Selective human endothelial cell activation by chemokines as a guide to cell homing

Abstract

An original model of organo-specific, immortalized and stabilized endothelial cell lines was used to delineate the part played by some chemokines (CCL21, CX3CL1, CCL5 and CXCL12) and their receptors in endothelium organo-specificity. Chemokine receptor expression and chemokine presentation were investigated on organo-specific human endothelial cell lines. Although the chemokines showed distinct binding patterns for the various endothelial cell lines, these were not correlated with the expression of the corresponding receptors (CX3CR1, CXCR4, CCR5 and CCR7). Experiments with CCL21 on peripheral lymph node endothelial cells demonstrated that the chemokine did not co-localize with its receptor but was associated with extracellular matrix components. The specific activity of chemokines was clearly shown to be related to the endothelial cell origin. Indeed, CX3CL1 and CCL21 promoted lymphocyte recruitment by endothelial cells from the appendix and peripheral lymph nodes, respectively, while CX3CL1 pro-angiogenic activity was restricted to endothelial cells from the appendix and skin. The high specificity of the chemokine/endothelium interaction allowed the design of a direct in vitro endothelial cell targeting assay. This unique cellular model demonstrated a fundamental role for chemokines in conferring on the endothelium its organo-specificity and its potential for tissue targeting through the selective binding, presentation and activation properties of chemokines.

Introduction

The endothelial cell controls the selectivity of invasive processes in two ways: through its phenotype, which differs from one organ to another, and through its microenvironment-mediated modulation of this phenotype, which reflects a physiological or pathological state.^1,2 Thus, according to the organ and the microenvironment, endothelial cells select specific populations of circulating cells via a multiple-step process leading adhesion to invasion.^3,4 This selectivity is regulated by a specific arrangement of cytokines, chemokines and adhesion receptors that guide circulating cells to specific locations. Chemokines, which are presented on the surface of endothelial cells via glycosaminoglycans (GAGs),^5,6 play an essential role in the selective regulation of lymphocyte homing by attracting circulating cells through a gradient and by activating integrins upon binding to seven transmembrane receptors.^7–9 Previous studies have highlighted the specific actions of some chemokines in particular tissues:⁷ CCL17 and CCL27 are specifically involved in lymphocyte trafficking into the skin,^10,11 CCL21 is associated with cell trafficking into the secondary lymphoid organs,¹² and CCL25 is associated with cell trafficking into mucosal tissues.¹³

A better understanding of the mechanisms that control the organo-selectivity of cell trafficking would be of great value in the design of immunotherapy, targeted drug therapy and cell therapy. To this end, we previously designed an in vitro cellular model by immortalizing endothelial cells from different organs and vessel types and stabilized their phenotypes.² The aim of the present study was to elucidate the mechanisms by which chemokines contribute to the organo-specific character of the endothelium. We focused on five endothelial cell lines (three from lymphoid organs and two from non-lymphoid organs, each organ presenting specific lymphocyte homing) and four chemokines (CCL5, CCL21, CXCL12 and CX3CL1). The chemokines CCL21 and, to a lesser extent, CX3CL1 show organo-specific activity by controlling lymphocyte trafficking into the lymphoid organs, while CCL5 and CXCL12 show no such activity.^12,14 CCL5 is a ubiquitous chemokine that plays an active role in recruiting leucocytes to inflammatory sites.¹⁵ CXCL12 is constitutively expressed in many tissues and is an essential regulator of haematopoiesis, lymphocyte homing, pre-B-cell growth, and angiogenesis.¹⁶ Finally, CX3CL1 is a unique chemokine that functions not only as a chemoattractant but also as an adhesion molecule.¹⁷ It is involved in many skin inflammatory diseases¹⁸ such as psoriasis, eczema and atopic dermatitis, and is associated with the terminal cutaneous nerve [nerve growth factor (NGF) receptor positive].¹⁹

Using our model and various activity tests, we demonstrated that chemokine action is endothelial cell origin-selective. On the one hand, CCL21 and CX3CL1 specifically induced a T4 activated cell line (CEMT4) lymphocyte adhesion on human peripheral lymph node endothelial cells, clone B3 (HPLNEC.B3), and human appendix endothelial cells (HAPEC), respectively, whereas CXCL12 and CCL5 did not display any specific action. On the other hand, CX3CL1 specifically promoted angiogenesis on HAPEC and human skin microvascular endothelial cells (HSkMEC). These observations prompted us to investigate whether chemokines could be used to target one endothelial cell and not another, and for this purpose an in vitro targeting model was developed.

Materials and methods

Endothelial cell lines

The five immortalized endothelial cell lines used in this study display the general characteristics of the in vivo endothelium phenotype [e.g. the presence of angiotensin converting enzyme (ACE), the von Willebrand factor and vascular endothelial (VE)-cadherin].

Three lines isolated from secondary lymphoid organs, namely human appendix endothelial cells, clone 1 (HAPEC.1), HPLNEC.B3 and human mesenteric lymph node endothelial cells (HMLNEC), and two lines from non-lymphoid tissues, namely human brain microvascular endothelial cells (HBrMEC) and HSkMEC, were compared.

Endothelial cell culture

Cells were cultured in OptiMEM (Invitrogen, Cergy Pontoise, France) supplemented with 2% fetal bovine serum (BioWest, Nuaillé, France), 40 μg/ml gentamycin (Invitrogen) and 0·05 μg/ml fungizone (Invitrogen). Cells were seeded at 2 × 10⁴ cells/cm², 48 hr before experiments. Cells were maintained at 37° in a 5% CO₂/95% air atmosphere, and after 24 hr cells were placed in serum- and antibiotic-free medium.

Reagents and antibodies

Recombinant human chemokines were obtained from R&D Systems (Abingdon, UK), as were goat anti-human chemokine polyclonal antibodies and mouse anti-human CCR7 monoclonal immunoglobulin G (IgG). The secondary antibody was fluorescein isothiocyanate (FITC)-conjugated rabbit anti-goat IgG (Sigma-Aldrich, Saint Quentin Fallavier, France). The rabbit anti-human CX3CR1 polyclonal immunoglobulin was from eBioscience (San Diego, CA). The mouse monoclonal antibodies anti-human CXCR4 and CCR5 were from BD Pharmingen (San Diego, CA). The rabbit anti-human CCR7 polyclonal IgG was from Epitomics (Burlingame, CA). For flow cytometry analysis, secondary antibodies were FITC-conjugated goat anti-mouse IgG (Sigma-Aldrich) and FITC-conjugated goat anti-rabbit IgG (BD Pharmingen). For fluorescence microscopy, secondary antibodies were Alexa Fluor (AF) 568-donkey anti-goat IgG and FITC-conjugated sheep anti-rabbit IgG (Sigma-Aldrich). Chondroitin sulfate A, C, D and E were obtained from Invitrogen and chondroitin sulfate B from Sigma-Aldrich.

Detection of chemokine receptors and chemokines in endothelial cells

For chemokine receptor detection, 5 × 10⁶ cells were washed twice with complete phosphate-buffered saline (cPBS) (1 mm CaCl₂ and 0·5 mm MgCl₂), 0·5% bovine serum albumin (BSA) [weight/volume (w/v)] (Sigma-Aldrich) and 0·1% NaN₃ (w/v) (Merck-Eurolab, Strasbourg, France). Endothelial cells were detached from 75-cm² culture plates using 0·5 mg/ml type I collagenase (Invitrogen) in a solution of cPBS and 0·5% BSA (w/v). After washing several times with cPBS, cells were incubated in the presence of 5 μg/ml primary antibody against human chemokine receptors for 30 min at 37°. Cells were fixed with a 2% (w/v) paraformaldehyde (PFA; Merck-Eurolab) solution in cPBS for 10 min at 37°. The reaction was stopped by washing several times with cPBS solution and cells were incubated with 0·2 m urea (Merck-Eurolab) in cPBS for 10 min at room temperature, washed, and then incubated with FITC-conjugated goat anti-mouse IgG for 30 min at 4°.

For chemokine binding and presentation by endothelial cells, cells were prepared as described above for chemokine receptor detection and then incubated (v/v) with the appropriate chemokine solution for 30 min at 37°. Various concentrations of each chemokine were tested, and we chose the most effective concentration for the binding assay (CX3CL1, 0·5 nm; CXCL12, 0·8 nm; CCL5, 2·5 nm and CCL21, 20 nm). Cells were washed and fixed as described above. The cells were then incubated with 10 μg/ml goat anti-human chemokine antibodies for 30 min at 37°. Bound chemokine detection was achieved by a third incubation in the presence of FITC-conjugated rabbit anti-goat IgG for 30 min at 4°.

Endothelial cell stimulation by chemokines and adhesion

Endothelial cells were seeded in 24-well plates (Falcon, Grenoble, France) at 2 × 10⁴ cells/cm². After 48 hr of culture, endothelial cells were stimulated with chemokines for 5 hr. The cells were then washed twice with PBS and incubated (for 20 min at 4°) with a previously Paul Karl Horan 26 (PKH26)-labelled CEMT4 lymphocyte suspension at a ratio of five lymphocytes to one endothelial cell. After gentle washing, to protect the lymphocytes specifically bound to endothelial cells, cells were detached using a trypsin/ethylenediaminetetraacetic acid (EDTA) solution, washed and analysed by flow cytometry as described below. The data are expressed as the number of CEMT4 cells bound to 100 endothelial cells.

Flow cytometry analyses

The results were analysed by flow cytometry on a FACS Sort (Becton Dickinson, Sunnyvale, CA) and using cellquest software (Becton Dickinson).

For adhesion experiments, the ratio of PKH-labelled CEMT4 cells to unlabelled endothelial cells was determined after excitation at 488 nm and emission at 560 nm. The results are expressed as the adhesion ratio (number of CEMT4 cells fixed by 100 endothelial cells after chemokine stimulation/number of CEMT4 cells fixed by 100 endothelial cells without chemokine stimulation).

Immunostaining of CCL21 and CCR7 by fluorescence microscopy

Human endothelial cells were seeded at 2 × 10⁴ cells/cm² in 96-well plates. After 48 hr of culture in a 5% CO₂/95% air atmosphere, immunostaining was performed as described previously. For CCL21 binding inhibition by chondroitin sulfate, CCL21 was first incubated for 15 min at room temperature with a mix of chondroitin sulfate A, B, C, D and E (from 0·02 to 20 μg/ml of each) and then the mix was added to the cells. For CCR7–CCL21 co-localization, the antibodies were mixed and added together to endothelial cells at the same final concentration as for single labelling.

The results were analysed with axiovision 4·3 software (Carl Zeiss, Le Pecq, France).

In vitro angiogenesis assay

Angiogenesis was performed on Matrigel™-coated 24-well plates (BD Biosciences). The Matrigel contains extracellular matrix proteins and growth factors which permit endothelial cells to spontaneously form pseudo-vessels. After coating, cells were seeded at 2·5 × 10⁴ cells/cm². Once adhesion on the matrix had been achieved, cells were incubated in the presence of chemokines dissolved in OptiMEM. The rearrangement of the cells and the formation of pseudo-vessels were followed using an inverted microscope (Axiovert 200; Carl Zeiss). The qualitative analysis consisted of taking pictures at several time-points and comparing them with pictures of the controls incubated in the absence of chemokines, to evaluate the effect of the chemokines on vessel formation. This test allows the potential pro- or anti-angiogenic effect of chemokines on human endothelial cells to be determined.

Human endothelial cell proliferation test

Cell proliferation was assessed using Alamar Blue™ dye (Biosource, Nivelles, Belgium) according to the manufacturer’s instructions. Briefly, the cells were seeded on 96-well plates at 2 × 10³ cells/well in the presence or absence of the chemokines. After 24, 48 and 72 hr of cell culture, Alamar Blue™ dye was added to the medium and the cells were incubated for 3 hr at 37° in a 5% CO₂/95% air atmosphere. Cell proliferation was estimated using a spectrofluorimetric microwell-plate reader [Wallac Victor 1420 (Perkin Elmer, Courtaboeuf, France); λ excitation 560 nm; λ emission 590 nm] and is expressed as the percentage of growth compared with the control without chemokine. All experiments were performed in triplicate for each concentration and for each chemokine.

Carboxy fluorescein diacetate succinimidyl ester (CFDA SE) cell labelling

CFDA SE (Invitrogen) is a vital dye that can be used to label the cytoplasm after hydrolysis by cell esterases. The dye is inherited by cell division, not transferred to adjacent cells in a population. Consequently, upon co-culture of an unlabelled cell line with a labelled cell line, the two types of cell are distinguishable.

Briefly, trypsin-detached endothelial cells (1 × 10⁶) were washed with cPBS and incubated in CFDA SE (5 μm) solution diluted in cPBS for 10 min at 37°. After washing three times with 0·5% (w/v) BSA in cPBS, cells were mixed with unlabelled endothelial cells from another cell line, in the desired proportions, for seeding and were co-cultured in complete medium for 48 hr.

Chemokine substitution on fluorescent beads

Fluospheres (carboxylate-modified microspheres, 1·0 μm in diameter; red fluorescent λex = 580 nm, λem = 605 nm) were obtained from Invitrogen. Beads were substituted by chemokines according to the manufacturer’s instructions. Briefly, beads were mixed with CCL21 (4 nm; R&D Systems) in 50 mm 2 N-morpholino ethanesulfinic acid (MES) buffer, pH 6, and incubated for 15 min at room temperature. 1-Ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDAC) was then added to a final concentration of 7·4 mg/ml. The mixture was incubated for 5 hr at room temperature. Glycine was added to a final concentration of 100 mm and the mixture was incubated for 30 min at room temperature. Finally, beads substituted with chemokines were washed three times with PBS.

As a negative control, beads substituted with chemokines were heated at 60° for 15 min.

Chemokine targeting in flow conditions

HSkMEC and HPLNEC.B3 were seeded in co-culture on polystyrene tissue-culture slides (Nagle Nunc International, Rochester, NY) after fluorescent labelling of one of the two cell lines with CFDA SE (Invitrogen), as previously described. After 48 hr of co-culture, the cells were subjected to flow conditions under shear stress using the cell adhesion flow chamber CAF 10 (Immunetics, Boston, MA). The chamber was mounted on an Axiovert 200 epifluorescence inverted microscope (Carl Zeiss) for direct real-time visualization of the dynamic bead adhesion process using a ×10 objective. The microscope was coupled to an Axiocam high-resolution numeric camera (Carl Zeiss) directly linked to a computer equipped with the acquisition software axiovision (Carl Zeiss).

Chemokine binding was assessed by perfusing, through the flow chamber, the chemokine-substituted beads on the cell surface, using a withdrawal syringe pump (Harvard Apparatus, Les Ulis, France). Beads were injected in a laminar flow at a fixed flow rate of 50 μl/min for 5 min. The cell surface was then washed with OptiMEM medium for 10 min at a flow rate of 150 μl/min. After washing, pictures of the slide were taken at different time-points and the number of fixed beads on each cell line was quantified.

Results

Chemokines bind to and are presented by endothelial cells

The ability of chemokine molecules to bind endothelial cells was assessed after incubation of endothelial cells in the presence of chemokines and subsequent detection of their binding using cytochemical methods (Fig. 1a). CX3CL1 was mainly found to bind the brain-derived endothelial cell surface (Fig. 1a). The five endothelial cell lines bound CCL5 efficiently while CXCL12 was poorly detected on the cell surface in the conditions used. CCL21 was mainly found on HPLNEC.B3 among endothelial cells from lymphoid organs and on HBrMEC among endothelial cells from non-lymphoid organs.

Figure 1.

Chemokines bind differently to different endothelial cells, irrespective of the expression of their receptors. (a) Chemokine binding on fixed cells was detected using specific antibodies and analysed by flow cytometry. (b) Cell surface detection of chemokine receptors on fixed cells was achieved in the same way. Results are expressed as relative fluorescence intensity (rFI); data are from a typical experiment, performed in triplicate.

In order to determine whether the chemokine binding to endothelial cells is receptor mediated, we studied the expression of the receptors corresponding to the selected chemokines, i.e. CX3CR1, CXCR4, CCR5 and CCR7. Chemokine receptors on endothelial cells were detected by specific antibodies on fixed endothelial cells as shown in Fig. 1 (data from cytochemical labelling and flow cytometry detection). The results are given as the relative fluorescence intensity (rFI) calculated from the difference between the sample rFI and the isotypic control rFI.

All four receptors were detected in each endothelial cell line but their expression level varied according to the origin of the endothelial cell line. CX3CR1 was highly expressed on the three lymphoid organ-derived cell lines (HAPEC.1, HPLNEC.B3 and HMLNEC) and on skin endothelial cells but its expression was low on the brain endothelial cell surface. CXCR4 was expressed on all endothelial cells, with weaker expression on brain-derived endothelial cells. CCR5 was detectable on the surfaces of all endothelial cells tested, being expressed at particularly high levels on HAPEC and HMLNEC, and at low levels on HBrMEC. CCR7 was also expressed in the five lines but most strongly on the HMLNEC and HSkMEC surface.

Chemokines were able to bind each endothelial cell line tested but with distinct binding profiles which could not be correlated to their receptor expression level. Consequently, it was not possible to discern any clear relation between chemokine binding (Fig. 1a) and receptor expression (Fig. 1b). Because these data ruled out direct receptor chemokine recognition, we investigated another mechanism, namely presentation by extracellular matrix components.

CCL21 binds to the extracellular matrix on HPLNEC.B3

Using fixed HPLNEC.B3, we showed by fluorescence microscopy experiments that CCL21 is produced by endothelial cells and is detectable on the cell body (Fig. 2a). Its receptor, CCR7, showed a homogenous distribution on the surface, forming some receptor clusters in the cellular membrane. When the chemokine was added, its location seemed not to correspond to the receptor distribution. Indeed, higher levels of CCL21 binding could be detected, delineating the extracellular matrix around the cells (Fig. 2a). This allowed confirmation of the active binding of the chemokine on endothelial cells as previously shown by flow cytometry (Fig. 1a).

Figure 2.

Detection of CCL21 and CCR7 on human peripheral lymph node endothelial cells, clone B3 (HPLNEC.B3). (a) Fluorescence microscopy study of fixed HPLNEC.B3 for the expression of CCR7 and CCL21, in the absence (−CCL21) or presence (+CCL21) of recombinant CCL21 (20 nm), preincubated or not with chondroitin sulfate (200 ng/ml). (b) Co-detection of CCL21 and its receptor on fixed HPLNEC.B3, in the absence (−CCL21) or presence (+CCL21) of the chemokine. CCL21 and its receptor were detected using fluorescently labelled specific antibodies (red and green, respectively).

We previously observed the expression of GAGs (heparan sulfate and mainly chondroitin sulfate) on the HPLNEC.B3 surface. To determine whether the binding of the chemokine on endothelial cells was attributable to extracellular matrix components, we induced chemokine binding inhibition by preincubating CCL21 with soluble chondroitin sulfate. Figure 2a shows that this resulted in a clear reduction in the amount of CCL21 detected; this was quite similar to the endogenous chemokine level.

Furthermore, to augment our flow cytometry finding of a lack of correlation between chemokine binding and receptor expression, co-localization experiments with the chemokine CCL21 and its receptor CCR7 were carried out, and the data obtained are presented in Fig. 2b. These results showed that the distribution of CCR7 was quite distinct from that of the endogenous chemokine. In addition, when CCL21 was added exogenously to the cells and allowed to bind to the intact cell surface, CCR7 was observed to only partially co-localize with the bound CCL21.

These data suggest that CCL21 can bind to endothelial cells via its receptor and via extracellular matrix components such as GAGs.

To address the question of the significance of a second chemokine binding process in the specific activation of endothelial cells in an organ-, site- and state-dependent manner, the adhesion process of the leucocyte assay was chosen.

Chemokine-mediated endothelial cell-specific modulation of CEMT4 adhesion

The biological significance of the selective distribution and expression of chemokines as well as restricted chemokine binding to endothelial cells was assessed by investigating their effect on the leucocyte to endothelial cell adhesion process. For this assay, endothelial cells were exposed to chemokines and then submitted to an adhesion test as described in the Materials and methods. Various concentrations of each chemokine were tested according to the manufacturer’s instructions (between 0·1 and 30 nm) and we chose the most effective concentration for the activation experiment.

The results are shown in Fig. 3. Upon stimulation with CX3CL1, HAPEC.1 displayed a significantly increased adhesion capacity for CEMT4 lymphocytes (adhesion ratio = 9·5). This effect was clearly restricted to CX3CL1, whereas CXCL12, CCL5 and CCL21 had no effect on lymphocyte recognition by HAPEC.1.

Figure 3.

Chemokines selectively activate endothelial cell adhesion capacity. Lymphocyte adhesion on human appendix endothelial cells, clone 1 (HAPEC.1), human peripheral lymph node endothelial cells, clone B3 (HPLNEC.B3), human mesenteric lymph node endothelial cells (HMLNEC), human brain microvascular endothelial cells (HBrMEC) and human skin microvascular endothelial cells (HSkMEC) was estimated after stimulation with CX3CL1 (0·5 nm), CXCL12 (0·8 nm), CCL5 (2·5 nm) and CCL21 (20 nm). Endothelial cell adhesion capacity for CEMT4 lymphocytes was quantified by flow cytometry. The results are expressed as the number of CEMT4 cells bound to 100 endothelial cells. Data are from three independent experiments.

CCL21 specifically increased the adhesion of CEMT4 lymphocytes on HPLNECB3 (ratio = 5), in contrast to the other chemokines which were not able to modulate lymphocyte adhesion on HPLNECB3. In the case of HMLNEC, we observed a mixed profile and a reduced chemokine activation effect compared with peripheral lymphoid or mucosal derived endothelial cells. Indeed, only CCL5 was able to increase lymphocyte binding (ratio = 2·5).

The chemokines CX3CL1, CCL5 and CCL21 seemed to inhibit lymphocyte recognition by brain endothelial cells (ratio < 1), whereas CXCL12 had no significant effect. Interestingly, endothelial cells from the skin were not significantly affected by chemokines in terms of their lymphocyte binding activity.

The ability of CCL21 and CX3CL1 to increase lymphocyte adhesion to HPLNEC.B3 and HAPEC.1, respectively, clearly illustrates the selectivity of these chemokines according to endothelial origin.

Chemokines are organo-selective, pro-angiogenic factors: CX3CL1 specifically modulates angiogenesis in vitro

We performed a series of angiogenesis experiments with the aim of screening potential pro- or anti-angiogenic chemokine properties as a function of chemokine type in relation to endothelial cell type and origin. The findings are illustrated here by the results for CX3CL1 and its effect on endothelial cell angiogenesis kinetics (Fig. 4). Various chemokine concentrations were tested.

Figure 4.

CX3CL1 acts specifically on angiogenesis of human appendix endothelial cells, clone 1 (HAPEC.1) and human skin microvascular endothelial cells (HSkMEC). After 5 hr in the presence of CX3CL1 at 10 pm, the rearrangement of cells and the formation of pseudo-vessels of human appendix endothelial cells, clone 1 (HAPEC.1), human peripheral lymph node endothelial cells, clone B3 (HPLNEC.B3), human mesenteric lymph node endothelial cells (HMLNEC), human brain microvascular endothelial cells (HBrMEC) and human skin microvascular endothelial cells (HSkMEC) were followed and compared with the control without chemokine.

Compared with the control cells incubated on Matrigel without the chemokine, the presence of CX3CL1 at 10 pm significantly favoured the formation of pseudo-vessels for HAPEC.1 and HSkMEC (Fig. 4). This effect was also observed at 0·1 and 1 nm (data not shown). It slightly accelerated the rearrangement of HBrMEC but had no significant effect on HMLNEC and HPLNEC.B3.

Angiogenesis experiments in vitro were also performed by incubating CCL21, CCL5 and CXCL12, at various concentrations, with the five endothelial cell lines, but, interestingly, no significant modulation of vessel formation kinetics was observed (data not shown).

CCL21 acts as a specific peripheral lymph node endothelial cell growth factor

The proliferation test (Fig. 5) clearly suggested that CCL21 affected the growth of HPLNEC.B3, whereas it moderately (10%) influenced the growth of other endothelial cell lines (HAPEC.1, HMLNEC and HBrMEC) (Fig. 5a). This effect was concentration-dependent, with a maximal effect at 20 nm producing a 40% growth increase. CX3CL1 had a weak effect on the growth of the investigated endothelial cell lines (< 10%) except for HMLNEC at 10 nm and HBrMEC at 20 nm, where we observed an increase of 20% (Fig. 5b). These effects occurred at high concentrations of CX3CL1 compared with the active concentration observed in the adhesion assay and angiogenesis tests for HAPEC.1 (Figs 3 and 4).

Figure 5.

CCL21 acts specifically as a growth factor on human peripheral lymph node endothelial cells, clone B3 (HPLNEC.B3). CCL21 (a) and CX3CL1 (b) growth factor activity on endothelial cell lines was tested in dose–response experiments over 72 hr of culture. Cell density was assessed by Alamar Blue™ dye staining on a multiwall plate spectrofluorimeter. The effect on cell proliferation is expressed as the increase in the number of cells as a percentage of the control value without chemokine. Data are from three independent experiments.

CXCL12 increased by about 20% the growth of HMLNEC and HBrMEC when used at 1·2 and 2·5 nm, respectively, but no significant effect was detected on the other endothelial cell lines (data not shown). Finally, CCL5 showed no effect on the five tested endothelial cell lines (data not shown).

Specific peripheral lymph node endothelial cell targeting by CCL21

The direct selective interaction of the lymph node-specific CCL21 chemokine with peripheral lymph node endothelial cells was demonstrated in our in vitro assay which allowed their discrimination among other endothelial cells in a co-culture.

A monolayer of mixed endothelial cells, HPLNEC.B3 and HSkMEC, was obtained after 48 hr of co-culture. The cells were subjected to a flow containing a CCL21-substituted fluorescent bead suspension. The interaction was permitted by flushing the bead suspension for 5 min in conditions allowing the required shear stress. Final binding was assessed after washing for 10 min.

Figure 6 clearly shows the preferential binding of the CCL21-substituted beads to the peripheral lymph node-derived endothelial cells (67 ± 3% of beads fixed on HPLNEC.B3 relative to the total of beads in the field; Fig. 6e) as compared with the skin-derived endothelial cells (33 ± 2%; Fig. 6e). This effect was totally independent of the labelling procedure (Fig. 6a compared with Fig. 6c). The specificity of the binding was confirmed by the reduced effect observed in the controls obtained with heated CCL21-substituted beads (Fig. 6b and 6d).

Figure 6.

Demonstration of the in vitro targeting of endothelial cells: CCL21 specifically binds to peripheral lymph nodes endothelial cells in a mixed endothelial cell culture. Carboxy fluorescein diacetate succinimidyl ester (CFDA SE)-labelled endothelial cells from one line [human skin microvascular endothelial cells (HSkMEC) in (a) and (b) and human peripheral lymph node endothelial cells, clone B3 (HPLNEC.B3), in (c) and (d)] were mixed with unlabelled endothelial cells [HPLNEC.B3 cells in (a) and (b) and HSkMEC cells in (c) and (d)], co-cultured and submitted to chemokine recognition in flow conditions. Targeting is shown towards peripheral lymph node-derived endothelial cells in (a) (with unlabelled HPLNEC.B3) and (c) (with labelled HPLNEC.B3) and in comparison with the control (heat-inactivated beads) in (b) and (d). (e) A summary of this distribution, expressed as the percentage of CCL21-substituted beads fixed on HPLNEC.B3 compared with HSkMEC, according to the various flow conditions.

Discussion

Our results indicate that chemokines have various and selective activities on endothelial cells depending on their organ of origin. These activities, such as proliferation, lymphocyte adhesion and angiogenesis, are not correlated with receptor expression. Indeed, the measured amounts of chemokines able to bind to a given endothelial cell did not correlate with the expression level of their corresponding receptors. This confirms that, while chemokine receptors could represent a unique signal transduction pathway, they are not the only means of chemokine presentation. Although some chemokines are able to bind several receptors of the same family (e.g. CCL5 binds to CCR1, CCR2 and CCR5), this is not the case for all chemokines (e.g. CX3CL1). The possibility that this lack of correlation may be attributable to the binding of chemokines via the DUFFY receptor was ruled out by the use of several blocking antibodies (kindly provided by Dr Dominique Blanchard, INSERM U76, National Institute for Blood Transfusion, Paris, France; data not shown). Interestingly, however, we were able to demonstrate the involvement of GAGs in chemokine binding and selective inhibition using chondroitin sulfate. Cell surface GAGs were previously shown to immobilize chemokines in order to create the chemotactic gradient that orientates leucocyte migration.²⁰

Although the nature of the extracellular matrix components that bind chemokines was not determined in the present study, there is considerable evidence strongly suggesting a major role for GAGs.^5,21 As the cell surface GAG composition depends on the location and type of endothelium, GAGs may contribute to leucocyte recruitment selectivity and endothelium organo-specificity.⁵ Indeed, depending on the type of vessels and the localization in the body, vascular endothelial cells expressed various GAGs that differed in charge density, chain length and sulfation pattern.²² This variability could determine the type of chemokines presented at a given site and lead to selectivity of the type of recruited leucocyte.

These data was corroborated by our chemokine/endothelial cell stimulation data, which suggested a dual chemokine- and endothelial cell-specific response, thus validating the endothelium organo-specificity hypothesis. More precisely, CCL21 specifically acted on HPLNEC.B3 to increase adhesion capacity for CEMT4 lymphocytes; CX3CL1 specifically acted on mucosal endothelial cells (HAPEC.1) to produce a similar effect on CEMT4 lymphocyte adhesion. HMLNEC, a representative endothelial cell line for mesenteric lymph nodes, showed an intermediate response to both CX3CL1 and CCL21, as described for the homing process.⁴

The various possibilities for chemokine binding and presentation indicate that the crucial role of chemokines in the recognition of circulating cells by endothelial cells is more subtle than direct receptor–ligand signalling, and thus brings a higher degree of specificity to the endothelial repertoire.

The chemokine CCL21 is constitutively expressed in secondary lymphoid tissues such as lymph nodes, the appendix and the spleen.²³ CCL21 acts via the CCR7 chemokine receptor,²⁴ which is highly expressed in various lymphoid tissues and on blood T and B lymphocytes.²⁵ The inactivation of genes coding for CCR7 and CCL21 suggested their key role in the homing process.^14,26 The specificity of CCL21 for secondary lymphoid organs is confirmed here by its specific action in terms of lymphocyte binding to HPLNEC.B3.

Concerning CX3CL1, it was reported that this transmembrane chemokine²⁷ is expressed in mucosal sites such as the intestine and appendix.²⁸ It participates in leucocyte recruitment by interacting with its receptor, CX3CR1,²⁹ either in a soluble form or as a membrane-anchored molecule.³⁰ In this study, we showed that soluble CX3CL1 specifically favours the recognition of lymphocytes by mucosal lymphoid endothelial cells, but not by non-lymphoid brain and skin endothelial cells, where it is also expressed.^31,32

Our results highlight the diversity of chemokine action, although the molecular mechanisms involved, such as the induction of the expression of adhesion molecules³³ and chemokine receptors, as well as de novo chemokine secretion by endothelial cells,³⁴ require further investigation.

Of these various activities, one of the most important is the regulation of angiogenesis in normal or pathological conditions.^35,36 In this study, we have demonstrated that CX3CL1 is able to accelerate the kinetics of pseudo-vessel formation on the specific endothelial cell lines HAPEC.1 and HSkMEC. Expressed in mucosal sites but also in the brain and skin, CX3CL1 could act like a signalling molecule in intercellular communication within these organs. Our in vitro assay provides evidence for a pro-angiogenic role of CX3CL1,³⁷ specifically in organs where it plays an important role in cell–cell interactions. Indeed, although CX3CL1 does not have a significant effect on the growth of HAPEC, it induces the migration and differentiation of these endothelial cells, leading to the formation of pseudo-vessels in vitro. In contrast to CX3CL1, CCL21 is not able to modulate the angiogenesis process but clearly acts like a growth factor in its effect on peripheral lymph node endothelial cells (HPLNEC.B3).

This work confirms that the selective activities of chemokines towards endothelial cells, according to their organ of origin, provide specific markers which contribute to endothelium-based cell immunotherapy strategies.³⁸

This specificity allowed us to design a key assay using a mixture of two different endothelial cell lines, which showed that a chemokine is able to discriminate and target one endothelial cell type in a very selective manner. This validates the hypothesis of endothelium organo-specificity and opens up new possibilities for diagnostic and predictive assays for invasive pathologies based on the concept of the barrier role of the endothelium. The same principle can be applied to test the bioavailability of therapeutic molecules through endothelial receptor targeting and to design alternative methods for drug testing. Further work is needed to define the molecular cascade of events that intervene in the cell recognition process to determine the factors responsible for the specificity and to develop site-targeted therapies.

Glossary

Abbreviations

cPBS

complete phosphate-buffered saline

HAPEC.1

human appendix endothelial cells, clone 1

HBrMEC

human brain microvascular endothelial cells

HMLNEC

human mesenteric lymph node endothelial cells

HPLNEC.B3

human peripheral lymph node endothelial cells, clone B3

HSkMEC

human skin microvascular endothelial cells

PFA

paraformaldehyde

PLN

peripheral lymph node