Lymphocyte adhesion to high endothelium is mediated by two β₁ integrin receptors for fibronectin, α₄β₁ and α₅β₁

The recirculation of lymphocytes between peripheral blood and lymphoid organs is a well-characterised process in which cells leave the blood stream by crossing the walls of specialised, post-capillary venules lined with high endothelial cells (HEC) (Duivestijn and Hamann, 1989). The tissue specificity of these events is thought to be controlled by unique cell adhesion molecules (termed “homing receptors”) expressed by peripheral blood lymphocytes and complementary ligands (“vascular addressins”) on HEC. Although a number of candidate molecules on the lymphocyte surface have been identified, none of the proposed lymph node homing receptors is in fact specific to lymphocytes. The widespread distribution of LECAM-1, CD44 and LFA-1 on all leucocytes (Jutila et al., 1989) suggests that another receptor family may be responsible for the specificity of lymphocyte recirculation. Alternatively, a number of variably expressed molecules may either act in concert or be differentially activated, resulting in the extravasation of individual leucocyte populations at distinct sites.

Members of the β₁ integrins, lymphocytes, high endothelium, cell adhesion. integrin family of heterodimeric receptors (also known as the VLA family) have been reported to mediate the adhesion of different cell types to extracellular matrix components such as collagens, fibronectin and laminin (for review, see Hemler, 1990; Albelda and Buck, 1990). In addition, α₄β₄ is unique within this family as it also mediates various types of cell-cell adhesion such as T- to B-cell interactions (Clayberger et al., 1987; Takada et al., 1989), helpersuppressor cell interactions (Groux et al., 1989) and the homotypic aggregation of lymphocytes (Campanero et al., 1990). It has also been reported that peripheral blood lymphocytes express α₄β₁, α₅β₁ and α₆β₁ at moderate levels and at low levels (Hemler, 1990). In the absence of activation, the α₁ andβ₂ subunits are missing from these cells. In contrast, peripheral blood neutrophils do not express any detectable β₁ integrins (Hemler, 1990). It is therefore possible that in contrast to the widely distributed adhesion molecules LECAM-1, CD44 and LFA-1, some β₁ integrins might act as lymphocyte-specific homing receptors or play a crucial role in lymphocyte extravasation. In fact an α₄ subunit containing integrin has been shown to function as a Peyer’s patch homing receptor in mice (Holzmann et al., 1989). A further unanswered question is whether lymphocyte homing occurs by direct lymphocyte-HEC interaction or by an indirect cell-cell adhesion mediated by intercellular matrix macromolecules (Hemler, 1990).

To address this possibility we have adopted an in vitro model of lymphocyte homing using HEC cultured from rat lymph nodes (Ager, 1987). Cultured HEC express several differentiated properties uniquely associated with HEV such as phenotypic markers (Ise et al., 1988) and secretion of HEC-specific sulphated macromolecule (s) (Andrews et al., 1983; Ager and Martin, unpublished). Cultured HEC express recognition molecules for recirculating lymphocytes but not for immature lymphocytes (Ise et al., 1988). Following adhesion to the surface of HEC a subpopulation of lymphocytes invade the HEC and occupy a position underneath the endothelial layer in a manner analogous to the diapedesis that occurs in vivo (Ager and Mistry, 1988). Cultured HEC therefore provide a novel in vitro assay for the properties of vascular endothelium that mediate not only the adhesion but also the subsequent transmigration of lymphocytes.

Using this and other models of lymphocyte homing we have previously proposed a role for HEC-associated fibronectin in the binding of unactivated lymphocytes (Ager and Humphries, 1990). Lymphocyte adhesion was blocked by both CS1 peptide (from the major cell adhesion site in the type III connecting segment) and RGD-containing peptides. In order to study the receptors used by lymphocytes to bind to HEC we have developed a xenogeneic model in which the adhesion of human lymphocytes to HEC cultured from rat lymph nodes is measured. CS1 peptide and RGD-containing peptides gave similar profiles of inhibition of human lymphocytes to those found previously using rat cells. Monoclonal antibodies to the α₄ and α₅ subunits of ft integrin receptors blocked adhesion by 60% and 65%, respectively. Simultaneous inhibition of all ft integrins gave slightly higher inhibition at 70%. These results suggest that lymphocytes employ one or both of two different β₁ integrin receptors for fibronectin to bind to HEC.

The following monoclonal antibodies were obtained as gifts: TS2/7 (anti-α₁, culture supernatant from M. E. Hemler, Dana-Farber Cancer Institute, Boston, MA, USA); 12F1 (anti-α₁1 mg/ml from V. L. Woods, University of California, San Diego, USA); J143 (anti-α₃, 2 mg/ml from A. P. Albino, Memorial Sloan Kettering Institute, New York, USA); 8F2 (anti-α₄, ascites fluid from C. Morimoto, Dana-Farber Cancer Institute, Boston, MA, USA); mAb 16 (anti-o, 12.5 mg/ml from S. K. Akiyama, Howard University Cancer Center, Washington, DC, USA); GoH3 (anti-α₆, culture supernatant from A. Sonnenberg, Netherlands Red Cross Blood Transfusion Service, Amsterdam, Netherlands); mAb 13 (anti-α₁, 11 mg/ml from S. K. Akiyama, see above); MHM24 (anti-α₁, ascites fluid from A. McMichael, Institute for Molecular Medicine, John Radcliffe Infirmary, Oxford, UK); mAb 15 (anti-α₃, 5 mg/ml from M. Ginsberg, Scripps Clinic, La Jolla, CA, USA): 439-9B (anti-α₄, ascites fluid from S. J. Kennel, Oak Ridge National Lab, Oak Ridge, TN, USA); Hermes-1 and Hermes-3 (anti-CD44, culture supernatants from S. Jalkanen, Department of Medical Microbiology, Turku University, Turku, Finland); mAb 29 (anti-CD15, culture supernatant from N. Hogg, Macrophage Lab, Imperial Cancer Research Fund, London, UK).

Other antibodies were obtained from commercial sources: T3 (anti-CD3), T4 (anti-CD4), T8 (anti-CD8) and FITC-conjugated F(ab)₂ fragments of rabbit anti-mouse Ig from DAKO, UK; LeuM3 (anti-CD14) and Leu8 (anti-LECAM-1) from Becton Dickinson, UK; W6-32 (anti-HLA), FITC-anti-human Ig and FfTC-F(ab)₂ fragments of rabbit anti-rat IgG from Serotec, UK.

CS1 peptide was used as a diagnostic probe for activity of the type III connecting segment of fibronectin and CS4 peptide was used as an inactive control. GRGDS was used to indicate RGD-dependent activity and GRDGS was used as control. Peptides were synthesized and purified as described previously (Humphries et al., 1986). The CS peptide sequences are as follows: DELPQLVTLPHPNLHGPEIL-DVPST (CS1); FRRTTPPTTATPIRHRPRPYPPNVGEE (CS4).

Peripheral blood lymphocytes (PBL) were isolated from heparinized blood by separating mononuclear cells on Histopaque 1077 (Sigma, UK). After washing in Dulbecco’s PBS containing 1% foetal calf serum (FCS), the mononuclear cells were resuspended in RPMI 1640 containing 10% heat-inactivated FCS (RPMI 10) at 5×10⁶ cell/ml and incubated in 90 mm diameter Petri dishes (5 ml/dish Nunc, Gibco) for 60 min at 37°C to remove plastic-adherent cells.

Monocytes were purified from the mononuclear cell fraction obtained above on a Nycodenz monocyte gradient (Nycomed, UK). Briefly, mononuclear cells were layered on the Nycodenz solution and centrifuged for 15 min at 250 g. Lymphocytes were pelleted while monocytes remained at the interface. Neutrophils were isolated from the red-cellcontaining pellet obtained using Histopaque 1077. Briefly, a solution of 2.5% gelatin in 0.9% saline at 4°C was layered on the top of the pellet, mixed by inversion and left to stand. After 30 min at room temperature, red blood cells sedimented and neutrophils remained in the supernatant. Remaining red blood cells in the supernatant were removed by lysis in 0.83% NH4CI for 5 min at 37°C. All cells were washed twice and resuspended in assay medium (HEPES-buffered RPMI 1640 containing 1% FCS: RPMI 1) at 10⁷/ml.

Primary cultures of high endothelial cells (HEC) were established from cervical lymph nodes of individual AO or (AO × DA)p! rats. Confluent cultures were serially passaged using trypsin/EDTA and plated at 50% of confluent density as described previously (Ager, 1987). Three different strains of HEC were used between 13th and 20th passage in this study. Previous studies have shown that the interactions between lymphocytes and HEC are independent of passage number (Ager and Mistry, 1988). HEC were detached using 0.025% EDTA in PBS and resuspended at 10⁷ cells/ml in RPMI 10 for FACS analysis.

A 50 μl sample of a cell suspension (10⁷ cell/ml) was stained with 50 μl of primary monoclonal antibodies diluted in Dulbecco’s PBS containing 1% FCS for 60 min at 4°C. After washing three times, cells were stained with 50 μl of secondary antibody for 45 min at 4°C. After washing again, cells were fixed with 1% formaldehyde in Dulbecco’s PBS and analysed on a FACScan (Becton Dickinson, UK) using Consort 30 software. Purified leucocytes were characterised using antibodies against CD3, CD4 and CD8 for T cells, human Ig for B cells, CD14 for monocytes and CD15 for neutrophils. Cells were also stained for a variety of different adhesion molecules that have been implicated in lymphocyte homing (sources of mAbs mentioned above).

HEC were plated in 30 mm diameter wells of 6-well tissue culture trays (Nunc) at 10⁵ per well and grown to confluence over 3 days. HEC were preincubated in assay medium (RPMI 1640 containing 1% FCS) for 30 min at 37°C and then incubated with PBL (10⁷ cell/ml; 2 ml/well) for 2 h at 37°C in a humidified atmosphere of 5% CO2 in air. Non-adherent cells were removed by aspiration and the HEC layer was washed five times with Dulbecco’s PBS plus 1% FCS. Wells were analysed by phase-contrast microscopy for human PBL-rat HEC interactions. HEC-adherent lymphocytes were detached by incubating the endothelial layer using 0.025% EDTA in PBS for 5 min at 37°C. PBLs were stained for subsets and homing receptors and analysed by FACS as described above. Incubation of lymphocytes in PBS/EDTA did not affect the expression of phenotypic markers and adhesion molecules as determined by FACS analysis.

Previous studies have shown that rat lymphocytes exhibit one of two different morphologies when bound to HEC. Type I cells are phase-light, spherical and stain uniformly with toluidine blue, while type II cells are phase-dark and flattened, in which nuclear and cytoplasmic detail are seen. Electron microscopic analysis has shown that type I lymphocytes are bound to the upper surface of HEC and type II cells are underneath the endothelial layer (Ager and Mistry, 1988). We have analysed HEC-adherent PBL for the distinct morphologies associated with either type I (surface-bound) or type II (migrated) lymphocytes.

HEC were plated at l(r per well in 8-well glass tissue culture chamber slides (Lab-Tek slide, ICN, UK) and grown to confluence over 3 days. PBL were incubated with HEC for up to 2 h at 37°C (10⁷ cells/ml; 300 μl well) after which unbound PBLs were removed by aspiration, the plastic gasket was detached from the slide and the cell layer was washed five times in 50 ml wash solution (D-PBS plus 1% FCS prewarmed to 37°C), fixed for 30 min at 37°C in 3% glutaraldehyde, stained with 0.1% toludine blue and examined by high-power light microscopy. For each sample, nuclei of adherent PBL (either surface-bound or migrated) and HEC were counted at ×1000 magnification in 25 fields of view. The total number of PBL per HEC was calculated and means ± s.d. are presented. The effects of monoclonal antibodies to various lymphocyte surface molecules were studied by pre-incubating PBL with the mAbs in assay medium for 30 min at 4°C as described below for the radioisotope assay.

HEC were plated at 5 × 10³/well in 96-well tissue culture trays (Nunc, Gibco, UK) and grown to confluence. The adhesion assay was performed essentially as described previously (Ager and Humphries, 1990). PBL were labelled with 10 μCi/ml [³H]leucine ([4,5-³H]leucine, 60 Ci/mmol, Amersham International, UK) in leucine-free MEM (containing 5% FCS dialysed against PBS) and resuspended in assay medium. [³H]leucine-labelled PBL were incubated with HEC for 60 min at 37°C (10⁷ cells/ml; 100 μl/well), non-adherent cells were removed by aspiration and the cell layer was washed five times with 100 /d/well wash solution. Wells were checked by phase-contrast microscopy before HEC and adherent PBL were solubilized in 100 μ1 1 M NH4OH. Samples were processed for liquid scintillation counting using Optiphase‘Hisafe II’ (LKB, UK) and counted on a Beckman LS1801 β - counter. The number of adherent PBL was directly proportional to the radioactivity measured. Bound radioactivity was expressed as a percentage of total plated to give the % lymphocyte adhesion. Results are means ± s.d. (n=4).

The effects of peptides on lymphocyte adhesion were measured using the isotope assay described above. Peptides were dissolved in assay medium at twice the final concentration and 50 μl was added to quadruplicate wells of HEC. An equal volume of PBL at twice the final concentration was added immediately and the adhesion assay continued as above. The effects of monoclonal antibodies on lymphocyte adhesion were also measured using the radioisotope assay. PBL were resuspended to 10⁷ cells/ml in mAb solution and preincubated for 30 min at 4°C. Following warming to 37°C PBL plus mAb were added to the HEC layer and the assay continued as above.

A 50 μl sample of PBL suspension at 10⁷ cells/ml in RPMI 10 was incubated with an equal volume of anti-integrin mAbs or with 50 nM phorbol myristic acetate in 96-well tissue culture trays. PBL were checked for clustering after 2h and 18 h at 37°C in 5% CO₂ in air.

Cells were counted in suspension by electronic particle counting and viability was monitored by trypan blue exclusion.

Purified leucocyte populations were characterised using FACS analysis. The differences in size and granularity between lymphocytes, monocytes and neutrophils allowed these 3 cell populations to be separated from one another using 2-dimensional forward versus side scatter analysis on the FACScan (Griffin et al., 1990; data not shown). Analysis of electronically gated populations of the 3 cell types showed that the PBL populations consisted of 65–70% CD3⁺, 50-57% CD4⁺, 18–24% CD8⁺ cells and 3–15% slg⁺ cells. PBL did not stain with the antibody LeuM3, which recognises CD 14. Purified monocytes were 71% positive for CD 14 and negative for lymphocyte markers. Neutrophils were 50% positive for CD 15 and exhibited neither lymphocytenor monocyte-specific markers (data not shown).

The distributions of the α₄β₁ andα₅β₁ integrin receptors for fibronectin on lymphocytes, monocytes and neutrophils were determined and compared with the distribution of three other molecules known to play a role in lymphocyte homing, LFA-1, CD44 and LECAM-1. Virtually all PBLs (97%) expressed the common ft integrin subunit, 88% of PBL expressed the α₄ subunit and 82% the a₅ subunit (Fig. 1). PBLs did not express the α₁ orα₂ subunits but were 20% positive for α₃ and 67% positive for α₆ (data not shown). Monocytes also expressed α₄, α₅ and β₁ subunits, although the level of fluorescent staining was consistently lower than that for lymphocytes (Fig. 1). The staining profiles for α₅ and β₁ integrin subunits demonstrated populations of high and low-expressing cells in both PBL and monocyte populations. The staining profile for α₄ subunit also showed high- and low-expressing cells in the PBL population; however, monocytes stained uniformly for the α₄ integrin subunit. PBL did not express either theβ₃ or the β₄ subunits of other integrin families. Neutrophils did not express either the α₄ or the α₅ integrin subunits. Analysis using an antibody to α₁ subunit showed that neutrophils did not in fact express any members of the β₁ integrin family (Fig. 1).

In contrast, CD44, LFA-1 and LECAM-1 were widely distributed on all the leucocyte subsets analysed. The majority (>95%) of lymphocytes, monocytes and neutrophils expressed both the CDlla and the CD18 subunits of the β₂ integrin family (Fig. 1 and data not shown). Lymphocytes and monocytes contained populations of high- and low-LFA-1 expressing cells. In contrast, neutrophils stained uniformly for CDlla and CD 18. The mean levels of fluorescence were similar to those for lymphocytes and monocytes. CD44 was widely distributed on >95% of all lymphocytes, monocytes and neutrophils. Lymphocytes and monocytes contained high- and low-CD44 expressing cells. Neutrophils stained uniformly for CD44; however, the mean level of fluorescence was considerably lower than that of either lymphocytes or monocytes (Fig. 1). LECAM-1 was also expressed by the majority (>85%) of lymphocytes, monocytes and neutrophils. All three populations were stained equally using the Leu8 antibody.

The antibodies against human integrins, CD44 and LECAM-1, used above did not cross-react on rat tissues, as assessed by indirect immunofluorescence staining and FACS analysis of rat lymph node cells and cultured HEC.

When human PBL were incubated with cultures of rat HEC, lymphocytes adhered preferentially to the endothelial layer within the first 60 min rather than to uncoated areas of tissue culture plastic. After longer incubation (up to 3 h) HEC-adherent lymphocytes adopted one of two distinct morphologies, which were readily identified by light microscopy. Adherent PBL were either 4–5 μm diameter phase-light cells or larger (>10 μm diameter) phase-dark cells. Analysis of fixed and stained preparations of HEC-adherent PBL showed that the small cells stained uniformly with toludine blue whereas nuclear and cytoplasmic details were visible in the larger, flattened lymphocytes. The identification of two morphologically distinct populations coincides with previous results using rat lymphocytes in which it was shown that the small, uniformly stained, type I lymphocytes were bound to the upper surface of HEC and the larger, flattened, type II lymphocytes were underneath the HEC layer. Quantitation of PBL adhesion to rat HEC using the radioisotope assay showed that there was significant interexperimental variation. In this study the basal level of PBL adhesion after 60 min ranged from 2.5 to 14.6% of the cells plated. About 10% of HEC-adherent PBL demonstrated the morphology of type II, migrated lymphocytes after 120 min of incubation. This was less than half the number found in parallel experiments using rat lymphocytes (data not shown).

In order to study PBL that bound to HEC layers further, all adherent cells (PBL and HEC) were collected in suspension following incubation in 0.025% EDTA in PBS and analysed by FACS. HEC are larger than PBL and were gated out for FACS analysis. HEC-adherent PBL showed identical profiles on forward versus side scatter analysis as the PBL population that was plated (data not shown). However, not all lymphocyte subsets bound equally to the HEC layers. After 120 min of incubation there was an enrichment for B-over T-lymphocytes and the adherent T-cell population was enriched for CD8⁺ over CD4⁺ cells, as found previously using rat lymphocytes (Ager and Mistry, 1988). As shown in Table 1 the number of CD3⁺ cells was reduced from 72% in the plated population to 54% in the adherent fraction and the percentage of slg⁺ cells was increased fourfold from 4% to 16%. In the adherent T-cell population, the number of CD8⁺ cells at 23% was similar to the population plated (22%), whereas the CD4⁺ population was significantly reduced from 57% to 27%. HEC-adherent PBLs were slightly enriched in CD14⁺ cells from 0.3 to 2%.

Inclusion of CS1 peptide in the adhesion assay had a dose-dependent inhibitory effect on PBL adhesion to rat HEC. The control peptide, CS4, had little effect on lymphocyte adhesion (Fig. 2A). Inhibition with CS1 was detectable at 3.7 μM (0.01 mg/ml) and saturated at 37 μM and above. At saturating doses, PBL adhesion was reduced from 9.2 ± 0.5% in the presence of CS4 to 5.5 ± 0.1% in the presence of CS1. This represented a maximal inhibition of 43%. CS1 and CS4 peptides tested up to concentrations of 1.0 mM had no effect on the adhesion of HEC to tissue culture plastic or on lymphocyte viability as reported previously (Ager and Humphries, 1990).

Inclusion of GRGDS in the adhesion assay had a dose-dependent inhibitory effect on PBL adhesion to rat HEC at concentrations above 20 μM (Fig. 2B). The blocking activity of GRGDS saturated above 600 μM. The control peptide, GRDGS, had no effect when tested up to 2 μM. The level of adhesion in the presence of GRDGS was not significantly different from that in the complete absence of peptides. In comparison with the control peptide, 60 μM GRGDS inhibited adhesion by 52% and 2 mM GRGDS by 71%. As found previously using rat lymphocytes GRGES also had a dose-dependent inhibitory effect on PBL adhesion to rat HEC. The effects of GRGES were not significantly different from those using GRGDS (Fig. 2B). GRGDS, GRGES and GRDGS at concentrations up to 2 mM had no effect on the adhesion of HEC to tissue cultute plastic or on PBL viability as reported previously (Ager and Humphries, 1990). The inhibitory effects of CS1 and GRGDS peptides on the adhesion of human PBL to rat HEC are comparable with those reported previously using rat lymphocytes. Antibodies to the known receptor for CS1 are not available for analysis of rat lymphocytes. The demonstration that human PBL bind to rat HEC via CS1 and RGD peptide-dependent pathways provides a suitable model for determining the role of the CS1 receptor, α₄β₁ integrin. In addition, the contribution of the RGD-dependent fibronectin receptor, α₅β₁ could also be determined in order to investigate the role of fibronectin as an adhesive ligand for lymphocytes. Attempts to study the role of fibronectin by testing the ability of polyclonal anti-fibronectin antibodies to block rat lymphocyte adhesion to rat HEC were hampered by their ability to detach HEC from the culture dish (data not shown). Thus, by studying human PBL-rat HEC interactions and using antibodies against human integrin receptors for fibronectin that do not cross-react on rat tissues, direct effects of the antibodies on the endothelial layer will be avoided.

Monoclonal antibodies directed against the α₄, α₅ and ft integrin subunits, which have previously been shown to inhibit cell adhesion to fibronectin, all showed dosedependent inhibition of PBL-HEC adhesion. Using the radioisotope assay, the inhibitory effects of mAb 13 (anti-ft) were detectable at 1 in 3000 dilution and those of mAb 16 (antiα₅) and 8F2 (anti-α₄) at 1 in 300 (Table 2). Using saturating concentrations of each of the antibodies there was some variation between experiments in the absolute level of inhibition. However, in 5 separate experiments the inhibitory activities were consistently ranked as follows: anti-β₁anti-> α₅ >antiap (Table 3). Mean levels of inhibition were 70% for antiβ₁66% for anti-α-₅, 59% for anti-α-₄ antibodies. Combinations of antibodies to α₄ and α₅ subunits did not inhibit lymphocyte adhesion further than with the β₁ subunit antibody alone (data not shown).

Antibodies to Class I MHC antigens (which are uniformly expressed by PBL) gave less than 10% inhibition (Fig. 3). The inhibitory effects of α₄, α₅ and β₁ antibodies on PBL adhesion to HEC were not due to the promotion of lymphocyte clustering, which would reduce the number of lymphocytes available for adhesion to HEC. Incubation of PBL with saturating concentrations of these antibodies for up to 18 h did not induce lymphocyte clustering, which was found after incubating lymphocytes for 2 h with PMA (data not shown). Pre-incubation of rat HEC with antibodies to α₅β₁ and α₄ subunits prior to the adhesion assay had little effect on the subsequent binding of PBL (inhibition <10% for all 3 antibodies, data not shown), which is consistent with the lack of cross-reactivity of these anti-human antibodies on rat integrins as shown above by FACS analysis.

Monoclonal antibodies directed against other integrin subunits were also tested for effects on PBL-HEC adhesion. Antibodies to α₁α₂β₃ and β₄ subunits all gave less than 15% inhibition of PBL adhesion, which is consistent with their absence from PBL by FACS analysis. Anti-α₃ antibodies inhibited PBL adhesion in a dose-dependent manner, which saturated at a maximum of 30% (Fig. 3). Antibodies to the β₂ subunit also inhibited PBL adhesion to rat HEC; however, the level of inhibition was consistently less than that seen with anti-β₁ antibodies (Table 3 and Fig. 3). Inclusion of anti-β₂ antibodies did not inhibit PBL adhesion over and above that seen with the β₁ subunit antibody alone (Table 3).

The inhibitory effects of a₄, a_s and fa subunit antibodies were also detectable using a second assay of PBL-rat HEC adhesion, the microscope assay, in which the number of PBL bound per HEC is measured. Using this assay, the levels of inhibition were similar to those found using the radioisotope assay at 73% for anti-)β₁, 72% for anti-α₅ and 63% for anti-a-₄ (Tables 3 and 4). Light microscope analysis of HEC-adherent PBL identified two distinct morphologies previously associated with either surface-bound (type I) or migrated (type II) lymphocytes. In order to determine whether the integrin antibodies preferentially blocked lymphocyte binding to the upper surface or lymphocyte transmigration, the numbers of type I and type II lymphocytes in the presence and absence of antibodies were determined. As shown in Fig. 4 and Table 4, α₄, α₅ and β₁ subunit antibodies all blocked binding of lymphocytes to the surface of HEC. Antibodies to α₅ and β₁ also blocked the transmigration of lymphocytes; however, anti-α₄ antibodies had little effect on this population (Table 4).

The tissue-specific migration or homing of lymphocytes involves an undetermined number of cell-cell and cellmatrix interactions, which together are referred to as lymphocyte-EC recognition. Thus far, several different lymphocyte surface molecules have been implicated in homing to lymphoid organs; the selectin, LECAM-1, the integrins, LFA-1 and VLA-4, and CD44. The precise roles of each of these different adhesion receptors are not completely elucidated, although, by analogy, with neutrophil-EC interactions (Lawrence and Springer, 1991) the selectins are thought to mediate the initial, reversible binding of lymphocytes to the luminal surface of EC and the integrins to mediate the stabilisation of adhesive contacts and the subsequent transendothelial migration of lymphocytes.

Using a new in vitro model of lymphocyte homing we have previously identified an adhesive ligand on HEC using anti-adhesive synthetic peptides. CS1 peptide, which encodes the major cell-adhesion domain in the type IIICS of fibronectin, supported the adhesion of rat lymphocytes and blocked adhesion to cultured HEC by at least 50%. The receptor for CS1 is the integrin oqft. In this study a heterologous in vitro adhesion assay was developed to study the functional aspects of α₄β₁β₁ receptors in lymphocyte-endothelial cell interactions. The essential features of lymphocyte-HEC interactions that we have described previously were retained using human PBL. HEC-adherent lymphocytes adopted one of two differing morphologies, which are associated either with lymphocytes bound to the upper surface of HEC or lymphocytes that have migrated to a position underneath the endothelial layer. The relative binding of CD4⁺-, CD8⁺- and B-lymphocytes to cultured HEC was similar to that found previously using rat lymphocytes (Ager and Mistry, 1988). The differential binding of lymphocyte subsets to high endothelium has been reported using other models of lymphocyte homing (Kimpton et al., 1989; Stevens et al., 1982); however, the molecular basis of this remains to be determined. The effects of anti-adhesive synthetic peptides on PBL adhesion to cultured HEC (see below) were similar to those previously reported using rat lymphocytes. Thus the adhesion receptors that lymphocytes use to bind to cultured HEC cross-react functionally between rat and human cells. Inter-species cross-reactivity of lymphocyte homing receptors has also been reported using the frozen section assay (Wu et al., 1988) and in vivo assays of homing (Binns and Licence, 1990). After the establishment and characterisation of the heterologous human PBL-rat HEC adhesion assay, functional analysis of β₁ integrin receptor-ligand interactions was carried out employing the purified peptide ligands of these receptors and integrin subunit-specific monoclonal antibodies.

CS1 and GRGDS both showed dose-dependent inhibition of PBL adhesion to HEC reaching maxima of 40% and 60%, respectively. A modified form of the latter ligand (GRGES) also exhibited dose-dependent inhibition. The effects of GRGES were equivalent to those of GRGDS as found previously using rat lymphocytes (Ager and Humphries, 1990). The activities of CS1 and GRGES suggest a role for α₄β₁ integrin, since both peptides are diagnostic inhibitors of fibronectin IIICS adhesive activity (Mould et al., 1991). However, the activity of GRGDS cannot be readily attributed to a specific integrin receptor.

In order to examine further the role of α₄β₁ and to explore the site of action of RGD-containing peptides, we have used antibodies to the α₄ subunit and to other fibronectin binding β₁ integrins. The use of monoclonal antibodies against human integrins that did not crossreact on rat integrins allowed effects on human lymphocytes to be investigated independently of any effects on rat endothelium. FACS analysis of PBL confirmed that the majority (80-90%) of cells expressed both α₄β₁ and α₅β₁ integrins. The integrin α₁β₁, which has been reported to recognise the RGD-dependent central cell-binding domain of fibronectin (Elices et al., 1991), was expressed by a subpopulation of lymphocytes (30%). The prototype member of the VLA family, α₅β₂ was not expressed on fresh PBL, neither was α₂β₁ integrin, as reported previously (Hemler,1990). Monoclonal antibodies to β₁, α₄, α₅ and α₃ subunits, which have all been shown to inhibit cell adhesion to fibronectin, gave dose-dependent inhibition of lymphocyte adhesion to rat HEC; the effects were ranked β₁> α₅α₄> α₃. Antibodies to the other integrin subunits, not expressed by PBL, and to MHC Class I, which is uniformly expressed by PBL, were without effect. Antibodies to α₄ and α₅ subunits gave similar levels of inhibition at 55–65%, while β₂ antibodies consistently gave the highest levels of inhibition at 65–75%. Combinations of antibodies to α₄ and α₅ subunits did not inhibit adhesion further, indicating that maximal levels of inhibition were achieved with the fa subunit antibody alone. In addition, these results suggest that the α subunit activities were associated with β₁ integrin subunits rather than with alternative β chains such as β_p, which is known to hybridise with the α₄ subunit (Holzmann and Weissman, 1989). The dominant roles of both α₄β₁ and α₃β₁ in lymphocyte-HEC adhesion strongly suggest that fibronectin is a ligand on HEC. The additional activity of α₃β₁ integrin also supports this suggestion, although α₃β₁ is known to recognise other extracellular matrix molecules in addition to fibronectin (Elices et al., 1991). Although α₄β₁ integrin is known to recognise an alternative cell surface ligand, V-CAM-1, the activity of CS1 peptide has been mapped to a different site on α₄β₁ and thus is an indicator of fibronectin, rather than V-CAM-1, binding (Pulido et al., 1991).

Taken together, these data provide evidence that lymphocytes use at least two different fibronectin receptors for adhesion to cultured HEC. If this is the case, it is curious that blocking either of the receptors gives almost maximal inhibition of lymphocyte adhesion compared to that obtained by blocking all fa integrins. The formation of stable, adhesive contacts between lymphocytes and cultured HEC is an active process, which is blocked by pre-treatment of the lymphocyte with the protein kinase inhibitor, staurosporine (Hourihan and Ager, unpublished). Previous reports using the frozen section assay have demonstrated the requirement for an intact lymphocyte cytoskeleton in order to bind to HEC (Woodruff et al., 1977). It is therefore possible that simultaneous engage-ment of α₄β₁ and α₅β₁ by immobilised fibronectin in the HEC layer is, in fact, required by the lymphocyte in order to bind to HEC in this model. Attempts to identify the potentially overlapping roles of α₄β₁ and α₅β₁ showed that combinations of antibodies to the α₄ and α₅ subunits or CS1 and RGD-containing peptides did not inhibit adhesion further than with either antibody or peptide alone, suggesting that a single antibody or peptide can give almost maximal inhibition. There is a growing body of evidence to suggest that activation of T-lymphocytes via cross-linking of surface CD3 can upregulate the affinities of integrins and other adhesion receptors for their respective ligands (Dustin and Springer, 1989; Shimizu et al., 1990; Spertini et al., 1991). As found using other models of lymphocyte homing, lymphocyte adhesion to cultured HEC is independent of antigen-receptor engagement (May and Ager, 1992). The proposal that cross-linking of α₄β₄ and α₅β₁ on the lymphocyte surface by immobilised fibronectin results in the formation of stable, adhesive contacts either via fibronectin itself or via altered affinities of other adhesion receptors, e.g. LFA-1, is a possibility in this model. The observation that antibodies to fa integrins also inhibited adhesion and that the effects were overlapping rather than additional to those of Pi integrins supports this proposal. In addition, these results suggest that ft integrins may play a more important role than ft integrins in lymphocyte-HEV interactions.

The fact that antibodies to α₄ and α₅ subunits blocked binding to the upper surface of HEC layers suggests that fibronectin is localised there. The additional observation that antibodies to α₅, but not α₄, blocked the transmigration of lymphocytes in this model suggests different roles for these two adhesion domains in fibronectin: α₅β₂ CSl recognition mediates binding to the upper surface of HEC and α₅β₁-fibronectin recognition mediates the subsequent transendothelial migration of bound lymphocytes. In support of this proposal, we have already reported that CS1 peptide preferentially blocked binding of lymphocytes to the upper surface of HEC layers rather than transendothelial migration in this model (Ager and Humphries,1990). In addition, lymphocytes bound to the upper surfaces of cultured HEC are exclusively distributed to the peripheral edges of HEC, not over the cell body (Ager and Mistry, 1988). Together, these results suggest preferential localisation of CS1 domain-containing fibronectin at the peripheral edges of HEC. The successful transmigration of bound lymphocytes may then depend on α₅β₁-mediated recognition of fibronectin in the intercellular or subendothelial matrix of HEC. The results presented here raise important questions about the precise distribution of fibronectin in high endothelial venules in vivo and it will be important to determine whether HEC-associated fibronectin, in particular the CS1 domain, is accessible to circulating leucocytes.

The molecular basis of the selective homing of lymphocytes, but not other leucocytes, to lymph nodes is not completely worked out. The selection of lymphocytes for entry into lymph nodes could occur either during the initial adhesion step or during the subsequent transendothelial migration phase. As shown here, LECAM-1, which is the prototype lymph node “homing receptor”, is widely distributed on the majority of circulating leucocytes and it is known to function on these other types of leucocyte. A soluble form of LECAM-1 has been shown to inhibit the migration of neutrophils into sites of inflammation (Watson et al., 1991) and antibodies to LECAM-1 block the initial, reversible adhesion or “rolling” of neutrophils on the luminal surface of inflamed EC (von Andrian et al., 1991). Although there is evidence that lymphocyte LECAM-1 and neutrophil LECAM-1 may recognise different ligands on endothelial cells (Picker et al., 1991) and thus regulate their distinct migration pathways via differential adhesion to HEC, it is possible that lymphocyte LECAM-1 on its own may not function as the homing receptor for lymph nodes and that additional signals will be required. Studies of lymphocyte interactions with HEV in vivo have shown that lymphocytes initially attach to HEC via their surface microvilli (Anderson and Anderson, 1976) and that this is a reversible, adhesive interaction (Bjerknes et al., 1986). Lymphocytes that remain attached to HEC enter apical clefts between individual HEC, adopt a motile morphology and subsequently penetrate the endothelial lining to enter the lymph node. In this context, fibronectin may function to stabilise the initial reversible, adhesive contacts between lymphocytes and HEC that precede transendothelial migration and thus fibronectin may be crucial for the successful homing (i.e. entry) of lymphocytes into lymph nodes. There are several observations that support a role for fibronectin in the selective migration pathways of lymphocytes. The CS1 sequence of fibronectin is subject to alternative splicing (Hynes, 1985) and thus is a possible source of tissue specificity in lymphocyte migration, e.g. fibronectin expressed by HEC contains the CS1 sequence whereas fibronectin expressed by other types of endothelia does not. As we have shown here, the selective expression of β₂ integrins by lymphocytes and not neutrophils contrasts with the widespread distribution of other homing-associated adhesion molecues on these two populations. Thus, α₄β₁ integrin on peripheral blood lymphocytes and the CSl-containing splice variant of fibronectin on HEC may contribute to consitutive migration of lymphocytes, but not neutrophils, into organised lymphoid tissues. The use of cultured high endodothelium may allow the precise roles of ft integrin? and the other homing-associated adhesion molecules, LECAM-1, CD44 and LFA-1, in the initial binding and subsequent transendothelial migration of lymphocytes to be delineated.

We thank Cheryl Holt for endothelial cell culture, Dr Helen Hourihan for assistance with the migration assay and Jackie Jolley for secretarial assistance. This work was supported by the Medical Research Council (UK) and the Wellcome Trust (UK).