin-dependent adhesion of lymphocytes to HEC. We have already identified an important role for the fibronectin receptor, VLA-5, on the lymphocyte surface in mediating adhesion to cultured HEC, although its effect overlapped with that of VLA-4 rather than operating in addition (Szekanecz et al., 1992). There are no antibodies available that will detect the CS1 sequence in fibronectin, therefore it is not possible to determine its localisation in the HEC layer. The precise role of VLA-4 in the binding and transendothelial migration of lymphocytes is currently under investigation and we have shown recently that the affinity of VLA-4 for CS1 peptide is upregulated on lymphocytes that have migrated across the HEC layer (Hourihan et al., 1993). Results presented here suggest that the initial binding is mediated via VCAM-1 on the surface of HEC. The distribution of VCAM-1 throughout the walls of HEV that we have observed raises the possibility that VCAM-1 may also mediate the transmigration of lymphocytes via alterations in VLA-4 affinity. Studies of the distribution of VCAM-1 and the CS1-containing splice variant of fibronectin will be important for determining the roles of these two ligands for VLA-4 in the binding and transendothelial migration of lymphocytes. Studies using non-specialised (umbilical vein) EC in culture have shown that VCAM-1 is expressed at the upper surface of EC and that antibodies to VCAM-1 do not inhibit transendothelial migration of lymphocytes (Oppenheimer-Marks et al., 1991). Lymphocytes that transmigrate this type of EC use the LFA-1 adhesive pathway, which correlates with the fact that they are recently activated (Masuyama et al., 1990). As reported previously in this and other studies using rat models (Tamatani et al., 1991), we could detect no effect of anti-CD18 antibodies on lymphocyte-HEC interactions, which correlates with the fact that we are studying the interactions between unactivated lymphocytes and HEC. The fact that doses of HP2/1 that completely blocked adhesion of lymphocytes to immobilized CS1 and VCAM-1 only partially blocked adhesion to HEC suggests that additional pathways of adhesion operate in this model.

Together our results identify VCAM-1 as a CS1 peptide-inhibitable ligand expressed by peripheral lymph node high endothelium that supports the binding of freshly harvested, unactivated lymphocytes. The mechanism of CS1 peptide inhibition of VCAM-1 remains to be determined; however, these results indicate that caution should be applied when interpreting results of adhesion studies using CS1 peptide as an indicator of fibronectin. Further studies are required to determine the precise role of VCAM-1 in the migration of lymphocytes to peripheral lymph nodes via HEV; however, our results suggest that this adhesive pathway may not be restricted to sites of inflammation. The use of drugs that interfere with this pathway to target inflammation may consequently have unwanted, immunosuppressive sideeffects.

We would like to thank Graham Preece for expert technical assistance, Dr Eric Bell for reading the manuscript, Rita Findon for secretarial assistance and Joe Brock for the figures. This work was supported by project grants from the Arthritis & Rheumatism Council (A33) and the Medical Research Council (G900377CA), UK.