Synthetic Colloids Attenuate Leukocyte-Endothelial Interactions by Inhibition of Integrin Function

Human cells were derived from human umbilical veins or from blood donations of healthy volunteers after obtaining institutional and ethical committee approval (Ethical Committee of the Faulty of Medicine, University of Tuebingen, Tuebingen, Gemany) and informed consent as previously described.21,22,39,40Human umbilical venous endothelial cells (HUVECs) were harvested from umbilical cords by collagenase treatment (0.1% collagenase A; Boehringer, Mannheim, Germany) and cultured in endothelial cell growth medium (PromoCell, Heidelberg, Germany) on collagen-coated cell culture dishes (Falcon Biocoat; Becton Dickinson Labware, Bedford, MA) or rectangular coverslips (Kindler, Freiburg, Germany) coated with rat tail collagen I (Falcon Biocoat). Confluent HUVECs of the first and second passage were used for the experiments. Polymorphonuclear neutrophils (PMNs) were isolated from freshly drawn citrated blood by density gradient centrifugation at 1,700 rpm on a discontinuous Percoll gradient with 63% and 72% Percoll in phosphate buffered saline (1.130 g/ml Percoll; Amersham Pharmacia Biotech, Uppsala, Sweden). The bottom layer, containing the PMNs, was collected, and contaminating erythrocytes were removed by hypotonic lysis in 10% NH⁴Cl on ice. After a final wash with phosphate-buffered saline (Sigma, St. Louis, MO), the PMN pellet was resuspended in cold Medium 199 (Sigma) at 5 × 10⁷/ml until immediately before the flow experiment. As determined by trypan blue exclusion, staining for vWF and cell surface antigen analysis both separation protocols yielded functionally intact, nonactivated HUVECs and PMNs at greater than 90% purity.22,39,40 

To address the effects of various colloids on endothelial cell adhesion molecule expression, HUVECs were incubated with 5 mg/ml of HES 450 (molecular weight in kilodaltons), HES 200, HES 130, HES 70, dextran 70, dextran 40, or gelatin dissolved in medium. This concentration resembles plasma concentrations during the first 6 h after bolus infusion of 500–1,000 ml of the tested volume expanders.41–43HES 450 (degree of substitution 0.7), HES 200 (degree of substitution 0.5), HES 130 (degree of substitution 0.4), and dextran 70 were purchased from Fresenius Kabi (Bad Homburg, Germany). HES 70 (degree of substitution 0.5) was obtained from Baxter (Erlangen, Germany), dextran 40 was obtained from Delta-Pharma (Pfullingen, Germany), and urea-linked gelatin was obtained from Aventis Pharma (Bad Soden, Germany). Two hours after exposure to colloids, HUVECs were activated with tumor necrosis factor α (TNF-α; Sigma) at 0.5 ng/ml for 4 h or were left nonactivated so that a total colloid incubation of 6 h was achieved in all experiments (table 1). Because activation with TNF-α does not result in sustained up-regulation of P-selectin on human endothelium,44we examined its expression on thrombin-activated HUVECs. Therefore, in separate experiments, HUVECs were activated with 2 units/ml thrombin (Sigma) during the last 2 min of the 6-h colloid incubation. Within each activation, the colloid-pretreated HUVECs were compared with untreated controls without colloids (table 1). At the end of each incubation period, the cells were washed and stained with saturating amounts of fluorochrome-conjugated monoclonal antibodies against vascular cell adhesion molecule 1 (VCAM-1), E-selectin (both fluorescein conjugates; Southern Biotechnologies Associates, Birmingham, AL), ICAM-1 (phycoerythrin-conjugate; Immunotech, Marseille, France), and P-selectin (primary antibody immunoglobulin [Ig] G1, mouse anti-human and secondary antibody Fab²IgG, goat anti-mouse, phycoerythrin-conjugate; Immunotech). After antibody staining, the HUVECs were trypsinized and immediately analyzed for adhesion molecule expression in a flow cytometer (FACSort; Becton Dickinson, San Jose, CA).

To determine the mechanisms by which certain synthetic colloids may inhibit leukocyte-endothelial interactions, we examined the effects of HES 200, HES 130, dextran 70, dextran 40, and gelatin on PMN adhesion under a postcapillary shear stress of 2 dyn/cm². Adhesion of PMNs to untreated, cytokine-activated HUVECs and colloid-pretreated, cytokine-activated HUVECs were compared. To examine immediate effects on bond formation between the interacting cells, a third experimental group consisting of colloid-pretreated HUVECs and cotreated PMNs was introduced. In this group, the PMNs were resuspended in a colloid-supplemented medium immediately before the adhesion assay (table 1). To guarantee constant PMN concentrations (10⁶cells/ml in Medium 199) in all experiments, dry substances of the colloids were dissolved in medium instead of using colloid solutions. Cytokine activation and colloid pretreatment were identical to the protocol described above, with one modification. Because we also wanted to address early effects of colloids on bond formation between leukocytes and the endothelium rather than endothelial pretreatment alone, we increased the colloid concentration to 10 mg/ml in all functional assays to account for the higher initial plasma concentrations of volume expanders during infusion.41–43PMN adhesion was quantified in a parallel plate flow chamber as previously reported.21,22,39,40In brief, HUVEC-covered coverslips were placed into a heatable flow chamber, rinsed with Medium 199, and perfused at 37°C with the PMN suspension for 10 min at a shear stress of 2 dyn/cm². The flow rate Q  in milliliters per second necessary to produce the desired shear stress τ^w  could be calculated according to the formula Q = (h  ²w τ^w)/6μ, in which h  is the chamber height (0.015 cm), w  is the width (1.25 cm), and μ is the viscosity of the perfusate (0.007 poise for isolated PMNs in Medium 199, 0.007–0.009 poise for colloid-supplemented medium) as measured in a cone and plate viscometer (DV-III+; Brookfield Engineering Laboratories, Middleboro, MA). Using phase-contrast microscopy (20× objective, DMIRB; Leica, Bensheim, Germany), PMN adhesion was determined from digitized 10-s video recordings of five different fields of view at the end of each 10-min perfusion period. According to Lawrence et al. ,19,23PMNs were defined as rolling when traveling below 40 μm/s. A PMN moving less than one cell diameter in 10 s was defined to be firmly adherent. As a measure for adhesion efficiency,20,27the rolling fraction was calculated as [(No. of rolling cells) × 100]/(No. of rolling cells + No. of firmly adherent cells). Using this equation, it could be distinguished whether inhibition of PMN adhesion occurred by inhibition of selectin-mediated rolling interactions or inhibition of integrin-mediated firm adhesion to endothelial ICAM-1.27,28The mean rolling velocities were determined from more than 25 individual PMN-velocity profiles for each experimental condition as derived from a customized software for image recognition (CellTracker; Christof Zanke, University of Tuebingen, Tuebingen, Germany).

The medians of fluorescence intensity (MFIs) were calculated from 5,000 single events detected by the flow cytometer for each sample of five experiments. For statistical analysis, an analysis of covariance (ANCOVA) was performed to determine whether adhesion molecule expression was influenced by the cell culture (modeled as random effect), TNF-α activation, colloid pretreatment (continuous factors), or whether an interaction between TNF-α–activation and colloid pretreatment occurred. To analyze effects of individual colloids, ratios between colloid-pretreated samples and untreated controls (colloid/control) were calculated for each of the various colloid preparations. Significant colloid effects were assumed for those ratios whose 95% confidence intervals did not include 1. Firm adhesion and rolling fractions were analyzed from the mean of five randomly chosen fields of view of six experiments. Effects of colloid treatment (either pretreatment of HUVECs with and without cotreatment of PMNs) were compared by ANCOVA and post hoc  paired t  tests using the Bonferroni-Holm correction for multiple comparisons. For presentation, ratios (colloid/TNF-α–activated control) were calculated, which are presented as means and their corresponding 95% confidence intervals. All analyses were performed using the statistical software package JMP (SAS Institute Inc., Cary, NC).

Nonactivated HUVECs expressed small constitutive amounts of ICAM-1 but did not show expression of endothelial selectins or VCAM-1. After activation with TNF-α, a 10-fold up-regulation of E-selectin, ICAM-1, and VCAM-1 was observed. P-selectin increased by 60% after activation with thrombin (ANCOVA, P < 0.05; not shown).

On nonactivated HUVECs, small increases in P-selectin, ICAM-1, and VCAM-1 were observed for some starch preparations (ANCOVA, P < 0.05). HES 450 increased all three adhesion molecules by 10–20%. HES 200 and HES 130 increased P-selectin by 10%, whereas HES 70 increased ICAM-1 by 10% (fig. 1).

None of the tested colloids reduced the expression of any adhesion molecule on activated HUVECs. In contrast, a 21% increase in E-selectin and a 15% increase in ICAM-1 were noted after treatment with gelatin. Dextran 70 increased the expression of both adhesion molecules by 11% and 15%, respectively (fig. 2; ANCOVA, P < 0.05).

Activation of HUVECs with TNF-α largely increased PMN adhesion under a postcapillary shear stress of 2 dyn/cm²by 10- to 20-fold (460 ± 35 PMNs/mm²) when compared with nonactivated HUVECs (not shown). Pretreatment of HUVECs with HES 200, HES 130, gelatin, dextran 70, and dextran 40 had no significant effect on firm adhesion, rolling fraction, or rolling velocity (fig. 3and tables 2 and 3). In contrast, cotreatment of PMNs immediately before perfusion showed significant reductions in firm adhesion for all colloids tested (fig. 3; ANCOVA, P < 0.05). Dextran 40 and dextran 70 reduced adhesion by 31% and 51%, respectively. Treatment with gelatin resulted in a 46% decrease. HES 200 and HES 130 attenuated firm adhesion by 43–50%. Together with the decrease in firm adhesion, twofold increased rolling fractions were observed (table 2; ANCOVA, P < 0.05), indicating that rolling interactions were maintained whereas the transition to firm adhesion became impaired as soon as the colloids were added to the PMN perfusate. Accordingly, the rolling velocities were slow and showed no relevant differences between the treatment groups (table 3). As determined by viscometry, the changes in viscosity were small after addition of the colloids (0.007–0.009 poise) and did not correlate with the observed reduction in firm adhesion.

The recruitment of leukocytes to areas of infection or tissue damage is one of the most important functions of the innate immune system.14Because postcapillary wall shear stress opposes cell interactions on quiescent endothelium, up-regulation of endothelial cell adhesion molecules and binding to their leukocyte counterreceptors are prerequisites for the emigration of leukocytes into areas of inflammation.13–17When considering the slow turnover rate of endothelial cells and the prolonged intracellular storage of synthetic colloids such as HES,40,45it is, therefore, most important that none of the tested colloids attenuated leukocyte adhesion by direct down-regulation of endothelial cell adhesion molecules. Furthermore, because PMN adhesion to colloid-pretreated HUVECs remained unchanged, it seems that colloid treatment of the endothelium did not interfere with endothelial chemokines relevant for adhesion. The unchanged expression of E-selectin and ICAM-1 on HUVECs pretreated with HES 200 and HES 130 is consistent with previous studies on diafiltered medium-molecular-weight starch fractions or HES 200.38,40However, direct effects of other synthetic colloids on endothelial cell adhesion molecule expression have not yet been investigated. In addition, the direct effects of synthetic colloids on VCAM-1 have not been studied, except for measurements of soluble adhesion molecules in patients.8Expression of P-selectin, in contrast, has been suspected to become attenuated by HES 450.38However, this assumption is based on the finding that HES 450 decreased the release of vWF in HUVECs after activation with thrombin. Because vWF and P-selectin are stored in the same intracellular granules, it was hypothesized that the decreased release of vWF after pretreatment with HES 450 may indicate a concomitant decrease in P-selectin.38We measured P-selectin directly, and our results clearly argue against this hypothesis. Because selectins bind to carbohydrate structures, it has been further suspected that specific polysaccharides such as HES interfere with selectin binding by mechanisms involving molecular weight, substitution with hydroxyethyl groups, or both.1,4However, the well-preserved rolling interactions in our experiment show that this is not the case. On colloid-pretreated HUVECs, no interferences with PMN adhesion were shown for any colloid tested. In addition, the functional adhesion assay revealed that the minor increase in certain adhesion molecules after pretreatment with dextran 70 or gelatin did not translate to an increased adhesion.

The finding that gelatin increased E-selectin and ICAM-1 on activated HUVECs by 21% and 15% without increasing PMN adhesion is in accord with previous results on cell interactions at defined E-selectin site densities. In that investigation, the rolling of transfected cells was not changed to a relevant extent unless E-selectin site density was more than doubled.18In our study, the relatively large upper confidence interval (1.8-fold adhesion) found on gelatin-pretreated HUVECs may reflect some greater degree of endothelial activation when compared with the other colloids. However, this finding was not statistically significant, and its clinical relevance is clearly offset by the 46% decreased adhesion after cotreatment of the PMNs. On nonactivated HUVECs, gelatin slightly increased the basal expression of ICAM-1 but did not induce any up-regulation of inducible E-selectin. This suggests that gelatin is a relatively weak stimulus for endothelial cells that does not result in substantial expression of adhesion molecules unless the endothelium is activated by other agonists. The mechanism by which gelatin might have contributed to TNF-α–induced endothelial activation remains unknown. Although gelatin contains proteins of the extracellular matrix, such as collagen, which have been shown to induce signal transduction pathways in different cell populations,46,47we are not aware of any study that has addressed the effects of gelatin on endothelial signaling.

In contrast to the negligible effects of synthetic colloids on endothelial cell adhesion molecule expression, cotreatment of PMNs with colloids before perfusion immediately reduced adhesion regardless of the colloid type. The finding that biochemically different polysaccharides and polypeptides caused similar inhibition suggests that these synthetic macromolecules act by means of a direct, unspecific inhibition of PMN binding rather than by interfering with endothelial cell activation. Because firm adhesion is largely controlled by functional changes in integrin avidity and affinity rather than quantitative cell surface expression,27–30we abstained from measuring integrin expression by flow cytometry. To measure integrin function, we calculated rolling fraction instead, which is an established parameter for adhesion efficiency and integrin activation.20,27,28The twofold increased rolling fractions show that the presence of dextrans, HESs, and gelatin in the perfusate immediately impaired the integrin-dependent transition of rolling into firm adhesion, whereas selectin-mediated rolling interactions were not affected.

These findings are in accord with a previous report on static adhesion of T cells to mouse endothelioma cells.48Examining dextran 40, the authors observed a dose-dependent reduction in T-cell adhesion only when the lymphocytes were pretreated with dextran 40 but not after pretreatment of the endothelium. Using flow cytometry and confocal microscopy, they showed that dextran 40 inhibits the adhesion-induced clustering of LFA-1, thereby attenuating the activation-dependent avidity of this integrin without any change in quantitative surface expression. For HES, however, conflicting findings have been reported. Consistent with our recent results, endothelial pretreatment alone with HES 200 for up to 48 h did not attenuate PMN adhesion to cytokine-activated HUVECs under flow.40In contrast, others reported a 24% reduced PMN-transmigration across HES-pretreated HUVECs during 3 h of activation with a chemotactic stimulus in a static adhesion assay.7However, unlike dynamic adhesion assays, static assays lack selectin function which operate only under flow conditions.18Because integrin activation is a short-lived process peaking during preceding selectin interactions,26,30static assays and prolonged activation with chemotactic agents do not necessarily mirror leukocyte-endothelial interaction in vivo . Therefore, our findings obtained at postcapillary shear stress and the observation of an altered integrin clustering after dextran 4048strongly suggest that all types of synthetic colloids impair leukocyte adhesion by a common inhibition of integrin function.

In contrast to many studies in vivo , our results are derived from comparisons to medium and not to crystalloid or albumin controls. Unlike in vivo , however, there was no need for a control of hemodilutional effects in our in vitro  assay. Because purified dry substances of the different colloids were used, the PMN concentrations remained constant before and after adding the colloids. In vivo , PMN adhesion is mainly determined by molecular interactions of adhesion molecules but also seems to be modulated by unspecific mechanisms like erythrocyte aggregation.49Although erythrocyte rheology might be altered differently by crystalloids and colloids,50these interferences did not play a role for the adhesion of isolated PMNs in our assay. Most importantly, the addition of colloids did not affect the concentrations of nutrients and cations as confirmed by measuring sodium, potassium, calcium, and magnesium in colloid supplemented medium. Dilution of the medium with crystalloids, in contrast, alters the concentrations of these molecules. Regarding the importance of divalent cations for selectin- and integrin-mediated interactions,28,51comparisons to crystalloids in vitro , therefore, result in confounding effects rather than representing a valid control for interactions of adhesion molecules. For similar reasons, we abstained from using an albumin group. Albumin exerts its controversially discussed antiinflammatory effects in vivo  through a variety of mechanisms, such as binding to the interstitial matrix, interferences with erythrocyte deformability and shear stress, or through transport of fatty acids, nitric oxide, and metals, with resultant scavenging of reactive oxygen species.52–56Regarding leukocyte sequestration into sites of infection, these mechanisms are rather unspecific when compared with adhesion molecule function. Adhesion molecule expression, in contrast, might be decreased or even increased by human albumin.8,38,57,58In accord with the well-documented proinflammatory effects of certain albumin batches,59,60we have shown previously that different albumin preparations activate E-selectin, ICAM-1, and VCAM-1 differently depending on their manufacture.58These observations show that the effects of albumin preparations are not the same among each other and largely question the relevance of human albumin as a valid control for molecular interactions in adhesion assays.

Although flow chamber-based, in vitro  models cannot simulate all parameters of the human microcirculation, a number of experiments with blocking antibodies have shown that the mechanisms of PMN adhesion in the parallel plate flow chamber are largely identical to those that have been observed in vivo .15–23,31,33,34,61Therefore, our finding of a common, direct inhibition of integrin-dependent PMN adhesion to cytokine-activated endothelium in vitro  might have several relevant implications for leukocyte sequestration in vivo . First, the immediate inhibition of PMN adhesion suggests that adhesion will also be immediately inhibited in vivo  and does not require pretreatment before a microcirculatory insult. Second, the observation that all types of synthetic colloids attenuated adhesion similarly indicates that the type of volume expander does not play a major role in the acute inhibition of leukocyte-endothelial interactions. Finally, because colloids reduced adhesion by a PMN-dependent mechanism rather than interfering with the slowly regenerating endothelium, it seems that the duration of this effect will be largely determined by the intravascular elimination of the colloid. Regarding the susceptibility of traumatized patients to both leukocyte-related tissue injury and infection,37we conclude that the clinically relevant effect of synthetic colloids on leukocyte sequestration depends on the regulation of integrin function and thus their pharmacokinetic profiles, but not on specific interactions with the endothelium.

The authors thank Alice Mager and Christof Zanke (Technical Assistants, Department of Anaesthesiology and Intensive Care Medicine, Tuebingen, Germany) for their help with the adhesion assays and Joan Robertson, M.D. (Family Practitioner, Stuttgart, Germany), for copyediting the manuscript.