Prednisolone inhibits cytokine-induced adhesive and cytotoxic interactions between endothelial cells and neutrophils in vitro

Abstract

We assessed whether prednisolone influenced the ability of human polymorphonuclear neutrophils (PMN) to adhere to and cause lysis of human umbilical vein endothelial cells (HUVEC) in vitro (as measured by the release of ⁵¹Cr). Pretreatment of the endothelium with IL-1β or tumour necrosis factor-alpha (TNF-α) caused prominent endothelial E-selectin expression and endothelial hyperadhesiveness for neutrophils, as well as PMN-mediated cytotoxicity. All these processes were dose-dependently reduced when prednisolone was added to the assay system. This protective effect remained when HUVEC alone were pretreated with the drug prior to washing and cytokine activation. Likewise, when HUVEC cytotoxicity was induced by the nitric oxide (NO) donor S-nitroso-acetyl-penicillamine (SNAP), prednisolone reduced cell injury significantly. In contrast, prednisolone did not interfere with signalling systems between TNF-α-stimulated HUVEC and quiescent PMN such as IL-8 generation and release of cytosolic Ca^{2 +} in the PMN. Thus, in this in vitro model of vasculitis, prednisolone dose-dependently reduced cytokine-induced E-selectin expression and HUVEC hyperadhesiveness for neutrophils, as well as reducing neutrophil-dependent cytotoxicity against HUVEC via NO-dependent steps.

INTRODUCTION

Activated polymorphonuclear neutrophil (PMN) granulocytes may adhere to and interact with the vascular endothelium resulting in damage or killing of the latter [1–4]. This process plays an important part in the pathogenesis of, e.g. vasculitides associated with rheumatic diseases [5]. It is well known from in vitro systems that adhesion molecules, release of toxic oxygen metabolites as well as proteolytic enzymes, alone or in concerted action, from activated PMN are pivotal for this process [6,7]. Moreover, we have recently shown that released nitric oxide (NO) is important for neutrophil-mediated injury of human umbilical vein endothelial cells (HUVEC) [8].

The proinflammatory cytokines IL-1β and tumour necrosis factor-alpha (TNF-α) are known to stimulate endothelial cells (EC) to express adhesive and activation molecules for leucocytes [9–12]. One consequence of this is the activation of the cytotoxic capacity of PMN, which may result in injury of the EC [13–15]. This process is dependent on the expression of adhesion molecules and is probably associated with released NO [13]. TNF-α also induces production of high levels of IL-8 by HUVEC that may be part of the signalling system between these cells and neutrophils [13,16,17]. Moreover, TNF-α-treated HUVEC also confer a transient rise of [Ca^{2 +}]_i in PMN [13], possibly indicating a receptor and phospholipase C-mediated signalling event.

We have recently described that the anti-rheumatic disease-modifying drugs auranofin, gold sodium thiomalate and sulphasalazine are potent inhibitors of endothelial cytotoxicity induced by PMN stimulated by the calcium ionophore A23187, the chemotactic peptide N-formyl-methionyl-leucyl-phenylalanine (fMLP) and the physiologically occurring lipoxygenation product of arachidonic acid, lipoxin A₄ (LXA₄) [18]. Others have reported that glucocorticoids inhibit PMN adhesive, chemotactic and secretory responses, as well as the generation of arachidonic acid metabolites [19–23]. These reactions may be of significance for the clinical effects of the drugs.

Based on these observations we wanted to study whether prednisolone influenced: endothelial expression of E-selectin; endothelial hyperadhesiveness for neutrophils and neutrophil-mediated injury of HUVEC induced by activation of HUVEC with IL-1β or TNF-α in vitro.

MATERIALS AND METHODS

Chemicals

Chemicals were obtained as follows: bovine haemoglobin, heparin, human serum albumin (HSA; essential fatty acid free), prednisolone, TNF-α and Triton X-100 from Sigma Chemical Co. (St Louis, MO); IL-1β (5 × 10⁷ U/mg) from Boehringer Mannheim (Mannheim, Germany) and S-nitroso-N-acetyl-penicillamine (SNAP) from Wellcome Research Labs (Beckenham, UK); ⁵¹Cr from Du Pont Co. (Wilmington, DE) and endothelial cell growth supplement from Becton Dickinson Labware (Bedford, MA); fetal calf serum (FCS), HEPES, penicillin, streptomycin, sodium pyruvate, non-essential amino acids, RPMI 1640 and Hanks’ balanced salt solution (HBSS) from Gibco (Paisley, UK); collagenase (type 3) from Worthington (Freehold, NJ), Sephadex G25 and Percoll from Pharmacia Fine Chemicals (Uppsala, Sweden); EDTA from Merck (Darmstadt, Germany), BCECF/AM and Fura-2/AM from Molecular Probes (Eugene, OR). Ninety-six-well microtitre plates (Immunolon) were obtained from Dynatech Labs (Chantilly, VI), and 24-well polystyrene plates (2 cm²/well) and tissue culture plastic ware were from Nunc (Roskilde, Denmark). A prednisolone stock solution of 100 mm was prepared in DMSO and aliquots were frozen. Working dilutions were prepared by dissolving the stock solution in HBSS + HSA 0·4%.

Antibodies

Murine (unlabelled) MoAbs directed against E-selectin (H18/7) and murine isotype-matched non-binding control MoAbs were from Becton Dickinson Immunocytometry Systems (San Jose, CA), FITC-conjugated rabbit anti-mouse immunoglobulin antibodies from Dako A/S (Copenhagen, Denmark).

EC cultures

HUVEC were obtained from human umbilical veins by treatment with 0·2% collagenase as previously described [24,25]. Cells were suspended in culture medium (RPMI 1640 with glutamine, FCS 20%, heparin 90 μ g/ml, endothelial growth supplement 50 μ g/ml, HEPES 0·012 m, penicillin 100 U/ml, streptomycin 100 μ g/ml, sodium pyruvate and non-essential amino acids) and grown in 75-cm² tissue culture flasks precoated with 2% gelatine. The culture medium was changed the following day. HUVEC were trypsinized when confluent, resuspended in medium and seeded into 96- or 24-well microtitre plates (for adherence or cytotoxicity assays, respectively) or 25-cm² tissue culture flasks (for the adhesion molecule assay) and utilized upon achieving confluence.

Neutrophil preparations

PMN were obtained from healthy donors by a one-step discontinuous Percoll gradient centrifugation as previously described [26]. The purified neutrophils (> 95% purity and viability) were resuspended in HBSS with HSA 0·4%. PMN to be stained with the fluorescent probe BCECF/AM (for the adherence assay) were resuspended in Ca^{2 +}- and Mg^{2 +}-free HBSS and incubated with BCECF/AM (2 μm) for 20 min at 37°C. PMN were washed twice and transferred to regular HBSS (with Ca^{2 +} and Mg^{2 +}) with HSA 0·4% [27,28].

Endothelial expression of E-selectin

Endothelial expression of E-selectin was assessed as previously described [28,29]. Briefly, HUVEC were incubated with prednisolone for 30 min at 37°C. Subsequently, HUVEC were incubated with IL-1β, TNF-α or medium for 3 h. After trypsinization, washing and resuspension in PBS, HUVEC were incubated with either an anti-E-selectin MoAb (H18/7) or a control MoAb for 30 min, and then with a secondary FITC-conjugated antibody. Finally, HUVEC were washed and fixed with 1% formaldehyde. Fluorescence was analysed in a Becton Dickinson (Mountain View, CA) FACScan flow cytometer.

In vitro PMN adherence assay

Adherence of PMN to HUVEC monolayers was assessed as previously described [27,28]. HUVEC were treated with prednisolone (or its solvent) for 30 min. This time period was chosen because pretrial experiments showed no difference in the amount of adherent PMN for pretreatment periods with prednisolone ranging from 30 min to 120 min. Moreover, glucocorticoids have been reported to exert an immediate impairment of cytokine-induced activation of HUVEC in the absence of any pretreatment period [30]. Consequently, all other assays were run with a 30-min treatment period with prednisolone. Subsequently, either IL-1β or TNF-α was added. After incubation at 37°C for 3 h (which is the optimal time for the induction of adhesive properties of these cytokines) [28], HUVEC were washed three times, and prednisolone (or the solvent) was added again. BCECF-stained PMN (2 × 10⁵; in assay buffer) were added to each well and adherence was analysed after 10 min. Subsequently, fluorescence was determined in a microtitre plate fluorimeter (Fluoroskan II; Labsystems, Helsinki, Finland) (detailed in [28]). Tests were run at least in triplicate. The evaluation of the adherence assay with regard to possible interfering factors, e.g. PMN aggregation, etc., has been described previously [31].

⁵¹Cr-release cytotoxicity assay

⁵¹Cr-release cytotoxicity assay was performed as described previously [25]. In brief, HUVEC monolayers were loaded with ⁵¹Cr. IL-1β or TNF-α was added for 4 h or 24 h, respectively. After washing the HUVEC monolayer was covered with 1·25 × 10⁶ PMN in HBSS. Subsequently, after incubation of HUVEC and PMN together for 4 h, the radioactivity of the supernatants was counted. Injury of the HUVEC was expressed as percent specific ⁵¹Cr release, as previously described [25]. Prednisolone was either added together with the PMN (the complete system) or used to incubate the HUVEC for 30 min (and subsequently washed away) prior to activation with IL-1β or TNF-α. All assays were performed in duplicate. Details of the spontaneous release, intra- and interassay variations and other technical aspects have been given previously [13,25].

PMN [Ca^{2 +}]_i changes

PMN [Ca^{2 +}]_i changes were assessed as previously described [32]. PMN were treated with 1 μm Fura-2/AM for 40 min, washed and resuspended in HBSS. After calibration of the system [32,33], first passage HUVEC were suspended in HBSS, incubated for 2 h with TNF-α and added to the PMN suspension in the fluorimeter at a final concentration of 5 × 10⁵ HUVEC/ml. Prednisolone was added before TNF-α. HUVEC treated with HBSS alone served as controls.

IL-8 production from HUVEC

IL-8 production from HUVEC was assessed as previously described [13]. HUVEC were stimulated with TNF-α for 4 h. The IL-8 concentration in combined freeze–thawed cell supernatants and cell lysates was analysed using an ELISA technique (Medgenix Diagnostics, Fleurus, Belgium).

Neutrophil lysate preparations

Extracts were prepared either by lysing neutrophil suspensions with three repetitive freeze–thaw cycles or by ultrasonicating PMN suspension for 20 s followed in both methods by centrifugation (10 000 g; 10 min) to sediment cellular debris. We found no difference in the cytotoxic capacity of PMN lysate prepared by either method.

Preparation of oxyhaemoglobin and measurement of NO release

Solutions of oxyhaemoglobin (HbO₂) were prepared freshly each day and oxygenated and deoxygenated as previously described [8]. The concentration of HbO₂ was determined spectrophotometrically and absorption spectra (400–600 nm) were obtained for each preparation. NO release from SNAP was assessed spectrophotometrically by the NO-dependent formation of met-haemoglobin from HbO₂. Release of NO was measured as pmol met-haemoglobin formed per minute during the linear phase of the response using an e= 19·7 mm⁻¹ cm⁻¹[34].

Assessment of viability of HUVEC

Assessment of viability of HUVEC prior to use or after separate incubations with the drugs was performed by trypan blue exclusion, and release of either lactate dehydrogenase or ⁵¹Cr. No decreased viability under these experimental conditions could not demonstrated.

Statistical analysis

Data were analysed using the Statistica software package and Student’s two-tailed t-test for paired samples. All statistical analyses are based on at least three separate experiments, performed in triplicate, with PMN and HUVEC from different donors.

RESULTS

HUVEC expression of E-selectin

The first question was whether prednisolone would affect the expression of one of the early adhesion molecules on HUVEC up-regulated by TNF and IL-1, namely E-selectin. There were no signs of E-selectin surface expression, as analysed by flow cytometry, on resting HUVEC (i.e. the fluorescence of anti-E-selectin antibody-labelled unstimulated HUVEC did not differ from either HUVEC labelled with the control antibody or non-labelled HUVEC; data not shown). Incubation of EC with IL-1β or TNF-α for 3 h resulted in a dose-related expression of E-selectin (Fig. 1a,c). The HUVEC responses were homogeneous, in that no subpopulations of less responsive cells were detected. When HUVEC were treated with prednisolone for 30 min prior to and during subsequent incubation with 5 U/ml IL-1β or 10 ng/ml TNF-α, dose-dependent reductions of the expression of E-selectin were noted (Fig. 1b,d).

Fig. 1.

Effect of prednisolone on human umbilical vein endothelial cell (HUVEC) expression of E-selectin, assessed by flow cytometry. HUVEC were activated with IL-1β (a,b) or tumour necrosis factor-alpha (TNF-α) (c,d) for 3 h. Subsequently, dispersed HUVEC were labelled with fluorescent antibody to E-selectin. (a) HUVEC activated with IL-1β or medium (control). (c) HUVEC activated with TNF-α or medium (control). (b,d) HUVEC were treated with prednisolone, at concentrations indicated, for 30 min prior to and during subsequent activation with IL-1β 5 U/ml (b) or TNF-α 10 ng/ml (d). The Figure depicts one representative experiment out of three with similar results.

PMN adhesion to HUVEC

Based on previous experiments, where a MoAb to the PMN binding epitope on E-selectin blocked adhesion of PMN to TNF- or IL-1-stimulated HUVEC (cf. [13]), we asked whether the inhibitory effect of prednisolone on E-selectin expression would translate into reduced PMN adhesion. The spontaneous adhesion of PMN to HUVEC (incubated in medium alone for 180 min) was 3·5 ± 0·4% (mean ± s.e.m., n = 16). When HUVEC were stimulated with IL-1β or TNF-α, the adhesiveness for PMN increased in a time- and dose-related fashion, with a 10·2-fold increase of 5 U/ml for IL-1β and a 9·0-fold increase of 10 ng/ml for TNF-α and 3 h incubation (Fig. 2a).

Fig. 2.

Effects of prednisolone on adhesion of polymorphonuclear neutrophils (PMN) to human umbilical vein endothelial cell (HUVEC) monolayers. (a) HUVEC were activated with IL-1β or tumour necrosis factor-alpha (TNF-α) at indicated concentrations for 3 h. Subsequently, HUVEC were washed three times, then PMN were added and allowed to adhere for 10 min. The results are given as the relative increase of adherent PMN compared with no stimulation, i.e. medium alone. (b) Spontaneous adhesion. HUVEC were treated with prednisolone or medium alone for 3 h and 30 min. PMN were added, and thus exposed to prednisolone (if added) for 10 min. (c) Activated adhesion. HUVEC were treated with prednisolone at indicated concentrations, or medium alone for 30 min prior to and during subsequent activation with IL-1β (5 U/ml) or TNF-α (10 ng/ml) for 3 h. Subsequently, HUVEC were washed three times, prednisolone or medium was added again. Finally, PMN were added and allowed to adhere for 10 min. The results of (b) and (c) are given as the change of adherence when compared with cells not treated with prednisolone. Mean and s.e.m. values for the number of separate experiments given at the bottom of the bars, run in triplicates. *P < 0·05; **P < 0·01; ***P < 0·001 compared with medium-treated controls.

Treatment of the HUVEC system with prednisolone reduced dose-dependently the IL-1β-, as well as TNF-α-induced hyperadhesiveness (Fig. 2c), with significant inhibition occurring at 10 μm. At the maximal tested concentration, 100 μm, prednisolone caused a 27·6% and a 34·5% reduction of adherence induced by IL-1β and TNF-α, respectively. The spontaneous adhesion of PMN to HUVEC was unaffected by prednisolone (Fig. 2b).

Cytokine- and PMN-mediated endothelial cytotoxicity

In order to test whether prednisolone affected the cytotoxicity of PMN for EC, induced by cytokine treatment of HUVEC [13], we used an in vitro model for vasculitides. When quiescent PMN were co-incubated with unactivated HUVEC, a spontaneous cytotoxicity of 1·0 ± 0·1% (n = 24) occurred. This ⁵¹Cr release was not in any significant manner affected by addition of prednisolone in the concentrations used for the assays described below (data not shown).

Endothelial monolayers pretreated with IL-1β or TNF-α, rinsed and then co-incubated with PMN, exhibited a significantly increased cytotoxicity. The optimal concentrations for these cytokines were 10 U/ml and 100 ng/ml, causing 3·8- and 7·9-fold increases in cytotoxicity, respectively, compared with unactivated HUVEC (Table 1). The dose and time response curves for these two cytokines have previously been reported [13]. Based on earlier results HUVEC were exposed to IL-1β for 4 h and to TNF-α for 24 h [13].

Table 1.

The effect of IL-1β and tumour necrosis factor-alpha (TNF-α) on neutrophil (PMN)-dependent endothelial cytotoxicity

Treatment	Concentration	Incubation time	Percent cytotoxicity	n
PMN + HBSS		4 h	1·0 ± 0·1	24
		24 h	1·2 ± 0·5	34
PMN + IL-1β (U/ml)	10	4 h	3·8 ± 0·3***	22
PMN + TNF-α (ng/ml)	100	24 h	9·5 ± 0·6***	34

Human umbilical cord venous endothelial cells were exposed for prednisolone 30 min, washed and then H₂O₂ or neutrophil lysate was added. Mean and s.e.m. values for the number (n) of separate experiments, run in duplicates.

^***P < 0·05 compared with cytotoxicity induced by H ₂O₂ or lysate alone.

First, we assessed whether prednisolone would inhibit the cytokine-induced cytotoxicity in the complete system. Second, we examined whether prednisolone protected EC by incubating the HUVEC monolayers alone with prednisolone for 30 min, followed by washing and addition of IL-1β for 4 h or TNF-α for 24 h to the monolayers; subsequently we rinsed the HUVEC and then co-incubated them with PMN.

Prednisolone significantly reduced IL-1β- and TNF-α-induced cytotoxicity in the complete system in a dose-dependent way (Fig. 3a). When HUVEC were incubated with the maximal prednisolone concentration used here, 50 μm, cytotoxicity induced by IL-1β or TNF-α was reduced by 88% and 73%, respectively. When HUVEC alone were preincubated with prednisolone, the effect of the two cytokines was reduced in a similar manner to that observed in the complete system (Fig. 3b). Thus, prednisolone protects HUVEC from neutrophil-dependent cytotoxicity induced by IL-1β and TNF-α.

Fig. 3.

Effect of prednisolone on cytokine-induced cytotoxicity. (a) Human umbilical vein endothelial cells (HUVEC) were exposed to either 10 U/ml of IL-1β (□) for 4 h or to 100 ng/ml of tumour necrosis factor-alpha (TNF-α; hatched bar) for 24 h, rinsed, and then unstimulated polymorphonuclear neutrophils (PMN) together with prednisolone were added (the complete system). (b) HUVEC were treated for 30 min with prednisolone, rinsed and then exposed to IL-1β (□) for 4 h or to TNF-α (hatched bar) for 24 h; subsequently unstimulated PMN were added. Mean and s.e.m. values for five experiments, run in duplicates. *P < 0·05; **P < 0·01; ***P < 0·001 compared with controls, i.e. cytotoxicity induced by IL-1β or TNF-α without prednisolone.

Release of intracellular Ca^{2 +} in PMN induced by activated HUVEC

Based on our previous findings that TNF-α-treated HUVEC confer a transient rise of [Ca^{2 +}]_i in Fura-2-loaded neutrophils [13], thereby indicating that PMN may recognize an activating molecule on cytokine-activated EC, we studied whether prednisolone interfered with such TNF-α-induced activation of neutrophils. HUVEC monolayers were treated with prednisolone for 30 min, washed, treated for 2 h with TNF-α, and then added to the PMN. However, prednisolone (50 μm) was unable to affect this TNF-α effect in any significant way (Fig. 4).

Fig. 4.

Effect of prednisolone on polymorphonuclear neutrophils (PMN) [Ca^{2 +}]_i, as reflected by changes of Fura-2 fluorescence. Human umbilical vein endothelial cells (HUVEC), treated with 100 ng/ml of tumour necrosis factor-alpha (TNF-α) for 2 h with or without pretreatment by prednisolone 50 µm for 30 min, were added (arrows) to PMN loaded with Fura-2. The figure depicts one representative tracing out of three separate experiments run in duplicate with similar results.

TNF-α-induced endothelial production of IL-8

We have previously shown that IL-8, a potent neutrophil agonist, causing up-regulation of membrane adhesion receptors in neutrophils and release of oxygen radicals [35,36], is produced in high levels by TNF-α-activated HUVEC [13,17]. These experiments suggested that this rise in IL-8 was involved in the stimulus response coupling of cytokine-driven PMN cytotoxicity, since a MoAb against IL-8 blocked the rise of [Ca^{2 +}]_i in the PMN induced by TNF-α-treated HUVEC [13]. In order to examine whether the inhibitory effect of prednisolone on cytokine-induced PMN cytotoxicity might relate to a modulation of IL-8 production, we assessed if prednisolone interfered with TNF-α-induced HUVEC production of IL-8 in vitro.

IL-8 was produced at high levels by TNF-α-activated HUVEC (45 ± 14 pg/ml from quiescent HUVEC versus 877 ± 192 pg/ml after 100 ng/ml of TNF-α for 4 h) (cf. [13]). Prednisolone (50 μm for 30 min prior to stimulation with TNF-α) did not reduce the levels of IL-8 produced by TNF-α-treated HUVEC (736 ± 151 pg/ml; n = 5–6).

Endothelial cytotoxicity mediated by H₂O₂ or PMN lysates

Based on the finding that prednisolone protected HUVEC from cytotoxic PMN, we asked whether prednisolone interfered with the two major cytotoxic systems of PMN: the release of oxygen radicals (e.g. H₂O₂) or the release of granule constituents (e.g. elastase, collagenase or defensins). Hence, we added such substances to unstimulated HUVEC. Based on previous studies we chose to use H₂O₂ at a concentration of 50 μm and PMN lysates as previously described [7]. These additions conferred a 8·0 ± 1·3% and 6·1 ± 2·2% cytotoxicity, respectively. When HUVEC were pretreated with prednisolone for 30 min and subsequently washed before addition of H₂O₂ and PMN lysates, cytotoxicity was significantly reduced (Table 2).

Table 2.

The effect of prednisolone on H₂O₂ and polymorphonuclear neutrophil (PMN) lysate-induced endothelial cytotoxicity

	Cytotoxicity, % of control	n
H2O2 50 μm	100	4
Lysate (1·25 × 106 PMN)	100	4
H2O2+ prednisolone 10 μm	35·8 ± 4·9*	4
Lysate + prednisolone 10 μm	58·9 ± 6·1*	4

Human umbilical cord venous endothelial cells were exposed for prednisolone 30 min, washed and then H₂O₂ or neutrophil lysate was added. Mean and s.e.m. values for the number (n) of separate experiments,run in duplicates.

^*P <0·05 compared with cytotoxicity induced by H₂O₂ or lysate alone.

Cytotoxicity of HUVEC induced by the NO donor SNAP

We then asked if prednisolone might affect one major mediator of HUVEC injury, identified in previous experiments with the current assay system, namely NO [13]. To address this question, we assessed whether prednisolone interfered with NO-induced cytotoxicity by adding the NO donor SNAP to HUVEC, in order to achieve NO generation, in the presence or absence of prednisolone. SNAP (0·2 nm) produced 750 ± 90 pmol NO per minute (mean ± s.e.m.; n = 3) (data not shown). The dose–response curve for SNAP has previously been reported [8]. In the absence of prednisolone, SNAP induced a significant cytotoxicity (4·1 ± 0·3%). Incubation of HUVEC with prednisolone (at 50 μm for 30 min, followed by washing before the addition of SNAP) significantly reduced cytotoxicity (to 2·2 ± 0·4%) (data not shown). Thus, cytotoxicity of HUVEC conferred by the NO donor SNAP was partly inhibited when HUVEC were treated with prednisolone.

DISCUSSION

We have recently shown that anti-rheumatic gold compounds are able to reduce IL-1β-induced endothelial expression of E-selectin and intercellular adhesion molecule-1 (ICAM-1), as well as reducing endothelial hyperadhesiveness for neutrophils, in vitro[28]. Here, prednisolone conferred a significant reduction of E-selectin expression induced by IL-1β or TNF-α. This is in line with the finding that another glucocorticoid, dexamethasone, impairs EC expression of E-selectin induced by endotoxin [37]. Moreover, we found that our findings translated into an inhibitory effect on the function of E-selectin, in that treatment of EC with prednisolone reduced endothelial hyperadhesiveness for neutrophils induced by IL-1β or TNF-α. We then assumed that another E-selectin-dependent function between EC and PMN, the cytotoxic process [13], might be affected.

We have recently demonstrated that IL-1β and TNF-α are powerful inducers of PMN-dependent cytotoxicity for HUVEC in vitro[13]. HUVEC cytotoxicity induced by these cytokines is associated with NO produced by HUVEC and/or PMN and requires divalent cations [13], but is not directly related to the release of oxygen radicals or granule constituents (cf. [7]). Moreover, TNF-α-treated HUVEC confer a transient rise of [Ca^{2 +}]_i in previously quiescent PMN, possibly activating PMN via a surface receptor and phospholipase C-mediated event [13]. Finally, TNF-α induced increased production of the chemokine IL-8 from HUVEC [13], a powerful activator of neutrophils and up-regulator of membrane adhesion receptors in neutrophils [35,36]. Against this background it appeared conceivable that prednisolone might interact with one or several steps of these reactions.

In this study, stimulation of HUVEC with IL-1β or TNF-α induced a consistent and significant increase in PMN-mediated cytotoxicity. The cytokine-mediated cell injury was highly reproducible and consistent, with very small day-to-day variations. Although endothelial cell injury comprised less than 10% of the HUVEC in the well, it may be biologically significant. Moreover, the maximal cytokine-induced endothelial cytotoxicity was three-to-four-fold higher than that observed in previous studies with PMN stimulated by fMLP and LXA₄[7].

We found in this in vitro model that prednisolone inhibited the effect of IL-1β and TNF-α as potent promoters of cytokine-mediated neutrophil-dependent cytotoxicity for HUVEC by effects on the effector cells, PMN, as well as on the target cells, HUVEC. The mechanisms of action were partially analysed and indicated that prednisolone protected HUVEC from three important mechanisms of PMN-induced cytotoxicity, the first two being the release of H₂O₂ or granule constituents (cf. [7]), and the third the release of NO [8,13].

We know from earlier studies that cytokine-induced PMN-dependent cytotoxicity is reduced by a scavenger of NO (HbO₂), as well as by two specific inhibitors of the NO-synthase (L-NMMA or L-NAME) and augmented by the specific substrate for the NO-synthase, l-arginine [13]. Against this background we analysed the effect of prednisolone on endothelial cytotoxicity induced by the NO donor, SNAP. The relative reduction of SNAP-induced cytotoxicity conferred by prednisolone was almost 50%. A 50% reduction of the endothelial injury might be highly relevant in an in vivo setting of an active vasculitis. Increased HUVEC resistance towards NO may be one of the mechanisms responsible for the reduction of cytokine-induced endothelial cytotoxicity conferred by prednisolone.

In our system, prednisolone did not interfere with the PMN-activating properties of TNF-α-treated HUVEC, measured as the release of Ca^{2 +} in PMN. Thus, the inhibitory effect of prednisolone on cytokine-induced cytotoxicity appears not to include interference with Ca^{2 +}-dependent endothelial–PMN cell-to-cell signalling. Moreover, prednisolone did not affect HUVEC production of the PMN-activating cytokine IL-8, indicating that prednisolone does not interfere with IL-8 gene activation systems in HUVEC in response to TNF-α.

For many years prednisolone has been suggested to stabilize cell membranes [38,39]. The observations presented here that endothelial cells resisted the cytotoxic effects of H₂O₂, PMN lysates and NO might be taken as a proof of this time-honoured idea. It might also be that prednisolone confers resistance against putative apoptosis-promoting signals from H₂O₂ and lysates. This issue is presently under investigation.

Thus, prednisolone inhibits neutrophil-dependent endothelial cell injury in a dose-dependent manner, by effects on PMN as well as on HUVEC. This effect of prednisolone is complex, related to interference with cytokine-induced adhesion between PMN and HUVEC, reduction of endothelial expression of E-selectin and an inhibitory effect on HUVEC cytotoxicity induced by NO. The concentrations of prednisolone used in this study are comparable to those of methylprednisolone obtained during intravenous pulse steroid treatment [40]. Intravenous high-dose steroid treatment is a frequently used treatment modality for inflammatory disorders such as systemic vasculitis, and the suppressive effects of prednisolone noted in our in vitro system may explain some of the beneficial effects of intravenous pulse steroid treatment in vivo.