Granulocyte adhesion to endothelium in culture

Theproduction of an acute inflammatory response is dependent upon the localized adhesion of granulocytes to the endothelial cells of small blood vessels, particularly venules, adjacent to the stimulus. Following adhesion, granulocytes migrate between endothelial cells to reach the site of the reaction. To study the interactions between blood cells and vascular cells under defined conditions we have used a model system in which suspensions of leukocytes are rotated over monolayers of cultured endothelium and adhesion to the vascular cells can be assessed microscopically and counted by using 51Cr-labelled leukocytes.

With this technique we showed previously that adhesion of leukocytes to endo-thelium or other substrata was markedly affected by flow rate, that leukocytes pre-ferentially adhered to endothelial cells, in comparison with several other cell types or serum-coated glass, and that adhesion required the presence of divalent cations but not plasma-derived cofactors (Beesley et al. 1978).

We have now studied the effects of other blood cells on granulocyte adhesion and investigated possible modulating factors produced by endothelial cells or granulo-the effects of some drugs and vasoactive agents on adhesion, and attempted to localize their sites of action.

Unless otherwise stated experiments were carried out at 37 °C, by rotating 0·5 ml of a suspension of autologous 51Cr-labelled granulocytes (z x 10°/ml) in serum-free HEPES buffered Dulbecco’s medium (DMEM; Dulbecco & Freeman, 1959) for 30 min over 13-mm serum-coated glass coverslips supporting a confluent monolayer of porcine aortic endothelial cells. Before experiments the endothelial cells were rinsed free of serum-containing growth medium and incubated in DMEM for 30 min. At the end of the experiments the suspensions of granulocytes were removed and each coverslip was rinsed by passing 4 times through an air/saline interface. Adherent granulocytes were counted by measuring the radioactivity associated with the coverslips using a Packard Autogamma Spectrometer. Details of the cell culture and of experimental methods, including the preparation from heparinized pig blood of platelet-rich or cell-free plasma and purified (> 80%) radiolabelled granulocytes, have been presented previously (Pearson, Carleton, Hutchings & Gordon, 1978; Beesley et al. 1978). Erythrocytes were purified by adding 1-to z-ml samples of packed red cells to a 25 mm x 3 mm glass-wool column in a Pasteur pipette, previously rinsed with 10 ml of phosphate-buffered saline, eluting with 3-4 ml of phosphate-buffered saline, and passing the eluate through a second similar column. After this procedure there was no detectable contamination of erythro-cytes with white cells.

To examine whether variations in adhesion caused by various treatments were due to effects primarily on the endothelium or on the granulocytes, the adhesion of granulocytes to serum-coated glass was measured in some experiments, in parallel with adhesion to endothelium.

Because of inter-experimental variability in the absolute numbers of granulocytes adhering under control conditions, the results have been expressed as a percentage of the control value obtained in the same experiment. Significant differences between groups of observations were revealed by analysis of variance of the original data, and contrasts between means were evaluated by computing t-values from the residual variation.

These were either added at the beginning of incubations, or granulocytes or endothelial cells were separately preincubated at 37 °C (usually for 30 min) in DMEM with the drug, and rinsed once before use. Synthetic prostaglandins (PG) 1₂ and 6-oxo-PGF_1α were gifts from Dr P. Walton (ICI, Macclesfield, U.K.). Other PGs and thromboxane (Tx) B2 were given by Dr J.E. Pike (Upjohn Co., Kalamazoo, MI U.S.A.). Paroven (O-β-hydroxyethyl)rutosides) and (+)-Catechin were donated by Dr T. B. Pulvertaft (Zyma, Macclesfield, U.K.). Other compounds were obtained from commercial sources.

The maximum concentration of each drug tested was selected on the basis of its established activity in other aspects of the blood-vascular system. Where a drug was ineffective, only the highest concentration tested is shown, and where a drug had a substantial effect at the maximum selected concentration, lower concentrations were usually tested.

Preliminary light-microscopic examination of granulocyte adhesion to endothelium, whether by phase-contrast examination of living cells, or study of fixed and stained preparations (Fig. 1), showed that the adherent blood cells appeared either compact and rounded, or spread out. Subsequent investigation (see the accompanying paper; Beesley et al. 1979) revealed that the rounded cells were adhering to the upper surface of the endothelial monolayer, whereas flattened cells with obvious multilobed nuclei had migrated beneath the monolayer. In all of the following experiments, using radiolabelled granulocytes, the values for adhesion quoted include adherent and migrated cells, the latter usually comprising about 10 % of the total.

The presence of a 100-fold excess of purified erythrocytes (as in whole blood) significantly and consistently enhanced the adhesion of granulocytes to endothelium, but did not affect the adhesion of granulocytes to serum-coated glass.

The adhesion of granulocytes suspended in heparinized (5 i.u./ml) cell-free plasma varied somewhat with individual batches of plasma. Adhesion to endothelium, as we have previously noted (Beesley et al. 1978) was usually similar to DMEM control values, and never significantly lower. In some experiments, however (as in the example shown in Table 1), adhesion to endothelium was significantly higher in cell-free plasma than in DMEM. In parallel experiments, using the same batch of granulocytes and plasma, adhesion of granulocytes to serum-coated glass was not increased.

In all experiments granulocytes suspended in heparinized platelet-rich plasma adhered to endothelium in significantly lower numbers than granulocytes in DMEM or cell-free plasma. This decreased adhesion was not affected by pretreating the platelet-rich plasma with I mM aspirin (15 min at 20 °C; data not shown). The adhesion of granulocytes in flatelet-rich plasma to serum-coated glass, however, did not differ from that found in either DMEM or cell-free plasma (Table 1).

Granulocytes were suspended in DMEM that had been previously incubated over endothelium for 30 min, and their adhesion, either to a separate endothelial mono-layer or to serum-coated glass, was tested. Granulocytes suspended in this conditioned medium adhered to endothelium in similar numbers to granulocytes in fresh DMEM. In some paired experiments, however, their adhesion to serum-coated glass was significantly increased (e.g. 130±11% of control value 100±6% (n=6); 0·001< P < 0·01). This pattern was more consistently demonstrated if granulocytes were suspended in cell-free plasma that had been preincubated over endothelial cells, rather than DMEM. In such experiments, although adhesion to fresh endothelial cells was again not different from that found when granulocytes were suspended in fresh cell-free plasma, adhesion to serum-coated glass was significantly greater in all experiments (e.g. 139±7% of control value I oo±9 % (n = 6); o·oo1 < P < o·o 1).

The adhesion of granulocytes in fresh DMEM was also tested on endothelium that had been preincubated with either DMEM, cell-free plasma or platelet-rich plasma for 30 min, and then rinsed. Adhesion was greatly increased after pre-incubation with either plasma, the values obtained being I oo±11 % (n=6) for DMEM; 223 ± 19% (n= 5) for cell-free plasma; and 296 ± 35 % (n=4) for platelet-rich plasma. Both plasma figures were significantly different from DMEM (P < 0·001).

Paroven, clinically used for the treatment of venous insufficiency, has been shown to reduce capillary permeability apparently by a direct action on endothelium (Comel & Laszt, 1972). When granulocytes in DMEM were incubated with endothelial cells that had been preincubated with Paroven for 30 min, then rinsed, there was a dose-related increase in adhesion to the Paroven-treated cells (Table 2). The chemically related drug (+)-Catechin also increased adhesion, although this result was significant only at the level 0·05 < P < 0· 1. Adhesion of granulocytes to serum-coated glass was not enhanced in the presence of Paroven nor was granulocyte adhesion to endothelium enhanced if the blood cells were preincubated with Paroven.

The addition of cyclic AMP (cAMP) itself (up to 10^-4 M) did not affect adhesion to endothelium, but 10^-4 M dibutyryl cAMP significantly increased adhesion. Adenosine, which stimulates adenylate cyclase in many cell types, also increased adhesion. Papaverine, a phosphodiesterase inhibitor, did not affect adhesion when added alone; neither did it potentiate the increase caused by adenosine (data not shown). ADP also slightly but significantly increased adhesion, probably because it is broken down at the endothelial cell surface to give adenosine (Lieberman, Lewis & Peters, 1977; Pearson & Gordon, unpublished results). Preincubation of granulocytes with 10^-5 M noradrenaline, another adenylate cyclase stimulant, also increased adhesion (Table 3).

Granulocyte adhesion to serum-coated glass was increased when adenosine or dibutyryl cAMP was present, but preincubation of endothelium with these agents, then rinsing, had no effect on granulocyte adhesion.

Several prostaglandins were found to be capable of increasing the adhesion of granulocytes to endothelium, but the effects varied in magnitude in different experiments. For example, in 6 separate experiments, 10^-6 M PGE2 significantly increased adhesion in 2, increased adhesion slightly but not significantly in 2, and had no effect in 2.

Inhibition of adhesion was never found with any of the prostaglandins tested. Paired samples in experiments showing sensitivity to exogenous prostaglandins (see Table 4) indicated that pro-adhesiveprostaglandins acted in a dose-dependent manner, and that the most active compound was 6-oxo-PGF1a;, TxB2 did not significantly increase adhesion and the PG endoperoxide analogue U44619 was also ineffective. The effects of pretreating endothelium with 10^-3 or 10^-4 M aspirin were tested in several experiments and also found to be variable; granulocyte adhesion was sometimes reduced and sometimes unaltered (data not shown). Preincubating granulocytes with aspirin (10^-3 M for 30 min) had no effect on their adhesion to endothelium.

The variability encountered when testing the effects of prostaglandins or aspirin on granulocyte adhesion to endothelium was not found when testing adhesion to serum coated glass. Adhesion was consistently enhanced in the presence of 10^-6 M PGE2 (e.g. control, 100±7% (n=12); test, 135±n% (n=6); 0·001<P<o·o1). Surprisingly, aspirin-treated granulocytes adhered in significantly greater numbers than untreated cells to serum-coated glass (control, 100 ± 10% (n=6), aspirin-treated cells 167 ± 15 % (n=6); P<o·oo1).

Granulocyte adhesion to cultured endothelium can be significantly altered by the presence of other blood cells. The enhancement of adhesion produced by erythrocytes, unlike the effects of any other treatment, was accompanied by a large increase in the proportion of migrated cells, and a possible mechanism is discussed in the accompanying paper (Beesley et al. 1979). In contrast, the presence of blood platelets (which had no effect on migration) markedly decreased adhesion to endothelium. This is unlikely to be a purely physical effect: platelets themselves adhered very poorly to endothelium, and much better to serum-coated glass, but granulocyte adhesion to serum-coated glass was not altered by the presence of platelets. The inhibition by platelets of granulocyte adhesion to endothelium was not affected by pretreating the platelet-rich plasma with aspirin. This drug acetylates fatty acid cyclo-oxygenase and thus blocks the synthesis from arachidonic acid of platelet prostaglandins and thromboxanes, consequently inhibiting the secretion of platelet dense granule constituents such as ADP and serotonin. Therefore, if a factor secreted by platelets is responsible for their effect, it cannot be a product of the cyclo-oxygenase pathway and is unlikely to be a constituent of dense granules. The possibility remains that such a factor might be derived from another granule population, the contents of which are more readily released (Witte et al. 1978), or from the action of fatty acid lipoxygenase, which is not blocked by aspirin, and produces biologically active derivatives of arachidonic acid (Goetzl & Gorman, 1978). Alternatively, platelets may remove or inactivate a pro-adhesive factor produced by the endo thelium. In this context, it is of interest that the presence of platelets inhibits granulocyte migration (Nelson, McCormack & Fiegel, 1978).

That endothelial cells may produce a pro-adhesive factor when incubated with plasma was also suggested by the observation that granulocyte adhesion to endothelium, but not to serum-coated glass, was sometimes significantly greater in cell-free plasma than in DMEM. The clearest demonstration of this was provided by testing the adhesion of granulocytes, in fresh DMEM, to endothelial cells preincubated with either platelet-rich or cell-free plasma; in these experiments adhesion was 2-to 3-fold greater than when the endothelial cells were preincubated with DMEM. These results indicate that exposure to plasma for 30 min stimulates the production of pro adhesive activity by the endothelial cells, which is sustained for a sufficient time after removal of plasma to produce a substantial increase in adhesion to those endothelial cells in a subsequent test using granulocytes in DMEM.

Granulocytes suspended in cell-free plasma that had been preincubated with endothelium also adhered in significantly greater numbers to serum-coated glass than did granulocytes in fresh plasma. Adhesion was, however, increased only 1·2-to I· 5-fold in these experiments, suggesting either that the bulk of the pro-adhesive activity was associated with the endothelial cell surface rather than being freely soluble, or that the activity was unstable in plasma. The latter conclusion was supported by finding that the adhesion of granulocytes in plasma to endothelium was not further increased if the plasma had been pre-incubated with endothelial cells.

The flavonoid drugs Paroven and (+)-Catechin, the former of which is used to treat venous insufficiency, increased granulocyte adhesion to endothelium and our experiments showed that this effect was due to the binding of drugs by endothelial cells and not to an effect on leukocytes. This is consistent with the current hypothesis that these drugs bind, together with calcium ions, to endothelial surface glycoproteins (Hladovec, 1977); we have previously shown that divalent cations stimulate granulocyte adhesion (Beesley et al. 1978).

Tests with dibutyryl cAMP suggested that increasing granulocyte cAMP levels potentiated adhesion, and this conclusion was strengthened by the observation that compounds known to stimulate adenylate cyclase (adenosine, noradrenaline and several prostaglandins) also increased adhesion. Thromboxane B2 and a PGH2 analogue, which are not cyclase stimulants, did not increase adhesion. MacGregor, Macarak & Kefalides (1978) reported that granulocyte adhesion to endothelium (measured in a static system) was slightly reduced by adding 10 μM cAMP; we found that cAMP itself had no effect and, indeed, exogenous cAMP is generally recognized as being ineffective in other systems, except at high concentrations, because it does not readily penetrate cell membranes. Higgs, Moncada & Vane (1978) have presented preliminary data showing that infusion of PG 1₂ reduced the number of granulocytes rolling along venules of the hamster cheek pouch. This could have been caused by increased granulocyte adhesion elsewhere in the vascular bed, or due to a local inhibition of adhesion; if the latter, there is apparently some discrepancy with our results, presumably due to differences in methodology or animal species, and further work is needed to clarify this. The increase in granulocyte adhesion that we observed with added prostaglandins was variable when endothelium was the sub-stratum but reproducible when serum-coated glass was used, which again suggested that different batches of endothelium were producing variable amounts of factor(s) that affected granulocyte adhesion (and hence altered the effects of exogenous modulators). Our cultured endothelial cells produce PGl₂, and this synthesis is stimulated by plasma (MacIntyre, Pearson & Gordon, 1978). The amount produced is variable (Gordon & Pearson, 1978), and variable amounts of PGE2 are also formed (Ager, Pearson & Gordon, unpublished work); all these observations are consistent with the concept that prostaglandins are partly responsible for the modulation of granulocyte adhesion by endothelial cells. In addition, platelets are known to bind PGl₂ with high affinity (Siegl, Smith & Silver, 1978), which might in part explain their effect on granulocyte adhesion.

Variable prostaglandin production by endothelium (and also perhaps by granulocytes) could also explain some other observations: for example, in our previous study (Beesley et al. 1978) we established that granulocyte adhesion to endothelium did not require any plasma cofactors, but during the more detailed experimen’s of our present study we sometimes observed a stimulation of adhesion by plasma, compared with DMEM (see Table 1). In such cases, pretreatment of endothelium with aspirin (100 /tM) prevented this stimulation: for example, in a paired experiment to that in Table 1, with aspirin-treated cells, adhesion in DMEM was 100 ± 5 % (n= 6), and adhesion in cell-free plasma was 84 ± 6% (n= 6). In contrast, treatment of granulocytes with aspirin increased adhesion to serum-coated glass, perhaps suggesting that production of an anti-adhesive prostaglandin from granulocytes had been blocked, although we have not yet identified such a product.

To summarize, we have shown that other blood cells can significantly modulate the adhesion of granulocytes to endothelial cells, and have also found several chemical agents that affect adhesion, including some prostaglandins. By comparing granulocyte adhesion to endothelium and to serum-coated glass we have been able to localize the apparent site of action of some agents. The modification of adhesion we have achieved has usually not been great, from about 80 to 120 % of control values, and only rarely

< 50% or > 200%, Hoover, Briggs & Karnovsky (1978), using a static collecting assay, have also found only slight changes (up to about 150% of controls) in granulo cyte adhesion to cultured endothelium on pretreating either cell type with a variety of chemotactic agents. Few if any of these differences are as marked as the preferential adhesion of granulocytes to endothelium by comparison with other cellular substrata (Beesley et al. 1978; Hoover et al. 1978; MacGregor et al. 1978), and we therefore conclude that the major specificity of the interaction lies with the types of surface glycoprotein expressed by endothelial cells, rather than any soluble pro-adhesive factors released from them.

Our main conclusion, however, is that leukocyte-endothelial interaction is complex, and each cell type responds to the other or to factors released from it. There is suggestive evidence for this from studies in vivo, where the temporary adhesion of a leukocyte to endothelium can subsequently render that cell adhesive to other leuko-cytes (Atherton & Born, 1972). We found that the effects of several agents on granulo cyte adhesion to endothelium were not paralleled by their effects on adhesion to serum-coated glass, and we therefore conclude that data derived solely from studies of granulocyte adherence to inert substrata will not always reflect the pattern of granulocyte-endothelium interaction.

Finally, because of the mutual interdependence of the 2 cells’ responses we believe that it is important to evaluate in more detail the individual responses of each cell type (for example, the patterns of prostaglandin production by endothelial cells or leuko-cytes in response to stimuli; the role of plasma components) before attempting a fuller understanding of their interaction.

This study was supported by grants from the Arthritis and Rheumatism Council and the Medical Research Council.