Unique Role of the Chemokine Domain of Fractalkine in Cell Capture

The chemokine fractalkine (FK) has two structural features that make it unique in the chemokine family: a CX 3C motif and an extended carboxyl terminus that anchors it to the cell surface. This mucin-like stalk or an equivalent spacer is required for FK to mediate the adhesion of cells expressing its receptor, CX 3CR1. To determine whether the ability of FK to act as a cell adhesion molecule is due to the unique presentation of a chemokine domain on a stalk or to properties of the chemokine domain itself, we created a series of chimeras in which other soluble chemokines (RANTES (regulated on activation normal T cell expressed), monocyte chemoattractant protein 1, macrophage inflammatory protein 1β, secondary lymphoid tissue chemokine, and interleukin 8) were fused to the mucin stalk. When tested in a static-cell adhesion assay, many of these chemokine chimeras demonstrated activity equivalent to that of FK. In flow assays, however, none of the chimeras captured cells as efficiently as FK. Interestingly, FK captured cells expressing either CX 3CR1 or the viral receptor US28. Cells bound to FK without rolling or detaching, whereas the interleukin 8 and monocyte chemoattractant protein 1 chimeras induced primarily cell rolling and detaching, respectively. In binding studies, FK has a significantly slower off-rate from its receptors than any of the other chemokine chimeras had for their cognate receptors. We conclude that presentation of a chemokine atop a mucin-like stalk is not, in and of itself, sufficient to capture cells. The unique ability of FK to mediate adhesion under flow may be a function of its slow receptor off-rate.

In the development of the inflammatory response, leukocytes interact with vascular wall endothelial cells in a multistep, sequential process that includes rolling, firm adhesion, and diapedesis (1,2). Firm adhesion is thought to be dependent on integrin binding, a process that can be regulated by the activity of the chemokine family of soluble proteins (3)(4)(5)(6). However, recent studies have demonstrated that fractalkine (FK), 1 a novel member of the chemokine family, can capture leukocytes in an integrin-independent manner (7)(8)(9).
FK is a structurally unique molecule in which a chemokine domain is located atop a mucin stalk connected to a transmembrane domain (10). It is expressed on the surface of endothelial cells (10) and neurons (11) and is up-regulated by pro-inflammatory cytokines, such as lipopolysaccharide, interleukin-1, and tumor necrosis factor-␣ (9,10,(12)(13)(14). We and others have recently shown that under physiological flow conditions FK efficiently captures leukocytes and cell lines transfected with CX 3 CR1, the FK receptor (7,9). The cell capture function of FK is not dependent on integrin activation or on activation of G proteins by CX 3 CR1 (7), but it is dependent on the presentation of the FK chemokine domain on a stalk (8,9,15). These studies revealed an unusual role for FK in directly mediating cell adhesion and raised the question of whether other chemokines presented in a similar manner would also directly support cell capture and adhesion.
To determine whether the ability of FK to act as a cell adhesion molecule was due to its unique presentation atop a rigid stalk or rather to properties of the FK chemokine domain itself, we created a series of chimeras in which other soluble chemokines were fused to the FK mucin stalk. In this study, we have examined the ability of these chemokine/stalk chimeras to capture cells expressing the appropriate cognate receptors in both static and flow adhesion assays. Chemokine Chimera Constructs-A shuttle vector was generated containing the cDNA encoding the extracellular domain of FK with an additional carboxyl-terminal 6-His epitope. The cDNA for human FK was obtained from the American Type Culture Collection (ATCC, Manassas, VA; IMAGE clone 44145G). The primers 5Ј-ttcgcggccgccaccatggctccgatatctctgtcgtggctgc and 5Ј-ttcggcgcgcctttaatgatgatgatgatgatgattctgcttctgcctccgggtggcagcctgggcg were used to add 5Ј-NotI and 3Ј-AscI * This study was supported by National Institutes of Health Grants HL-63894 and HL-52773 (to I. F. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To allow replacement of the FK chemokine domain with other chemokines, a unique BamHI restriction site was introduced at the sequence coding for the junction of the chemokine domain and the mucin stalk (after glycine 100). The primers were 5Ј-gccctaactcgaaatggcggcaccttcgagaagcag and 5Ј-ctgcttctcgaaggtgccgccatttcgagttagggc. The QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used to generate a conservative change in the amino acid sequence, replacing threonine 102 with serine. The modified FK was subcloned into JH1104R (a gift from R & D Systems). The cDNAs for the other chemokines were PCR-amplified in the process adding flanking 5Ј-NotI and 3Ј-BamHI sequences and removing the stop codon. The cDNAs were obtained from the following sources: RANTES (ATCC number 105475), MIP-1␤ (ATCC number 185267), and MCP-1 (GenBank accession number X14768) were from ATCC, and SLC was a generous gift from Dr. Jason Cyster (University of California, San Francisco). cDNA for IL-8 was amplified from tumor necrosis factor-␣ (20 h, 10 ng/ml)stimulated human umbilical vein endothelial cells by reverse transcription-PCR with the Pro-Star first-strand reverse transcription-PCR kit (Stratagene). The primers used were (5Ј and 3Ј): RANTES, 5Ј-aaggaaaaaagcggccgctaatatgaaggtctccgcggcacgcc and 5Ј-cgcggatccgctcatctccaaagagttg; MIP-1␤, 5Ј-aaggaaaaaagcggccgctaatatgaagctctgcgtgactgtcc and 5Ј-cgcggatccgttcagttccaggtcatacacg; MCP-1, 5Ј-aaggaaaaaagcggccgctaatatgaaagtctctgccgccc and 5Ј-cgcggatccagtcttcggagtttgggtttgc; SLC, 5Ј-aaggaaaaaagcggccgctaatatgatggctcagtcactg and 5Ј-cgcggatcctggccctttaggggtctgtgaccg; and IL-8, 5Ј-attcgcggccgcatgacttccaagctggtcgcggccgcatgacttccaagctgg and 5Ј-attcggatccctcagccctcttccggatccctcagccctcttc.
These chemokines were inserted into the JH1104R shuttle vector in frame with the mucin stalk of FK. At R & D Systems, these constructs were subcloned into an insect cell line expression vector, and protein was generated and purified using the polyhistidine epitope tag. All chimeras were tested for chemokine activity as described under "Results." Cell Culture and Transfection-The murine pre-B cell line (300-19) (16, 17) was a generous gift from Dr. G. La Rosa (Millennium Pharmaceuticals, Cambridge, MA). Cells were grown and transfected as described (17). The human CX 3 CRI receptor cell line was derived from a cDNA containing the Ade-745, Cyt-839 haplotype. This corresponds to the receptor isoform Ile-249, Thr-280 that has a reported affinity for FK of 4.5 pM (18). To select for lines expressing high numbers of receptors, cells were incubated with a fluorescently labeled anti-FLAG epitope antibody (M1) to detect the receptor amino-terminal FLAG epitope and to perform sorting on a fluorescence-activated cell sorter (FACSVantage, Becton Dickinson, Franklin Lakes, NJ). The surface expression of the chemokine receptors was assessed by flow cytometry as described previously (17,19,20).
Intracellular Calcium Measurements-300-19 cells were centrifuged and suspended at a density of 10 7 cells/ml in growth medium containing 3 M Indo-1AM (Molecular Probes, Eugene, OR) and incubated for 60 min at room temperature. Calcium mobilization was measured with a Hitachi F-2000 spectrophotometer, as described previously (19,21). Experiments were performed at 37°C with constant mixing in cuvettes each containing 1.0 -1.5 ϫ 10 6 cells in assay buffer (Hanks' balanced salt solution, 10 mM HEPES, 1.6 mM CaCl 2 , pH 7.4). Data are represented as a ratio of emission wavelengths (410/490 nm).
Static Adhesion Assay-The assay was performed as described (7), with the following modification to quantitate the number of adherent cells. 300-19 cells were washed with adhesion buffer (RPMI, 1.0 mg/ml bovine serum albumin, 10 mM HEPES, pH 7.4) and resuspended to a density of 10 7 cells/ml in adhesion buffer containing the fluorescent dye BCECF-AM (final concentration 1 M, Molecular Probes). Cells were incubated at 37°C for 30 min and then washed and resuspended in adhesion buffer to a density of 3.0 ϫ 10 6 cells/ml. Adherent cells were quantitated by measuring fluorescence on a Bio-Rad Molecular Imager FX (Hercules, CA). Slides were imaged with the fluorescein isothiocyanate, medium-intensity settings, and the resultant images were analyzed by densitometry with the Quantity One software (Bio-Rad). The mean fluorescence intensity in each well was calculated. Dilutions of input cells in solution were analyzed in parallel to calibrate fluorescence as a function of cell number.
Flow Chamber Adhesion Assay-Cells were perfused over a 20-ϫ 3-mm area of a 60-mm 2 tissue culture dish in a laminar flow chamber (Glycotech, Rockville, MD) as described (7).
Analysis of Adherent Cells-After a 3-min perfusion, the entire substrate-coated region, equaling 10 fields, was captured on video. The number of adherent cells was determined by capturing video frames with Apple Video Player on a PowerMacintosh G3 and analyzing the images with IMAGE 1.62 (National Institutes of Health) obtained on the Web. To assess the behavior of cells that tethered, the video was analyzed for the initial 60 s of flow (see also Ref. 22). Every cell that tethered (i.e. came to a complete stop, even briefly) during this period was monitored for 30 s after tethering. Cells were considered "detached" if they left the surface and re-entered the bulk fluid flow. Rolling cells were defined as those that moved more than one cell diameter. "Arrested" cells remained stationary during the measurement periods.
Calculation of Chemokine/Stalk Coating Density on Flow Adhesion Dishes-Flow adhesion dishes were coated as described above. The laminar flow chamber was placed on the dish, and the adhesion buffer was flowed over the coated area for 20 s to mimic the conditions used in the cell adhesion assay. The dish was then removed from the flow chamber, excess liquid was removed by aspiration, and vacuum grease (Dow Corning Corporation, Midland, MI) was "painted" around the substrate coated area with a cotton swab. 4ϫ gel running buffer (bromphenol blue (0.004%), Tris base (120 mM, pH 6.8), glycerol (20%), SDS (6%), and 2-mercaptoethanol (10%)) (30 l) was added to solubilize the chemokine/stalk chimera coating the plate (15 min, room temperature). Samples were run on an 8% SDS-polyacrylamide gel and analyzed by Western blotting with the ECL system (Amersham Pharmacia Biotech, Piscataway, NJ). Primary antibodies were specific for the chemokine domain of each stalk chimera. All primary antibodies were used at 0.1 g/ml except anti-SLC, which was used at 1.0 g/ml. Secondary antibodies were selected to be specific to the primary antibody. Films of each gel were scanned, and the bands were analyzed by densitometry. The amount of chemokine protein was calculated by comparing the bands from solubilized chemokine/stalk against standard concentrations of chemokine/stalk run on each gel.
Receptor Labeling with Chemokine/Stalk Constructs and Analysis by FACS-Cells were washed and resuspended at 10 7 /ml in static adhesion buffer and then added to an equal volume of buffer containing ligand (2ϫ concentration of chemokine/stalk Ϯ soluble chemokine) in a 96-well V-bottom microtiter plate (final volume, 100 l) (Costar, Corning Inc., Corning, NY). Plates were incubated for 60 min at room temperature for ligand binding. When the assay called for delayed addition of soluble chemokines, this was done at 30 min. All subsequent steps were done on ice in FACS buffer (phosphate-buffered saline, bovine serum albumin (1%), sodium azide (0.1%)). Cells were washed by spinning the plates (500 ϫ g) for 3 min and then inverting and tapping the plate to remove the liquid from the cell pellet. Cells were labeled first with an antihistidine monoclonal antibody (4 g/ml, 60 min) and then with an anti-mouse fluorescein isothiocyanate-conjugated secondary antibody (1:100, 30 min) (Zymed Laboratories Inc.).
Chemokine Binding-Radioligand binding assays were performed as described (23) except that 1.25 ϫ 10 5 cells were used for each data point, and binding was done at 4°C for 24 h.
Statistical Analysis-Statistics for all assays were done with the Mann-Whitney test and Instat software (GraphPad Software, San Diego, CA) for Macintosh.

RESULTS
To determine whether if the unique architecture of FK could completely explain its unusual ability to capture cells, we constructed a series of chimeras in which different chemokines were fused to the extracellular domain of the mucin-like stalk of FK (Fig. 1). Each chimera had a series of six histidine residues added to the extreme carboxyl terminus to immobilize the protein on glass slides coated with antihistidine antibodies. A panel of 300-19 pre-B cells stably expressing epitopetagged cognate receptors for FK and each of the chimeric chemokines was developed. Chemokine receptor expression was comparable in each of these cell lines, with the exception of US28, which was expressed at approximately half the level of the other receptors (Fig. 2). A binding isotherm using 125 I-FK and the CX 3 CR1-expressing cell line revealed 1.4 Ϯ 0.2 ϫ 10 5 receptors/cell (not shown).
The ability of each chimera to interact with its receptor was first assessed by measuring agonist-dependent changes in intracellular calcium levels. Full-length human and murine FK (FK/stalk) and the smaller, soluble chemokine domain-only portion of FK were essentially equipotent in their ability to mobilize intracellular calcium in CX 3 CR1-transfected cell lines (Fig. 3A). Similar results were obtained for the MCP-1/stalk versus MCP-1 on CCR2-transfected cells (Fig. 3B), for the MIP-1␤/stalk versus MIP-1␤ on CCR5-transfected cells (Fig. 3C), and for SLC/stalk versus SLC on CCR7-transfected cells (Fig.  3D). IL-8 and IL-8/stalk gave similar results on CXCR1-transfected cells (not shown).
We began the adhesion studies by examining each of the chemokine/stalk chimeras in static assays (Fig. 4). FK captured cells expressing CX 3 CR1 and US28, but not cells expressing other chemokine receptors (e.g. CCR5), consistent with previous reports (8). The other chemokine/stalk chimeras varied in their ability to capture receptor-expressing cells. Some of these receptor-ligand pairs (MCP-1/stalk:CCR2; IL-8/stalk:CXCR1) were as potent as FK at mediating static adhesion. All pairs except MIP-1␤/stalk:CCR5 mediated adhesion of cells above background levels. The ability to capture cells depended on the specific ligand-receptor pair. Thus, the MCP-1/stalk captured cells expressing CCR2 better than those expressing US28. Similarly, the RANTES/stalk captured cells expressing US28 better than cells expressing CCR1, even though CCR1 was expressed at a higher level than US28. In all cases, the addition of excess soluble chemokine (100 nM) reduced or eliminated adhesion of cells to the tethered chemokine (not shown). These data demonstrate that, under static conditions, at least a portion of the cell capture function of FK can be mimicked by chimeras that express other chemokines atop the FK mucin stalk.
To determine whether the chemokine/stalk chimeras could capture cells under physiological flow conditions, we used a parallel-plate adhesion assay. At a shear rate of 0.8 dyn/cm 2 , FK captured cells expressing CX 3 CR1 more efficiently than those expressing US28, and the RANTES/stalk chimera failed to capture cells expressing CCR1, CCR5, or US28 (Fig. 5A). Similar results were obtained for MCP-1/stalk and US28 cells, although some adhesion of CCR2 cells by MCP-1/stalk remained. In addition, the capture of CXCR1 cells by the IL-8/ stalk remained robust. At a shear rate of 1.2 dyn/cm 2 , however, only FK/stalk captured cells well (Fig. 5B). As previously demonstrated for FK-mediated adhesion of cells expressing CX 3 CR1, the IL-8/CXCR1 adhesion was not reduced by pertussis toxin treatment or by EDTA (not shown) (7,8).
Next we attempted to determine the coating density of the chemokine/stalk chimeras presented on the antihistidine antibody. As assessed from Western blots, the calculated density of FK was approximately 400 molecules/m 2 (60 pg/mm 2 ). The "active" concentration was probably lower, because it is unlikely that all the protein was presented in a functional form by the antibody. The other chimeras coated at similar densities. Assuming that the surface area of one face of an endothelial cell has an area of 10 3 m 2 (24), the density of FK in our assays would be equivalent to 4 ϫ 10 5 molecules/cell. Although the actual density of FK on endothelial cells has not been reported, other adhesion proteins are expressed at comparable levels. For example, thrombin-activated endothelial cells express Pselectin at 50 molecules/m 2 or a total number of 5 ϫ 10 4 molecules/cell (25). It is also interesting to note that we can see adhesion, albeit with reduced numbers of cells, to FK coated at 10% of this concentration (not shown).
We observed during these flow adhesion assays that the behavior of captured cells depended on the specific chemokine/ stalk substrate. We therefore further analyzed the interaction between the receptor-transfected cells and the chemokine/stalk chimeras. A single field of view in the flow chamber was selected, and cells that became tethered during an interval of 60 s were observed for an additional 30 s and classified as to whether they were arrested (remained stationary), rolled, or detached and re-entered the flow. Of the chemokine/stalk chi- meras examined, only FK was able to capture and retain cells (Fig. 6). In the case of the IL-8/stalk chimera, the great majority of the cells that were "captured" subsequently rolled on the substrate. In the case of MCP-1/stalk and RANTES/stalk, cells adhered only briefly, and the majority detached during the observation period. Thus, only in the case of FK did cells attach and remain firmly adherent for the entire 30 s. In addition, both US28-and CX 3 CR1-expressing cells showed a similar phenotype when interacting with FK (i.e. they arrested without rolling).
These data indicated that cell adhesion properties of FK were not simply due to the presentation of a chemokine-like domain at the top of a rigid stalk and thus suggested unique features of the interaction between FK and its receptor. To test this notion, we first asked whether cells expressing chemokine receptors could be "labeled" with the appropriate chemokine/ stalk chimera. A variant of this approach, using epitope-tagged FK, has been used to evaluate receptor expression on cells (8,9). FK/stalk bound to cells expressing both CX 3 CR1 and US28 but not to wild-type, untransfected cells (Fig. 7). The RANTES/ stalk labeled cells expressing US28 but failed to label cells expressing either CCR1 or CCR5. MCP-1/stalk and SLC/stalk also failed to label cells expressing their cognate receptors, but IL-8/stalk did bind to cells expressing CXCR1. These data indicated that the chemokine/stalk chimeras that supported cell adhesion under flow also bound to cells expressing the appropriate receptors with sufficient affinity to resist dissociation during the time and under conditions necessary to perform this flow cytometry assay.
We next asked whether the ability of FK to capture and retain cells under flow was due to a slow off-rate of the chemokine from its receptor. To address this question, we allowed the chemokine/stalk chimeras to bind to cells expressing the appropriate receptors and then added excess soluble chemokine either simultaneously with the chemokine/stalk or midway through the labeling period (Fig. 8). Excess soluble FK blocked binding of the FK/stalk chimera when the two were added simultaneously but not if the soluble FK was added 30 min after the FK/stalk. In contrast, addition of the soluble chemokine reduced binding of the IL-8/stalk significantly even when added after 30 min. FK binding to US28 gave a result intermediate between those obtained with FK-CX 3 CR1 and IL-8-CXCR1. In contrast, RANTES/stalk binding to US28 was almost completely eliminated by the addition of soluble RANTES, even after 30 min. Similar experiments could not be performed for the other chemokine/stalk-receptor pairs, because the initial binding affinity was too low (see Fig. 7). FIG. 4. Static adhesion of chemokine receptor cells. 300-19 cells stably expressing chemokine receptors were assayed for adhesion to antibody-tethered chemokine/stalk chimeras. Cells were loaded with the fluorescent dye BCECF-AM and allowed to adhere for 30 min. Nonadherent cells were washed off, and the number of adherent cells was determined from the mean fluorescence in two wells. Each point represents an independent assay, and the bars represent the means. The following ligand receptor pairs showed no significant difference (p Ͼ 0.05) from wild-type, untransfected cells adhering to FK: FK-CCR5, mFK-US28, MIP-1␤-CCR5, MCP-1-CXCR1, and IL-8-CX 3 CR1. The others were significantly different (p Ͻ 0.05), indicating specific adhesion. Adhesion of US28-expressing cells to FK/stalk or RANTES/stalk was significantly greater than to MCP-1/stalk or mFK/stalk (p Ͻ 0.05).
FIG. 5. Flow adhesion by chemokine/stalk chimeras. 300-19 cells stably expressing chemokine receptors were perfused over antibody-tethered chemokine/stalk at a wall shear stress of 0.8 or 1.2 dyn/cm 2 . Cells adhering after 3 min were counted in 10 or more fields, and the mean value was determined. Each point represents an independent assay, and the bars represent the means. A, 0.8 dyn/cm 2 , there was no significant difference between FK and mFK in capturing CX 3 CR1-expressing cells. All others were significantly lower than FK-CX 3 CR1 (p Ͻ 0.005). B, 1.2 dyn/cm 2 , all were lower than FK-CX 3 CR1 (p Ͻ 0.01), except US28, which could not be analyzed statistically, because it contains only one point.

DISCUSSION
In this study, we have created a series of chimeras in which soluble chemokines were fused to the mucin-like stalk of FK to test the hypothesis that the ability of FK to capture cells is due to the unique presentation of its chemokine-like domain. Under static conditions, a number of the chemokine/stalk chimeras bound cells expressing the cognate receptors. Under flow conditions, however, only FK/stalk and IL-8/stalk successfully captured cells. The majority of the CXCR1-expressing cells rolled on the IL-8/stalk chimera and were not firmly adherent. Finally, the ability of FK to capture cells expressing CX 3 CR1 and US28 correlated well with the slow off-rate of the chemokine from these two receptors. We conclude that presentation of chemokines atop a rigid mucin-like stalk is not sufficient to confer the ability to capture cells under flow conditions. Evidence from many groups supports a multistep model for leukocyte emigration from the bloodstream. The initial step is selectin-mediated rolling of unactivated cells along the vessel wall, followed by capture and diapedesis through the endothelium. One conclusion from the current study is that static adhesion assays are less discriminatory than simple flow assays in identifying molecules capable of capturing cells. Thus, under static conditions, human and murine FK/stalk, RAN-TES/stalk, and MCP-1/stalk had comparable abilities to bind cells, a result similar to that observed with various chemokine/ stalk:receptor pairs (8,26). However, under conditions of low shear, only the FK/stalk interactions remained robust, and all the other chemokine chimeras had significantly reduced adhesion. We were further able to distinguish between human and murine FK/stalk in their ability to capture cells at a slightly higher flow. Interestingly, the IL-8/stalk interaction represented cell rolling rather than firm adherence. Although rolling is typically attributed to selectin-mediated interactions, there FIG. 8. FK/stalk has a slower off-rate than other chemokines. 300-19 cell lines were labeled with chemokine/stalk (10 nM), and excess soluble chemokine (100 nM) was added at the beginning of the assay or at 30 min into the 60-min incubation. This was followed by labeling with an antipolyhistidine primary antibody and a fluorescein isothiocyanate-conjugated secondary antibody. Cells were then analyzed by fluorescence-activated cell sorting. These data are representative of two assays.
FIG. 6. Action after tethering. 300-19 cell lines were perfused over antibodytethered chemokine/stalk at a wall shear stress of 0.8 dyn/cm 2 . To determine the fate of cells that tethered, video was analyzed for the initial 60 s of flow. Every cell that tethered during this period was monitored for 30 s after the time of tethering. Cells were considered "detached" if they left the surface and re-entered bulk fluid flow. Rolling cells were defined as those that moved more than one cell diameter. "Arrested" cells remained stationary. Two assays were quantitated, and the means are shown (ϮS.D.). is evidence that the integrin ␣ 4 ␤ 1 (VLA-4) also mediates rolling (27). Similar to what was seen with ␣ 4 ␤ 1 (27), increased concentrations of IL-8/stalk induced firm adhesion, rather than rolling, in a subpopulation of the cells (not shown). Because IL-8 and other chemokines bind well to heparin sulfate proteoglycans, it is possible that they too contribute to rolling along the endothelium (28 -30).
US28, a seven-transmembrane protein encoded by cytomegalovirus, is also a high-affinity receptor for FK (31). Although it is a promiscuous receptor that binds many chemokines, Kledal et al. (31) found that US28 binds FK preferentially with subnanomolar affinity and speculated that it has been "optimized" through evolution to promote viral entry through binding to FK on the host cell surface. Consistent with these data, we found that, although cells transfected with US28 adhered to many chemokines (FK/stalk, RANTES/stalk, MCP-1/stalk) under static conditions, only FK/stalk captured US28-expressing cells under flow conditions. Of the two receptors for FK, CX 3 CR1 appeared to capture cells more efficiently than US28. A caveat to this conclusion, however, is that US28 was expressed at somewhat lower levels than CX 3 CR1 on the 300-19 cells. Whether or not the binding of US28 to FK actually contributes to the entry of cytomegalovirus into mammalian cells remains to be determined.
The three-dimensional structure of the chemokine-like domain of FK was recently solved by Mizoue et al. (32), and several features were found that distinguish it from the other chemokines whose structures are known. For example, FK exists as a monomer at high concentrations, and the CX 3 C motif forms a "bulge" in the structure. It is not known if these structural features are important for FK's unique ability to act as a cell adhesion receptor; however, the only other chemokine/ stalk chimera to support adhesion was IL-8, a CXC chemokine. It is also interesting to note that a potential N-glycosylation site is present within the CX 3 C motif ( 33 NX(S/T)) in human FK, but not in murine FK. In our studies, murine FK failed to support adhesion of US28-expressing cell lines. Mutation of this asparagine or treatment with N-glycosidases may reveal whether these functional differences are due to glycosylation of human FK in this critical region.
The ability of FK to capture and retain cells under flow conditions suggested that the off-rate from its receptor might be considerably slower than typical chemokine-receptor interactions. Indeed, preliminary experiments revealed that we could actually label cells expressing CX 3 CR1 or US28 with epitope-tagged FK and that the binding was sufficiently robust to be easily detected by flow cytometry. Competition binding experiments revealed that, after 30 min at 22°C, a substantial portion of the binding of FK/stalk to its receptor could no longer be displaced by soluble FK, whereas they competed equally well when added simultaneously. The stalk domain of FK was not critical in determining the off-rate, because incubation with soluble FK also blocked subsequent labeling with excess FK/ stalk (not shown). Overall, we saw a good correlation between those chemokine/stalk chimeras that supported cell capture and those whose receptor binding was poorly reversible. This was particularly evident in the case of US28, in which FK had a very slow off-rate, whereas a different ligand, RANTES/stalk, was almost fully reversible. Furthermore, soluble FK blocked adhesion of US28 cells to RANTES/stalk, but RANTES/stalk did not block the adhesion of US28 to FK/stalk (not shown). The reason why the FK/CX 3 CR1 interaction is less reversible than the interaction of other chemokine/receptor pairs is not clear but may represent dimer or higher-order cooperative interactions between receptors, as has been reported for CCR2, CCR5, and CXCR4 (33)(34)(35).
It is important to note that cell adhesion represents the overall affinity of the cell for the immobilized ligand and is thus a much more complex parameter than the equilibrium dissociation constant. In general, however, the range of reported dissociation constants for FK/CX 3 CR1 (30 -740 pM) suggests that it has a higher affinity than the other chemokine/receptor pairs in this study (8,15,36). In our hands, chemokine/receptor pairs had dissociation constants as follows: FK/CX 3 CR1 (100 Ϯ 40 pM), IL-8/CXCR1 (1100 Ϯ 400 pM), and MCP-1/CCR2 (260 -650 pM) (17,21). Others have reported a range of 1100 -3600 pM for IL-8/CXCR1 and 35-680 pM for MCP-1/CCR2 (37)(38)(39)(40). The wide range of values in the literature probably reflects differences in experimental systems and assay methods. It is possible, therefore, that FK's unique ability to capture cells is due in part to a higher affinity for its receptor than those seen in other chemokine/receptor pairs. In summary, we have shown that the ability of FK to act as a robust cell adhesion receptor is not simply due to its unique architecture, with prominent display of the chemokine-like domain. These results may be relevant to the question of whether the binding of other chemokines to heparin sulfate proteoglycans on the surface of endothelial cells functionally mimics FK. Although the basis for cell adhesion by FK is not revealed in these studies, it is clear that the chemokine-like domain is crucially important for this unique function. Whether or not this is due to its CX 3 C motif remains to be determined.