Identification of CX3CR1 A CHEMOTACTIC RECEPTOR FOR THE HUMAN CX3C CHEMOKINE FRACTALKINE AND A FUSION CORECEPTOR FOR HIV-1*

Fractalkine is a multimodular human leukocyte chemoattractant protein and a member of the chemokine superfamily. Unlike other human chemokines, the chemokine domain of fractalkine has three amino acids between two conserved cysteines, referred to as the CX3C motif. Both plasma membrane-associated and shed forms of fractalkine have been identified. Here, we show that the recombinant 76-amino acid chemokine domain of fractalkine is a potent and highly specific chemotactic agonist at a human orphan receptor previously named V28 or alternatively CMKBRL1 (chemokine b receptor-like 1), which was shown previously to be expressed in neutrophils, monocytes, T lymphocytes, and several solid organs, including brain. CMKBRL1/ V28 also functioned with CD4 as a coreceptor for the envelope protein from a primary isolate of HIV-1 in a cell-cell fusion assay, and fusion was potently and specifically inhibited by fractalkine. Thus CMKBRL1/V28 is a specific receptor for fractalkine, and we propose to rename it CX3CR1 (CX3C chemokine receptor 1), according to an accepted nomenclature system.

The chemokine superfamily consists of specific leukocyte chemoattractant proteins that can be classified by structure into four groups, designated CC, CXC, C and CX 3 C, depending on the number and spacing of conserved cysteines (1)(2)(3)(4)(5). The only known CX 3 C chemokines are human fractalkine (4) and its apparent mouse counterpart, named neurotactin (5). The C subgroup also only contains one known member in mouse and human, whereas both CC and CXC subgroups each have many members. Unlike other chemokines, fractalkine and neurotactin are both multimodular proteins consisting of four distinct domains: a chemokine domain N-terminal to an extended mucin-like stalk; a transmembrane domain, which tethers the molecule to the plasma membrane; and a cytoplasmic domain. After transfection of HEK 293 cells, human fractalkine can be detected on the cell surface, and the cells adhere to human monocytes, T cells, and neutrophils. In addition, a truncated 95-kDa "shed" form, which can be found in the supernatant, induces chemotaxis of monocytes and T cells (4). Consistent with this, the recombinant chemokine domain of mouse neurotactin (amino acids  induces in vitro chemotaxis of human neutrophils and T cells and in vivo accumulation of mouse neutrophils, monocytes, and T cells (5).
Based on sequence similarity between CC chemokines and the chemokine domain of fractalkine (4,5), we reasoned that the putative receptor that mediates its action might be most closely related to known CC chemokine receptors (CCR 1 ; 6 -12), which are all members of the seven-transmembrane domain G protein-coupled receptor superfamily. In this regard, CMK-BRL1 (chemokine ␤ receptor-like 1) (13), a previously identified "orphan" receptor also known as V28 (14), stood out as a particularly good candidate because it has ϳ40% amino acid sequence identity to most CCRs; its gene is located on chromosome 3p21-3pter near a cluster of CCR genes (13)(14)(15); it is expressed in human neutrophils, monocytes, and T cells, which are targets for fractalkine and/or neurotactin (13,14,16), and its RNA, as well as RNA for fractalkine, are both present constitutively at high levels in human brain. CMKBRL1/V28 appears to be the human orthologue of RBS11, a rat orphan receptor expressed in leukocytes and brain (17).
Since identification of the chemokine receptor CXCR4 as the first HIV-1 coreceptor, several other chemokine receptors and chemokine receptor-like proteins have been shown to have similar activity for HIV-1 and in some cases for HIV-2 and simian immunodeficiency virus (18). Because of its sequence relationship to chemokine receptors, CMKBRL1/V28 has previously been tested and shown to be a coreceptor for certain isolates of HIV-2 (19). Here we identify functional interactions of CMKBRL1/V28 with HIV-1 and demonstrate that it is a receptor highly specific for fractalkine.

EXPERIMENTAL PROCEDURES
Production and Use of Anti-CX 3 CR1 Antiserum-A synthetic peptide corresponding to amino acids 1-27 of the CMKBRL1/V28 sequence was produced by FMOC (N-(9-fluorenyl)methoxycarbonyl)chemistry, conjugated to KLH, and injected into rabbits. Preimmune sera and sera harvested after serial boosts were tested for reactivity with various cell lines by flow cytometry using a previously published method (20) with two modifications. A donkey anti-rabbit phycoerythrin-conjugated F(abЈ) 2 (Jackson ImmunoResearch Laboratory, West Grove, PA) was used as the second antibody, and seven-amino actinomycin D (Sigma) was used to gate dead cells. Sera harvested after the third immunization was used for all experiments at a dilution of 1:50.
Creation of Cell Lines Expressing Chemokine Receptors-The cloning of a 1.6-kilobase pair cDNA named clone 74.2 coding for CMKBRL1, construction of recombinant pBluescript and pREP9 plasmids containing this cDNA, and creation of HEK 293 cell lines stably transfected with pREP9-CMKBRL1 have been described previously (13). In addi-* 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. tion, the cDNA was excised from Bluescript using BamHI (5Ј) and XhoI (3Ј) sites in the polylinker and subcloned between the corresponding sites of the mammalian expression vector pcDNA3 (Invitrogen, Carlsbad, CA), and additional HEK 293 cell lines were derived using previously described methods (13). The production of HEK 293 cell lines stably expressing CCR1, CCR5, and CCR6 have been previously described (9,11,21,22). A CCR6 cell line was a generous gift of F. Liao and J. Farber (NIAID, Bethesda, MD). HEK 293 cell lines were maintained in DMEM with 10% fetal bovine serum containing streptomycin 100 g/ml and penicillin 100 units/ml, supplemented with 1 mg/ml G418.
Chemotaxis-HEK 293 cells were washed twice with phosphatebuffered saline, resuspended in serum-free RPMI 1640 containing 0.1% bovine serum albumin, then loaded in a total volume of 25 l into the upper compartment of a microchemotaxis chamber (Neuroprobe, Cabin John, MD). A final volume of 32 l of buffer Ϯ fractalkine was placed in the lower compartment. The two compartments were separated by a polyvinylpyrollidone-free polycarbonate filter with 10-m pores coated for 1 h with mouse collagen type IV (Life Technologies, Inc.) at 20 g/ml. The chemotaxis chamber was incubated at 37°C, 100% humidity, and 5% CO 2 for 5 h. The filter was then removed, rinsed with phosphatebuffered saline, and stained with Diff-Quick. The number of cells in 5 high power fields was counted for each data point. All conditions were tested in triplicate.
Cell-Cell Fusion Assay-The CMKBRL1 cDNA was subcloned into the StuI site of the vector pSC59, which contains a synthetic vaccinia early/late promoter (23). pSC59-based plasmids encoding CXCR4 (pYF1-fusin (24)) and CCR5 (pGA9-CKR5 (25)) and details of the HIV-1 Env-mediated cell-cell fusion assay have been described previously (26). Briefly, effector cells were prepared by coinfecting HeLa cells with recombinant vaccinia viruses encoding bacteriophage T7 RNA polymerase (vP11T7gene1) and the indicated HIV-1 Envs. In some experiments, HIV-1 Env was expressed by transfecting cells with pSC59based plasmids instead of infecting with recombinant vaccinia virus. The target cells were prepared by transfecting NIH 3T3 cells with pSC59-based plasmids expressing the indicated chemokine receptors and coinfecting with a recombinant vaccinia virus encoding CD4 (vCB-3) and a recombinant vaccinia virus containing the ␤-galactosidase reporter gene linked to the T7 promotor (vCB21R-LacZ). After transfection and infection, cells were maintained at 31°C overnight. The following day, 1 ϫ 10 5 effector cells were mixed with an equal number of target cells and incubated together for 2.5 h at 37°C. The cells were then lysed in 0.5% Nonidet P-40 and assayed for fusion by measuring ␤-galactosidase activity. For all experiments, the low background signal obtained using target cells that lacked coreceptor was subtracted for presentation of the results with each Env.
We created an additional set of HEK 293 cell transfectants using a pcDNA3 expression plasmid and screened them plus the first generation of transfectants for cell surface expression of CX 3 CR1 based on reactivity with a polyclonal anti-CX 3 CR1 rabbit antiserum. We selected two cell lines transfected with the pcDNA3 construct named 3.2 and 3.3 for further study because they exhibited the highest level of specific staining, which was consistently ϳ10-fold greater than either untransfected HEK 293 cells or HEK 293 cells expressing CCR5 and CX 3 CR1-transfected cells stained with preimmune serum ( . Consistent with this, we did not detect endogenous RNA by Northern blot analysis for CXCR1, CXCR2, CCR1-5, or CCR8 in these cells. As previously reported, HEK 293 cells stably transfected with CCR1, CCR5, or CCR6 responded to MIP-1␣, MIP-1␣, and liver activation-related chemokine, respectively (8,9,21,22), but we observed no responses by any of these cell lines or by untransfected HEK 293 cells to fractalkine ( Fig. 1B and data not shown). Thus, we infer from this gain-of-function result that fractalkine is a specific agonist acting at CX 3 CR1 in this system.
For both the 3.2 and 3.3 cell lines, the relationship between fractalkine and calcium flux response was concentration-dependent. We carried out a detailed study for the 3.2 cell line (Fig. 1C) and found that the threshold, half-maximal, and saturating concentrations were 0.5, 2, and 10 nM, respectively. This is consistent with the potency range that we and others have reported for other chemokines acting at other chemokine receptors for this response. However, the potency of natural forms of soluble fractalkine and membrane-anchored fractalkine could differ, either in this system or at endogenous CX 3 CR1.
When the cells were pretreated with ineffective chemokines at 100 nM, there was no effect on the magnitude or kinetics of the fractalkine-induced calcium flux response, suggesting that other chemokines are not antagonists at CX 3 CR1 (Fig. 1B, top  tracings). In contrast, when cells were sequentially stimulated with the same concentration of fractalkine, no response to the second application was observed (Fig. 1B, top right tracing), suggesting homologous desensitization of the receptor.
Fractalkine Is a High Affinity Ligand for CX 3 CR1-To directly examine the interaction of fractalkine with CX 3 CR1, radioligand binding was performed (Fig. 2). 125 I-Fractalkine specifically bound to CX 3 CR1-but not to CCR5-transfected HEK 293 cells ( Fig. 2A). In pilot experiments we observed that the ratio of specific to nonspecific binding was significantly higher in assays carried out at 23 versus 4°C (data not shown), and all subsequent experiments were done at the higher temperature and in the presence of azide to prevent receptor internalization. The addition of excess unlabeled chemokines revealed that fractalkine competed highly effectively for the labeled site on CX 3 CR1, but RANTES, MCP-3, eotaxin, and MIP-1␣ did not compete at all (Fig. 2B). These results are consistent with the specificity observed in the calcium flux assay. Consistent with this, we were unable to observe specific binding of 125 I-labeled MIP-1␣ or MCP-3 to CX 3 CR1-expressing HEK 293 cells using conditions where high affinity binding was observed with CCR1-and CCR2B-expressing HEK 293 cells, respectively (data not shown).
Unlabeled fractalkine competed in a concentration-dependent manner for the 125 I-fractalkine-labeled site on the 3.2 cell line with an IC 50 ϭ 1.6 Ϯ 0.2 nM (n ϭ 3) (Fig. 2C). Maximal binding varied by less than 5% in these three experiments. In one comparative experiment, the IC 50 values observed for the 3.2 and 3.3 cell lines varied by less than 10% (data not shown). In direct binding studies, fractalkine bound to the 3.2 cell line in a saturable manner (Fig. 2D). Scatchard analysis of the data were consistent with a single binding site, with a K d ϭ 0.74 Ϯ 0.36 nM and 62,000 Ϯ 7800 sites/cell (n ϭ 3) (Fig. 2D).

CX 3 CR1 Is a Chemotactic
Receptor-To test the biological function of CX 3 CR1, we carried out chemotaxis assays using the 3.2 cell line stably transfected with CX 3 CR1 cDNA. Fractalkine was able to significantly induce concentration-dependent transmigration of CX 3 CR1-expressing cells across a filter in a modified Boyden chamber assay of chemotaxis, but RANTES had no activity (Fig. 3). The dose-response curve was bellshaped, which is typical for chemotaxis, and the optimal concentration and EC 50 were 100 and 20 nM, respectively. Thus, by gain-of-function criteria we infer that fractalkine is a chemotactic agonist at CX 3 CR1 in this system. CX 3 CR1 Is a Cofactor for HIV-1 Env-dependent Cell Fusion-To determine whether CX 3 CR1 functions as a coreceptor for HIV-1, we used a vaccinia-based ␤-galactosidase reporter assay for Env-dependent cell-cell fusion, which models the viral envelope-target cell fusion step in HIV-1 infection (26.). This assay was used to identify CXCR4 as the first HIV-1 coreceptor (24), and subsequently several other chemokine receptors and chemokine receptor-like proteins have been shown to have similar activity using the same or related assays (see Ref. 18 for review and primary references). HIV-1 strains vary in the coreceptors they can use for cell entry, and this determines strain cytotropism. Among other classification schemes, classification of HIV-1 strains according to their ability to infect primary macrophages and use CCR5 (M-tropic or R5 strains) versus transformed T cell lines and CXCR4 (TCL-tropic or X4 strains) has been particularly useful. Some primary isolates can infect both cell types and use both coreceptors and are classified as dual-tropic strains. CX 3 CR1 was able to reconstitute fusion activity in target NIH 3T3 cells expressing CD4 when mixed with effector cells expressing Env from the dual-tropic primary isolate 89.6, inducing approximately half the level of fusion found for the same Env tested with CXCR4-or CCR5-expressing target cells (Fig.  4). In contrast, no significant level of fusion was detected with Envs from three R5-tropic isolates or with Env from a X4-tropic isolate. Thus, by gain-of-function criteria, we infer that CX 3 CR1 has HIV-1 Env fusion coreceptor activity. CX 3 CR1 was analyzed for its sensitivity to inhibition of Envmediated fusion by various chemokines. Fractalkine, but not any other chemokines tested, was able to inhibit fusion mediated by 89.6 Env-expressing effector cells with CD4-and CX 3 CR1-expressing target cells (Fig. 5A). Fractalkine inhibited the ␤-galactosidase signal to background levels. To determine whether the inhibitory effect of fractalkine on fusion was specific for CX 3 CR1 and not other HIV-1 coreceptors, its effect on fusion mediated by CXCR4 and CCR5 with the 89.6 Env was assayed in a parallel experiment. No significant inhibition of fusion was seen with either CXCR4 or CCR5 (data not shown). Fractalkine inhibited CX 3 CR1-dependent fusion activity for the 89.6 Env in a concentration-dependent manner (IC 50 ϭ 5 Ϯ 1 nM, n ϭ 3 experiments; Fig. 5B). This value correlates well with the IC 50 for homologous competition of fractalkine binding to CX 3 CR1-expressing HEK 293 cells and the EC 50 values for fractalkine induction of calcium flux. Complete inhibition was seen at concentrations of 100 nM or greater.

DISCUSSION
In summary, we have shown by four different assays, direct radioligand binding, ligand-induced calcium flux and chemotaxis, and ligand inhibition of HIV-1 coreceptor activity, and by both gain and loss of function criteria that fractalkine is a highly specific ligand for the previously identified orphan receptor CMKBRL1/V28, thus justifying renaming it as CX 3 CR1 according to guidelines of an accepted nomenclature system. While this work was under review, Imai et al. (27) also reported that fractalkine is a functional ligand for the same receptor and also renamed it CX 3 CR1. In addition to showing as we have here that the receptor mediates chemotactic responses to a soluble form of fractalkine, they also showed that it directly mediates cytoadhesion to cells expressing fractalkine on the surface.
Since RNA for CX 3 CR1 is expressed in neutrophils, monocytes/macrophages, and T lymphocytes, it is an excellent candidate for the cellular receptor involved in fractalkine-dependent signal transduction in these cell types that have previously been described; however, specific blocking agents will have to be developed before this can be specifically tested. Based on its in vitro properties, future research will be directed toward examining the role of CX 3 CR1 in specific leukocyte trafficking in vivo. Like other leukocyte chemoattractants and chemoattractant receptors, fractalkine and its receptor have the potential to function beneficially, for example in host defense and tissue repair, as well as harmfully, for example in immunologically mediated inflammatory diseases. In this regard, our identification of CX 3 CR1 may be useful in future research aimed at evaluating the fractalkine signal transduction pathway as a target for development of potential anti-inflammatory therapies.
Also while this paper was in review, Rucker et al. (28) reported that CX 3 CR1 functions as a weak fusion cofactor for a small percentage of HIV-1 Envs, including the 89.6 Env that we have tested, the X 4 -tropic Env from strain 92UG024.2, and the Env from strain 93ZR001.3, which is of undefined tropism. Our results are in support of this and extend an earlier report regarding HIV-2 usage of this molecule as a coreceptor (19). In addition, our results show that the functional interaction of CX 3 CR1 with HIV-1 Env can be blocked potently, effectively, and specifically by fractalkine. The importance of this coreceptor to HIV-1 transmission and pathogenesis in vivo remains speculative, as is the case for many of the coreceptors identified by in vitro assays; however, based on its high expression in brain (13,14,17), it could conceivably play an important role there. Studies of CX 3 CR1 usage by primary brain isolates of HIV-1 are needed to further address this hypothesis.
Having identified fractalkine as a ligand for this receptor, it remains possible that, as for other chemokine receptors, additional agonists specific for CX 3 CR1 may exist, including novel members of a putative CX 3 C chemokine family not yet discovered or even CC chemokines. It is also possible that CC chemokines may be agonists at CX 3 CR1 in other cellular contexts such as primary leukocytes or for other functions not tested here.