Differential inhibition of polymorphonuclear leukocyte recruitment in vivo by dextran sulphate and fucoidan

The selectin-mediated rolling of leukocytes along the endothelial cells is a prerequisite step followed by firm adhesion and extravasation into the inflamed tissue. This initial contact can be suppressed by sulphated polysaccharides. We have studied the effect of sulphated polysaccharides on the ultimate polymorphonuclear leukocyte (PMN) recruitment and plasma leakage in rabbit skin in response to intradermal injection of various inflammatory mediators. PMN infiltration evoked by various PMN chemoattractants (FMLP, C5a desArg, LTB4 and IL-8) was significantly inhibited after intravenous injection of dextran sulphate (25 mg/kg), heparin (2 × 90 mg/kg) or fucoidan (1 mg/kg). PMN-dependent plasma leakage was equally well reduced by the different sulphated polymers. Vascular permeability induced by histamine or thrombin acting via a PMN-independent mechanism was not reduced. Fucoidan was the only polysaccharide able to suppress IL-1-induced PMN infiltration for 60–70%. Local administration of dextran sulphate had no effect on PMN-dependent plasma leakage. Differential inhibition of PMN recruitment was determined after injection of dextran sulphate or fucoidan depending on the type of insult. Therefore, these results suggest that different adhesion pathways are utilized during PMN recruitment in vivo in response to chemoattractants and IL-1.


Introduction
An inflammatory reaction (e.g. adult respiratory distress syndrome, ischaemic reperfusion injury or arthritis) is characterized by leukocyte emigration and plasma protein leakage from the blood stream into the extravascular injured tissue. Leukocyte recruitment is designed as a multistep process in which several types of cell adhesion molecules are involved. 1 '2 The initial contact of leukocytes with postcapillary endothelial cells, described as rolling, is mediated by the selectin family of adhesion molecules. 3 '4 The subsequent firm adhesion and transmigration through the endothelial cell lining into the inflamed tissue is mediated by leukocyte fi2-integrins (CD 11/CD 18) and endothelial immunoglobulin-like molecules (ICAM-1 and PECAM-1). [5][6][7] The selectin family consists of three closely related cell surface molecules: L-selectin (CD62L, MEL-14, LAM-1), P-selectin (CD62P, GMP-140) and E-selectin (CD62E, EIAM-1). Lselectin, originally described as a lymphocyte homing receptor (MEL-14) 8 is constitutively present on PMNs, monocytes and most of the lymphocytes. PMN activation results in a rapid removal of L-selectin from the cell surface 9 (shedding). P-selectin normally exists preformed in a-granules of platelets and in Weibel-Palade bodies of endothelial cells. After stimulation with histamine, thrombin or oxygen radicals P-selectin is mobilized within minutes on the endothelial cell surface. 1 E-selectin is expressed on endothelial cells via de novo mRNA and protein synthesis after stimulation with various inflammatory agents, i.e. LPS, IL-1 and TNE 11 In vitro and in vivo studies have shown that leukocyte L-selectin as well as endothelial Pand E-selectin are involved in leukocyte rolling supporting the final leukocyte recruitment.9,12-15 A common characteristic for the three selec-Material and Methods tins is their weak binding via their extracellular Animals Ca2+-dependent (C-type) lectin domain to sialylated, fucosylated carbohydrates, like the tetra-Male New Zealand White rabbits (2.5-3.5 kg saccharide sialyl Lewis X (SLeX). 6 Nevertheless body weight)were purchased from Penet Farm, high affinity glycoprotein ligands for all three Moorsel, Belgium. selectins have been described. 4 This suggests that in vivo each selectin preferentially inter-Reagents acts with its specific ligand(s). A mucin-like Pselectin glycoprotein ligand (PSGL-1) and a Calcitonin gene-related peptide (CGRP), cytofibroblas growth factor-related E-selectin ligand chalasin B (Cyt B), Evans blue dye, FMLP and 2-(ESL-1) are described on endothelial cells, lv '8 mercapto pyridine-N-oxide (MERe) were pur-Various ligands for L-selectin are described, chased from Sigma, Chemical Co. (St Louis, MO, depending on the type of endothelial cells: the USA). Thrombin (Thr)(Topostasine(R)) was from sulphated glycoproteins Sgp50 (GlyCAM-1) and Hoffmann-La Roche (Basel, Switzerland).
Sgp90 (identical to the sialomucin CD34) 19 15 min) injected In-labelled PMNs and 125I-human serum albumin (5/,Ci mixed with 2 ml of a 2.5% Evans blue dye solution to visualize sites of plasma protein leakage). Inflammatory mediators were injected intradermally (i.d.) in 0.1 ml volumes (diluted in sterile pyrogen-free saline), each agent having six replicate injection sites per animal (according to a balanced site pattern).
In all in vivo experiments 't-----0' refers to the i.d. injection of various inflammatory mediators plus CGRP. IL-1 acting via a protein synthesisdependent mechanism, was injected 3.5 h before the other inflammatory agents and before CGRP, in order to get a maximal response. 28 After measuring PMN accumulation and plasma leakage over a period of 1 h, a blood sample was taken by cardiac puncture. Rabbits were killed by an overdose of sodium pentobarbital and the injection sites were excised with a 17 mm diameter punch. The skin samples were counted in a gamma-counter with automatic spill-over correction (Cobra 5005, Packard). PMN infiltration is expressed as the number of 1111n-PMNs/site determined by comparing 11 lin. counts per site with Ill In-counts per PMN after labelling. Plasma extravasation is expressed in terms of equivalents of tl plasma by dividing skin sample 125I-counts by 25I-counts in 1/,1 plasma. Calculation of the percentage of circulating radiolabelled PMNs at the end of the experiment was based on total cell-associated activity injected and total blood volume of the recipient rabbit (7.4% expressed as percentage of the body weight). In comparison to dextran sulphate a higher In control-treated animals (dextran T500, dose of heparin was needed to inhibit leukocyte 25 mg/kg), PMN accumulation and plasma leakrolling. 34 The dose of heparin (90 mg/kg), age dose-dependently increased after local injec-which was required to inhibit leukocyte rolling tion of the PMN chemoattractants FMLR C5a for >90%, was not sufficient to significantly desArg, LTB4 and IL-8 (Fig. 1). PMN infiltration suppress PMN infiltration in the rabbit skin induced by IL-1 was not associated with detect-measured over a period of 1 h (data not shown). able oedema formation. Histamine and throm-Therefore two bolus injections of heparin bin evoked an increase in microvascular plasma (90 mg/kg) were injected i.v. (15 rain before leakage without a significant increase in PMN and 15 min after i.d. injection of the inflammaaccumulation. Intravenous injection of dextran tory agents). As shown in Fig. 3, this dose of sulphate (25 mg/kg, 30 rain before the start of heparin resulted in a significant but incomplete the experiment), completely abolished PMN inhibition of PMN infiltration and plasma leakinfiltration and plasma extravasation in response age induced by the PMN chemoattractants. to the PMN chemoattractants. This concentra-PMN recruitment in response to IL-1 was not tion of dextran sulphate had been shown to significantly suppressed. Again plasma leakage inhibit leukocyte rolling in rabbit mesenteric induced by histamine and thrombin was not venules for more than 90%. 24 In contrast, PMN affected. accumulation in response to IL-1 was not The ligand activity of fucoidan for L-selectin suppressed by dextran sulphate and it did not was described to be at least 100-fold more affect the PMN-independent increase in vascular potent per unit weight than heparin. 35 Therepermeability induced by histamine and thromfore, 1 mg/kg fucoidan was injected 15 rain bin. The inhibition of PMN extravasation by before the start of the experiment. This dose of dextran sulphate was not due to an effect on fucoidan significantly inhibited PMN infiltration the circulating number of In-labelled PMNs for 60-70% (after background substraction)and measured at the end of the experiments (dex-oedema formation for 50-60% in response to tran sulphate" 51 4-3% vs dextran T500: the chemoattractants (Fig. 4). In contrast to the 49 + 3%). results obtained after administration of dextran IL-1 was injected i.d. 3.5 h before the other sulphate or heparin, fucoidan also reduced PMN inflammatory mediators and before CGRP recruitment after IL-1 injection. Fucoidan did (10-11 mol/site), but also 3 h before the intra-not affect the percentage of circulating Invenous injection of dextran sulphate. To ex-PMNs as measured at the end of the experiment clude the possibility that a time-related (fucoidan" 48 -+-8% vs control: 54 -+-8%).
phenomenon could be responsible for the lack Protamine sulphate (i.v., 25 mg/kg), also of effect of dextran sulphate on the IL-l-induced shown to inhibit leukocyte rolling in mesenteric PMN infiltration, dextran sulphate was also postcapillary venules, 4 could not be admininjected just before the i.  into account that the blood volume in the rabbit is about 7.4% of its body weight. 31 After recalcification (Ca2+/Mg2+, 0.5 mM) and addition of Cyt B (10 g/mD, PMNs were stimulated with FMLP (10 -6 M). As shown in Fig. 6, PMN aggregation in the presence of dextran T500 (300 #g/ml) was not altered as compared to control aggregation. Although the maximal aggregation was not altered in the presence of dextran sulphate, the aggregation was delayed. Similar results were obtained with heparin (data not shown). In contrast, fucoidan (20 g/mD had no effect on PMN aggregation. Under the same experimental conditions, anti-CD18 sup-pressed PMN aggregation by more than 70% (R 15.7 10/Jg/ml, vs mouse IgG).

Discussion
Rolling of leukocytes along endothelial cells is mediated by various selectins: L-selectin on leukocytes and Pand E-selectin on activated endothelial cells. All   Since the increase in vascular permeability in response to histamine and thrombin was not affected by dextran sulphate, we can exclude possible dextran sulphate-induced changes in local blood flow. The complete inhibition of PMN infiltration by dextran sulphate was comparable with the more than 90% reduction of leukocyte rolling in rabbit mesentery venules, obtained by Ley et al. 24 The presence of sulphate on glycans is crucial since other negatively charged or neutral po!ysaccharides had no effect on leukocyte rolling. 4 The possible involvement of sulphated glycoconjugates during leukocyte-endothelial cell interaction is demonstrated at different levels.
The interaction of sulphated and phosphated polysaccharides (i.e. dextran sulphate, fucoidan and PPME (a yeast polysaccharide of mannose and mannose-6-phosphate) with L-selectin has been demonstrated in vitro and suggested in ViVO. [37][38][39] Rapid sulphate uptake by high endothelial cells in the venous microvasculature is correlated with increased lymphocyte accumulation, most likely mediated by the homing receptor L-selectin. 8'4 An internal pool of heparin-like chains, recognizing L-selectin was found in cultured venular endothelial cells. 22 The sulphatation of the endothelial cell ligand Gly-CAM-1 was demonstrated to be essential for 41 the binding to L-selectin. The sulphated form FIG. 6. Effect of dextran sulphate (300/g/ml, ,) or dextran T500 (300/g/ml, A) and fucoidan (20/g/ml, T, on PMN aggregation, measured as percentage change in light transmission vs control ((C)). Anti-CD18 (R15.7, 10/g/ml) (0) was added to investigate CD18-dependent aggregation.
of SLe x (6'-sulphated SLeX), capping the Gly-CAM-1 structure, binds to L-selectin with higher affinity than SLeX. 42 Tyrosine sulphatation of the P-selectin glycoprotein ligand-1 (PSGL-I) structure is in addition to the appearance of fucosylated, sialylated glycans important for the binding of P-selectin but of no importance for 43 the recognition of E-selectin.
Inhibition of PMN infiltration after administration of heparin was similar to the effect of dextran sulphate, although less pronounced. The lower potency of heparin to reduce PMN extravasation is consistent with its lower potency to block leukocyte rolling. 4 Administration of high doses of heparin as antiinflammatory agent may be irrelevant, because of its strong anticoagulant activity. Recently, heparin oligosaccharides without significant anticoagulant activity have been demonstrated to inhibit Pand L-selectin-dependent PMN infiltration in a murine model of thioglycollate-induced peritoneal inflammation. s In contrast to dextran sulphate, fucoidan (1 mg/kg) not only suppressed PMN and plasma protein accumulation in response to PMN chemoattractants but also attenuated IL-l-induced PMN recruitment for about 60-70%. It is important to know that fucoidan had no effect on the number of circulating PMNs. Similar results were obtained by Lindbom et al. in rat mesenteric microvessels using intravital microscopy. 44 They achieved a 70% inhibition of leukocyte adhesion, via a rolling-dependent, CD11/CD18-independent mechanism after administration of 1 mg/kg of fucoidan. Fucoidan has been demonstrated to attenuate leukocyteand plasma protein extravasation into cerebrospinal fluid of rabbits challenged with pneumococcal antigen. 26 Our in vivo results have shown a different inhibitory pattern of fucoidan and dextran sulphate for IL-l-induced PMN accumulation.
The exact mechanism is until now not known. Different adhesion pathways or different binding sites may be the reason for a complete inhibitory effect by fucoidan and only a partial reduction after administration of dextran sul-dextran sulphate at the PMN surface or due to phate. IL-l-activated venular endothelium may an interference on the lectin-dependent PMN express a ligand for L-selectin with binding aggregation. characteristics different from that expressed All three selectins are possible candidates to after administration of PMN chemoattractants, interact with the sulphated polysaccharides. This ligand may be E-selectin, as suggested by Differential involvement of Pand E-selectin in Kishimoto et al., 45 or a new until now not well addition to L-selectin may be important in PMN defined, inducible ligand as demonstrated by recruitment in rabbit skin after injection of Spertini et al. 46 Differences in binding to L-PMN chemoattractants or IL-1. E-selectin medselectin have been described for PPME and iates leukocyte rolling in IL-l-treated rabbit fucoidan. 7 Both PPME and fucoidan are able to mesentery venules. 15 A cytokine-inducible liblock L-selectin-mediated murine lymphocyte gand for L-selectin is described on nonlymphoid and PMN adhesion to lymph node high en-cells. 22 P-selectin is translocated within minutes dothelial venules (HEV) in vitro, but only to the cell surface after stimulation with thromfucoidan is capable to inhibit L-selectinbin, histamine or oxygen radicals. 1 Recently a mediated leukocyte rolling in venules of rat cytokine upregulation of P-selectin with a time mesentery. 25 Nevertheless identical binding course similar to that of E-selectin expression characteristics to Pand L-selectin were dewas also found. 51'52 scribed for fucoidan and dextran sulphate, Participation of P-selectin in PMN infiltration 47 48 whereas no binding was found to E-selectin.
in response to the different inflammatory media-In contrast to identical binding characteristics, tots in this skin model is still a question mark. functional differences between fucoidan and The sulphated polysaccharides, like dextran dextran sulphate were found after in vitro sulphate, heparin and fucoidan are shown to incubation of PMNs followed by stimulation, disrupt P-selectin-mediated PMN interaction in The rate and extent of formyl-peptide-induced vitro. 3,53 Until now, there is no direct evidence L-selectin shedding is consistently reduced after to suggest that P-selectin is expressed on treatment of PMNs with fucoidan as compared endothelial cells after stimulation with pure with dextran sulphate-treated PMNs. 49 PMN chemoattractants such as FMLP or IL-8. It is unlikely that the sulphated polysacchar-Under those conditions histamine is not reides are acting on the level of the PMN /2leased as secondary mediator in response to ILintegrins (CD11/CD18). As demonstrated by Ley 1 or PMN chemoattractants. 54 In our model the et al., dextran sulphate up to concentrations of sulphated polysaccharides had no effect on 1 mg/ml did not alter CD1 lb/CD18 expreshistamine-or thrombin-related increase in mision. 3 Furthermore, both fucoidan and dextran crovascular permeability. Nevertheless, neurosulphate failed to inhibit CD1 l b/CD18-depenpeptides such as calcitonin gene-related peptide dent PMN adhesion under stationary conditions (CGRP), locally injected in normal human skin, in vitro. 33 '44 In addition, selectin-mediated rolltranslocate P-selectin within 15 min to the ing of leukocyte is blocked by sulphated polyluminal membrane. 55 Since in our model CGRP saccharides, whereas anti-CD18 antibodies had is co-injected to enhance the local blood flow, no effect. 1'5 In our model dextran sulphate did the role of P-selectin during PMN infiltration not inhibit IL-l-induced PMN accumulation, can therefore not be excluded. whereas monoclonal antibodies against CD18 The relative role of the selectins may depend reduced PMN accumulation in response to IL-1 on the particular mediators and on the inflamin this rabbit skin model. A non-specific in-matory model being studied. The selectincrease in charge density on the PMN-surface by mediated adhesion pathways of leukocytes on dextran sulphate is unlikely to be the cause of endothelial cells are partially redundant, parthe inhibition, because a similar inhibitory tially separate. 56'57 The combined involvement effect of dextran sulphate would then be of different selectins was demonstrated in vivo expected after IL-1 injection. CD18-dependent by Mulligan in two models of lung inflammation PMN aggregation induced by FMLP was delayed using selectin antibodies. Acute lung injury in the presence of dextran sulphate, however induced by Cobra Venom Factor, measured after not suppressed. In contrast, fucoidan had no 30 min, was mediated by both Land P-selectin, attenuating effect on PMN aggregation. Our whereas IgG immune complex-induced injury, results are in agreement with the observations measured after a period of 4 h was Land Eof Bazzoni et al., who found a reduced PMN selectin-dependent. 58 Recent studies on animals aggregation in the presence of heparin meagenetically deficient in these selectin have shed sured at 3 min after stimulation3 The delay in light on the temporal sequence of selectin aggregation may be due to sterical hindrance of functions, as well as on their individual con-tributions in acute inflammatory models. 59 Early after injury leukocyte rolling was defective and PMN emigration was delayed in P-selectin knock-out mice, whereas later on L-selectinmediated rolling appeared to be more prominent. 56'6 Discrimination between the different selectins can also be made on the level of oedema formation. In a model of contacthypersensitivity plasma leakage was increased via a L-selectin-dependent mechanism with only a minor role for P-selectin. 61'62 Which adhesion pathways are involved during PMN recruitment evoked by the chemoattractants and by IL-1 in the rabbit skin is at present not clear. Different infiltration pathways in response to IL-1 or to the chemoattractant FMLP were also demonstrated by Wakelin et al. 63 With mAb against PECAM-1 IL-l-induced PMN extravasation across rat mesenteric microvessels could be suppressed, whereas anti-PECAM-1 had no effect on FMLP-evoked PMN infiltration. There is, however, no evidence that dextran sulphate or fucoidan can interfere with the PECAM-1 adhesion pathway.
The exact mechanism behind the inhibition of PMN infiltration and oedema formation by the sulphated polysaccharides is not clear. It cannot be excluded that dextran sulphate and heparin might interfere with PMN-derived polycations, such as elastase, cathepsin G or defensins to reduce PMN recruitment. 64'65 However, the direct inhibition of elastase by dextran sulphate is unlikely, since LTB4-induced PMNdependent increase of plasma leakage in the hamster cheek pouches was significantly suppressed by dextran sulphate, whereas administration of elastase inhibitors had no effect. 65 The role of other naturally occurring cationic peptides, derived from PMNs, dnring the induction of plasma leakage cannot, however, be excluded.
Nevertheless, since the recent selectin-carbohydrate binding studies, there is increasing and attractive evidence to suggest that the antiinfiltrating properties of these sulphated polysaccharides are attributed to their interference with the selectins. 25 '38 Although the different selectin adhesion molecules can commonly bind to SLe x and related compounds, the knowledge of the specific binding characteristics of the carbohydrate structure with each selectin, can contribute to a more efficient anti-adhesion therapy.
We have demonstrated that sulphated glycans may be involved during PMN infiltration and plasma leakage in the rabbit skin unrelated to the CD1 l b/CD18 adhesion pathway. Since dextran sulphate and fucoidan differently inhibit PMN infiltration in vivo, depending on the type of inflammatory stimulus, different sulphated glycans may be involved during PMN recruitment.