Adhesion-independent synergy of monocytes and endothelial cells in cytokine production: regulation of IL-6 and GM–CSF production by PAF

Co-Cultures of monocytes (MO) and endothelial cells (EC) were studied for their capacity to synergize in the production of interleukin-6 (IL-6) and granulocyte-macrophage colony-stimulating factor (GM–CSF), two cytokines potentially important in vascular physiopathology. Resting monocytes produced detectable amounts of IL-6 but no GM–CSF, whereas confluent EC produced significant quantities of GM–CSF, but minimal IL-6. In co-cultures without stimuli, additive synthesis of both cytokines was observed. When EC were pretreated, however, with either PAF, TNF or both stimuli, before addition of MO, synergistic production of IL-6 was observed. In contrast, GM–CSF production was not enhanced by coculture of monocytes with activated EC. When either cell population was fixed with paraformaldehyde or killed by freeze-thawing before addition to the co-culture, cytokine levels reverted to those produced by the unaffected population alone. On the other hand, separating the two cell populations by a cell-impermeable membrane in transwell cultures did not affect the synergistic production of the cytokines. Taken together, our data suggest that EC and MO can synergize in response to stimuli by producing IL-6 and that this synergy is dependent on the integrity of both cell populations, but independent of cell-cell contact.


Introduction
Vascular endothelium interacts closely with leukocytes and constitutes an important cellular element in inflammatory and immunologic responses. This interaction is in part regulated by cytokines and several other mediators derived from leukocytes which act on endothelial cells (EC). [2][3][4] Among these, platelet-activating factor (PAF) can be produced by a variety of cells such as monocytes, neutrophils and endothelial cells. 5-This phospholipid is a potent pro-inflammatory mediator and causes a variety of effects, including microvascular leakage, contraction of smooth muscle and activation of neutrophils, macrophages and eosinophils. 9-We have demonstrated that PAF can induce suppressor cell activities in mononuclear leukocytes in vitro, ]2'13 as well as stimulate IL-1, TNF and IL-6 production in monocytes and alveolar macrophages. 4 '5 In cultured human EC, PAF was shown to induce shape changes, release of prostacyclin and vasodi-1617 k r latation.
PAF is also proposed as a ey playe in leukocyte adhesion to the endothelium by 56 Mediators of Inflammation Vol 5 1996 tethering leukocytes and inducing subsequent activation and adhesion stages.
Interleukin-6 (IL-6) has been shown to be a multifunctional cytokine that regulates, among other things, immune responses, acute phase reactions and haematopoiesis. It has also been reported to induce EC proliferation. 9 IL-6 is synthesized by EC following stimulation with IL-1, LPS or TNF-(, 2-22 but other cells also produce IL-6, including monocytes, macrophages and fibroblasts. [23][24][25][26] EC can also synthesize GM-CSF, either constitutively or more commonly after stimulation with IL-12v-31 or TNF. 32 The production of GM- 33 34 CSF by fibroblasts, T lymphocyt.es, mast 35 36 cells and monocytes/macrophages has also been demonstrated. GM-CSF is an important glycoprotein in the interaction of leukocytes with EC since it jromotes the adherence of monocytes to EC, augments accessory cell function of monocytes and stimulates certain monocyte effector activities involved in microbial and tumoricidal killing. 8 In the present work, we examined the potential synergy between EC and (C) 1996 Rapid Science Publishers monocytes in their production of IL-6 and GM-coated flasks 42 for 60 min at 37C. After removal CSF, using a co-culture system, of nonadherent cells, containing 95-98% lymphocytes, the adherent cells were further incubated Materials and Methods for 20 min at 37C in the presence of monosodium EDTA and medium (1:1). This procedure Reagents: PAF (1-O-alkyl-2-acetyl-sn-glycero-3yielded a cell population consisting of more than phosphocholine) was obtained from Bachem 95% pure monocytes as determined by non-Fine Chemicals (Torrance, CA); TNF-a was purspecific esterase staining. chased from Amersham and indomethacin was obtained from Sigma Chemical Co. (St Louis, Endothelial cells and monocyte co-cultures: Con-MO). fluent EC were detached from flasks, washed and resuspended in Iscove medium containing 5% Isolation of endothelial cells: EC were isolated FBS. Aliquots of 0.2 x 10 EC were preincubated from human umbilical vein as described pre-for 24h with either medium or graded conviously. 39'40 Briefly, umbilical veins from indivicentrations of PAF. EC were then washed and dual cords were cannulated, washed with PBS, incubated in the presence or absence of TNF-a disodium EDTA (5 x 10-4M; Sigma, St Louis, (25U/ml) and indomethacin (1 M) for 4h in MO), and treated with 0.2% collagenase type V Iscove medium supplemented with 0.25% BSA. (Sigma) in PBS. After an incubation of 20 min at Cells were washed again and incubated with 37C, detached cells were collected by washing 1 x 10 monocytes in Iscove medium containing twice with PBS, centrifuged and resuspended in 5% FBS. After a 24h co-culture, cell-free supercomplete medium consisting of Iscove medium natants were harvested and assayed for the (Flow, McLean, VA) supplemented with 10% presence of IL-6 and GM-CSF. In preliminary heat-inactivated foetal bovine serum (FBS, Flow), experiments, these conditions were found to be penicillin (167U/ml), streptomycin (2001.tg/ml), optimal for cytokine production. amphotericin B (4 I.tg/ml), and 37.5 l.tg/ml EC growth supplement (CR-ECGS; Collaborative IL-6 and GM-CSF assay: The activity of IL-6 was Research Inc., Lexington, MA). These primary determined using IL-6-dependent cells of the culture cells were plated on 75-cm 2 tissue culture murine hybridoma cell line B9 as described. 43 '44 flasks (Falcon Oxnard, CA), fed three times a Briefly, serially diluted samples were dispensed week and usually became confluent in 4-7 days. into fiat-bottomed 96-well microtitre plates When cultures reached confluence, flasks were (Limbro, McLean, MO). B9 cells, resuspended in rinsed with disodium EDTA and then incubated RPMI 1640 medium + 5% FBS and 5 x 10-5M for 2-3 min with EDTA containing 0.05% trypsin 2-mercaptoethanol, were added (5 x 103 cells/ (Gibco, Grand Island, NY) to detach the cells, well) and the plates were incubated at 37C for EC were washed, resuspended in complete 72 h. Proliferation of B9 cells was measured by a medium and seeded in flasks precoated with colorimetric assay using MTT (5mg/ml; Sigma). 0.1% gelatin (Sigma)in Iscove medium.
Plates were then read at 595nm using the EC were characterized by the detection of BioRad Microplate Reader (BioRad, Richmond, factor VIII-like antigen (von Willebrand's factor) CA). The activity of IL-6 in test supernatants was by indirect immunofluorescence performed with calculated on the basis of cellular growth a rabbit anti-vW factor antiserum (Calbiochem obtained in the presence of various concentra-Behring, La Jolla, CA). tions of a rIL-6 standard, using probit analysis of dilution curve data. Immunoreactive human GM-Preparation of monocytes.. Peripheral blood was CSF was measured using the enzyme-linked  Ficoll-Hypaque (Ficoll 400, Pharmacia, Uppsala, Sweden; Hypaque sodium, Winthrop Results Laboratories, Aurora, Ontario). After centrifugation for 30 min at 400 x g at room temperature, Unstimulated EC produced minimal quantities peripheral blood mononuclear leukocytes of IL-6, and their stimulation with TNF-a (25 U/ (PBML) were collected at the interface, washed ml) had a negligible effect (Fig. 1). In contrast, twice in PBS and resuspended in Iscove medium unstimulated monocytes produced significant containing 10% FBS and antibiotics. The cell. susamounts of IL-6. When these monocytes were copension was composed of approximately 75-85% cultured with resting EC, an additive effect was lymphocytes and 15-20% monocytes. Monocytes observed. In contrast, when they were co-culwere separated by adherence to microexsudate-tured with TNF-treated EC, a synergistic three- EC were preincubated in the absence or presence of graded concentrations of PAF for 24 h, washed and exposed to medium or TNF-(25 U/ml) for 4h. After washing, EC were co-cultured with unstimulated monocytes for 24h. Cell-free supernatants were then harvested for measurement for IL-6 activity. Data represent mean -t-S.E.M. of four to 12 experiments. Statistically significant effects were noted as: *p < 0.05. I--, with TNF; <, without TNF.
fold stimulation of IL-6 production was observed ( Fig. 1). Since PAF is involved in leukocyte-EC interactions, we examined whether PAF could modulate IL-6 production in EC-MO co-cultures. EC were pretreated with graded concentrations of PAF (10 -16-10-6M) before addition of monocytes.
As shown in Fig. 2, PAF alone was at least as effective as TNF alone in stimulating IL-6 production in the co-cultures, with a maximal effect at 10-M PAF. When EC were pretreated with PAF and subsequently stimulated with TNF, their coculture with monocytes resulted in a synergistic six-fold enhancement of IL-6 production.
In order to analyse the cellular requirements for the synergistic response to PAF and TNF, cocultures were compared to EC or MO populations alone, as well as to co-cultures in which one of the cell populations was either killed by freeze-thawing (to preserve membrane structures) or fixed by paraformaldehyde treatment.
As shown in Fig. 3, killed or fixed EC could no 58 Mediators of Inflammation Vol 5 1996 longer synergize with MO and IL-6 production reverted to that of MO alone. Similarly, killed or fixed MO could no longer synergize with EC and IL-6 levels fell to the minimal production observed with EC alone.
In contrast to IL-6, GM-CSF production was mainly observed in resting or stimulated EC, whereas monocytes alone produced negligible quantities of the cytokine (Fig. 4). Enhanced synthesis of GM-CSF was observed in co-cultures of untreated EC and MO but, whereas stimulated EC alone showed doubling of their GM-CSF production, addition of MO to stimulated EC had no further effect. The main cellular source of GM-CSF production was made evident in the co-cultures with one cell population killed by freezethawing or fixed with paraformaldehyde. Co-cultures of intact MO with fixed or killed EC showed little GM-CSF production, whereas cocultures of intact EC with killed or fixed MO had responses similar to those of EC alone. EC were washed and either fixed with paraformaldehyde (0.05%) for 30 min, or lysed by three cycles of freezing (liquid nitrogen) and thawing (37C). Thereafter EC were washed and incubated with monocytes. Alternatively, monocytes were fixed or lysed in the same manner and then washed and added to intact EC. After 24h of co-culture, cell-free supernatants were harvested and assayed for IL-6 activity. Data represent mean -t-S.E.M. of three experiments. Statistically significant effects were noted as: *p < 0.0001" **p < 0.0005, in comparison to monocytes alone. ,, Medium; [], TNF.
Because synergy in IL-6 production was dependent on live and fully functional MO and EC, and intact membranes did not substitute for live cells, we tested the requirement for actual cell-cell contact for the observed synergistic response. We used Costar transwell culture vessels in which MO were separated from EC by a polycarbonate, cell-impermeable membrane. As shown in Table 1, synergy in IL-6 production was maintained in spite of the physical separation of the two cell populations.

Discussion
Our studies described here present evidence for contact-independent synergy in production of IL-6 by co-cultures of human MO and EC.
Maximal effects were observed when resting MO were added to PAF-primed and TNF-treated EC, ## #_T_# FIG. 4. Effect of prior lysis or fixation of either cell population on the production of GM-CSF by EC:MO co-cultures. After treatment for 24h with either medium or PAF (10-1M)+TNF-(25 U/ml), EC were washed and either fixed with paraformaldehyde (0.05%) for 30 min, or lysed by three cycles of freezing (liquid nitrogen) and thawing (37C). Thereafter EC were washed and incubated with monocytes. Alternatively, monocytes were fixed or lysed in the same manner and then washed and added to intact EC. After 24 h of co-culture, cell-free supernatants were harvested and assayed for GM-CSF activity. Data represent mean -t-S.E.M. of three experiments. Statistically significant effects were noted as: *p < 0.005; #p < 0.0005, in comparison to EC alone. ,, Medium; [], PAF and TNF. but a significant concentration-dependent enhancement of IL-6 production was also observed in co-cultures of MO and EC, when the latter were pretreated with PAF alone.
Since PAF can induce its own synthesis in EC 45 and since we have shown PAF to stimulate IL-6 production in EC, monocytes and alveolar 46 47 macrophages, it is possible that the observed synergy may be due, in part, to the action of de novo synthesized, EC-derived PAF and MO. This is unlikely, however, since EC retain most of the synthesized PAF in a cell-associated form and cell-cell contact was shown not to be necessary for the observed synergy. Moreover, EC mem- branes alone were not sufficient to trigger MO synthesis of IL-6, as demonstrated by the freezethaw experiments: MO cultured with lysed, but otherwise intact PAF-primed EC produced no more IL-6 than MO alone. In contrast to IL-6, GM-CSF production was mainly derived from EC and their stimulation with PAF and TNF resulted in enhanced production of the cytokine. Since GM-CSF can increase IL-6 production by monocytes, 2 it remains to be tested whether GM-CSF plays a role in the synergy in IL-6 production observed in our cocultures. The transwell experiments suggest that a soluble factor from either of the two cell types, or from both cell types, is essential for the upregulated production of cytokines. Further studies will help determine the nature of such factor(s).
In summary, exposure to EC and the inflammatory stimuli PAF and TNF resulted not only in their enhanced production of GM-CSF, but also in their ability to synergize with MO in augmented IL-6 production. Interestingly, this synergy did not require cell-cell contact between EC and MO, in contrast to many other synergistic events which depend on cell contact and involve various combinations of cell adhesion molecules.