The chemokine SDF-1/CXCL12 binds to and signals through the orphan receptor RDC1 in T lymphocytes.

Combined phylogenetic and chromosomal location studies suggest that the orphan receptor RDC1 is related to CXC chemokine receptors. RDC1 provides a co-receptor function for a restricted number of human immunodeficiency virus (HIV) isolates, in particular for the CXCR4-using HIV-2 ROD strain. Here we show that CXCL12, the only known natural ligand for CXCR4, binds to and signals through RDC1. We demonstrate that RDC1 is expressed in T lymphocytes and that CXCL12-promoted chemotaxis is inhibited by an anti-RDC1 monoclonal antibody. Concomitant blockade of RDC1 and CXCR4 produced additive inhibitory effects in CXCL12-induced T cell migration. Furthermore, we provide evidence that interaction of CXCL12 with RDC1 is specific, saturable, and of high affinity (apparent KD approximately 0.4 nM). In CXCR4-negative cells expressing RDC1, CXCL12 promotes internalization of the receptor and chemotactic signals through RDC1. Collectively, our data indicate that RDC1, which we propose to rename as CXCR7, is a receptor for CXCL12.

In concert with surface adhesion molecules, chemokines are secreted proteins that govern the migration of distinct leukocyte subsets to sites of inflammation and to their specific niches in lymphoid organs (1)(2)(3). Two main subfamilies are distinguished according to the position of the first two cysteines, which are separated by one amino acid (CXC chemokines) or are adjacent (CC chemokines) (4,5). Chemokines mediate their functions by binding to chemokine receptors (CKRs), 6 which belong to the large family of heptahelical G protein-coupled receptors. CKRs share Ͼ20% sequence identity, and their interactions with chemokines are highly promiscuous and often redundant (6).
The orphan receptor RDC1, which was originally cloned on the basis of its homology with conserved domains of G protein-coupled receptors (7,8), was primarily believed to act as a receptor for vasointestinal peptide (9), a possibility later dismissed (10,11). In fact, combined phylogenetic and chromosomal location studies suggested that RDC1 represents a CKR structurally close to CXC receptors (12)(13)(14), pointing to CXC chemokines as potential ligands. The RDC1 gene maps to mouse chromosome 1 and human chromosome 2, where the genes encoding CXCR1, CXCR2, and CXCR4 are localized (12,15,16,48). Like CXCR4, the only known receptor for the chemokine stromal cell-derived factor 1 (SDF-1)/CXCL12, RDC1 can serve as coreceptor for certain genetically divergent human immunodeficiency virus (HIV) and simian immunodeficiency virus strains, in particular for the HIV-2 ROD, an X4-tropic isolate (17). This suggests that both CKRs interact with a conserved domain of the viral envelope glycoprotein gp120, which displays some similarities with chemokines, for binding to CKRs and signaling in T cells (18,19). Here we provide evidence that RDC1 and CXCR4 also share CXCL12 as a natural ligand.
RDC1 and CXCR4 Expression Vectors and Functional Assays-The human RDC1, ⌬RDC1, and CXCR4 cDNAs were cloned into the pTRIP vector (kindly provided by Dr P. Charneau, Institut Pasteur, Paris, France) and expressed in cells as specified in the legends to Figs. 2-5. Briefly, transient expression was achieved either by the calcium phosphate-DNA co-precipitation method or by using Nucleofector TM technology (Amaxa Biosystems, Cologne, Germany), whereas stable expression was obtained following a lentiviral-based strategy as described (20) or by electroporation with 20 g of linearized pcDNA3 containing RDC1 and selection with G418 (Sigma-Aldrich). The ⌬RDC1 product was engineered by polymerase chain reaction and lacks the last C-terminal 39 residues (325stop). Competition assays of mAbs with chemokines or RDC1-(7-21) for binding to CKRs were adapted from (21). Briefly, competition of mAb (used at 2 or 5 g/ml) binding by chemokines or RDC1- (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21) was performed at 4°C for 90 min in PBS containing 1% bovine serum albumin, 0.2% ␥Fc, and 0.1% NaN 3 . Bound mAbs in treated cells were quantified by flow cytometry as follows: [receptor geometric mean fluorescence intensity (MFI) of treated cells/receptor geometric MFI of untreated cells] ϫ 100. Specific binding of CXCL12 to RDC1 was evaluated using the biot-CXCL12 in A0.01 T cells that lack endogenous RDC1 and CXCR4 at both mRNA (data not shown) and protein (see Figs. 2B and 5A) levels. These cells also lack cell surface heparan sulfates that are known to bind CXCL12. For saturation bind-ing experiments, RDC1-expressing A0.01 cells (2.5 ϫ 10 5 ) were maintained at room temperature for 90 min in binding buffer (PBS containing 0.2% bovine serum albumin and 0.1% NaN 3 ) in the presence of increasing concentrations of biot-CXCL12. Following a washing step with ice-cold binding buffer, specifically bound biot-CXCL12 was subsequently revealed by incubating cells at 4°C with 1 g/ml streptavidin (SAv)-PE conjugate (BD Biosciences) in binding buffer for 20 min. Cells were then washed once and fixed in PBS containing 2% paraformaldehyde. Results were analyzed by flow cytometry. Specific binding was estimated by subtracting from the total binding the nonspecific binding determined either in the presence of unlabeled CXCL12 at 10 Ϫ6 M (in RDC1-expressing cells) or following the binding of biot-CXCL12 to parental cells. For competition experiments, 7.5 ϫ 10 Ϫ10 M biot-CXCL12 was used as the tracer in the presence of increasing concentrations of untagged CXCL12. Binding parameters (K D and IC 50 ) were determined with the Prism software (GraphPad Software Inc., San Diego, CA) using non-linear regressions applied to one-site models.  A, upper sections, PBLs were stained with 10 g/ml anti-RDC1 or anti-CXCR4 mAb followed by Texas Red-conjugated horse anti-mouse Ab and then labeled with a FITC-conjugated anti-CD3 mAb. Lower sections, PBLs were stained at 4°C with 2 g/ml specific mAb for RDC1 or CXCR4 in the absence (Ϫ) or presence (ϩ) of 10 M CXCL12 and then incubated with a FITC-conjugated horse anti-mouse Ab. Images were taken with structured illumination microscopy. No RDC1 or CXCR4 labeling was detected when PBLs were incubated with isotype control Abs (data not shown). Typical receptor fluorescence, 4Ј,6-diamidino-2-phenylindole (DAPI)-positive nuclei (blue, lower panels), and differential interference contrast images are shown for 3-5 representative cells. Cells (n Ͼ 100) from two independent donors were analyzed. B, chemotaxis of PBLs was determined using a Transwell system in response to 30 nM CXCL12 or 60 nM CCL21 after 90 min of incubation in the presence (ϩ) or absence (Ϫ) of anti-RDC1 and/or anti-CXCR4 mAbs. Chemokines were added to the lower chamber only, whereas mAbs were added to both chambers. Transmigrated cells recovered in the lower chamber were stained with an anti-CD3 mAb and counted by flow cytometry. Results (mean Ϯ S.E.) are from three independent experiments and are expressed as percentage of input CD3 ϩ T cells that migrated to the lower chamber. The fraction of CD3 ϩ T cells migrating in response to CXCL12 was not modulated in the presence of the mouse isotype control Abs (65.9% Ϯ 14.5% and 72.3% Ϯ 9.0%, for IgG2a and IgG1, respectively).
in competition binding experiments. Receptor internalization and chemotaxis assays were performed as described (20). Statistical analyses consisted of unpaired two-tailed Student's t tests and were conducted with the Prism software.
Immunofluorescence-PBLs (2 ϫ 10 5 ) were plated in RPMI medium supplemented with 1% bovine serum albumin and 20 mM HEPES onto polylysine-coated glass coverslips for 3 h at 37°C and fixed in PBS with 4% paraformaldehyde. Cells were stained at room temperature with the anti-CXCR4 or anti-RDC1 mAb (10 g/ml) for 30 min followed by FITC-or Texas Red-conjugated horse antimouse Ab (Vector Laboratories, Burlingame, CA). After three washes with PBS, cells were incubated with the FITC-conjugated anti-human CD3 mAb and finally mounted in Vectashield medium containing 4Ј,6-diamidino-2-phenylindole (Vector Laboratories). For competitive experiments, cells were incubated at 4°C with a mAb to CXCR4 or RDC1 (2 g/ml) in the presence of 10 M CXCL12. Fluorescence was imaged on a Zeiss microscope (Oberkochen, Germany) using a Plan Apochromat 100ϫ, 1.4 oil immersion objective. Images were collected with a cooled charge-coupled device camera (Axiocam MRm), piloted by Axiovision imaging software (Zeiss). Optical sectioning was performed according to the structured illumination principle using the ApoTome system (Zeiss).

CXCL12-induced Chemotaxis of T lymphocytes Is Inhibited by an
Anti-RDC1 mAb-Like CXCR4, the RDC1 gene is ubiquitously expressed in non-hematopoietic and hematopoietic tissues, and its mRNA is found in brain, heart, kidney, spleen, and PBLs (7,17,22). We investigated by immunofluorescence whether RDC1 is expressed in leukocytes using a mAb raised against human RDC1 (9C4). Leukocytes including CD3 ϩ T lymphocytes (Fig. 1A) were stained either with the 9C4 anti-RDC1 mAb or 12G5 mAb, which recognizes an epitope in the second extracellular loop of CXCR4 (23)(24)(25). As expected, the concomitant addition of 10 M CXCL12 abolished staining of CXCR4 with mAb 12G5. Interestingly, a molar excess of CXCL12 also prevented the 9C4 mAb from binding to T lymphocytes (Fig. 1A, lower sections). As cell migration features chemokine responses, we asked whether the 9C4 FIGURE 2. CXCL12 competes for binding of the 9C4 mAb to RDC1. A, alignment of the amino acid sequences of human RDC1 and CXCR4. Amino acid residues with identity (*), high similarity (:), and moderate similarity (.) are marked. Transmembrane domains are unlabeled, whereas intracellular and extracellular domains are highlighted in blue and red, respectively. A synthetic peptide (framed residues, RDC1- (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)) was used to screen hybridomas producing anti-RDC1 mAbs. B, RDC1-(7-21) prevents binding of the 9C4 mAb to RDC1. Left section, cell surface expression level of CXCR4 in A0.01 T-cells expressing the receptor after transduction was determined by flow cytometry using the 12G5 anti-CXCR4 mAb (12G5, red histogram). No labeling was detected using the corresponding isotype-matched control Ab (CTRL; gray histogram) or the anti-RDC1 9C4 mAb (green histogram). Right section, binding of the 9C4 mAb to RDC1 in A0.01 T-cells expressing the receptor after transduction is shown in the absence (9C4 untreated; red histogram) or presence of 10 Ϫ7 M (dotted histogram) or 10 Ϫ5 M (blue histogram) of RDC1- (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21). No labeling was detected using the corresponding isotype-matched control Ab (CTRL; gray histogram) or the anti-CXCR4 12G5 mAb (data not shown), and binding of the 9C4 mAb was not affected by pre-incubating cells with an excess of RDC1-(7-21) (data not shown). C, HEK 293T cells expressing CXCR4 or RDC1 were generated by the calcium phosphate-DNA co-precipitation method and were assessed for the binding of the 9C4 or 12G5 mAb in the absence (untreated; mAb could interfere with the CXCL12-stimulated chemotaxis of T lymphocytes. The chemotactic effect of 30 nM CXCL12 was inhibited by 40 -50% in the presence of the 9C4 anti-RDC1 mAb or the 12G5 anti-CXCR4 mAb (10 g/ml). When added together, both mAbs produced an additive inhibitory effect that extends up to 70% of the CXCL12induced chemotaxis (Fig. 1B). By contrast, migration of lymphocytes in response to CCL21 (Fig. 1B) or CXCL13 (data not shown) was not affected by the addition of anti-RDC1 and anti-CXCR4 mAbs. These findings suggest that RDC1 is expressed at the surface of T lymphocytes and contributes to cell migration toward the CXCL12 gradient. However, the physiological role of RDC1 in T cell migration remains to be determined. Interestingly, CXCL12 displaces the binding of the anti-RDC1 mAb, which reacts with the N terminus of RDC1. Sequence alignment reveals that the N termini of CXCR4 and RDC1 display amino acid similarities and encompass acidic and aromatic residues ( Fig. 2A) that are known to be critical for CXCR4 to bind human immunodeficiency virus gp120 and CXCL12 (26,27). These findings suggest that CXCL12 interacts with the N terminus of RDC1.
CXCL12 Binds to RDC1-To confirm that the 9C4 mAb specifically recognizes RDC1, we first assessed whether the synthetic peptide RDC1-(7-21), which encompasses the epitope recognized by the 9C4 mAb, prevents binding of the mAb to the receptor. In A0.01 cells that lack endogenous expression of RDC1 and CXCR4, ectopic expression of RDC1 restored binding of the 9C4 mAb (Fig. 2B, right section, red histogram and Fig. 5A, right section) but not that of the anti-CXCR4 12G5 mAb (data not shown and Fig. 5A). Conversely, ectopic expression of CXCR4 promoted binding of the 12G5 mAb but not that of the anti-RDC1 9C4 mAb (Fig. 2B, left section, and see Fig. 5A, left section). When pre-incubated with RDC1-(7-21), the 9C4 mAb (5 g/ml) was found to be impaired in its ability to recognize RDC1 in a dose-dependent manner (Fig. 2B, right section). We further evidenced that the 9C4 anti-RDC1 mAb bound to HEK 293T cells expressing the receptor (Fig. 2C, right section, red histogram) but not to the parental cells (data not shown) or to CXCR4-expressing (Fig. 2C, left section, green histogram) or CCR5-expressing (data not shown) cells. Conversely, the anti-CXCR4 12G5 (Fig. 2C, right section) or anti-CCR5 2D7 (data not shown) mAb did not label cells that expressed RDC1 alone. These findings indicate that the 9C4 mAb specifically recognizes RDC1. We then investigated the interaction between CXCL12 and RDC1 by assessing the ability of the chemokine to compete with the binding of the 9C4 mAb to RDC1. CXCL12 displaced the 9C4 as well as the 12G5 mAbs used at 2 g/ml from binding to their respective receptors in a dose-dependent manner (Fig. 2, C-D). In contrast, CXCL12 did not modulate the binding of the anti-CCR5 mAb 2D7 on CCR5-expressing HEK 293T cells, and macrophage inflammatory protein-1␤/CCL4 did not prevent binding of mAbs to RDC1 or CXCR4 (data not shown). Truncation of the N terminus of CXCL12, which occurs upon cleavage by proteases, results in analogs that display reduced affinity for CXCR4 and a limited ability to trigger intracellular signaling (5,27,28). CXCL120-(4 -67), which lacks the first three Lys 1 -Pro 2 -Val 3 residues, displaced binding of the mAb 12G5 to CXCR4, but less efficiently than full-length CXCL12 did (Fig. 2D, left section). This CXCL12 derivative also moderately competes for the binding of the 9C4 mAb to RDC1 (Fig. 2D, right section), thus providing evidence of the involvement of the N terminus of CXCL12 in the binding to RDC1. To directly assess binding of CXCL12 to RDC1, we performed binding experiments using biot-CXCL12 (Fig.   3). The binding of biot-CXCL12 to cells expressing RDC1 was revealed with the SAv-PE conjugate and then quantitated by flow cytometry (Fig.  3, A and C). We found that binding of biot-CXCL12 to RDC1-expressing A0.01 cells was specific, dose-dependent, saturable, and of highaffinity (apparent K D ϭ 0.4 Ϯ 0.1 nM; Fig. 3, A and B). This affinity is approximately one order of magnitude higher than that estimated for CXCR4 (27). Nonspecific binding of biot-CXCL12 assessed either in A0.01 parental cells or in the presence of CXCL12 at 10 Ϫ6 M was found to be minimal (Fig. 3B). When added at increasing concentrations, untagged CXCL12 displaced binding of 7.5 ϫ 10 Ϫ10 M biot-CXCL12 to RDC1-expressing A0.01 cells in a dose-dependent manner with IC 50 ϭ 5 Ϯ 1.8 nM (apparent K i ϭ 1.8 Ϯ 0.7 nM; Fig. 3, C and D). Overall, this demonstrates that RDC1 is a high affinity receptor for CXCL12.
CXCL12 Induces Down-regulation of RDC1-Receptor internalization is a widely described process that occurs upon chemokine stimulation (29). We thus investigated CXCL12-induced internalization of RDC1. Treatment of pre-B 300.19 cells expressing human RDC1 with increasing concentrations of CXCL12 induced a dose-dependent internalization of RDC1 (Fig. 4, A and B). Similarly, 2 ϫ 10 Ϫ7 M CXCL12 induced down-regulation of RDC1 expressed on dermal fibroblasts (Fig.  4, C and D) or on Chinese hamster ovary K1 or A0.01 cells (data not shown). By contrast, RDC1 was not internalized in response to two other CXC chemokines, CXCL8 and CXCL13 (data not shown). Agonist-induced CKR internalization generally relies on the phosphorylation of C-terminal Ser/Thr residues that, in turn, promotes binding of ␤-arrestins for targeting to clathrin-coated pits (30). The C-terminal sequence of RDC1 contains multiple Ser/Thr residues ( Fig. 2A) that could serve as structural determinants for receptor internalization.  4D shows that truncation of the RDC1 C terminus (⌬RDC1) abolished CXCL12-induced internalization, indicating that endocytosis requires the integrity of this domain. These results support the view that Ser/Thr residues of the RDC1 C terminus are subjected to phosphorylation by second messenger-dependent kinases and/or G protein-coupled receptor kinases in response to agonist. This finding is reminiscent to those reported for the internalization of other CKRs including CXCR4 (31)(32)(33).
CXCL12 Promotes Cell Migration through RDC1-We tested the ability of CXCL12 to trigger chemotaxis of A0.01 T-cells expressing either RDC1 or CXCR4 (Fig. 5A). Addition of the chemokine resulted in a dose-dependent migration of RDC1-positive cells (Fig. 5B). The magnitude of this process was comparable with that of CXCR4-positive cells, with a maximum migration obtained at 10 Ϫ8 M CXCL12. CXCL12 also triggers chemotactic signal through the truncated ⌬RDC1 receptor (data not shown), indicating that an intact receptor C terminus is not required for chemotaxis. This finding highlights that the structural determinants required for RDC1-mediated chemotaxis may differ from those involved in receptor internalization, as reported for other CKRs (34,35). The 9C4 mAb strongly inhibited migration of RDC1-expressing cells but did not modify that of cells expressing CXCR4 (Fig. 5C). Conversely, the anti-CXCR4 12G5 mAb abolished chemotaxis of cells expressing CXCR4 but had no effect on CXCL12-promoted migration of cells expressing RDC1 alone. These results confirm that CXCL12 triggers chemotactic signals directly through RDC1. CCL21, CXCL8, CXCL13 (Fig. 5D), CCL4, RANTES (regulated on activation of normal T cell expressed and secreted)/CCL5 (data not shown), or CXCL12-(4 -67) (Fig. 5D) did not promote chemotaxis of RDC1-expressing cells.
Taken together, our data indicate that, in cells ectopically expressing RDC1, CXCL12 binds to and signals via RDC1 and further suggest that activation of RDC1 by CXCL12 requires the intact N terminus of the chemokine.
Conclusions-Collectively, our data demonstrate that the orphan receptor RDC1 is a high affinity receptor for CXCL12. Currently, six CXC receptors have been characterized and, according to the nomenclature for CKR families (6), we propose that RDC1 be renamed as CXCR7. The highly conserved sequences of RDC1 and CXCL12 among species suggest that CXCL12/RDC1 interactions may have conserved functional significance. However, the responsiveness of RDC1 to CXCL12 is intriguing because of the nearly identical phenotypes of CXCR4-and CXCL12-deficient mice. The notion that both CXCR4 and CXCL12 knock-out mice die perinatally and display profound defects in hematopoiesis and cerebellar development was generally interpreted as a monogamous relationship between CXCR4 and CXCL12 (36,37). The interaction between CXCL12 and RDC1 opens questions about the role of this receptor during embryonic life. RDC1 may function in organogenesis and hematopoiesis in concert with CXCR4. Alternatively, but not exclusively, RDC1 may exert distinct activities to those of CXCR4. To this end, it would be important to understand the spatio-temporal regulation and distribution of RDC1 in embryonic tissues.
In post-natal life, as shown for CXCR4, our results are also consistent with a possible role of RDC1 in CXCL12-mediated regulation of CD34 ϩ stem cell homing in the bone marrow and leukocyte trafficking (38 -41). Accumulating evidence shows involvement of the CXCL12/ CXCR4 axis in inflammatory diseases, tumor metastasis, and angiogenesis (42)(43)(44)(45). The reported expression of RDC1 in tumor endothelial FIGURE 5. CXCL12 triggers chemotactic signals through RDC1. A, cell surface expression levels of CXCR4 and RDC1 in nucleoporated A0.01 T cells were determined by flow cytometry using the 12G5 anti-CXCR4 or 9C4 anti-RDC1 mAb (red histograms). Nonspecific binding was assessed after incubation of cells with the corresponding isotype-matched control Ab (CTRL; gray histograms) or in the presence of the 9C4 anti-RDC1 (in CXCR4expressing cells; left section, green histogram) or 12G5 anti-CXCR4 (in RDC1-expressing cells; right panel, green histogram) mAb. A representative experiment of five independent determinations is shown. B, concentration-dependent chemotaxis of A0.01 T-cells expressing either CXCR4 (open bars) or RDC1 (filled bars) is shown in response to CXCL12. C, A0.01 T-cells expressing RDC1 or CXCR4 were treated for 90 min with the indicated Abs and then tested for CXCL12 (10 nM)-induced chemotaxis. D, CXCR4-or RDC1-expressing A0.01 T-cells were assayed for chemotaxis in response to 10 nM CXCL12 or N-terminal truncated CXCL12-(4 -67), 30 nM CXCL8, 60 nM CCL21, or 100 nM CXCL13. B-D, results are expressed as the percentage of input A0.01 cells that migrated to the lower chamber and represent the mean Ϯ S.E. of 3-5 independent experiments. The parental A0.01 T cell line was unresponsive to CXCL12. *, p Ͻ 0.05; **, p Ͻ 0.005 compared with cells that migrated spontaneously (B and D) or in the presence of CXCL12 and the appropriate isotype control Ab (C). cells of brain and peripheral vasculature (46) or in Kaposi's sarcomaassociated herpesvirus dermal microvascular endothelial cells (47) also supports the view that pathological processes might implicate the CXCL12/RDC1 axis.