Fractalkine and CX₃CR1 Mediate a Novel Mechanism of Leukocyte Capture, Firm Adhesion, and Activation under Physiologic Flow

Resting PBMCs were isolated from whole blood by centrifugation over lymphocyte separation medium (Organon Teknika, Durham, NC) as described previously (11). IL-2–activated PBLs were generated by culturing PBMCs in RPMI containing 10% FBS and 400 U/ml IL-2 (R & D Systems, Inc., Minneapolis, MN) for 5 d as described previously (9) and harvesting the nonadherent cells.

Human umbilical vein endothelial cells (HUVECs) were obtained from Clonetics Corp. (San Diego, CA), and grown in EGM™ medium (Clonetics Corp.). ECV-304 cells were obtained from the American Type Culture Collection (Rockville, MD) and the generation of FKN-expressing ECV-304 cells (ECV-FKN) has been described previously (9). ECV-304 and ECV-FKN cells were maintained in M199 medium (GIBCO BRL, Gaithersburg, MD) supplemented with 10% FBS. K562 cells were obtained from the American Type Culture Collection and maintained in RPMI supplemented with 10% FBS. Cells were grown to confluence on 25-mm glass coverslips before use in flow assays. HUVECs used in flow assays were activated with TNF-α (100 ng/ml; R & D Systems, Inc.) for 12 h.

For stable transfection of CX₃CR1 in K562 cells, the expression plasmid pCAGGS-Neo-V28 was transfected into K562 cells as described previously (9). Control neomycin-expressing cells were generated by transfecting pcDNA3.1⁻ (Invitrogen, Carlsbad, CA) into K562 cells. After selection with 500 μg/ml G418 for 3 wk, drug resistant cells were used unselected. Expression of CX₃CR1 was assessed by RNAse protection using the hCKR-5 and hCKR-6 probe sets (PharMingen, San Diego, CA) as described by the manufacturer. CX₃CR1 is identical to V28 in the probe set. K562 cells expressed no mRNA for chemokine receptors CCR1, CCR2, CCR3, CCR4, CCR5, CCR7, CCR8, CXCR1, CXCR2, CXCR3, CXCR4, and CX₃CR1. Cell surface expression of CX₃CR1 was also tested by indirect immunofluorescence (IF) as described below. To inhibit signal transduction through CX₃CR1, cells were treated with PTX at 500 ng/ml for 30 min at 37°C. To inhibit integrin function, cells were preincubated with 5 mM each of EDTA and EGTA in flow buffer (see below) for 15 min before use in flow assays, and both EDTA and EGTA were added to the flow buffer. Integrins were also blocked by preincubation of K562-neo or K562-CX₃CR1 cells for 10 min with the anti-β1 antibody 4B4 (anti-CD29, IgG1) (12) or the anti-β2 antibody H52 (anti-CD18, IgG1) (13) before use in flow assays. Antiintegrin antibodies were used as ascites fluid diluted 1:100.

FKN, thymus and activation regulated chemokine (TARC), and EBI1 ligand chemokine (ELC) fused with secreted placental alkaline phosphatase (FKN-SEAP, TARC-SEAP, and ELC-SEAP), and a control SEAP protein were produced as described previously (9, 14, 15). Monocyte chemotactic protein (MCP)-1–SEAP was also produced essentially as described previously (14). In brief, the DNA fragment containing the coding region of MCP-1 was amplified from MCP-1 cDNA by PCR. After digestion with SalI and XbaI, the MCP-1 fragment was subcloned into SalI–XbaI sites of pDREF-SEAP(His)₆-Hyg vector in order to express MCP-1 as a soluble fusion protein linked through five amino acid residues (Ser-Arg-Ser-Ser-Gly) with SEAP tagged with six histidine residues [(His)₆]. 293/EBNA-1 cells (Invitrogen) were transfected with the expression vector pDREF–MCP-1–SEAP(His)₆ using lipofectamine (GIBCO BRL). The transfected cells were incubated for 3–4 d in DMEM supplemented with 10% FBS. The supernatant containing each SEAP(His)₆ chimera was collected by centrifugation, filtered (0.45 μm), and stored at 4°C after adding 20 mM Hepes (pH 7.4) and 0.02% sodium azide. 25-mm round glass coverslips were coated with fusion proteins as described previously (9). In brief, 10 μg/ml anti-SEAP antibody 8B6 (Sigma Chemical Co., St. Louis, MO) in buffer (50 mM Tris, pH 9.5) was placed on the coverslip and incubated overnight at 4°C. The coverslips were blocked in PBS with 1% BSA at room temperature for 2 h. After washing, 10 nM FKN-SEAP, SEAP, MCP-1–SEAP, TARC-SEAP, or ELC-SEAP was added and incubated at room temperature for 1 h. Coverslips were washed three times before use in the flow chamber.

Parallel plate flow chamber assays were performed as described previously (16). In brief, protein-coated glass coverslips were assembled in a parallel plate laminar flow chamber and placed on the stage of an inverted phase-contrast microscope. K562-neo or K562-CX₃CR1 cells were suspended at a density of 10⁶ cells/ml in flow buffer (PBS containing 0.75 mM CaCl₂, 0.75 mM MgCl₂, and 0.5% bovine serum albumin). Cell suspensions were perfused through the chamber for a 5- or 10-min period at a wall shear stress of 0.25 dynes/cm² via a syringe pump (Harvard Apparatus, South Natick, MA). At the end of the 5- or 10-min perfusion period, flow buffer was perfused through the chamber at 1.85 dynes/cm² for 3 min and at 10 dynes/cm² for 3 min, and multiple ×100 fields were visualized. The entire experiment was recorded using a CCD video camera (Hitachi Denshi, Ltd., Woodbury, NY) and a Sony SuperVHS video recorder (Sony Electronics, Inc., San Jose, CA). For some experiments, cell suspensions were initially passed through the chamber at a low shear stress (0.25 or 0.5 dynes/cm²), followed by flow buffer at progressively increasing wall shear stresses of 0.5, 1.0, 1.85, 5.0, 10, and 20 dynes/cm² at 2-min intervals to determine the stability of adhesive interactions. The number of adherent cells was determined by analyzing the videotapes and counting 16 fields each with a 0.16-mm² area.

PBMCs and IL-2–activated PBLs were characterized by three-color flow cytometry as described previously (17) on a FACStarPlus^® (Becton Dickinson, San Jose, CA) using CD3-FITC, CD4-PE, CD8-Cy5, CD14-PE, CD16/56-PE, and CD19-FITC directly conjugated antibodies generously provided by Coulter Corp. (Hialeah, FL). Resting PBMCs and IL-2– activated PBLs that bound to FKN-SEAP–coated coverslips were characterized by two-color IF using CD3-FITC, CD4-PE, CD8-FITC, CD14-PE, CD16/56-PE, and CD19-FITC antibodies. In brief, coverslips were incubated with 50 μl of mAbs diluted in PBS with 1% BSA for 15 min at room temperature and washed. Coverslips were mounted on glass slides and analyzed by IF on an epifluorescent microscope.

Cell surface expression of FKN was assessed by indirect IF using mAb 1D6 to the mucin domain of FKN as described previously (9). Expression of FKN on HUVECs was determined by two-color flow cytometry using mAb 1D6 and directly conjugated CD31-PE mAb kindly provided by Coulter Corp. (Hialeah, FL) as described previously (17). Cell surface expression of CX₃CR1 was also determined by indirect IF. In brief, 10⁶ cells were incubated with 5 nM chemokine-SEAP fusion proteins in PBS containing 0.05% NaN₃ for 30 min at room temperature and washed. After incubation with anti–alkaline phosphatase mAb 8B6 and washing, cells were incubated with fluorescein-conjugated goat anti–mouse Ig (Cappell, Gaithersburg, MD), washed, and analyzed by flow cytometry.

To determine whether FKN can mediate capture of free flowing leukocytes under conditions of shear stress encountered in vivo, freshly isolated PBMCs were perfused over immobilized FKN-SEAP fusion proteins at physiologic shear stresses. FKN-SEAP (but not SEAP alone) captured leukocytes at wall shear stresses up to 1.85 dynes/cm² (Fig. 1). The capture of leukocytes by FKN was rapid (<33 ms), and no rolling was observed. Despite diminished efficiency of capture with increasing wall shear stress, there was no significant difference in the rate of capture of monocytes compared with lymphocytes. Thus, FKN-mediated leukocyte capture at physiologic wall shear stresses.

To test if FKN could support the firm adhesion of normal human leukocytes, both resting PBMCs and IL-2–stimulated PBLs were perfused over immobilized SEAP fusion proteins at 0.25 dynes/cm² and allowed to capture for 5 min. Firm adhesion was defined as the ability of cells to resist detachment at a wall shear stress of 10 dynes/cm². TARC-SEAP and ELC-SEAP were used as controls. CCR4, the receptor for TARC, is selectively expressed on CD4⁺ T cells (14) and CCR7, the receptor for ELC/MIP-3β, is selectively expressed on activated T and B cells (15, 18, 19). Although resting PBMCs and IL-2– stimulated PBLs adhered firmly to purified FKN-SEAP (Fig. 2,B), they were neither captured by (data not shown) nor adherent to (Fig. 2,B) control proteins. Virtually all captured cells remained firmly adherent throughout the experiments (>10 min in duration). As determined by IF, the majority of firmly bound PBMCs were either CD14⁺ monocytes (62.5%) or CD8⁺ T cells (30%; Fig. 2 C). Resting NK cells bound to immobilized FKN, but the percentage of bound NK cells varied widely between experiments (2.5–15%). Of the IL-2–activated PBLs bound to FKN, 37.5% were NK cells (CD16⁺ and/or CD56⁺), 56.3% were CD8⁺ T cells, and only 6.3% were CD4⁺ T cells. Thus, FKN mediated the firm adhesion of resting monocytes, resting and activated CD8⁺ T cells, and resting and activated CD16/56⁺ NK cells under physiologic flow conditions.

Soluble FKN-SEAP fusion proteins bind to and transduce signals through CX₃CR1 to mediate a Ca²⁺ flux and chemotaxis (9). Immobilized FKN induced the spreading of bound monocytes (Fig. 2 A), but had no effect on the shape of resting PBLs. Furthermore, FKN induced pseudopod formation in the majority of IL-2–activated PBLs. Thus, FKN can activate resting monocytes and IL-2–activated PBLs.

Normally, CX₃CR1 mRNA is expressed primarily by CD16⁺ NK cells and at low levels by CD8⁺ T cells and CD14⁺ monocytes (9). As for other chemokine receptors, IL-2 enhances CX₃CR1 expression on both CD4⁺ and CD8⁺ T cells (9). Since these were the very cell types that were captured by and firmly adherent to purified FKN before becoming activated, FKN-CX₃CR1 interactions probably mediated these events. Thus, the ability of FKN-transfected ECV-304 transformed endothelial cells (ECV-FKN; Fig. 3) to capture and induce stable arrest of CX₃CR1-transfected K562 erythroleukemia cells (K562-CX₃CR1; Fig. 3) was examined. CX₃CR1-transfected and control cells were allowed to bind to ECV-FKN cells at 0.25 dynes/cm² for 10 min and exposed to increasing shear stresses up to 20 dynes/cm² (Fig. 4 A). K562-CX₃CR1 cells remained bound to ECV-FKN monolayers at wall shear stresses up to 10 dynes/cm², and began to release at 20 dynes/ cm². Thus, FKN on endothelium interacts with CX₃CR1 on leukocytes to mediate their capture and firm adhesion.

To test the role of integrins and signal transduction through CX₃CR1, the effect of PTX, EDTA/EGTA, and antiintegrin antibodies on FKN-mediated firm adhesion was tested. Neither PTX (known to inhibit Ca²⁺ mobilization and chemotaxis in CX₃CR1 expressing cells; 9) nor EDTA/ EGTA (to disrupt integrin function; 20, 21) had an effect on FKN-mediated capture or firm adhesion (Fig. 4,A). Function-blocking antiintegrin mAbs to β1 (4B4; 12) and β2 (H52; 13) subunits also had no effect on the firm adhesion of K562-CX₃CR1 cells to ECV-FKN cells under flow (Fig. 4 B). Therefore, FKN-mediated capture and firm adhesion occurred independently of integrin activation and PTX-sensitive signal transduction through CX₃CR1.

To further confirm the specificity of FKN-CX₃CR1– mediated adhesive interactions, the ability of immobilized FKN-SEAP to induce the firm adhesion of CX₃CR1- expressing versus nonexpressing cells was compared. FKN-SEAP, but not SEAP or MCP-1–SEAP, could support firm adhesion of K562-CX₃CR1 under physiologic wall shear stresses (Fig. 4 C). Thus, FKN–CX₃CR1 interactions alone are sufficient to mediate firm adhesion under flow.

Although the above studies indicate that FKN can support firm adhesion of leukocytes under conditions of flow, they were performed with either purified fusion proteins or with transfected cells. To determine whether the FKN interaction is physiologically relevant, the level of FKN expression on TNF-activated primary HUVECs was tested. TNF-α induced the expression of FKN on the surface of a subset of CD31⁺ HUVECs (Fig. 5, A and B). FKN expression was induced by 1 h, and was maximal by 6–12 h, in a dose-dependent manner with peak expression at a dose of 100 ng/ml TNF-α (data not shown). Subsets of TNF-activated HUVECs and ECV-FKN cells expressed similar levels of FKN (Fig. 5,C), indicating that ECV-FKN cells express physiologically relevant levels of FKN. To further confirm this point, the ability of TNF-activated HUVECs to induce stable arrest of K562-CX₃CR1 cells was tested. CX₃CR1-transfected and control cells were allowed to bind to resting and TNF-activated HUVECs at 0.5 dynes/cm² for 5 min and were exposed to increasing shear stresses up to 20 dynes/cm² (Fig. 5, D and E). K562-CX₃CR1 cells did not bind to unactivated HUVECs (Fig. 5,D), but did bind to a subset of TNF-activated HUVECs (data not shown). Although some K562-CX₃CR1 cells began to release from TNF-activated HUVECs at a shear stress of 5 dynes/cm², a subset of cells remained bound at wall shear stresses up to 20 dynes/cm² (Fig. 5 E). Thus, HUVECs can be induced to express sufficient levels of FKN to mediate leukocyte stable arrest.

In the current models of leukocyte migration, chemokines and their receptors transduce signals to the rolling leukocyte to induce cell arrest and firm adhesion by activating the adhesive capacity of integrins (1, 2, 5–7). This study demonstrates new roles for chemokines and their receptors in leukocyte migration, and describes a novel mechanism of leukocyte capture, firm adhesion, and activation mediated by the interactions of FKN with CX₃CR1.

We have shown that FKN alone on the endothelium can mediate leukocyte capture and firm adhesion of monocytes, CD8⁺ T cells, and CD16/56⁺ NK cells. Although CX₃CR1 mRNA is also expressed by IL-2 activated CD4⁺ T cells (9), they did not firmly adhere to immobilized FKN in these studies. The reason for this is not clear. Although CX₃CR1 on the leukocyte can serve as the receptor for FKN to mediate all of these functions, FKN may also interact with other receptors. In the current models of leukocyte recruitment, selectins capture circulating leukocytes in the primary adhesion step (1–4). The selectins recognize distinct, but closely related, sialylated carbohydrates on their receptors (22–25). For example, L-selectin recognizes mucin-like molecules with extensive O-glycosylation (22– 26). Since FKN has a mucin-like domain with many potential O-glycosylation sites, it may interact with selectins. However, in this study, selectins were not necessary for FKN-CX₃CR1 adhesion since K562 cells do not express detectable levels of L-, E-, or P-selectin (27 and data not shown). Thus, in this system, leukocyte capture by FKN was selectin independent.

Leukocyte firm adhesion, in the multistep model of leukocyte migration, requires activation by chemokines and is mediated primarily by the integrin family of adhesion molecules and their receptors (1, 2). In this study, FKN interactions with CX₃CR1 were sufficiently strong to mediate firm adhesion under conditions of high wall shear stresses (up to 20 dynes/cm²), without the involvement of integrins or other adhesion molecules. Combined with the ability of FKN to capture free-flowing leukocytes at physiologic wall shear stresses (1.85 dynes/cm²), this indicates that FKN and CX₃CR1 mediate leukocyte capture and firm adhesion under flow. However, FKN and CX₃CR1 may also act in concert with other effectors of the classical multistep pathways of leukocyte migration.

FKN appears to be unique amongst the chemokines in its ability to induce leukocyte firm adhesion. Neither TARC nor ELC fusion proteins were able to induce the firm adhesion of flowing PBMC. FKN naturally has a mucin stalk, whereas the other chemokines do not. The negative results with TARC and ELC may have been due to the fact that they lack mucin stalks. The extension provided by the mucin stalk may be a key factor in allowing FKN to mediate firm adhesion.

FKN expression is induced on primary cultures of human umbilical vein and pulmonary artery endothelium by the proinflammatory cytokines IL-1 and TNF-α (8, 10). Further, the mouse homologue of FKN, neurotactin, is expressed at low levels on brain endothelium in normal mice and upregulated on endothelium in inflamed brain in allergic encephalomyelitis (10), suggesting that the FKN pathway may be functional in leukocyte trafficking to inflamed brain. Although trafficking of lymphocytes to brain during peak inflammation in Sindbis virus–infected mice was blocked by antibodies to the β2 integrin LFA-1, lymphocyte entry into the brain during the early inflammatory response was not affected by antibodies to β1 integrins, VLA-4, or CD44 (28), indicating that the lymphocytes migrating to brain during early Sindbis virus infection used a different pathway of leukocyte migration. It is intriguing to hypothesize that the FKN pathway of circulating leukocyte adhesion to endothelium may play an important role in the accumulation of leukocytes on the endothelium of inflamed brain, facilitating subsequent diapedesis and migration.

In summary, we have shown that FKN can capture circulating leukocytes and induce their firm adhesion and activation. Furthermore, FKN-mediated firm adhesion is dependent on neither chemokine receptor signaling nor on integrins or other cell adhesion molecules, suggesting an alternate and novel regulatory mechanism for leukocyte trafficking. This novel FKN-mediated pathway may be particularly relevant for the recruitment of monocytes, CD8⁺ T cells, and NK cells to sites of inflammation.

We thank Leona P. Whichard, Jonathan Baron, and Dawn M. Jones for their technical assistance. We also thank Barton F. Haynes and Michael Krangel for helpful discussion and a critical review of this manuscript.

This work was supported by National Institutes of Health grants AR-39162, AI-26872, CA-54464, and HL-50985 and by the Shionogi Institute for Medical Science. L.A. Robinson is supported by the Pediatric Scientist Development Program through a grant from St. Jude Children's Research Hospital.

ELC

EBI1 ligand chemokine

FKN

fractalkine

HUVECs

human umbilical vein endothelial cells

IF

immunofluorescence

MCP

monocyte chemotactic protein

PTX

pertussis toxin

SEAP

secreted placental alkaline phosphatase

TARC

thymus and activation regulated chemokine