Abstract
STRL33/BONZO/TYMSTR is an orphan chemokine and HIV/SIV coreceptor receptor that is expressed on activated T lymphocytes. We describe an expression cloning strategy whereby we isolated a novel chemokine, which we name CXCL16. CXCL16 is an α (CXC) chemokine but also has characteristics of CC chemokines and a structure similar to fractalkine (neurotactin) in having a transmembrane region and a chemokine domain suspended by a mucin-like stalk. A recombinant version of CXCL16 fails to mediate chemotaxis to all known chemokine receptor transfectants tested but does mediate robust chemotaxis, high affinity binding, and calcium mobilization to Bonzo receptor transfectants, indicating that this is a unique receptor ligand interaction. In vitro polarized T cell subsets including Th1, Th2, and Tr1 cells express functional Bonzo, suggesting expression of this receptor in chronic inflammation, which we further verified by demonstration of CXCL16-mediated migration of tonsil-derived CD4+ T lymphocytes. CXCL16 is expressed on the surface of APCs including subsets of CD19+ B cells and CD14+ monocyte/macrophages, and functional CXCL16 is also shed from macrophages. The combination of unique structural features of both Bonzo and CXCL16 suggest that this interaction may represent a new class of ligands for this receptor family. Additionally, this chemokine might play a unique dual role of attracting activated lymphocyte subsets during inflammation as well as facilitating immune responses via cell-cell contact.
The chemokines comprise an ever growing family of small secreted proteins involved in a number of inflammatory and immunological processes including lymphocyte homing, suppression and stimulation of angiogenesis, suppression of HIV infection, and enhancement of CTL responses (1, 2, 3). All chemokines to date (with one exception noted) exhibit a conserved structure that is predominated by conserved disulfide bonds, Cys1-Cys3 and Cys2-Cys4 (1, 3). To date there are four subfamilies, based on the arrangement of N-terminal cysteine residues including the CC, CXC, and one member each of the C and the CX3C whereby, in each case, X represents any residue other than cysteine (1, 3). The selective binding of the chemokine subfamily members to G protein-coupled receptors (GPCRs) results in the subclassification of chemokine receptors as CCR1–11, CXC chemokine receptor (CXCR)41–5, XCR1, and CX3CR1 (1, 2, 4, 5, 6, 7, 8). The selectivity of these receptor ligand interactions will, in part, contribute to the specificity by which leukocyte subsets might preferentially localize to different extravascular sites (4, 9, 10). However, in many instances, as multiple receptors appear to be expressed on individual subsets it is likely that overlapping gradients of different chemokines and combinations of receptor ligand interactions might determine directionality of leukocyte trafficking from the circulation into tissues (9, 11).
Recently, several GPCRs have been cloned with no known natural ligands and are thus referred to as “orphan GPCRs” (12). One orphan receptor referred to as STRL33/BONZO/TYMSTR (referred to herein as Bonzo), simultaneously identified in three groups, was shown to be expressed on activated T cells, and can serve as an HIV/SIV coreceptor. Subsequent studies have shown Bonzo to be a receptor more frequently expressed on memory T cells with a preferential expression on CD8-positive T cells (but is expressed on CD4-positive cells as well) (13, 14, 15). As no known chemokines have been shown to interact with Bonzo, it has remained an orphan receptor, while sequence comparisons strongly suggest that it is a CCR. Here, we use a novel expression cloning strategy to isolate a cDNA that upon transfection into 293 cell recipients mediates a robust chemotactic response to Bonzo receptor transfectants. This cDNA encoded a novel chemokine, which we term CXCL16.
Although it is, in the strictest definition, a member of the CXC family, CXCL16 is distantly related to all known chemokines and might be phylogenetically closer to certain β chemokines than α chemokines. Additionally, it has a novel feature that is shared with the only known CX3C chemokine, neurotactin/fractalkine, in that sequence predicts that it is membrane bound and suspended by a heavily glycosylated mucin stalk (16, 17). We show that Bonzo/CXCL16 interactions define a unique receptor ligand pair as CXCL16 fails to bind all known chemokine receptors. We also show in this study that this ligand functionally interacts with activated lymphocytes and describe novel features of expression on leukocyte subsets. These data, along with the structural features of both Bonzo and CXCL16, indicate that this receptor ligand pair is unique among all members of this family known to date.
Materials and Methods
mAbs, isolation of primary cells, and cell lines
Medium for transfectants was standard DMEM with 10% FCS (Life Technologies, Gaithersburg, MD). L1-2 cells, a murine B cell lymphoma, were obtained from E. Butcher at Stanford University (Stanford, CA). Receptor transfectants expressing Bonzo were generated as previously described (18, 19). Chemokine receptor transfectants encoding CXCR5, CCR1–9, and CX3CR1 were maintained in RPMI 1649 medium (Life Technologies) with 0.88 g/L gentamicin (G418), 10% HyClone serum, 10 nM HEPES, 1% penicillin/streptomycin, 1% l-glutamine, 1 mM sodium pyruvate, and 55 nM 2-ME.
Chronically activated Th1, Tr1, and Th2 lymphocytes were prepared as previously described (20, 21) with the addition of generation of Tr1 subsets. For Tr1 lymphocytes IL-10 was used at 10 ng/ml. After initial activation in the presence of anti-CD28 and anti-CD3 mAbs Th1, Tr1, and Th2 lymphocytes were restimulated for 5 days with specific cytokines, but with the addition of anti-CD95 ligand (1 μg/ml) to prevent apoptosis. Activated Th1, Tr1, and Th2 lymphocytes were maintained in this way for a maximum of three cycles.
Tonsils were obtained from Massachusetts Eye and Ear (Cambridge, MA). Tissue was macerated with surgical scissors, mixed with DMEM (Life Technologies) and passed through a cell strainer (Becton Dickinson, Franklin Lakes, NJ). Cells were washed three times with PBS before staining. CD4+ T cells were isolated by positive magnetic selection using CD4 microbeads (Miltenyi Biotec, Auburn, CA) and the manufacturer’s instructions.
PBMC were isolated by density gradient centrifugation using Lymphoprep (Nycomed, Oslo, Norway). Monocytes were seeded into T75 flasks, allowed to adhere, and cultured for 10 days in RPM1 1640 supplemented with 2.5 mM HEPES, 20 μg/ml gentamicin, 2 mM l
MAbs reactive with Bonzo were generated by immunizing mice with L1.2 cells expressing high levels of transfected Bonzo, as previously described (1922). MAbs to CXCL16 were generated by immunization of BALB/C mice with the synthetic chemokine and screening for reactivity by ELISA.
FACS staining
Before the addition of primary Ab, PBMCs or tonsil cells were incubated in PBS with 5% True Clot human serum (Scantibodies, Santee, CA) and mouse IgG (Sigma) to prevent nonspecific staining. Cells were incubated for 30 min with specific mAbs or isotype controls (Sigma). FITC- or Cy5-goat anti-mouse IgG Abs (Jackson ImmunoResearch, West Grove, PA) were used as the secondary Abs at 1/200 dilutions. Abs against all surface Ags including CD4, CD14, and CD19 directly conjugated to PE were obtained from PharMingen (San Diego, CA). Cell staining was analyzed on a FACScan (Becton Dickinson) using the CellQuest program.
cDNA library expression cloning
An expression library was made from spleen mRNA purchased from Clontech Laboratories (Palo Alto, CA). cDNA was prepared as previously described (23) using synthesis reagents from Life Technologies with the exception that an ECOR1 adapter (Pharmacia, Piscataway, NJ) was ligated to the 5′ end to facilitate directional cloning into the elongation factor-1 (EF1)-based vector pcDEF3. cDNA was partitioned into 96 pools of 1000 colonies each, and plasmid DNAs were prepared as previously described (23).
Transfections were performed into 293T cells as described previously (23) with the exception that 60,000 cells/well were plated in collagen-coated plates (Becton Dickinson). Approximately 18 h after transfection, medium was changed to 0.5 ml/well of standard DMEM/10% FCS. Forty-eight hours after changing medium, supernatant was harvested, cell debris was removed by microcentrifugation, and the medium was then used in chemotaxis assays.
For the chemotaxis assays in the expression-cloning screen (and with macrophage supernatants), exponentially growing Bonzo transfectants were resuspended at a density of 1 × 107/ml in an assay buffer that consisted of DMEM supplemented with 10% bovine calf serum. The cell suspension (100 μl) was placed in the upper chamber of a 24-well chemotaxis plate (Costar; Corning Glass, Corning, NY), and 0.5 ml of the supernatant from each transfected well was placed in the lower chamber. The plates were then incubated for 6–24 h at 37°C. Numbers of migrating cells were quantitated on a FACScan (Becton Dickinson) using the acquisition phase at 30 s. cDNA pools that, upon transfection, mediated chemotaxis above background were subsequently enriched as previously described (23).
Chemotaxis with purified CXCL16 (described below) and L1-2 receptor transfectants was identical with the set-up for expression cloning with the exception that defined concentrations of chemokines were used and the buffer was changed to 50% RPMI, 50% M199 medium, and 0.5% BSA. Chemotaxis with in vitro-derived effector cells and isolated CD4+ tonsil lymphocytes was performed using 3-μm pore diameter gelatin-coated transwell inserts followed by growth of 2 × 105 ECV304 cells as previously described (20). An aliquot of 200 μl of cell suspension (input of 8 × 105 cells) was added to each insert. After 2 h, the inserts were removed and the number of cells that had migrated through the ECV304 monolayer to the lower well was counted for 30 s on a Becton Dickinson FACScan with the gates set to acquire the cells of interest. Using this technique 100% migration would be 25,000 cells for Th1/Th2 cells, where this number represents the cells in the lower well counted on the FACScan for 1 min. In all cases the data points were the result of duplicate wells, with the mean value shown and the error bars representing the sample SD.
Sequencing and sequence analysis
Sequencing of the entire cDNA insert was accomplished in conjunction with Seqwrite (Houston, TX) and the Tufts University sequencing core facility. Overlapping primers (originally using an SP6 primer for the 3′ end and a primer from the EF1 promoter for the 5′ sequence) made to both strands as sequence information was gathered, resulting in complete unambiguous sequence of both strands of the insert. Sequence comparison with known chemokines was performed with the Lasergene system (DNAstar, Madison, WI), using the Clustal method with a gap penalty of 10 and a gap length penalty of 10. Pairwise alignment parameters were: ktuple = 2, gap penalty-5, window = 4, and diagonals saved = 4.
Construction and purification of recombinant CXCL16 His-tagged protein
Fusion proteins consisting of regions of CXCL16 fused to a C-terminal 5× histidine (His) were made in the pEF1/V5-His vector from Invitrogen (Carlsbad, CA) by fusion with a PCR-generated fragment containing the entire extracellular domain of CXCL16. A 5′ primer starting at the initiation methionine with a BamHI site and a 3′ primer ending at Val155 with an XbaI site were used in the PCR. Primer sequences are available upon request. PCR inserts were purified from agarose gels along with vector DNA digested with BamHI and XbaI and recombinants isolated by standard procedures. Maxiprep (Qiagen, Chatsworth, CA) DNA was used in transient transfections of 10-cm plates using Lipofectamine (Life Technologies). Three days after transfection, supernatant was collected and centrifuged, and the clarified supernatant was then run over 5 ml wheat germ agglutinin conjugated to agarose in a column (Vector Laboratories, Burlingame, CA). PBS was used as the wash buffer and eluted with 100 mM acetic acid (pH 2.8). The 10-ml elution was brought to pH neutrality with 5 ml 1 M Tris base (pH 10.5) and run over an nickel-nitrilotriacetic acid-agarose column (Qiagen) washed with 50 mM NaH2PO4 (pH 8), 300 mM NaCl, and 20 mM imidazole. Elution was performed with 50 mM NaH2PO4 (pH 8), 300 mM NaCl, and 250 mM imidazole in 5 × 1-ml aliquots. Samples were dialyzed using a 3-ml 10,000 m.w. cutoff Slide-a-lyzer (Pierce, Rockford, IL) into PBS overnight. OD280 readings were taken to determine the protein concentration.
A Glycoprotein Deglycosylation Kit (Calbiochem, San Diego, CA) was used to deglycosylate 2 μg of CXCL16 protein. The standard denaturing protocol was followed, and 150 ng each of untreated and deglycosylase-treated protein was run on a 4–20% Tris-glycine gel (Novex, San Diego, CA). Protein was transferred onto a nitrocellulose membrane (Novex) and hybridized with a CXCL16 mAb (SD7). The NEN (Boston, MA) Renaissance system was used for detection.
Receptor binding assays
Exponentially growing L1.2 Bonzo transfectants were counted on the day of the assay and resuspended in binding assay buffer (BAB; 10 mM HEPES/1 mM CaCl2/5 mM MgCl2/0.5% BSA/0.05% sodium azide at 2.5 × 106/ml) at a density of 2.5 × 106/ml. Purified CXCL16 was labeled with 125I using a sodium iodide method (Amersham, Arlington Heights, IL) and diluted in BAB to 4 nM. Cold CXCL16 (in PBS at 5.3 μM) was diluted for competition in BAB to 0.4, 2, 4, 20, 40, and 400 nM. Reactions consisted, in triplicate (for a final volume of 100 μl) of 50 μl of cells (1.25 × 105 total), 25 μl 125I-CXCL16 (final concentration = 1 nM), and 25 μl cold CXCL16 serially diluted from 100 to 0 nM. The specific activity of the labeled ligand was 7.8 × 1010 nmol/cpm, total counts bound were 813.67, nonspecific counts were 222.33 resulting in 591.34 specific counts binding. For calculation of total 125I input, 50 μl cells was added to 25 μl 125I-CXCL16 and 25 μl BAB. All tubes were incubated for 1 h at room temperature. Cells were spun down at 3500 rpm and washed five times in BAB + 0.5 M NaCl. After the final wash, the cell pellet was resuspended in 100 μl wash buffer, and the 125I counts were calculated by a Cobra II Auto-Gamma scintillation counter (Packard, Meriden, CT) along with the total input sample. Binding data were calculated using a program written in Excel (L. Wu, unpublished observation).
Calcium flux assays
Bonzo/L1.2 transfectants or parental cells were washed once in PBS and resuspended in load buffer (HBSS, 20 mM HEPES, 2.5 mM probenecid, 0.1% BSA, and 1% FBS). Fluo-3 (Molecular Probes, Eugene, OR) was dissolved in 50% DMSO/50% pluronic acid and added to the cells at a final concentration of 4 μM. Cells were incubated for 1 h at 37°C. Then, cells were washed twice in load buffer and plated into 96-well assay plates at 300,000 cells per well. The plate was spun for 5 min at 1200 rpm to pellet cells on the bottom of the well. Chemokine (50 μl) was added to a separate 96-well plate at varying concentrations to achieve final concentrations as indicated in the figure. Ca2+ mobilization was then measured on a 96-well FLIPR System (Molecular Devices, Sunnyvale, CA).
Northern and Southern blots
Human multiple tissue mRNA blots (Human I, Immune System II) from Clontech Laboratories were used for Northern blot analysis. A 400-bp cDNA fragment representing the chemokine domain from the 5′ EcoRI site to a EcoRV site was used as the probe template and primed with random hexamers to produce an [α-32P]dCTP-labeled probe. Hybridization was performed according to the manufacturer’s instructions with four additional high stringency washes at 65°C in 0.1× SSC and 0.1% SDS, and then exposed to Kodak (Rochester, NY) XAR film with an intensifying screen. Blots were stripped and reprobed with a β-actin probe, the template of which was provided by the blot manufacturer.
For Southern blots, human genomic DNA (15 μg/reaction; Clontech Laboratories) was digested by BamHI, EcoRI, and HindIII and run on a 1.4% agarose gel. DNA was transferred to a Protran nitrocellulose membrane (Schleicher & Schuell, Keene, NH) and cross-linked with a Stratalinker. The probe template was the same as for the Northern blots. The blot was washed three times for 30 min at 60°C in 1× SSC and 0.1% SDS, and was exposed to film for 2 days at room temperature.
Results
Expression cloning reveals a novel chemokine
L1.2 Bonzo transfectants were tested in chemotaxis assays against a panel of all known chemokine receptor ligands (along with orphan chemokines), and no response was noted in any physiological dose range (data not shown). As this data suggested that Bonzo likely binds an unknown ligand, we subsequently devised a strategy whereby we could transfect pools of cDNAs, harvest supernatants, and ask whether we could detect a chemotactic response. As Bonzo had previously been shown to be highly expressed in spleen (13, 14), an expression library was constructed from spleen mRNA. Transient transfection of 96 pools (representing ∼1000 independent clones/pool) revealed one pool with a response above background. Subsequent rounds of enrichment increased the activity, and eventual deconvolution resulted in a single cDNA clone that, upon transfection, mediated a robust chemotactic response to Bonzo receptor transfectants (data not shown).
Sequence analysis reveals a novel chemokine
The isolated cDNA revealed a large open reading frame of 254 aa with an N-terminal hydrophobic predicted signal peptide of 27 aa (Fig. 1⇓A). The signal peptide is followed by 90 residues with a total of six cysteines, with an N-terminal CYC motif consistent with members of the α chemokine family (1, 3). The predicted chemokine domain is followed by a region that is largely composed of S/T/G/P residues (37/84 aa or 44%), which is the hallmark of a mucin domain (23), followed by a hydrophobic membrane spanning a region of 25 residues and a short 28-residue cytoplasmic tail, containing two additional cysteines in each of these regions. This overall structure is similar to the one CX3C chemokine, fractalkine/neurotactin (16, 17), while a C-terminal mucin-like region has also been observed for JE (murine monocyte chemoattractant protein-1) (24). Although this protein must be shed or proteolytically cleaved for chemotactic activity, there are no known dibasic cleavage sites before the transmembrane region.
Sequence of phylogenetic analysis of CXCL16. A, Predicted signal peptide and transmembrane regions are indicated by a bold underline. Past the signal peptide and before the line demarcating the mucin domain is the predicted chemokine domain. All cysteines are indicated in bold. B, Dendrogram showing phylogenetic relationship of CXCL16 with selected α chemokines as well as MIP-1α and MIP-1β. These sequence data are available from GenBank under accession number AF337812.
Alignment of this new sequence (using only the chemokine domain) with a number of other α chemokines shows it is distantly related to all CXC chemokines with the best homology with stromal cell-derived factor (SDF)-1 at 18.8% (Fig. 1⇑B). Additional searches with different regions of the protein sequence reveal no homology to any proteins in the database. Although the predicted protein has characteristics of both CXC and CX3C chemokines, similarities to CC chemokines exist as well, including 1) homology (closer than to any CXC chemokines) to macrophage-inflammatory protein (MIP)-1β at 22%; 2) similarity to Eskine/CTAK and Teck in having 27 residues between Cys 2 and 3, while this loop is generally no longer than 24 residues in all other chemokines (25); and 3) a relationship to secondary lymphoid chemokine (SLC)/6-C-Kine in having six cysteine residues in its chemokine domain (26, 27). Due to the CYC motif, we provisionally refer to this protein as a novel CXC chemokine termed CXCL16 (28) and we propose to rename STRL33/BONZO/TYMSTR as CXCR6.
Expression of CXCL16 RNA and Southern blot analysis
To examine the tissue distribution of CXCL16, Northern blots were probed with a fragment encompassing the chemokine domain (Fig. 2⇓A). A band of ∼2.4 kb was seen with the highest expression in lung, liver, fetal liver, spleen, and peripheral blood leukocytes. Lower levels of expression were also observed in kidney, pancreas, lymph nodes, and placenta. Reprobing these blots with a probe corresponding to the full length CXCL16 cDNA resulted in extreme multiple banding, most likely due to repeat sequences found in the 3′-most end of the cDNA and parts of the mucin sequences as well (data not shown). Additionally, prior exposure of the blot before more stringent washes revealed the presence of other sized RNA species in certain tissues, indicating the possibility that related genes are expressed in some of the tissues examined (data not shown).
Analysis of CXCL16 RNA expression and genomic organization. A, Membranes with poly(A)+ RNA from the indicated human tissues were hybridized with a chemokine domain CXCL16 probe (upper panels). β-Actin loading controls are shown in the panels below. The lower bands seen in muscle tissue lanes correspond to another β-actin isoform (Clontech Laboratories technical support, unpublished observations). B, Southern blot using the same probe, hybridized to genomic DNA digested with restriction enzymes as indicated.
Southern blots containing human genomic DNA digested with EcoRI, BamHI, or HindIII were also probed with the CXCL16 chemokine domain probe (Fig. 2⇑B). At moderate stringency, only one (BamHI and HindIII) or two (EcoRI) bands were observed, consistent CXCL16 existing as a single copy gene.
CXCL16 is expressed on the surface of leukocyte subsets and functional CXCL16 is shed from macrophages
Although sequence analysis suggests that CXCL16 can be presented as a cell surface molecule, we could not detect surface expression of CXCL16 on our 293 transfectants (data not shown). As Northern blot analysis showed high expression in peripheral blood leukocytes, we asked whether we could detect cell surface expression of CXCL16 on any leukocyte subsets. Staining was observed on subsets of CD19+ and CD14+ peripheral blood leukocytes, indicating that APCs including B cells and monocyte/macrophages express this chemokine as a cell surface ligand (Fig. 3⇓A). Although there was considerable variation in the percentage of each cell type expressing CXCL16, this general staining pattern was observed in all donors examined (Fig. 3⇓B). These observations raise the possibility that cell-cell contacts between activated T cell subsets known to express CXCR6 and APCs expressing CXCL16 might result in subset-specific immune responses. In addition to peripheral blood we also examined expression in tonsils and also found that subsets of CD19+ B cells stained specifically with anti CXCL16 mAbs (Fig. 3⇓C), suggesting that in settings of chronic inflammation CXCL16 might participate in cell-cell interactions as well.
CXCL16 is expressed on the surface of human monocytes and B cells and on tonsillar B cells A, PBMC from several donors were stained with an anti-CXCL16 mAb (SD7). Expression in one representative donor was observed in both monocytes (CD14+) and B cells (CD19+). IgG1 isotype controls are in the bottom panels. B, Percentage of CXCL16+ cells varied by donor. C, Expression of CXCL16 on CD19 subsets of freshly isolated tonsil lymphocytes.
We also asked whether functional chemokine could be shed from the surface of the leukocyte subsets expressing surface CXCL16. Cultured macrophages were propagated by adherence to plastic, and CXCL16 expression was still observed after several days in culture (data not shown). After replacement of medium, supernatants were collected in the presence or absence of inflammatory mediators and examined for chemotactic activity. No increase of cell surface staining was observed after 4- or 24-h incubation with LPS or TNF-α (data not shown). After 4 h, minimal chemotactic activity was seen above background, with no discernable difference between supernatants from unstimulated and stimulated cells. However, after 24 h, significant increase in activity was seen that was moderately increased by LPS and increased by ∼2-fold in the presence of TNF-α, whereas control orphan GPCR (GusB/CCR11) transfectants fail to chemotax (Fig. 4⇓). These data indicate that functional chemokine was shed into the medium, suggesting that either increases in gene expression or processing might contribute to observed increase in biological activity.
Chemotaxis induced by CXCL16 secreted from human macrophages. Supernatants taken from 4- and 24-h cultures of monocyte-derived macrophages induce chemotaxis of CXCR6/L1-2 transfectants. Levels of secreted CXCL16, as measured by chemotaxis, increased after 24-h stimulation with LPS and TNF-α, whereas control GusB (CCR11) transfectants fail to chemotax.
Activated T lymphocyte subsets express CXCR6 and functionally respond to CXCL16
CXCR6 was originally cloned from activated T lymphocytes, and recent reports have demonstrated the expression of CXCR6 on activated CD4+ and CD8+ T lymphocyte subsets along with NK cells (13, 14, 29, 30, 31). To look at expression and function of Bonzo in activated T cells, we generated T helper subsets by polarization in the presence of specific Th1, Th2, and Tr1 cytokines in conditions of repeated rounds of cytokine stimulation to mimic settings of chronic inflammation (21). Upon initial expansion in the presence of Th1, Th2, or Tr1 cytokines, we observed expression of CCR7 on all subsets, whereas CCR4 (as previously demonstrated) is only expressed on Th2 cells, and CXCL16 was low or undetected on all subsets (Fig. 5⇓A). Upon multiple rounds of cytokine stimulation, levels of CCR7 were reduced in all subsets, suggesting a transition to a memory phenotype, whereas CCR4 expression was greatly increased on Th2 cells. CXCR6 expression was also greatly increased on all three subsets, indicating that CXCR6 expression might mark effector cells in settings of chronic inflammation (Fig. 5⇓A). Upon examination of chemokine responsiveness, we observed that all three subsets subjected to multiple rounds of cytokine stimulation respond to RANTES, a ligand for both CCR5 and CCR1 (which marks all subsets), whereas only Th2 cells migrate in response to the CCR4 ligand macrophage-derived chemokine (MDC) (Fig. 5⇓B). All three subsets migrate with similar efficiency to CXCL16, showing that CXCR6 is functional on effector T cell subsets.
Expression and function of CXCR6 on effector cell subsets. A, Th1, Th2, and Tr1 effector cells were generated from umbilical blood CD4+ lymphocytes and stained after 1 cycle (primary stimulation) of cytokine stimulation with anti CCR7 mAb 7H12, anti CCR4 mAb 1G1, and anti-CXCR6 mAb 7F3. B, After two rounds of cytokine stimulation (secondary stimulation) mAb staining was repeated. C, After repeated rounds of stimulation, only Th2 cells migrate to MDC, whereas all three subsets show similar response to both RANTES and CXCL16. Chemokine concentrations were 100 nM for CXCL16 and 100 ng/ml of MDC and RANTES.
We next asked whether activated cells from a chronically inflamed tissue also express CXCR6 and respond to CXCL16. Staining of tonsil-derived lymphocytes demonstrates that a subpopulation of CD4+ T cells expresses CXCR6 (Fig. 6⇓A). Additionally, CXCL16 also mediates chemotaxis of the isolated CD4-positive cells. The observed chemotaxis is selectively blocked by anti-CXCR6 mAb 7F3, which indicates that this is a unique receptor ligand interaction (Fig. 6⇓B).
Expression and function of CXCR6 on tonsil CD4-positive lymphocytes A, CD4+ T cells were isolated from fresh tonsils and stained with anti-CXCR6 mAb 7F3 or an isotype control. B, Tonsil-derived CD4+ T cells chemotax to the indicated concentrations of CXCL16. Anti-CXCR6 mAb 7F3 reduced migration to background levels. CXCR6 L1.2 transfectants were simultaneously run in the same assay as a control.
Recombinant CXCL16 is heavily glycosylated
To further examine chemokine binding and receptor specificity, a recombinant form of CXCL16 was constructed into a fusion protein, which encompassed the entire predicted extracellular domain fused to a C-terminal polyhistidine sequence. Although CXCL16 has only one predicted N-linked glycosylation site (NETT at resides 168–171), we do predict a heavily glycosylated mucin sequence that is likely to be rich in O-linked glycans. The expression construct encoding recombinant CXCL16 was transfected into 293T cells and purified over successive wheat germ agglutinin and nickel-nitrilotriacetic acid columns. Western blotting with a polyclonal Ab to the chemokine domain detects a prominent species of Mr of ∼40 kDa, which is twice the size of the predicted protein backbone of Mr 19 kDa. Treatment of the recombinant protein with a mixture of deglycosidases including N-glycosidase F, endo-α-N-acetylgalactosaminidase, α2–3,6,8,9-neuraminidase, B1,4-galactosidase, and β-N-acetylglucosaminidase resulted in a significant increase in mobility and a shift in Mr from 40 to 23 kDa (Fig. 7⇓). This verifies the prediction that CXCL16, similar to fractalkine, is highly glycosylated (16).
Analysis of purified CXCL16 protein. Purified protein (150 ng) was resolved on a 4–20% Tris-glycine gel, transferred to nitrocellulose, and probed with a 1/5 dilution of a supernatant containing an anti-CXCL16 mAb. Mobility is decreased to a size roughly corresponding to the protein backbone after incubation in the presence of a glycosidase mixture including N-glycosidase F, endo-α-N-acetylgalactosaminidase, α2–3,6,8,9-neurominidase, β1,4-galactosidase, and β-N-acetylglucosaminidase.
Only CXCR6 functionally responds to CXCL16
Purified CXCL16 was tested in a chemotaxis assay and exhibited a robust response to CXCR6-L1-2 transfectants, showing a typical bell-shaped response with peak activity ranging from 10 to 50 nM (Fig. 8⇓A). CXCL16 was then tested in a chemotaxis assay against a panel of all known chemokine receptors including CCR1-9, CXCR1-5, and CX3CR1. Although known ligands for these receptors exhibited a robust response in this assay (data not shown), recombinant CXCL16 failed to mediate a response to any receptor other than CXCR6 at all concentrations tested (Fig. 8⇓B). Although this experiment does not exclude the possibility that CXCL16 might bind to these (and other nonchemokine-binding GPCRs), these data clearly illustrate that, for known chemokine receptors, only CXCR6 functionally responds to CXCL16. We next asked whether CXCL16 could specifically signal a rise in intracellular calcium in the CXCR6-transfected cell line. CXCL16 mediates a dose-dependent rise in intracellular calcium that is similar to that seen for SDF-1α, the ligand for CXCR4, which is expressed on L1-2 cells (Fig. 8⇓C). As a control, parental L1-2 cells only responded to SDF-1α, showing that CXCR6 can specifically signal in response to CXCL16.
Selective chemotaxis and calcium flux of L1.2 Bonzo transfectants in response to CXCL16. CXCL16 mediates selective chemotaxis to L1.2 CXCR6 (A and B) transfectants. A, Dose response of L1.2 Bonzo transfectants to CXCL16 in a chemotaxis assay. B, Results of chemotaxis assay using 10 nM CXCL16 in chemotaxis buffer and the following L1.2-transfected cell lines: CXCR1, CXCR2, CXCR3, CXCR4, CXCR5, and Bonzo. Chemotaxis buffer was used to measure background chemotaxis. Also tested (data not shown): CCR1, CCR2, CCR3, CCR4, CCR5, CCR6, CCR7, CCR8, and CCR9. All assays were also performed at 100 nM with similar results (data not shown). C, CXCL16 mediates specific calcium flux in only CXCR6/L1.2 transfectants. SDF-1α mediates calcium flux in both CXCR6 and control L1.2 cells, indicating a response to endogenously expressed CXCR4.
Radiolabeled CXCL16 was used as a probe to directly examine binding to the CXCR6-transfected L1.2 cells. Increasing concentrations of unlabeled CXCL16 competitively inhibited 125I-labeled CXCL16 binding, with an IC50 of 1 nM (Fig. 9⇓A), whereas labeled CXCL16 could not be inhibited by nonspecific chemokines (data not shown). Additionally, CXCL16 binding was not specific for control chemokine receptor transfectants including CCR6 and CCR7 (data not shown). Scatchard analysis demonstrates that binding is high affinity with an average KD of 1 nM and with ∼4000 binding sites/transfectant, indicating that CXCL16 is a high affinity, selective ligand for CXCR6 (Fig. 9⇓B). These data indicate that although CXCL16 might have a function in direct cell-cell interactions as has been postulated for fractalkine, it is clearly capable of functioning in a manner similar to more classical chemokines. Soluble CXCL16 might be rapidly shed from the cell surface and participate in recruitment of lymphocyte subpopulation to sites of immune and inflammatory responses as well. It is possible that mechanisms of shedding in a particular microenvironment may represent another level of control of CXCL16-mediated physiological responses.
Binding of CXCL16 to the CXCR6 chemokine receptor in L1.2-transfected cells. A, 125I-CXCL16 is specific for L1.2 CXCR6 transfectants. Panel shows binding activity of L1.2 CXCR6 transfectants with 1 nM 125I-CXCL16 with or without increasing concentrations of cold competitor. The calculated value of 50% inhibition is 1 nM CXCL16. B, Scatchard analysis of binding data. Results shown are the average of two experiments. The calculated Kd is 1 nM.
Discussion
We report the use of an expression cloning strategy to isolate a novel chemokine that mediates chemotaxis of receptor transfectants expressing the orphan GPCR, STRL33/Bonzo/TYMSTR (13, 14, 31), This chemokine contains an N-terminal CXC motif and is now named, according to recent systematic nomenclature (28), CXCL16; we also renamed STRL33/Bonzo/TYMSTR as CXCR6. This receptor ligand pair appears unique as no other receptors tested functionally respond to this new ligand. Although we have not ruled out the fact that CXCL16 might directly bind to other chemokine receptors (and other GPCRs as well), our data suggest that this interaction represents one of the few examples of exclusivity between a single ligand and chemokine receptor. It should also be noted that a murine cDNA encoding CXCL16 was also identified as a Bonzo ligand, and the human ortholog was also reported with essentially the same sequence as the one reported in this manuscript, with the exception of a longer signal peptide (32). This longer sequence is likely incorrect, as 1) we functionally cloned a shorter version (which is efficiently secreted) lacking this extended sequence; 2) this extended sequence does not align with the murine sequence; and 3) it fails to show a hydrophobic core commonly seen in observed signal peptides.
CXCR6 has close homology to CCR7 and CCR9, which is of interest as SLC and Teck, which share some structural similarity to CXCL16, bind to these receptors, respectively (4, 8, 20, 33). Additionally, CXCR6 has been localized to chromosome 3, which clusters with other CC receptors as well. CXCL16 has been identified on a first pass sequence of a genomic fragment from chromosome 17, meaning that CXCL16 would localize with CC, and not CXC, chemokines. This genomic fragment (AC015913) is 100% similar to the CXCL16 cDNA over bp 77–220 (residues 27–75). Based on the intron-exon structure of several other CC and CXC chemokines, this region corresponds to the predicted second exon of the chemokine region of CXCL16. Therefore, although we use the definition of the CYC motif in the N terminus to call CXCL16 a CXC chemokine, one might want to consider the possibility that this is a completely novel class of chemokine that does not fall into any single system of classification.
Although CXCR6 is phylogenetically related to chemokine receptors, it is structurally diverse in that it lacks cysteine residues in the N terminus and the third extracellular loop that are conserved in all other receptors to date (14). One might postulate that CXCR6, due to the loss of a critical disulfide bond, has a unique membrane topology and therefore might bind to a structurally unique chemokine. Interestingly, a shorter synthetic version CXCL16, truncated at S95 before the fifth and sixth cysteine residues, lacks the potency of the entire extracellular region and appears to be unstable at higher temperatures or pH (data not shown). In contrast, other chemokines that have been similarly truncated, such as Teck and fractalkine, retain activity (D. Soler, unpublished data), suggesting the possibility that the extra cysteine residues in the chemokine domain are critical for overall structure and binding to this novel receptor. We have shown by HPLC analysis that synthetic truncated CXCL16 shows deficiencies in efficient folding (data not shown), which might require these residues as well. Although these extra cysteine residues might be critical for efficient folding, it is also possible that, due to the unique features of CXCR6, this receptor ligand interaction might be completely unique and require residues that have not been required for other chemokines studied thus far.
CXCL16 is the second example of a chemokine that has the sequence of a transmembrane protein and would be predicted to be expressed on the cell surface. Interestingly, our attempts to show expression on the 293 cells (and Chinese hamster ovary cell transfectants) failed to demonstrate cell surface expression of CXCL16 and likely explains our success in the identification of a soluble active form derived from a membrane-bound chemokine. This suggests that in certain settings, the chemokine can be efficiently shed into the surrounding medium; accordingly, we did observe functional activity from supernatants derived from cultured macrophages. This activity was increased in the presence of LPS and TNF-α, indicating that CXCL16 may be rapidly shed from the cell surface and participate in recruitment of lymphocyte subpopulation to sites of both immune and inflammatory responses. It is possible that mechanisms of shedding in a particular microenvironment may represent a level of control of CXCL16-mediated physiological responses.
Additionally, we observed cell surface expression on subsets of peripheral blood leukocytes including CD19+ B cells and CD14+ monocytes. This expression pattern suggests that novel interactions might occur between APCs and CXCR6 expressed on activated T cell subsets, thus raising the possibility that immune responses may be facilitated by direct interactions mediated between cell types expressing this receptor ligand pair. This expression pattern also raises the possibility that other CD14-derived cells, such as monocyte-derived dendritic cells, might express this ligand as well. Along these lines, we have observed expression of CXCL16 in subsets (with a similar staining pattern to what we observe in macrophages) of monocyte-derived dendritic cells, cultured in the presence of IL-4 and GM-CSF (data not shown). This, to some extent, parallels observations showing expression of murine CXCL16 in spleen- and lymph node-derived dendritic cells and indicates similarity in observations between these studies (32). Conversely, expression of murine CXCL16 in B cells was not reported in that study although this does not preclude the possibility that expression might be seen there as well, as it may have yet to be examined.
Expression of the other known membrane-bound chemokine, CX3CL1 (fractalkine/neurotactin), has also been shown on dendritic cells in epidermis and lymphoid tissues (34, 35). CX3CL1 is also chemotactic for activated T cells and may play a role in recruitment and adhesion of T cells to secondary lymphoid tissues and directly interact with T cells in immune responses. Therefore, CXCL16 may represent a second member of the chemokine family that can play a dual role in recruitment and retention of lymphocyte subsets. Other chemokines, such as MDC, have been shown to be expressed by dendritic cells and play a role in attracting activated T cell subsets (36). MDC and fractalkine both map to chromosome 16q13 (35). Although CXCL16 might share some functional relationship with these chemokines, it appears to reside on chromosome 17 and, therefore, does not map to a chromosome encoding either these chemokines or CXC chemokines as well, further illustrating the unique nature of this new ligand.
We have observed CXCL16 expression on tonsil-derived CD19+s and CXCR6 expression on tonsil CD4+s. This observation raises the possibility that, within secondary lymphoid tissue, B cells expressing CXCL16 might be activated by a T cell-dependent process. MDC and fractalkine have also been shown to be expressed by activated B cells as well (36, 37), although expression of fractalkine was observed only at the RNA level. This is the first report of surface expression of a chemokine on a B cell suggesting that binding between this surface-encoded chemokine and its cognate receptor on T cells might contribute to T cell activation. These B cell/T cell interactions may occur after CXCR6+ T cell/dendritic cell interactions have been initiated, and it is possible that CXCL16 may play a role in both of these steps. As the cytoplasmic tail of CXCL16 has a number of potential phosphorylation sites, it will also be of interest to see whether interactions with CXCR6 result in signals transmitted through this chemokine, thus mediating signals in APCs in a novel manner.
Although chemokines have been classified on the basis of the structure surrounding the N-terminal cysteine residues, another classification has emerged as of late, subdividing chemokines into either homeostatic (lymphoid) or inflammatory, based on their site of production and cell types that respond to them. Certain chemokines are produced in lymphoid tissues such as BCA-1 (in stromal cells in B cell follicles) and SLC (in high endothelial venules and by stromal cells in T cells) (26, 27, 38). Others are more known to attract activated leukocytes and are induced by inflammatory stimuli including IL-8, RANTES, monocyte chemoattractant protein-1, MIP-1α, and eotaxin (among others) (39, 40, 41, 42), which are all known to be more highly expressed at sites of inflammation. Although many chemokines fall into these categories, some do not clearly fit either paradigm. MDC, in addition to having a role in secondary lymphoid tissue, has also been shown to be produced in inflamed lungs (43, 44). CXL16 might prove to be a dual-function chemokine as well, as its expression is shown here to be in secondary lymphoid tissues and in addition in nonlymphoid tissues, most notably in liver and lung tissues. Additionally, we show that activated T cell subsets including chronic Th1, Th2, Tr1, and CD4+ T cells derived from tonsil functionally respond to this chemokine. The observations of surface expression of CXCL16 on APCs suggest a potential homeostatic role in lymphoid tissue, whereas the functional interaction with polarized effector cells suggests a role in chronic inflammation as well.
As the sequencing of the genome becomes complete, the concept of “novel sequence identification” is rapidly becoming a thing of the past. There is now a new challenge to assign function of thousands of expressed sequenced clones where classification via conventional means is not possible. CXCL16 was represented in the database as an expressed sequence tag for 5 years, but due to lack of homology could not be identified by searching with related sequences. Although newer searching algorithms that use secondary structure and pattern predictions might solve some of this dilemma, the cloning of this chemokine raises the possibility that several other functionally relevant proteins of this class might also be identified in this manner. The isolation of this structurally unique chemokine and demonstration of binding to a novel chemokine receptor (the first to lack cysteine residues previously considered essential for function) breaks all known paradigms for known chemokine receptor/ligand interactions. Additionally, the observation of both surface expression of CXCL16 on APCs and shedding of active soluble material raises the possibility that this chemokine might play dual roles in inflammation and homeostasis.
Note added in proof.
While in revision it came to our attention that in an attempt to use expression cloning from monocytes to isolate novel receptors binding to OxLDL, a cDNA encoding a receptor termed SR-PSOX (scavenger receptor that binds phosphatidylserine and oxidized lipoprotein) was identified. This cDNA turns out to encode CXCL16, further illustrating the possibility of additional functional roles of this protein.
Acknowledgments
We thank Nasim Kassam for technical assistance in generation of hybridomas and Jose-Carlos Gutierrez-Ramos and Craig Gerard for critical reading of this manuscript. We also thank Keith Robison for initial identification of the identity of CXCL16 to SR-PSOX (see Note added in proof).
Footnotes
-
↵1 A.W., S.C.Z., and K.M. contributed equally to this work.
-
↵2 Current address: CuraGen Corporation, 322 East Main Street, Branford, CT 06405.
-
↵3 Address correspondence and reprint requests to Dr. Michael J. Briskin, Millennium Pharmaceuticals, Sidney Street, Cambridge, MA 02139. E-mail address: mbriskin{at}mpi.com
-
↵4 Abbreviations used in this paper: CXCR, CXC chemokine receptor; GPCR, G protein-coupled receptor; BAB, binding assay buffer; MDC, macrophage-derived chemokine; MIP, macrophage-inflammatory protein; SDF, stromal cell-derived factor; SLC, secondary lymphoid chemokine; EF1, elongation factor-1.
- Received November 27, 2000.
- Accepted February 5, 2001.
- Copyright © 2001 by The American Association of Immunologists
References
In this issue
