Transendothelial migration of effector T cells across inflamed endothelial barriers does not require heparan sulfate proteoglycans

Abstract

Leukocyte diapedesis is a chemotactic multistep process that requires optimal chemoattractant presentation by the endothelial barrier. Recent studies have described a critical role for heparan sulfate glycosaminoglycans (HSGAGs) in the presentation and functions of chemokines essential for lymphocyte interactions with the lymph node vasculature. We wished to test whether HS expression by a prototypic endothelial cell type, i.e. human umbilical vein endothelial cells (HUVECs), is critical for their ability to support neutrophil and lymphocyte adhesion and transendothelial migration (TEM) under shear flow. We found that HUVECs deposit HS GAGs mainly at their basolateral compartments in both their resting and inflamed states. We next inactivated the key enzyme involved in HS biosynthesis, exostosin-1 (Ext1). Silencing Ext1 resulted in a complete loss of HS biosynthesis; nonetheless, TNF-α and IL-1β stimulation of key adhesion molecules and inflammatory chemokines necessary for neutrophil or lymphocyte adhesion and TEM remained intact. Ext1 silencing reduced neutrophil arrest and markedly impaired TEM, consistent with a role of basolateral HS GAGs in directing neutrophil crossing of inflamed endothelial barriers. Strikingly, however, the TEM of effector T cells across identically Ext1-silenced HUVECs remained normal. Importantly, the biosynthesis of the main promigratory chemokines for effector T cells and neutrophils, respectively, CCL2 and CXCL1, and their vesicle distributions were also Ext1 independent. These results suggest that transmigrating neutrophils must respond to chemokines transiently presented by apical and basolateral endothelial HS GAGs. In contrast, effector T cells can integrate chemotactic TEM signals directly from intra-endothelial chemokine stores rather than from externally deposited chemokines.

Ext1, inflammation, leukocyte trafficking, neutrophils

Introduction

Neutrophil and lymphocyte recruitment to sites of inflammation involves multiple adhesive steps to cytokine-stimulated post-capillary venules; adhesion is followed by transendothelial migration (TEM) (diapedesis) (1, 2), a multistep process, tightly regulated by endothelium-associated chemoattractants (3–5). Among these chemotactic molecules, chemokines are small, structurally related chemotactic cytokines with shared biochemical properties. All chemokines signal through cognate G-protein-coupled receptors (GPCRs) that trigger numerous GTPases and protein tyrosine kinase targets on responding leukocytes (6, 7). These effector molecules coordinate the adhesive activities of leukocyte integrins as well as the actomyosin machineries critical for leukocyte motility, protrusive capacity and their ability to scan the apical (lumenal) endothelial surface and endothelial junctions for exit (diapedesis) signals (8).

Accumulating in vitro evidence has suggested that chemokines, as well as lipid chemoattractants, operate mainly in their immobilized states on their target receptors on the leukocyte surface and must therefore be presented by specific scaffolds on or near the apical and basolateral endothelial membranes (9–11). Whereas lipid chemoattractants may be directly adsorbed onto the plasma membrane of endothelial cells (ECs), all known chemokines contain positively charged C-terminal domains with high affinity to heparan sulfate glycosaminoglycans (HS GAGs) presented by heparan sulfate proteoglycans (HSPGs), and these chemokines use this recognition for correct presentation and function (11). It was therefore predicted that the adhesion and TEM of both neutrophils and T cells would share tight dependence on normal HS GAG expression by the endothelial barriers crossed by these leukocytes as they emigrate to target tissues.

Both resting and inflamed ECs express high levels of HSPGs, which comprise 50–90% of the total endothelial proteoglycans (12). HSPGs are ubiquitous and structurally diverse macromolecules, which interact with many cytokines, growth factors and extracellular matrix (ECM) components (13). Many chemokines contain HS-binding domains (14), and HSPGs were shown to be critical for the extracellular deposition and oligomerization of major leukocyte chemokines on both EC protein carriers and EC-derived basement membrane components (11, 15). Indeed, chemokines with mutations in specific amino acids that mediate HSPG binding promote chemotaxis in vitro but fail to recruit leukocytes to sites of inflammation (16).

In light of the well-established roles of chemokines and lipid chemoattractants in neutrophil and lymphocyte adhesion and extravasation (1, 4, 5, 7, 10, 17–21), in addition to the diverse expression of multiple chemokines and lipid chemoattractants by each inflamed endothelial barrier, it has been unclear, whether, when and where a given leukocyte must encounter various endothelium-associated chemoattractants to successfully cross endothelial barriers.

Recently, several in vivo studies provided compelling genetic evidence for the importance of HSPGs on ECs for chemokine presentation and function, promoting lymphocyte and neutrophil adhesion, and TEM (21, 22). Bao et al. described a novel approach to delete the entire HSPG repertoire from vascular cells by a transient conditional endothelium-targeted depletion of exostosin-1 (Ext1) (21), a critical enzyme in the biosynthesis of all HS chains (23). Presentation of the key CCR7-binding chemokine, CCL21, by both lymph node high endothelial venules (HEVs) and lymphatic endothelium was shown to depend on intact HSPG expression on these cells. Furthermore, the immobilization of CCL21 on both types of ECs was claimed to be critical for both its pro-adhesive and chemotactic functions (21, 24). However, the study of Bao et al. focused mainly on lymph node HEVs, lymphatic vessels and on the recruitment of naive T cells and DCs by these vessels. Hence, the contribution of HSPG scaffolds to functional presentation of inflammatory chemokines in inflamed flat ECs, comprising the vast majority of extralymphoid post-capillary venules, has remained elusive. In particular, it has been unclear how different pools of inflammatory chemokines produced by inflamed ECs support the high directionality of leukocyte TEM across these inflamed endothelial barriers.

In the present study, we used an in vitro approach to address some of these standing questions by genetically depleting all HS GAG moieties in a prototypic model of a flat endothelial barrier, human umbilical vein ECs (HUVECs), known to support, when properly inflamed, robust lymphocyte and neutrophil adhesion and TEM under physiological shear flow conditions (17, 19, 25–27). Ext1 ablation was achieved by introduction of small hairpin RNA (shRNA), which resulted in total loss of HS biosynthesis but normal expression of adhesion molecules and inflammatory cytokines as well as normal secretion of the ECM core proteoglycan, perlecan. This allowed us to dissect whether and how the adhesive and migratory properties of neutrophils and effector T cells are altered by the removal of HS GAGs from these ECs. As expected, the ability of HUVECs—inflamed with either TNF-α or IL-1β—to promote neutrophil adhesion and TEM under shear flow was impaired in the absence of HS expression, consistent with an essential role of HS scaffolds on these ECs in presenting TEM-promoting chemokines to neutrophils. Nonetheless, the ability of effector T cells to adhere to and transmigrate across the same inflamed endothelial barriers was not affected by ablation of HS biosynthesis. Because both production and distribution of inflammatory chemokines in intra-endothelial vesicles critical for T-cell TEM were independent of HS biosynthesis and expression, we conclude that effector T cells encounter and respond to these intra-endothelial chemokines directly without a requirement for HSPG-mediated chemokine immobilization on extracellular depots.

Methods

Antibodies and reagents

Anti-HS monoclonal mouse IgM clone 10E4 was purchased from Seikagaku Corporation (Tokyo, Japan). Mouse anti-human collagen IV (1.BB.753) was from Abcam (Cambridge, UK). Mouse anti-human CCL2 (23007), mouse anti-human CXCL1 (20326) and TNF-α were from R&D systems (Minneapolis, MN, USA). IL-1β and IL-2 were from PeproTech (London, UK). Mouse anti-human perlecan (CSI 001-76) was from Santa-Cruz (Dallas, TX, USA). Mouse anti-human intercellular adhesion molecule (ICAM)-1(HA58) and E-selectin (P2H3) were purchased from eBioscience (San Diego, CA, USA). Mouse anti-human CD3 (OKT3) and mouse anti-human CD28 (CD28.2) were from Biolegend (San Diego). The secondary antibodies goat anti-mouse IgM Alexa-488 or -647, goat anti-mouse IgG1 Alexa-488 or goat anti-mouse IgG2b Alexa-647 were purchased from Invitrogen (Carlsbad, CA, USA). BSA (fraction V), Ca²⁺-, Mg²⁺-free HBSS, Brefeldin A (BreA), fibronectin, phalloidin and Hoechst 33342 were purchased from Sigma-Aldrich (St Louis, MO, USA). WEB-2086 was purchased from Tocris (Bristol, UK).

Cells

HUVECs were cultured according to the supplier’s instructions (Promocell Heidelberg, Germany). All in vitro experiments with human leukocytes were approved by the Institutional Review Board of the Rambam Medical Center, in accordance with the provisions of the Declaration of Helsinki. Human neutrophils and T cells were isolated from citrate-anticoagulated whole blood of healthy donors by dextran sedimentation and density separation over Ficoll-Hypaque (28). Neutrophils were used within 2h, and peripheral blood T cells (>90% CD3⁺ T cells) were used for the production of effector lymphocytes by seeding on plates coated with anti-CD3 (OKT3; Biolegend) and anti-CD28 (CD28.2; Biolegend) and cultured for 9–12 days with IL-2 (29). Before experiments, effector lymphocytes were washed and kept overnight in fresh IL-2-containing medium.

Immunofluorescence

HUVECs were plated at confluence on glass-bottom dishes spotted with fibronectin (20mg ml⁻¹). ECs were stimulated with TNF-α (2ng ml⁻¹) or IL-1β (2ng ml⁻¹) for 16–18h or left unstimulated. Samples were fixed with PBS containing 4% (w:v) paraformaldehyde and 2% (w:v) sucrose. For intracellular immunostaining, fixed cells were permeabilized with saponin (0.1%, w:v). Fixed cells were extensively washed, blocked with 10% (v:v) goat serum and incubated with primary antibody, followed by incubation with secondary antibody, the nuclear dye Hoechst and the F-actin probe, phalloidin. Images were acquired using the Delta-Vision system (Applied Precision, Issaquah, WA, USA). Sections were acquired as serial z-stacks (0.2 μm apart) and were subjected to digital deconvolution (SoftWoRx, Applied Precision).

Generation of lentiviral vectors and gene transduction

Lentiviral particles were generated as previously described (30). Briefly, Lipofectamine 2000 (Invitrogen) was used to transfect 293T human embryonic kidney cells with the packaging plasmids pLP1, pLP/VSVG and pLP2 together with the pLKO.1 vector encoding Ext1 shRNA (targeting sequence: CCGGCAATTGTGAGGA CATTCTCATCTCGAGA TGAGAATGTC CTCACAATTGTTTTTG) or non-mammalian shRNA control (targeting sequence: CCGGCAACAAGAT GAAGAGCACCAACTCGAGTTGGTGCTCTT CATCTTGTTGT TTTTG) (Sigma-Aldrich). Supernatants containing lentiviral particles were collected 48h after transfection and were used to transduce HUVECs. Silenced HUVECs were selected using puromycin (2 µg ml⁻¹ for 48h), 24h after transduction.

RT–PCR

The detailed description of all RT–PCR analysis is included in the Supplementary Methods, available at International Immunology Online.

Analysis of neutrophil and effector T-cell migration under shear flow

HUVECs were plated at confluence on plastic-bottom dishes spotted with fibronectin (20mg ml⁻¹). ECs were stimulated with TNF-α (2ng ml⁻¹) or IL-1β (2ng ml⁻¹) for 16–18h. EC-coated plates were assembled in a flow chamber and washed extensively. All flow experiments were performed at 37°C. Leukocytes were perfused over the EC monolayer for 40 s at a force of 1.5 dyn cm⁻² (accumulation phase) and then left under constant shear (5 dyn cm⁻²) throughout the assay. Images were either acquired in real time (28) or were recorded at a rate of four frames per minute for analysis of migration with a Delta-Vision microscope equipped with ×20 phase contrast objective and motorized stage (Applied Precision). For migratory phenotype analysis, leukocytes accumulated in three fields were individually tracked throughout the assay and categorized as previously described (25). Crawling leukocytes moved at least three cell diameters over the EC surface without crossing through the monolayer, whereas transmigrating leukocytes crawled variable distances before crossing the endothelium (‘crawling’ and ‘TEM’). Arrests were defined as leukocytes arrested at a force of 1.5 dyn cm⁻² and remaining immovable for >10 s when subjected to a force of 5 dyn cm⁻² (28). Rolling and arrest categories were normalized to the leukocyte flux (28), whereas migratory phenotypes were calculated as fractions of leukocytes originally accumulated during the first 40-s phase. Neutrophil platelet-activating factor receptor (PAF-R) was blocked with the specific antagonist WEB-2086 at 10 µM for 15min, and the antagonist was maintained throughout the assay. For selective interference with vesicle production and transport, HUVECs were treated with BreA (5 μg ml⁻¹; 1h) and extensively washed. Recorded videos were analyzed offline using ImageJ (NIH).

Fluorescence quantification and statistical analysis

Mean fluorescence intensity was quantified using ImageJ (NIH). All data are reported as mean values ± SEM, SD or range (as indicated). Statistical analysis was performed using Prism 5 software (GraphPad Software, Inc., La Jolla, CA, USA) using Student’s t-test. P value <0.05 was considered significant.

Results

HUVECs deposit HSPGs mainly at their basolateral side in both their resting and inflamed states

To characterize HSPGs expression and secretion by HUVECs, we cultured the cells as a non-confluent layer on the ECM protein, fibronectin (Fig. 1A). Within 16–18h of culture, resting HUVECs secreted large amounts of prototypic basement membrane components, including collagen IV, perlecan and HSPGs detected by the pan HS-specific IgM mAb, 10E4, which recognizes N-sulfated glucosamine residues of the polysaccharide moieties, shared by all known HS GAGs (31). A recent study suggested that this sulfation rather than O-sulfation of the HS uronic acid moiety confers high-affinity binding of chemokines (32). We next stained either intact or permeabilized confluent HUVEC monolayers with the anti-HS IgM mAb, because without permeabilization, the accessibility of the IgM to the basolateral aspects of the endothelial monolayer was negligible (Fig. 1B). Using this approach, we detected significantly higher HS staining within the basolateral aspects than within the apical aspects of resting or cytokine-stimulated HUVECs (Fig. 1B). The HSPGs pattern, distribution and level of expression detected by the 10E4 mAb were similar for both TNF-α- and IL-1β-stimulated cells (Fig. 1B). Interestingly, HSPGs were also highly enriched at the basolateral side of resting fibroblasts (Supplementary Figure 1, available at International Immunology Online), suggesting that this enrichment of HSPGs at the basement membrane of HUVECs is not specific to endothelial cells. To further investigate the properties of the apically expressed endothelial HSPGs, confocal z-stack analysis was next performed. As the intact endothelial monolayer restricted the penetration of the IgM antibody to the basolateral side, this approach allowed us to study HS content and distribution exclusively on the apical endothelial surface. Interestingly, cultured HUVECs retained a mesh-like HSPG network enriched within the paracellular endothelial junctions extending up to 2 µm above the apical surface of these ECs, detected by staining of the cortical actin cytoskeleton (Supplementary Figure 2A, available at International Immunology Online). Importantly, the levels of these apically expressed HSPGs were not reduced by strong inflammatory stimuli, nor was their distribution altered (Supplementary Figure 2B, available at International Immunology Online).

Fig. 1.

HSPGs are localized mainly to the basement membrane in both resting and inflamed ECs. HUVECs immunostained for HSPGs with monoclonal mouse IgM (red), or for perlecan (red), collagen IV (red) and actin (green). Nuclear staining is shown in blue. (A) Immunostaining of resting non-confluent cells seeded on fibronectin for 45min or for 16–18h. (B) Immunostaining of either resting or TNF-α- or IL1-β- stimulated monolayers grown for 16–18h on fibronectin. The mean fluorescence intensity values shown are the mean ± SEM of eight fields of view from three independent experiments. Images were taken with a Delta-Vision microscope at ×60 magnification. No fluorescence could be detected in the presence of the corresponding isotype control primary antibodies. The scale bars represent 15 µm. N.S., not significant. ***P < 0.0001 by Student’s t-test.

Ext1 inactivation results in HS elimination with conserved proteoglycan deposition

To determine the function of HSPGs in leukocyte recruitment by inflamed ECs, we infected HUVECs with a lentivirus encoding shRNA targeted to the Ext1 gene, an enzyme involved in early stages of chain elongation in HS biosynthesis (13). Importantly, Ext1-specific silencing in HUVECs resulted in substantial decrease in Ext1 mRNA expression (Fig. 2A), and elimination of all HS biosynthesis, but did not affect the ability of these ECs to form monolayers under either resting or inflammatory conditions (Fig. 2B). Importantly, Ext1 silencing did not alter the amount or global distribution of the main endothelial proteoglycan, perlecan, secreted by HUVECs into the ECM (Fig. 2B).

Fig. 2.

Deletion of HSPGs by Ext1 silencing. (A) Relative Ext1 mRNA expression of sh-control- and sh-Ext1-silenced HUVECs stimulated with IL-1β for 16–18h. Values are the mean ± SEM from two experiments. (B) Sh-control- or sh-Ext1-treated HUVECs were stimulated with TNF-α or IL-1β and then immunostained for HSPGs (red) and perlecan (green); nuclear staining is shown in blue. Images from one representative experiment out of three are shown. Images were taken as in Fig. 1. The scale bars represent 15 µm.

Abrogated endothelial HS biosynthesis has no effect on chemokine packaging or on the induction of key adhesion molecules

In order to assess whether abrogation of HS biosynthesis retains the expression of the key adhesion and chemotactic molecules produced by inflamed HUVECs, we next compared the expression of ICAM-1 and E-selectin, the two most essential ligands for neutrophil and effector T-cell arrest on inflamed HUVECs (17, 19), on cytokine-stimulated Ext1-silenced ECs and their sham counterparts. Notably, Ext1-silenced HUVECs could normally express both adhesion molecules after either TNF-α or IL-1β stimulation (Fig. 3A and data not shown). Furthermore, ICAM-1 and F-actin distributions in variably inflamed HUVECs were insensitive to Ext1 silencing (Fig. 3A). To evaluate the function of endothelial HS in the production and distribution of the key chemokines involved in either effector T-cell or neutrophil migration across inflamed ECs, we next assessed by immunostaining of permeabilized HUVECs the expression and distribution of the two prototypic chemokines, namely, CCL2 and CXCL1, produced by these ECs in response to IL-1β stimulation. Both the expression and packaging of neutrophil- and effector T-cell-specific prototypic chemokines inside vesicles were not significantly affected in the absence of HSPGs, both in IL-1β- and in TNF-α-stimulated ECs (Fig. 3B). Previously, we showed that disruption of intracellular chemokine vesicles markedly reduced the TEM of effector T cells (19). Thus, our immunostaining results suggest that HS biosynthesis by inflamed ECs is not required for their ability to produce and package inflammatory chemokines in vesicles. To validate that endothelial HS is also not required for the early phases of cytokine stimulation of these chemokines, nor that of key adhesion molecules in HUVEC stimulated by either TNF-α or IL-1β, we performed quantitative transcriptional analysis of the key inflammatory chemokines and adhesion molecules studied above, as well as of the key mononuclear adhesion receptor, vascular cell adhesion molecule (VCAM)-1. As expected, complete loss of HS in Ext1-silenced HUVECs did not impair the early phase transcription of these trafficking molecules induced by either TNF-α or IL-1β (Supplementary Figure 3A and Supplementary Data, available at International Immunology Online).

Fig. 3.

HSPG deletion in HUVEC does not significantly affect ICAM-1 expression or chemokine expression and packaging in vesicles. (A) TNF-α- or IL-1β-stimulated sh-control- or sh-Ext1-pre-treated HUVECs were immunostained for ICAM-1 (red) and actin (green); nuclear staining is shown in blue. The mean fluorescence intensity values shown are the mean ± SEM of 10 fields of view from 3 independent experiments. Images were taken at ×40 magnification. The scale bars represent 20 µm. N.S., not significant. (B) IL-1β-stimulated sh-control- or sh-Ext1-pre-treated HUVECs were immunostained for HSPG (red) and CCL2 or CXCL1 (green); nuclear staining is shown in blue. The mean fluorescence intensity values shown are the mean ± SEM of 10 fields of view from 3 independent experiments. Images were taken using a Delta-Vision microscope at ×60 magnification. The scale bars represent 15 µm. N.S., not significant.

Endothelial HS elimination partially impairs neutrophil arrest on inflamed HUVECs

In light of the unimpaired ability of Ext1-silenced and cytokine-activated HUVECs to express the major inflammatory chemokines and adhesive molecules on the cell surface and inside cytoplasmic vesicles (Fig. 3A and B), we next assessed the effects of Ext1 silencing on the adhesive capacity of either TNF-α- or IL-1β-stimulated HUVECs. Despite their overall normal expression of adhesion receptors and chemokines, both TNF-α- and IL-1β-stimulated Ext1-silenced HUVECs sub-optimally supported neutrophil arrest under physiological shear stress conditions (Fig. 4A and B). In contrast, elimination of all chemokine vesicle packaging by BreA pre-treatment of inflamed HUVECs (Supplementary Figure 4A, available at International Immunology Online) totally inhibited the ability of inflamed HUVECs to arrest rolling neutrophils (Fig. 4C). This BreA pre-treatment was not toxic, because these ECs could promote normal neutrophil arrest and TEM if overlaid with exogenous chemoattractant (Supplementary Figure 4C, available at International Immunology Online). We also verified that this total elimination of neutrophil arrest was not the result of reduced endothelial expression of ICAM-1 or impairment in the cortical endothelial actin cytoskeleton (Supplementary Figure 4B, available at International Immunology Online). Notably, previous results from our laboratory have indicated this identical BreA pre-treatment of inflamed HUVECs to have no effect on effector T-cell arrest or crawling, consistent with the ability of the highly adhesive integrins expressed by these leukocytes to bypass chemokine signals (19). These results therefore collectively suggest that in the absence of normal HSPG expression, but with otherwise normal expression of adhesion molecules and intracellular chemokines, inflamed ECs are defective in their ability to arrest rolling neutrophils, a process mediated by both endothelium-presented HS-binding chemokines and lipid chemoattractants. Nonetheless, our immunostaining methods were not sensitive enough to detect sufficient levels of CXCL1 immobilized by either the apical or basolateral endothelial compartments on normal HS-expressing HUVECs (data not shown).

Fig. 4.

The effect of Ext1 silencing or BreA pre-treatment on neutrophil adhesion to inflamed HUVECs. Neutrophils were perfused on either sh-control- or sh-Ext1-pre-treated HUVEC monolayers stimulated for 16–18h with (A) TNF-α or (B and C) IL-1β, and the frequencies of their attachments to the different monolayers were analyzed. Where indicated, cytokine-stimulated HUVECs were pre-treated with BreA for 60min before neutrophil perfusion. Values shown represent the mean ± range of two fields. One representative experiment of six independent experiments is shown. *P < 0.05; **P < 0.01 by Student’s t-test. The difference in total percentage of adhesive events was not significant (N.S.).

Neutrophil TEM is markedly dependent on endothelial HS

In light of the much higher levels of HSPGs at the basolateral endothelial compartment than at the apical surface (Fig. 1), we reasoned that this pool of endothelium-associated HS may be necessary for neutrophil TEM under shear flow conditions. High HSPG content in this compartment as well as in the perijunctional endothelial compartments (Supplementary Figure 2B, available at International Immunology Online) could facilitate chemokine presentation to neutrophils that send protrusions through these junctions (17). Indeed, abrogation of all endothelial HS decreased neutrophil TEM by 3-fold in TNF-α- and IL-1β-stimulated HUVECs (Fig. 5A and B; and Supplementary Movies 1 and Supplementary Data, available at International Immunology Online). Importantly, this strong inhibitory effect was additive to the effect of HSPG abrogation on initial neutrophil arrest because TEM efficiencies were determined only for neutrophils that were stably arrested during the initial accumulation phase (Fig. 4A and B). Collectively, although endothelium-produced HS-binding chemokines such as CXCL1 were undetectable within the basolateral compartments of normal inflamed HUVECs with high HSPG content, these results suggest a critical role for Ext1-dependent endothelial HS expression for the functional presentation of these chemokines to transmigrating neutrophils.

Fig. 5.

Neutrophils require intact HSPG expression for TEM under shear flow. Neutrophils were perfused on sh-control- or sh-Ext1-pre-treated HUVECs stimulated for 16–18h with (A) TNF-α or (B) IL-1β, and the migratory phenotype of accumulated neutrophils was analyzed as described in Methods. Values are the mean ± SD in three fields. One representative experiment of four independent experiments is shown. ***P < 0.001 by Student’s t-test.

Adhesion and TEM of effector T cells are HS independent

The TEM of effector lymphocytes across either TNF-α- or IL-1β-stimulated HUVECs depends primarily on the chemokine CCL2, with a small contribution of CCL5 (19), all of which were reported to bind the HS moieties of HSPGs with high affinity (14). We therefore next determined the role of endothelial HS in directing the adhesion and TEM of these T lymphocytes over either sham- or Ext1-silenced HUVECs stimulated with TNF-α or IL-1β. Ext1-silenced HUVECs supported normal effector lymphocyte attachment, rolling, arrest and adhesion strengthening, (Fig. 6A and B), as well as shear-resistant crawling followed by successful TEM (Fig. 6C and D). These results support the notion that the arrest, adhesion, strengthening and crawling of these effector T cells on inflamed HUVECs, processes that depend on integrin activity on these T cells but not on chemokine signaling, do not require apical or basolateral endothelial HSPGs. Nevertheless, the TEM of effector T cells across variably inflamed ECs, although strictly dependent on chemokine activities within T cell–endothelial synapses (19), did not seem to require any endothelial HSPGs. Thus, HS-mediated presentation of endothelium-produced chemokines on the extracellular aspects of the endothelial barrier, although essential for neutrophil TEM, is not implicated in lymphocyte TEM across identically inflamed barriers under identical flow conditions.

Fig. 6.

Effector T cells do not require endothelial HSPG to arrest or transmigrate through inflamed ECs under shear flow. Effector T cells were perfused on either sh-control- or sh-Ext1-pre-treated HUVECs subjected to (A and C) TNF-α or (B and D) IL-1β stimulation for 16–18h. The frequencies of T-cell attachments to the endothelial monolayers, and the migratory phenotype of accumulated T cells, were analyzed as described in Methods . In A and B, the values shown are the mean ± range of two fields. In C and D, the values shown are the mean ± SD of three fields. One representative experiment of four to six independent experiments is shown.

Discussion

Chemokine immobilization on vascular ECs and endothelium-adherent platelets was suggested to be critical not only to prevent their dilution by blood flow but also to facilitate localized signaling to integrins and actin-remodeling GTPases on recruited leukocytes (18, 25, 33–35). Most chemokines share a C-terminal stretch of positively charged residues that recognize HSPGs with moderate affinities (11, 36). To address the function of endothelial HSPGs in leukocyte adhesion and TEM, we used Ext1-silenced HUVECs, which could no longer produce HS GAGs in spite of normal production of proteoglycan core proteins. Importantly, these cells formed normal monolayers, and in response to either TNF-α or IL-1β stimulation, normally up-regulated all adhesion molecules and inflammatory chemokines critical for leukocyte TEM. We demonstrated that although endothelium-expressed or -deposited HS GAGs are required for normal neutrophil TEM across these inflamed ECs, these GAGs are not required for the adhesion or TEM of effector T cells across the same inflamed endothelial barriers. The TEM of effector lymphocytes in this experimental system depended primarily on CCL2, with a small contribution from CCR1- and CCR5-binding chemokines (19), all of which avidly bind HSPGs (11, 14). Thus, the HSPG independence of effector T-cell TEM that we describe here is consistent with the conclusion that intracellular pools of inflammatory endothelial chemokines essential for TEM, rather than extracellular HSPG-deposited forms of these chemokines, can promote effector lymphocyte TEM.

Although HUVECs have been claimed to express a thinner glycocalyx in culture (37), we found that these cells in fact retain a mesh of HS GAGs on their apical side. Notably, the distribution of these HS GAGs on the apical endothelial surfaces was not modulated by inflammatory activation, in contrast to a recent study that indicated that TNF-α signals shed HSPGs from lung capillary ECs (38). This report suggested, however, that cremasteric post-capillary venules do not shed their HSPGs in response to similar signals, consistent with our in vitro findings in the present study.

Bao et al. provided compelling in vivo evidence that HSPGs are indispensable for immobilization and function of major lymph node chemokines required for lymphocyte adhesion to and crossing through blood and lymphatic vessels (21). In addition, HSPGs were found necessary for the endothelial presentation of CCL2, an inflammatory chemokine that is transported via lymphatics into inflamed draining lymph nodes and is transcytosed to the apical aspects of lymph node HEVs (21, 39). In spite of these findings, and the versatile contributions of HSPGs to different modalities of chemokine functions, it is still possible that many adhesive and extravasation processes of both neutrophils and T cells involve HS-independent but chemoattractant-dependent activities. For instance, several key lipid chemoattractants like leukotrienes, PAF, 2-AG and LPA are predicted to immobilize directly on the endothelial surfaces on which they are presented and are therefore expected to trigger both leukocyte integrin-mediated arrest, crawling and TEM independently of HS GAGs (40–44). Consistent with this notion, we found that silencing Ext1 in inflamed HUVECs inhibited neutrophil TEM to a greater extent than integrin-mediated arrest. Thus, although neutrophil TEM depends on endothelial HSPG, neutrophil adhesion to inflamed ECs may involve one or more lipid chemoattractants, as well as other integrin activation co-stimulatory molecules (5) that do not require apical endothelial HSPG.

Intra-endothelial HSPGs could also stabilize de novo–synthesized inflammatory chemokines and facilitate their intracellular packaging in storage granules, as reported for some leukocyte and endothelial proteoglycans (13, 45). However, HSPG deficiency in inflamed HUVECs did not affect the production or distribution of the main inflammatory chemokines synthesized by these ECs. Ext1 deficiency also did not affect the biosynthesis of another major GAG produced by these ECs, chondroitin sulfate (data not shown). Thus, our results indicate that HSPGs are not essential for the packaging of intra-endothelial chemokines inside endothelial vesicles or for their consumption by transmigrating effector T lymphocytes. Although highly enriched at the basolateral endothelial aspects and on perijunctional apical aspects of inflamed ECs, HSPGs were not required for trapping the vesicle-secreted chemokines within the synapses formed by the leading edge of transmigrating effector lymphocytes. On the other hand, HSPGs are essential for neutrophil TEM, probably because this TEM is much more dependent on inflammatory chemokines trapped in the dense basolateral endothelial pool of HSPGs. Such HSPG-dependent basolateral presentation of endothelial chemokines to GPCRs on neutrophil protrusions and on the leading edge of transmigrating neutrophils could explain the highly directional motility of these leukocytes from the apical to the basolateral compartments of the inflamed endothelial barrier. In addition, HSPG deficiency within inflamed ECs could also perturb neutrophil TEM by changes in EC morphology via an autocrine mechanism that involves the binding of endothelial secreted CXCL8 to endothelial GPCRs (46).

Taken together, our studies suggest that effector T cells can respond to inflammatory endothelial chemokines critical for their TEM in the absence of endothelial HSPGs. Our results also suggest that the highly enriched basolateral layer of endothelial HSPGs contributes to neutrophil emigration across the same inflamed cell barriers, possibly due to their ability to trap and present TEM-promoting HS-binding chemokines to transmigrating neutrophils. The functional significance of a highly enriched layer of HSPG underneath ECs is consistent with our recent quantification of HS distribution around post-capillary venules in different tissues, which indicate a much higher HS content at the basement membrane of these venules (47). Our in vitro results, however, do not resolve the in vivo importance of endothelial HSPG for neutrophil emigration across different types of inflamed post-capillary venules. Notably, HSPG biosynthesis around post-capillary venules is driven by multiple cell types in addition to ECs. For instance, mural cells associated with post-capillary venules may secrete their own HSPGs into the basement membrane of the ECs with which they associate (48, 49). Thus, total elimination of HS biosynthesis in the basolateral compartments of peripheral post-capillary venules would require simultaneous temporal Ext1 silencing in both ECs and pericytes. Such multicellular Ext1 silencing is currently not feasible, and so the importance of HS presentation of chemokines to neutrophil diapedesis in vivo is difficult to assess. Our recent results suggested that a subset of chronically inflamed skin blood vessels secrete much higher levels of HSPGs than their resting counterparts, reflecting the possibility of enhanced co-secretion of HSPGs by both endothelial and mural cells (47). Future studies will therefore be required to substantiate the contribution of endothelium- and pericyte-synthesized HSPGs for chemokine presentation and function in neutrophil emigration across post-capillary venules of differently inflamed tissues. These studies may provide new mechanistic insights regarding the mechanisms by which different neutrophil chemoattractants drive directional neutrophil migration from luminal to abluminal aspects of inflamed blood vessels. Such insights are necessary for the design and application of novel inhibitors of chemokine binding to HS GAGs for use in anti-inflammatory therapies (16, 50).

Funding

Israel Science Foundation (7112970101) to R.A.; Flight Attendant Medical Research Institute Foundation (7034640908) to R.A.; and Minerva Foundation, Germany (7108230301) to R.A.

Acknowledgements

We thank Dr Sara Feigelson for helpful discussions and Dr S. Schwarzbaum for editorial assistance. R.A. is an incumbent of the Linda Jacobs Chair in Immune and Stem Cell Research.