CCL25 and CCL28 promote α4β7-integrin-dependent adhesion of lymphocytes to MAdCAM-1 under shear flow

Alice Miles, Evaggelia Liaskou, Bertus Eksteen, Patricia F. Lalor, David H. Adams


Inflammatory bowel disease is characterized by the recruitment of lymphocytes to the gut via mucosal vessels. Chemokines are believed to trigger α4β1- and α4β7-integrin-mediated adhesion to vascular cell adhesion molecule-1 (VCAM-1) and mucosal addressin cell adhesion molecule-1 (MAdCAM-1) on mucosal vessels, although the contribution of each pathway and the chemokines involved are not well characterized. These interactions occur under conditions of hemodynamic shear, which is critical in determining how lymphocytes integrate chemokine signals to promote transmigration. To define the role of specific chemokines in mediating lymphocyte adhesion to VCAM-1 and MAdCAM-1, we studied the ability of immobilized chemokines to activate adhesion of human lymphocytes in a flow-based adhesion assay. Adhesion to immobilized MAdCAM-1 was α4β7 dependent, with no contribution from α4β1, whereas α4β1 mediated rolling and static adhesion on VCAM-1. Immobilized CC-chemokine ligand (CCL) 25 and CCL28 were both able to trigger α4β7-dependent lymphocyte arrest on MAdCAM-1 under shear, highlighting a potential role for these chemokines in the arrest of lymphocytes on postcapillary venules in the gut. Neither had any effect on adhesion to VCAM-1, suggesting that they selectively trigger α4β7-mediated adhesion. Immobilized CCL21, CCL25, CCL28, and CXC-chemokine ligand (CXCL) 12 all converted rolling adhesion to static arrest on MAdCAM-1 by activating lymphocyte integrins, but only CCL21 and CXCL12 also triggered a motile phenotype characterized by lamelipodia and uropod formation. Thus α4β1/VCAM-1 and α4β7/MAdCAM-1 operate independently to support lymphocyte adhesion from flow, and chemokines may act in concert with one chemokine triggering integrin-mediated arrest and a second chemokine promoting motility and transendothelial migration.

  • chemokines
  • integrins
  • T cells
  • mucosal immunity
  • inflammatory bowel disease

the chronic inflammatory bowel diseases (IBD), Crohn's disease and ulcerative colitis, are characterized by uncontrolled lymphocyte recruitment to the gut as a consequence of dysregulated lymphocyte adhesion and migration through mesenteric vessels (48, 58). Lymphocyte homing to the gut involves several adhesion molecules and chemokines, some of which display tissue specificity (1, 14). α4β7-Integrins are induced on T cells and B cells during activation by dendritic cells in gut-associated lymphoid tissues (GALT) (37, 51, 52, 69) and allow lymphocytes to bind mucosal addressin cell adhesion molecule-1 (MAdCAM-1), constitutively expressed on high-endothelial venules (HEV) in Peyer's patches and mesenteric lymph nodes and on postcapillary venules in the lamina propria (11). MAdCAM-1 contains a mucin domain that supports L-selectin-mediated rolling adhesion (5, 12, 47, 66) and two immunoglobin (Ig)-like domains that support α4β7-mediated rolling and firm adhesion in an L-selectin-independent manner (6). Naive lymphocytes express modest levels of α4β7, whereas effector/memory cells can be subdivided into α4β7high and α4β7low subsets, with the α4β7high populations preferentially binding to MAdCAM-1 (24, 62, 65). Under physiological conditions, α4β7+ lymphocytes preferentially migrate to the gut, whereas α4β1+ lymphocytes migrate to extraintestinal tissues. In Peyer's patches, initial capture from blood flow is mediated by L-selectin, and subsequent α4β7-mediated adhesion is required to convert rolling to arrest (4, 70). In contrast, homing of effector and memory lymphocytes to the lamina propria does not require L-selectin and is dependent on interactions between MAdCAM-1 and α4β7 (26, 31). Therefore, MAdCAM-1 is important in the physiological trafficking of both naive lymphocytes and memory/effector lymphocytes to the gut and associated lymphoid tissues.

MAdCAM-1 is aberrantly expressed in the liver of patients with liver disease complicating IBD (29), and expression is increased in the chronically inflamed gut in IBD (11), suggesting its involvement in disease pathogenesis. However, in IBD, the expression of vascular cell adhesion molecule-1 (VCAM-1), the endothelial ligand for α4β1, is also increased in the inflamed gut, where it supports lymphocyte recruitment both to the small bowel in models of Crohn's disease and to the colon in colitis (13, 59, 60, 67). Thus both α4β1 and α4β7 are important for effector cell recruitment to the inflamed gut, and both anti-α4 and anti-α4β7 antibodies have shown clinical efficacy in the treatment of Crohn's disease and ulcerative colitis (25, 27).

Chemokine receptors regulate effector T-cell trafficking to the gut by activating integrin-mediated adhesion and promoting lymphocyte motility (1). In humans, CC-chemokine receptor (CCR) 9 is selectively expressed by a proportion of circulating α4β7+ T cells and B cells, by most T cells in the small intestine, and a subset of T cells in the colon. The ligand for CCR9, CC-chemokine ligand (CCL) 25, is selectively expressed by epithelial cells in the small intestine, and the protein can be detected on mesenteric endothelium (35, 40, 55, 77, 78). The importance of CCL25 in lymphocyte homing to the gut is demonstrated in animal models, where blockade of CCL25 or CCR9 inhibits T- and B-lymphocyte recruitment to the small intestine (10, 35). Three other chemokines, which are not tissue specific, are also implicated in lymphocyte recruitment to the gut. CCL28 is predominantly expressed by epithelial cells within the colon, as well as at lower levels in the small intestine and other mucosal sites (54, 74). Its receptor CCR10 is expressed on IgA-secreting B cells in the small and large intestine and plays an important role in localization of these cells within the gut (41). No role for CCL28 in T-cell homing to the gut has yet been demonstrated, although it has been implicated in the recruitment of regulatory T cells to the inflamed liver (23). CCL21 is found at high levels in GALT (30), where it interacts with its receptor CCR7 on naive T cells (15) to activate β2-integrin-mediated arrest on HEV (71, 75). CCL21 can also activate α4β7-dependent adhesion to MAdCAM-1 in vitro (53, 73). The chemokine CXC-chemokine ligand (CXCL) 12 is constitutively expressed in secondary lymphoid tissues (8) and in gut epithelium, and its receptor CXC-chemokine receptor (CXCR) 4 is expressed at high levels on circulating lymphocytes (34). CXCL12 can activate α4β7-dependent adhesion to MAdCAM-1 in vitro (76), suggesting a potential role in triggering lymphocyte adhesion in Peyer's patches.

Thus both α4β1- and α4β7-integrins are implicated in the pathogenesis of IBD, and several chemokines could be involved in triggering integrin activation and lymphocyte recruitment (49). We compared the ability of specific chemokines to activate α4β1- and α4β7-integrins and stimulate human lymphocyte migration. The conditions imposed by blood flow require integrin activation to be a rapid event, as rolling cells only have a short time to interact with endothelial ligands (46). To model this in vitro, we compared the effects of the chemokines CCL21, CCL25, CCL28, and CXCL12 immobilized in glass capillaries on lymphocyte adhesion to coimmobilized MAdCAM-1 or VCAM-1 under conditions of physiological flow. We then extended the studies by using primary human hepatic endothelial cells transfected to express MAdCAM-1 as a more physiological substrate.


Isolation of peripheral blood lymphocytes.

Peripheral venous blood from normal donors or patients with IBD was collected into EDTA (BD Biosciences, Oxford, UK). Mononuclear cells were isolated by density centrifugation over Lymphoprep (Sigma Aldrich), washed, and suspended in RPMI-1640 (Invitrogen), supplemented with 60 μg/ml benzylpenicillin, 100 μg/ml streptomycin, 2 mM glutamine (all from Sigma Aldrich), and 10% (vol/vol) FCS (Invitrogen). Monocytes were depleted by incubation on tissue culture plastic for 30 min. The lymphocyte-rich supernatant was then transferred to a culture flask and rested overnight at 37°C before use in the flow-based adhesion assay or for FACS analysis.

Induction of MAdCAM-1 expression in human hepatic sinusoidal endothelium using adenoviral vectors.

An adenovirus-mediated gene transfer strategy was used to induce expression of MAdCAM-1 in primary hepatic sinusoidal endothelial cells (HSEC). Splice-variant MadCAM-1 and full-length (FL) MAdCAM-1 cDNA, which had been cloned into the pCDNA3.1 vector, were a kind gift of Dr. Nick Pullen, Pfizer, UK. Recombinant adenoviruses containing splice-variant-MAdCAM-1 and FL-MAdCAM-1 were produced using the AdEasy Adenoviral Vector System (Stratagene), according to manufacturer's instructions. Methodology incorporated use of the pShuttle-IRES-hrGFP-1 vector, allowing successful adenoviral infection of cells to be monitored by green fluorescent protein (GFP) expression. Blank vectors lacking the MAdCAM-1 constructs were used as controls. Pilot experiments confirmed the amount of virus required for optimal infection of primary cells [FL-MAd-GFP-Ad, multiplicity of infection (MOI) of 10, GFP-Ad MOI of 50, data not shown], with maximal GFP expression being observed 24–72 h postinfection (data not shown).

Lymphocyte adhesion to recombinant MAdCAM-1 and VCAM-1 or HSEC under flow.

A flow-based adhesion assay was used as previously described (43). Briefly, glass microslides with rectangular cross section and good optical qualities were precoated with 3-aminopropyltriethoxysilane (Sigma Aldrich) to facilitate the binding of protein to the glass. Microslides were incubated with optimal concentrations of recombinant human MAdCAM-1 protein (gift of Dr. N. Pullen, Pfizer UK) or recombinant human VCAM-1 (R&D Systems) at 5 μg/ml in PBS for 90 min at 37°C. Slides were then washed with PBS to remove unbound protein and blocked with 0.5% (vol/vol) BSA for at least 20 min before perfusion of lymphocytes.

Purified recombinant human chemokines CCL21, CCL25, CCL28, CCL11, and CXCL12 (all PeproTech) were coimmobilized on microslides with either MAdCAM-1 or VCAM-1 protein. Microslides were first coated with the appropriate adhesion molecule as above and then incubated with a predetermined optimal concentration of chemokine (1 μg/ml in PBS) for 120 min at 37°C. Microslides were washed and blocked with BSA before perfusion of lymphocytes.

Adhesion of lymphocytes to MAdCAM-1 expressed by hepatic endothelium was also measured using the flow-based adhesion assay. Endothelial cells were cultured in microslides, as previously described (45), and infected with recombinant adenoviruses containing FL-MAdCAM-1 or empty vector. Flow assays were performed 24–48 h after infection, and, on occasion, human chemokines were immobilized at predetermined optimum concentrations on the HSEC monolayers for 30 min before lymphocyte perfusion (20).

For all adhesive substrates, peripheral blood lymphocytes (PBL) were suspended in RPMI-1640 supplemented with 0.1% BSA and perfused over the immobilized protein at a constant shear stress of 0.05 Pa, which approximates the range found in postcapillary venules and hepatic sinusoids in vivo (38, 44). PBL were perfused for a period of 5 min, followed by perfusion with wash buffer for 1 min before analysis of lymphocyte adhesion. Lymphocytes captured via molecular interactions with the immobilized or cell-borne protein were visualized by phase-contrast microscopy (×100 magnification). Video clips were recorded of a minimum of 10 fields along the length of the microslide, and experiments were analyzed offline.

Analysis of lymphocyte adhesion.

Total cell adhesion was determined by counting all visible adherent cells in a number of fields of known dimension, which was then converted to give a value of adherent cells per millimeter squared and normalized according to the number of lymphocytes perfused (i.e., adherent cells/mm2 per 106 perfused). Adherent cells were classified as rolling or stationary, and rolling and static adhesion were expressed either as adherent cells per millimeter squared per 106 perfused or percentage of total adhesion, as indicated. For chemokine studies, static adhesion was further classified into subcategories: “phase-bright static” cells, which were phase bright, round cells, and “activated cells,” which had become activated and adopted an extended or flattened morphology and which had spread out on the adhesive surface (see Fig. 2). In the endothelial studies, adherent cells were again visualized microscopically under phase contrast and recorded for offline analysis. The number of adherent cells was converted to adherent cells per millimeter squared, corrected for the number of cells perfused. The pattern of adhesion was analyzed to determine the number of cells rolling, statically adherent, or transmigrated. Transmigrated cells are easily distinguished from cells migrating on the surface of the endothelial cell, as the former are phase dark and the latter phase bright.

Adhesion molecule blockade.

Adhesion-blocking antibodies were used against α4-integrin (HP2/1), L-selectin (Dreg 56) (both from Immunotech), β1-integrin (Ab216; Abcam), MAdCAM-1 (P1, kind gift of Dr. Nick Pullen, Pfizer UK), and β7-integrin (FIB504; BD Biosciences). PBL or endothelial cells were preincubated with optimal concentrations of these blocking antibodies (or an isotype-matched control antibody) for 20 min at 37°C, before perfusion over microslides coated with adhesion molecule.

Pertussis toxin treatment.

To inhibit chemokine-mediated signaling, PBL were incubated with 100 ng/ml of the Gα-protein inhibitor pertussis toxin (Sigma Aldrich) for 1 h at 37°C before perfusion.

Flow cytometry and Western blotting.

Four-color flow cytometry was used to examine the expression of chemokine receptors and integrins on lymphocytes. PBL were suspended in PBS-0.1% BSA, 0.5 mM MgCl2, and 1 mM CaCl2, and incubated with a blocking mixture of mouse, rabbit, and goat immunoglobulins (500 μg/ml) before being incubated with 60 μg/ml primary unconjugated MAb against α4β7 (ACT-1; Millenium) for 30 min at 4°C, washed in PBS, and then labeled with rabbit anti-mouse RPE-Cy5 (DAKO; 1:20 dilution), along with fluorochrome-labeled primary antibodies α4-FITC antibody (Serotec; 1:50 dilution) and CD3-RPE-Cy7 antibody (BD Biosciences, 1:20 dilution) for 30 min at 4°C, protected from the light. Chemokine receptor expression was detected using CCR7-phycoerythrin (PE), CCR9-PE, or CXCR4-PE (R&D Systems; 1:20 dilution), or with unconjugated goat polyclonal CCR10 (20 μg/ml), followed by donkey anti-goat-PE (1:50 dilution; both from AbCam). In all cases, control samples were labeled with isotype-matched control antibodies. To confirm expression of MAdCAM-1 by HSEC infected with adenoviruses, infected and control cells were trypsinized, washed, and resuspended in PBS. Cells were analyzed for expression of GFP by flow cytometry. All cells (PBL and HSEC) were analyzed on a Coulter Epics XL flow cytometer (Coulter Electronics) using Summit software (Cytomation). PBL populations were gated on, according to size and granularity and expression of CD3, and histograms produced from staining with isotype-matched control antibodies were used to assign bar regions for subsequent analysis of positive staining. GFP expression on HSEC was compared with signals from uninfected cells. Expression of MAdCAM-1 by HSEC was also confirmed by Western blotting. Protein lysates were generated from transfected HSEC (and Chinese hamster ovary cells) grown in six-well plates, according to standard methodology, and stored at −20°C until use. Protein samples were separated by electrophoresis on an 8% SDS-acrylamide gel and transferred from electrophoreses gels onto a Hybond nitrocellulose membrane (Amersham Biosciences). Bands were visualized with anti-MAdCAM-1 primary antibody (Pfizer), horseradish peroxidase-conjugated secondary antibody, and enhanced chemiluminescence reagents (Amersham Pharmacia Biotech). Enhanced chemiluminescence detection film (Amersham Biosciences) was exposed to the membrane and developed using a Kodak X-OMAT 1000 processor.


Chemokine receptor expression in α4β7+ PBL populations.

Before the functional assays, the expression of α4-integrins, CXCR4, CCR7, CCR9, and CCR10, was determined on blood lymphocytes (Fig. 1). α4-Integrin was expressed by ∼50% of peripheral blood CD3+ lymphocytes, and of these 15–20% were also positive for β7-integrin (Fig. 1). Comparison of chemokine receptor expression on CD3+ T cells and those subdivided into α4β7 positive and negative populations revealed that CXCR4 was expressed by 78% of CD3+ PBL and 74 and 75%, respectively, of the α4β7 and α4β7+ populations. Median channel fluorescence did not differ significantly between any population. Levels of CCR7 were reduced on α4β7 cells compared with the whole CD3+ population and the α4β7+ population (P < 0.001). CCR9 expression was increased on the α4β7+ population, but no significant differences were seen between CCR10 expression in either percentage of positive or median channel fluorescence values on the α4β7+ and α4β7 populations (Fig. 1).

Fig. 1.

Expression of CXC-chemokine receptor (CXCR) 4, CC-chemokine receptor (CCR) 7, CCR9, and CCR10 on circulating α4+ and α4β7+ T-cell populations. A, left: representative flow cytometry dot plots showing the coexpression of α4- and β7-integrins (x-axis) on CD3+ (y-axis) T cells from a single donor. Values in quadrant represent percent dual-positive cells and median channel fluorescent (MCF) staining intensity. A, right: pooled data from 4 normal donors showing the percentage of CD3+ cells staining positive for α4-, β7-, and α4β7-integrin. B: pooled data from 4 normal donors showing percent positive staining and MCF values for chemokine receptor staining on total peripheral blood CD3+ cells and cells gated into α4β7 negative and positive cells. Values are means ± SE.

MAdCAM-1 and VCAM-1 support lymphocyte rolling and firm adhesion.

Because MAdCAM-1 and VCAM-1 are both expressed in the inflamed gut, we compared their ability to support the adhesion of human PBL under conditions of physiological shear stress. Adhesion was classified as rolling adhesion, static adhesion, and “activated cells”, as described in materials and methods (Fig. 2A). Both adhesion molecules were able to support rolling and static adhesion in a concentration-dependent and shear-stress-dependent manner (data not shown). A significantly higher level of total lymphocyte adhesion was observed on VCAM-1 compared with MAdCAM-1 (Fig. 2B), but the total number of rolling cells was similar on both (Fig. 2C). The amount of static adhesion on VCAM-1 was significantly higher than that observed on MAdCAM-1 (Fig. 2D), and thus the relative proportion of static vs. rolling adhesion was greater on VCAM-1. This suggests that VCAM-1 supports a greater degree of spontaneous arrest of circulating PBL than MAdCAM-1 under physiological flow.

Fig. 2.

Comparison of lymphocyte adhesion to purified mucosal addressin cell adhesion molecule-1 (MAdCAM-1) and vascular cell adhesion molecule-1 (VCAM-1). A: representative image of appearance of lymphocytes adherent to MAdCAM-1 showing cells exhibiting “static” adhesion (a) and those that have become “activated” and remain phase bright (b) or have become phase dark (c). Peripheral blood lymphocytes (PBL) were perfused over either MAdCAM-1 or VCAM-1 protein (each at 5 μg/ml), at a constant shear stress of 0.05 Pa, and total adhesion (B), rolling adhesion (C), and static adhesion (D) were quantified. Values are expressed as adherent cells/mm2 per 106 perfused and are means ± SE of 3 separate experiments using different lymphocyte donors. Student's t-test indicated a significant difference between adhesion behavior on MAdCAM-1 and VCAM-1: *P < 0.05; **P < 0.005.

Lymphocyte adhesion to MAdCAM-1 is α4β7 dependent, whereas adhesion to VCAM-1 is α4β1 dependent.

To determine the receptors mediating adhesion to MAdCAM-1 and VCAM-1 under flow, PBL were pretreated with blocking antibodies against the α4-, β7-, and β1-integrin subunits and L-selectin (for adhesion to MAdCAM-1 only), or an isotype-matched control antibody before perfusion in the flow assay. Blockade of lymphocyte L-selectin had no significant effect on rolling or static adhesion on MAdCAM-1, because the recombinant MAdCAM-1 we used is not glycosylated and thus lacks L-selectin ligands (Fig. 3A). This allowed us to study integrin-mediated adhesion without any interference from L-selectin. Blockade of α4-integrin completely abolished rolling and reduced static adhesion on MAdCAM-1. Treatment with β7 antibodies also reduced rolling and static adhesion to minimal levels, whereas antibody blockade of β1-integrin had no effect on either rolling or static adhesion (Fig. 3A). In contrast, antibody blockade of lymphocyte α4-integrin abolished rolling on VCAM-1 and reduced static adhesion, whereas anti-β7-integrin had no significant effect on rolling or static adhesion (Fig. 3B). Blockade of the β1-integrin subunit significantly reduced both rolling and static adhesion on VCAM-1.

Fig. 3.

The effect of lymphocyte adhesion molecule blockade on lymphocyte adhesion to MAdCAM-1 (A) and VCAM-1 (B). PBL were pretreated with blocking antibodies targeting L-selectin (L-sel), α4-, β7-, or β1-integrin subunits, as indicated, before being perfused over immobilized MAdCAM-1 or VCAM-1 protein (5 μg/ml) at a constant shear stress of 0.05 Pa. Rolling adhesion and static adhesion were quantified. Values are expressed as adherent cells/mm2 per 106 perfused and are means ± SE of 5 separate experiments using different lymphocyte donors. Student's t-test indicated a significant difference between adhesion behavior in the absence and presence of antibody blockade: *P < 0.05; **P < 0.01.

Chemokine-mediated triggering of adhesion to MAdCAM-1 and VCAM-1.

To compare the ability of chemokines to trigger lymphocyte adhesion under flow, we coimmobilized either VCAM-1 or MAdCAM-1 on glass microslides with the chemokines CCL21, CCL25, CCL28, and CXCL12. This provides a simple model of the vessel wall in vivo, where chemokines are presented on the endothelial glycocalyx (72). The optimal doses of chemokines required to maximally stimulate lymphocyte binding was 1 μg/ml. After arresting on MAdCAM-1 or VCAM-1, some lymphocytes underwent shape change with the development of an “activated,” flattened morphology and appeared phase dark (Fig. 2A). These changes occurred rapidly after lymphocyte perfusion within minutes of initial exposure to chemokine. Figure 4 shows that coimmobilization of MAdCAM-1 with CCL21, CCL25, CCL28, or CXCL12 led to a significant decrease in the numbers of lymphocytes rolling, with a corresponding increase in static adhesion, indicating that all four chemokines can trigger the arrest of rolling lymphocytes on MAdCAM-1 (Fig. 4, A and B). A hierarchy of effects was observed so that CXCL12 had the most potent effect in reducing rolling, followed by CCL21, with CCL25 and CCL28 having comparable effects. Increases in static adhesion on MAdCAM-1 were observed with CCL28, CCL25, and CCL21 (Fig. 4B), again reflecting the triggering of lymphocyte arrest by these chemokines. CXCL12 had no effect on the level of static adhesion on MAdCAM-1, despite the fact that this chemokine had the most potent effects on lymphocyte rolling. However, >50% of cells exhibited activated, shape-changed morphology on MAdCAM-1 in the presence of CXCL12, suggesting that this chemokine drives migration and motility, whereas the others trigger stable adhesion (Fig. 4C). This is supported by the experiments using VCAM-1, where a similar increase in “activation” in the presence of CXCL12 was observed (Fig. 5C). Coimmobilization of CCL21, CCL25, or CCL28 with VCAM-1 did not decrease rolling or increase static lymphocyte adhesion (Fig. 5, A and B), suggesting that these chemokines specifically trigger adhesion to MAdCAM-1. We immobilized CCL11 as a negative chemokine control and showed it had no effect on lymphocyte, rolling, arrest, or shape change on MAdCAM-1 or VCAM-1 (data not shown), indicating that the effects of CCL21, CCL25, CCL28, and CXCL12 are chemokine specific.

Fig. 4.

The effects of CC-chemokine ligand (CCL) 21, CCL25, CCL28, and CXC-chemokine ligand (CXCL) 12 on lymphocyte adhesion to immobilized MAdCAM-1. The proportion of lymphocytes show rolling adhesion (A), static adhesion (B), or appear activated (C) on immobilized MAdCAM-1 (5 μg/ml), in response to the chemokines CCL21, CCL25, CCL28, and CXCL12 (all at 1 μg/ml). Values are expressed as percentage of total adhesion and are means ± SE of 4 separate experiments using different lymphocyte donors. Student's t-test indicated a significant difference between adhesion behavior in the presence of recombinant chemokine compared with adhesion to MAdCAM-1 alone: *P < 0.05; **P < 0.005.

Fig. 5.

The effects of CCL21, CCL25, CCL28, and CXCL12 on lymphocyte adhesion to immobilized VCAM-1. The proportion of lymphocytes show rolling adhesion (A), static adhesion (B), or appear activated (C) on immobilized VCAM-1 (5 μg/ml) in response to the chemokines CCL21, CCL25, CCL28, and CXCL12 (all at 1 μg/ml). Values are expressed as percentage of total adhesion and are means ± SE of 4 separate experiments using different lymphocyte donors. Student's t-test indicated a significant difference in activated cells in the presence of recombinant chemokine compared with VCAM-1 alone: *P < 0.05.

Chemokine-dependent promotion of lymphocyte adhesion to MAdCAM-1 is mediated via G protein signaling.

To confirm that the effects were due to chemokine receptor signaling, PBL were pretreated with pertussis toxin, which inhibited the effects of CCL21 or CCL25 on lymphocyte adhesion to MAdCAM-1. Figure 6 shows representative data for rolling behavior of lymphocytes on MAdCAM-1 in the presence of CCL25 and CCL21, confirming that the observed activating effects of the chemokines are abolished in the presence of pertussis toxin.

Fig. 6.

The effect of pertussis toxin (PTX) on chemokine-induced arrest of lymphocytes on immobilized MAdCAM-1. The proportion of lymphocytes undergoing rolling adhesion on immobilized MAdCAM-1 (5 μg/ml) is shown in the absence or presence of CCL21 (A) or CCL25 (B), both at 1 μg/ml. Where indicated, PBL were pretreated with PTX (100 ng/ml for 60 min) before perfusion over immobilized protein at 0.05 Pa. Values are expressed as percentage of adherent cells that rolled and are means ± SE from 5 replicate experiments using different lymphocyte donors. Student's t-test indicated a significant decrease in rolling in the presence of recombinant chemokine compared with MAdCAM-1 alone and increased rolling of PTX-treated cells in the presence of chemokine compared with untreated lymphocytes: *P < 0.05, **P < 0.005.

Adhesion of lymphocytes to endothelial-expressed MAdCAM-1 is also triggered by CCL21, CCL25, CCL28, or CXCL12.

We investigated the ability of chemokines to trigger binding of lymphocytes to endothelial-expressed MAdCAM-1 by inducing expression in human hepatic endothelial cells using adenoviral vectors. Figure 7 shows that the vector containing FL-MAdCAM-1GFP resulted in high levels of MAdCAM-1 expression in cultured HSEC, confirmed by Western blotting to detect MAdCAM-1 protein in cells transduced with MAdCAM-1 vectors at different MOI (Fig. 7). Optimal transgene expression was achieved at an MOI of 10, and so this viral concentration was used for subsequent adhesion assay experiments with FL-MadCAM-1.

Fig. 7.

Induction of MAdCAM-1 expression in hepatic sinusoidal endothelial cells (HSEC) following adenoviral infection. A: the expression of MAdCAM-1 (arrow) was examined in protein lysates prepared from HSEC following infection with FL-MAd-GFP-Ad [multiplicity of infection (MOI) 5; lane 1], FL-MAd-GFP-Ad (MOI 10; lane 2), control GFP-Ad vector (MOI 50; lane 3). Lysates from Chinese hamster ovary cells transfected with SV-MAdCAM-1 (lane 4) and FL-MAdCAM-1 (lane 5) were included as a positive control. B: representative flow cytometry histograms from two different primary isolates of human hepatic endothelial cells 24 h postinfection with FL-Mad-GFP (MOI 10, shaded histogram) compared with uninfected cells (open histogram).

We then carried out flow-based adhesion assays on HSEC transduced with control or MAdCAM-1 vectors using PBL isolated from normal individuals or patients with IBD who have higher levels of expression of α4β7 in PBL (49). We saw no significant adhesion of normal lymphocytes to resting endothelium, while induction of MAdCAM-1 expression permitted leukocyte adhesion, which was inhibited by >60% by treatment of the endothelium with MAdCAM-1 blocking antibody (P1, Fig. 8). Blockade of lymphocyte α4β7-integrin also resulted in a modest reduction of adhesion. When the binding of lymphocytes from patients with IBD was compared with that seen with normal individual patients, PBL bound in higher numbers to MADCAM-1 expressing endothelium (58 vs. 42/mm2 per 106 cells perfused).

Fig. 8.

Induction of MAdCAM-1 expression in endothelial cells results in enhanced binding of lymphocytes. PBL were perfused over untransfected HSEC (control) or cells transduced to express MAdCAM-1 (MAd) at 0.05 Pa. Values are means ± SE of number of adherent cells from 2 representative experiments. Student's t-test indicated a significant increase in adhesion to MAdCAM-1 expressing cells (*P ≤ 0.01), which was inhibited by treatment with an anti-MAdCAM-1 antibody (P1, $P ≤ 0.01).

Immobilization of recombinant chemokines on the transduced endothelial cells resulted in a change in the adhesive behavior of captured lymphocytes (Fig. 9). While total numbers of adherent lymphocytes were unchanged in the presence of chemokine, both CCL28 and CXCL12 [stromal cell-derived factor (SDF)] resulted in a significant increase in the numbers of patient lymphocytes migrating, and there was a trend for increased migration in the presence of CCL25, which was abolished when chemokine-dependent signaling was blocked by pertussis toxin treatment.

Fig. 9.

The effects of CCL21, CCL25, CCL28, and CXCL12 on lymphocyte adhesion to MAdCAM-1 expressing endothelial cells. PBL from inflammatory bowel disease patients (A and B) or normal individuals (C and D) were perfused over MAdCAM-1 expressing HSEC at 0.05 Pa. Where indicated, HSEC were preincubated with recombinant chemokine (10 μg/ml, 30 min) before perfusion of PBL. Values are means ± SE of number of adherent cells in representative experiments (n = 3 for normal donors and n = 6 for inflammatory bowel disease donors). Student's t-test indicated no significant effects of chemokines on total numbers of adherent cells, but significant increases in number of transmigrating cells in the presence of CCL28 (*P = 0.07) and CXCL12 (*P = 0.012). Treatment of PBL with PTX significantly reduced the number of migrating cells in the presence of CCL25 ($P = 0.007) compared with untreated PBL. E: representative video images of lymphocytes interacting with MAdCAM-1 expressing HSEC. The panels demonstrate the increased numbers of activated, shape-changed lymphocytes binding to HSEC in the presence of CCL25 and how this is reduced by pretreatment of the PBL with PTX.


Crohn's disease and ulcerative colitis are characterized by uncontrolled lymphocyte recruitment to the gut, and there is evidence that both α4β7/MAdCAM-1- and α4β1/VCAM-1-dependent pathways are involved (13, 25, 27, 57, 59, 60, 67). In the present study, we examined the relative contribution of α4β1 and α4β7 and the role of specific chemokines in regulating the binding of human lymphocytes to VCAM-1 and MAdCAM-1 under physiological flow. These studies are important because they not only provide information about receptor usage during interactions with gut adhesion molecules, but will also inform therapeutic strategies based on inhibition of VCAM-1- and/or MAdCAM-1-mediated recruitment in IBD (9).

Both MAdCAM-1 and VCAM-1 immobilized in glass microslides were able to support lymphocyte rolling and static adhesion from flow in a concentration-dependent manner. Although α4β1 can bind to MAdCAM-1, we found restricted integrin usage with all detectable adhesion to MAdCAM-1 mediated by α4β7, and all binding to VCAM-1 mediated by α4β1. Because we used purified MAdCAM-1 protein that lacks the mucin side chains required for selectin binding, we were able to compare the role of the two integrins directly in the absence of L-selectin-mediated events. A high proportion of lymphocytes interacting with VCAM-1 arrested, whereas up to 50% of cells interacting with MAdCAM-1 demonstrated rolling adhesion and failed to arrest. This could mean that MAdCAM-1 is less efficient at supporting static adhesion or that arrest mediated by α4β7 is highly dependent on a chemokine-mediated signal. The latter appears to be true because, although all of the chemokines tested increased the proportion of lymphocytes arrested on MAdCAM-1, their addition had no effect on VCAM-1-mediated arrest. This is surprising, given previous reports of SDF-1 activation of T-cell adhesion to VCAM-1 under flow (18), but may be explained if a subset of circulating cells expressing high-affinity α4β1 can bind without the need for chemokine-mediated integrin activation (17). However, it is important to note that the nature of adhesion to an immobilized substrate can depend on the concentration of adhesive ligand present or the efficiency by which it immobilizes to the surface (42). Thus, although we performed dose-response experiments to determine optimal protein concentrations, it is possible that VCAM-1 may immobilize more efficiently to glass microslides and thereby promote increased static adhesion.

We next compared the ability of different chemokines to trigger lymphocyte adhesion under shear stress. CCL21, CCL25, CCL28, and CXCL12 were all able to trigger arrest of rolling lymphocytes on MAdCAM-1 under conditions of physiological flow. Because lymphocyte adhesion to MAdCAM-1 in our system is α4β7 dependent, this implies that all four chemokines can activate α4β7-mediated adhesion. We noted a hierarchy of effects. CXCL12 was the most potent, followed by CCL21, then CCL25 and CCL28, which had similar effects. However, it is possible that different chemokines could be differentially immobilized and presented on the adhesive substrate, so any head to head comparisons must be seen in the light of this caveat. These effects were inhibited by pertussis toxin and were not seen when an irrelevant chemokine, CCL11, was immobilized, demonstrating that they are chemokine receptor dependent. Previous studies have reported the ability of CCL21 and CXCL12 to activate α4β7-dependent lymphocyte adhesion (53, 76), but this is the first report to demonstrate that CCL25 and CCL28 can trigger α4β7-mediated adhesion to MAdCAM-1 under flow.

After arresting from flow, adherent lymphocytes can be subdivided based on their morphology into those that are round and phase bright and those that have become activated and adopted an extended or flattened morphology with spreading on the protein surface. The first group represents cells that are adherent but not motile, whereas the latter have undergone cytoskeletal reorganization, indicative of motile/migratory behavior. Chemokines are critically involved in lymphocyte arrest from flow (2, 64), and, in addition, shear stress itself can activate α4β1-integrin-mediated adhesion and modulate the effects of chemokine-mediated arrest (3). It is thus essential to study the effects of chemokines under conditions of physiological shear stress. Both affinity and avidity changes have been implicated in chemokine-induced activation of integrin-mediated adhesion (46). Affinity changes result from alterations in integrin conformation, which affects the ligand binding site, whereas avidity changes involve fluidity in the plasma membrane and cytoskeletal rearrangements, which lead to integrin clustering on the cell surface (16). The static cells that maintained their round morphology in our study may represent cells that arrested via high-affinity, integrin-ligand interactions, whereas shape change and flattening are a consequence of cytoskeletal rearrangements and integrin clustering. This results in the formation of lamellipodial extensions at the leading edge and a uropod at the trailing edge (22) that permit translocation and motility (33).

The relative contribution of chemokines to adhesion triggering, shape change, and motility on different substrates is not well understood (63). We saw low but detectable levels of lymphocyte shape change and spreading on both MAdCAM-1 and VCAM-1 in the absence of chemokines, as has been reported previously for VCAM-1 and ICAM-1 (33, 43). We were surprised that, although all four chemokines triggered arrest of rolling lymphocytes on MAdCAM-1, only CXCL12 and CCL21 increased the proportion of motile/shape-changed lymphocytes. This suggests that CXCL12 and CCL21 not only activate high-affinity α4β7, but also induce avidity changes via cytoskeletal rearrangements and integrin clustering. This is consistent with studies showing that CXCL12 can trigger a rapid reorganization of the actin cytoskeleton and induce motility in T cells (68, 76) and can promote transendothelial migration under conditions of shear stress (64). We did not see shape change in response to CCL25 and CCL28, suggesting that these chemokines may be less effective in inducing cytoskeletal reorganization, and a second signal may be required to promote migration. Only CXCL12 triggered shape change in lymphocytes that had arrested on VCAM-1, confirming previous reports that CXCL12 triggers rapid α4β1 microclustering, leading to enhanced avidity for VCAM-1 under flow (28). The inability of any of the chemokines to trigger α4β1-mediated arrest from rolling may reflect the ability of VCAM-1 to support spontaneous arrest of circulating cells expressing high-affinity α4β1 (17). Thus only cells with constitutively activated α4β1 can be captured on VCAM-1, and this binding is not further activated by an additional chemokine-mediated signal. It is also possible that distinct lymphocyte subpopulations were activated to adhere by specific chemokines, and a true head-to-head comparison is only valid, if highly defined subpopulations of lymphocytes matched for integrin and chemokine receptor expression are compared, and this was beyond the scope of the present study.

Our findings indicate that, while CXCL12 can activate both α4β7- and α4β1-mediated lymphocyte adhesion, the effects of CCL25 and CCL28 are restricted to α4β7. This could be explained if expression of CCR9 and CCR10 was restricted to α4β7+ T cells. CCR10, however, is expressed by very few β7-integrin-positive CD4+ or CD8+ T lymphocytes (36), and gene expression by β7+ and β7 CD4+ memory T cells (61) suggests that CCR9 and CCR10 are restricted to β7+ (78) and β7 cells, respectively (36). Our cytometry data confirm low levels of CCR9 and CCR10 on PBL. This, combined with the ability of these cells to respond to immobilized CCL25 and CCL28, suggests that such chemokines can efficiently trigger adhesion of the small proportion of cells that express the relevant receptor. The agonist potency of chemokines to activate integrins under flow correlates with the density of cell surface chemokine receptors (21). Thus the higher levels of CCR7 and CXCR4 expression observed in the α4β7+ subset and the α4+ population overall may explain the more potent adhesion-promoting effects of CCL21 and CXCL12.

In vivo CCL21 and MAdCAM-1 are both expressed in Peyer's patches and mesenteric lymph nodes, where they support T-cell homing to GALT (4, 11, 73, 75). The ability of CCL21 to activate α4β7-mediated arrest on MAdCAM-1 under flow would promote this. CXCL12 is also constitutively expressed in secondary lymphoid tissues (7), but current evidence suggests it is not involved in the arrest of lymphocytes on HEV (71), but rather that it regulates transendothelial migration once CCR7-mediated activation of integrin-dependent adhesion has occurred (19, 56). However, we have shown here that CXCL12 can activate α4β7-mediated arrest of T cells on MAdCAM-1 under flow. Thus it is possible that, in Peyer's patches HEV, where MAdCAM-1 is constitutively expressed, CXCL12 plays an additional role in activating arrest, as well as in transmigration. CXCL12 is also constitutively expressed by gut epithelial cells, suggesting it may also drive transendothelial migration across MAdCAM-1+ vessels in the lamina propria (50).

CCL25 is found on epithelial cells within the small intestine, but not in the colon or other mucosal tissues within the gastrointestinal tract (40). CCL25 protein, but not mRNA, can be detected in small intestinal endothelial cells (55), suggesting it may be secreted by epithelial cells and then presented on intestinal venules, where MAdCAM-1 is constitutively expressed (11). This is consistent with a role in triggering integrin-mediated arrest on lamina propria vessels (40). In support of this, we have shown that CCL25 can trigger α4β7-dependent arrest of rolling lymphocytes on MAdCAM-1 under flow. CCL28 is predominantly expressed by epithelial cells in the colon, as well as at lower levels in the small intestine and at several other mucosal sites (54, 74). Its receptor CCR10 is expressed on IgA-secreting B cells in both the small and large intestine (39, 41), and the importance of CCL28 in B-cell localization to the colon and small intestine is accepted (32), although a role in T-cell homing to the gut has not been demonstrated. Our finding that CCL28 can trigger lymphocyte arrest on MAdCAM-1 under flow provides further evidence for a role in lymphocyte homing to the gut. However, although we have reported a population of CCR10+ α4β7+ T cells in the inflamed human liver (23), CCR10 is absent from the majority of memory T cells within the gut (41). Furthermore, analysis of memory CD4+ T cells in blood shows that CCR10 is expressed by the CLA+ subset of T cells that display skin tropism, and not by β7+ cells (61). It thus remains to be determined whether CCL28 is involved in T-cell homing to the gut.

This study reveals distinct functions for chemokines involved in lymphocyte recruitment to the gut and suggests that cooperative interactions between different chemokines may be required for efficient recruitment. The results have implications for the design of novel therapies aimed at inhibiting specific chemokines or α4-mediated adhesion in IBD.


We are grateful for the financial support provided by grants from the Endowment Fund of the former United Birmingham Hospitals and an unrestricted grant from Pfizer UK. B. Eksteen was supported by Fellowships from Core and the Medical Research Council. E. Liaskou is a Marie-Curie doctoral student EC 20996 TRIFID.


We thank Dr. Nick Pullen for the kind gift of reagents.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract