Cloning and Characterization of a Novel Promiscuous Human β-Chemokine Receptor D6*

Members of the chemokine family of chemotactic peptides interact with their target cells through heptahelical cell surface receptors. An understanding of the biochemistry and expression patterns of these receptors is therefore central to our overall understanding of the roles played by chemokines in both physiological and pathological processes. To date, eight receptors for the β-chemokine subfamily have been described. We have recently cloned a novel murine β-chemokine receptor and report here the identification and characterization of a highly homologous human gene termed human D6 (hD6). This is a promiscuous β-chemokine receptor and appears to be able to bind the majority of members of the β-chemokine family. It is, however, specific for this family and shows no detectable affinity for members of the α-chemokine or the C or CXXXC chemokines. Unlike the majority of other chemokine receptors, human D6 does not appear to be able to flux calcium following ligand binding, thus it is currently not clear if this novel receptor is indeed a signaling receptor. Human D6 is expressed in a range of tissues including hemopoietic cells although it appears not to be ubiquitously expressed in hemopoietic cells. Human D6, unlike some other β-chemokine receptors, appears not to be able to function as an entry co-factor for human immunodeficiency virus type 1 (HIV-)1 on CD4-expressing cells.

The chemokine family of peptides is defined on the basis of sequence homology and on the presence of variations on a conserved cysteine motif (1,2). The family can be subdivided on the basis of this motif into two major subfamilies, in which members of each contain four characteristic cysteine residues. This subdivision therefore defines the CC or ␤-chemokine family in which the first two cysteines are juxtaposed, and the CXC or ␣-chemokine family in which there is an intervening amino acid between the first two cysteines. Further, two subfamilies have recently been described, the C family, which has only two cysteines in the mature protein (3,4), and the CXXXC family, which has three intervening amino acids between the first two cysteine residues of the mature protein (5). The sole member of the C family cloned to date is Lymphotactin, and Fractalkine is the only member of the CXXXC family identified thus far.
Chemokines display a range of in vitro and in vivo functions ranging from pro-inflammatory activities on a range of cell types to proliferative regulatory activities. All functions of the chemokine family are believed to be signaled into a responsive cell using members of the G-protein-coupled heptahelical receptor family (6). To date, a number of CC and CXC chemokine receptors have been cloned. In general these receptors are specific for the respective subfamilies, however, they do display complex and overlapping ligand binding profiles. Thus far, four ␣-chemokine receptors (CXCR1-4) have been described (7)(8)(9)(10)(11), and an additional eight ␤-chemokine receptors are now known (12)(13)(14)(15)(16)(17)(18)(19)(20)(21). Other chemokine receptors identified to date include the DARC receptor (22,23), which recognizes both ␣and ␤-chemokines and a number of virally encoded heptahelical receptors (24 -26).
In addition to their likely roles in inflammatory processes, a number of chemokine receptors have recently been implicated as co-factors with CD4 in mediating entry of HIV-1 1 into CD4 positive cell types. CCR5 appears to be the predominant receptor for mediating entry into monocytes and T cells by M-tropic HIV-1 strains and SIV (27)(28)(29), whereas the ␣-chemokine receptor CXCR4 appears to be a major determinant of entry of T cell-tropic strains of HIV-1 into T cells (10,11).
We now report the identification and characterization of a novel human ␤-chemokine receptor that has high structural and functional homology to a murine receptor, which we have recently described (30). This receptor, which we have designated human D6, is a promiscuous ␤-chemokine receptor and appears to bind the majority of members of the ␤-chemokine family. In contrast to some other CC chemokine receptors, hD6 does not appear to support HIV-1 or SIV entry into CD4expressing cells. We have, so far, been unable to demonstrate signaling through human D6 following ligand binding, and in keeping with the agreed restrictions on chemokine receptor nomenclature, designation of human D6 as CCR9 awaits such data.

MATERIALS AND METHODS
Chemokines-All chemokines were purchased from PeproTech (London, United Kingdom) except human HCC-1 and murine JE, which were purchased from R&D systems (Abingdon UK). The murine MIP-1␣ used in these studies is a nonaggregating variant with wild-type activity as described previously (31). All peptides were maintained at a concentration of 0.1 mg/ml in phosphate-buffered saline.
Cloning of Human D6 -Fragments of human D6 were generated by degenerate oligonucleotide-primed PCR using the following primers designed from the murine D6 receptor (30) in two regions that show conservation of amino acid sequence between other human and mouse chemokine receptors: at the 5Ј end, hD6TCM, 5Ј-TG(C/T) GGI ATC TT(C/T) TT(C/T) AT(C/T) ACI TG(C/T) ATG-3Ј; or hD6DKY, 5Ј-GAC AA(A/G) TA(C/T) CTI GA(A/G) AT(C/T) GTI CA(C/T) GC-3Ј; at the 3Јend, hD6HCC, 5Ј-GTA CAG IAC IGG IGT (A/G)CA (A/G)CA (A/G)TG-3Ј. PCR was performed using variable amounts of MgCl 2 (from 1.25 to 2.25 mM), 0.3 mM of each dNTP, and 6 ng/l of one of the 5Ј oligonucleotides and hD6HCC. Genomic DNA template isolated from human lymphocytes was used at 1 g per reaction. Reactions were incubated for 94°C for 1 min, 50 -52°C for 1 min, and 72°C for 2 min. Products of expected size were cloned into pCRScript (Stratagene, La Jolla, CA) and sequenced. Two different products were cloned: hD6, and a fragment identical to the previously cloned orphan heptahelical receptor, RDC-1. 5Ј and 3Ј rapid amplification of the hD6 cDNA ends (RACE) was performed using RACE kits from Life Technologies, Ltd. (Paisley, UK), using oligonucleotide primers designed from the sequenced fragment. Reverse transcribed U937 cDNA was used as a template for the RACE reactions, which had previously been shown to express hD6. Once the start and stop codons were identified, the full-length gene was amplified with Pfu polymerase (Stratagene) in three separate reactions from human lymphocyte DNA. These products were cloned into pCRScript and fully sequenced. The three products were identical.
Generation of CHO and human osteosarcoma Cells Stably Expressing hD6 -CHO cells were maintained in special liquid medium (Life Technologies, Ltd.) supplemented with 4 mM glutamine and 10% fetal calf serum (TCS Biologicals, Buckingham, UK). The full-length cDNA for hD6 in the pcDNA3 expression vector was transfected into CHO cells as described previously (32). Stably transfected cells were selected in 1.6 mg/ml Geneticin. Hos.CD4 transfectants were generated essentially as described previously (28,33). Briefly, hD6 was ligated into the pBABEpuro retroviral vector (34), and this construct was used to transduce the Hos.CD4 human osteosarcoma cell line stably expressing the human CD4 cDNA. Transduced cells were selected in 1 g/ml puromycin and cell surface expression of hD6 confirmed by radioiodinated MIP-1␣ binding and Scatchard analysis (see below).
Expression Analysis-hD6 expression in poly(A)ϩ mRNA from human tissue was assessed using blots obtained from Cambridge Bioscience (Cambridge, UK). Full-length hD6 cDNA was labeled with [␣-32 P]dCTP using the Ready To Go labeling kit (Pharmacia Biotech Inc., St. Albans, UK) and unincorporated nucleotides were removed using NICK-Sephadex G-50 columns (Pharmacia). The probe was denatured by heating at 100°C for 3 min prior to hybridization, which was performed with Express-Hyb (Cambridge Bioscience) according to instructions obtained with the blots. Filters were exposed to Kodak X-Omat x-ray film and stripped in 0.1% SDS prior to reprobing with a murine ␤-actin cDNA probe (D. Jarmin, Beatson Institute). RT-PCR was performed using the RNA PCR core kit (Perkin-Elmer). For each RNA sample tested, reverse transcription was done in two tubes, one with reverse transcriptase (RT) and one without. 1.2 g of heat denatured total RNA was used per PCR reaction, which had previously been treated with DNase I to remove genomic contaminants (according to Life Technologies, Inc. RACE kits), followed by phenol/chloroform extraction, ethanol precipitation and 70% ethanol wash. Reactions with RT was split into two tubes with one aliquot used to amplify actin mRNA and the other hD6. Lack of hD6 product from the reaction without RT was used to confirm lack of DNA contamination. Genomic DNA from human lymphocytes (T. Jamieson, Beatson Institute) was used as a positive control for hD6 amplification. Specific PCR primers used were: hD6sense 5Ј-CGT TCA TGC TCA GCC CTA C-3Ј with hD6antisense 5Ј-CTG GAG TGC GTA GTC TAG ATG C-3Ј (expected size, 447 base pairs); and actin-sense 5Ј-TCC ATC ATG AAG TGT GAC GT-3Ј with actin-antisense 5Ј-TAC TCC TGC TTG CTG ATC CAC-3Ј (expected size, 246 base pairs). PCR proceeded for 30 cycles of 94°C for 1 min, 58°C for 1 min, 72°C for 1.5 min.
Receptor Binding Studies-All receptor binding studies were carried out as ligand displacement analyses essentially as described previously (30,35) with cells being plated at 2 ϫ 10 5 cells/well in 24-well plates overnight at 37°C prior to binding analysis. Cells were incubated in azide-containing medium (pH 7.4) with a constant amount of radioiodinated murine MIP-1␣ and displacement analysis carried out using increasing concentrations of unlabeled ligand as outlined in the relevant figures. Data was analyzed by the LIGAND program (36).
Co-receptor Analysis-Co-receptor analysis was carried out exactly as described previously (28,33). Briefly, Hos.CD4 cells and the variants expressing CCR5 or hD6 were plated at 10 4 cells/well in 24-well plates and incubated overnight. The following day, cells were infected with luciferase reporter virus pseudotyped by Envs of macrophage-tropic HIV-1 (JRFL), T-cell line adapted HIV-1 (HXB2), SIV (SIV-1a11), or with A-MLV env. Luciferase activity was measured 4 days later as described (33).

RESULTS AND DISCUSSION
Cloning and Sequence Analysis of Human D6 -Cloning of human genes related to murine D6 has taken advantage of the inclusion of the coding region for chemokine receptors within a single genomic exon. PCR primers designed as outlined under "Materials and Methods" from murine D6 were used to generate fragments of a novel human receptor-like gene from genomic DNA. These fragments were subsequently used to design oligonucleotide primers for use in RACE PCR reactions to isolate the 5Ј and 3Ј sequences of the cDNA using human U937 monocytic RNA as a template, known to be positive for hD6 expression. Pfu polymerase was used to generate a fulllength receptor cDNA of 1181 base pairs encompassing an open reading frame of 1152 base pairs (GenBank TM accession number: Y12815). The cDNA encodes a protein of 384 amino acids in length (Fig. 1), displaying the characteristic seven transmembrane-spanning domains (dashed lines) and four conserved cysteine residues (underlined C) involved in maintaining receptor structure. hD6 bears a single putative N-linked glycosylation site at the amino terminus (NSS, underlined) and a number of potential phosphorylation sites at the carboxyl terminus, which is rich in serine and threonine residues. The hD6 cDNA displays 71% identity and 86% similarity to the murine D6 receptor, and in common with this protein, it has an alteration in the highly conserved DRYLAIVHA motif seen in other chemokine receptors changing it to DKYLEIVHA in human and murine D6, introducing an additional charged residue into the site (Fig. 1, bold, italicized, and underlined). These particular changes are not observed in any other G-proteincoupled receptors although there is a certain amount of variance in this motif in nonchemokine receptors. Whereas the extent of homology between human and murine D6 (71% identity) is marginally lower than that observed for the murine and human orthologues of other chemokine receptors (generally around 80%), comparison of Southern blots probed with hD6 with the same blots probed at low stringency with murine D6 suggests that the murine D6 gene hybridizes most strongly to this human gene rather than to other receptors in the human genome, implying that these genes are the closest sequence homologues between these two species (data not shown). hD6 has ϳ40% identity and ϳ50% similarity to the eight other known human ␤-chemokine receptors, which is a consistently lower homology than these receptors show to one another, implying that hD6 is a more divergent member of this family. However, the amino terminus of hD6, a region believed to be important in ligand binding, contains a number of tyrosines and acidic residues that are also seen in the other ␤-chemokine receptors. One potentially important sequence difference between D6 and the other receptors is the alteration of a highly conserved aspartic acid residue in the second transmembrane domain to an asparagine (Fig. 1, labeled #). This aspartic acid is seen in almost all heptahelical receptors with a few signaling competent virally encoded proteins and DARC being the notable exceptions. When this residue is mutated in hCCR5 or other heptahelical receptors certain intracellular effects, such as Ca 2ϩ flux, are no longer initiated, although ligand binding is usually unaffected (37)(38). This is likely to significantly alter the signaling properties of hD6 (see below). Finally, it is of note that the 3Ј sequences of hD6 are identical to an expressed sequence tag derived from a human placental cDNA library, and thus hD6 represents the full-length cDNA corresponding to this expressed sequence tag (GenBank TM accession number R82383).
Analysis of the Expression of hD6 in Human Tissues-Tissue blots containing representative mRNA samples from various human tissues were probed with full-length hD6. Two transcripts were detected of around 4 and 6 kilobases, and as shown in Fig. 2A, expression of the receptor was seen to be very high in the placenta, lower in the liver, with weak but detectable expression seen in the lung and thyroid. On prolonged exposure of these blots, weak expression is also detectable in a range of other tissue types, particularly small intestine and the mucosal lining of the colon. However, it should be mentioned that the contribution of resident leukocytes to this weak expression cannot be discounted, although we have been unable to detect significant expression of hD6 in peripheral blood leukocytes using Northern blotting (see Fig. 2A and below for a discussion of this point). In contrast to the murine D6 receptor (30), hD6 does not appear to be expressed in the spleen; however, it is possible that this could be a reflection of the relatively different cellular composition of the murine and human spleens. Surprisingly, it has proved difficult to detect expression of hD6 in hemopoietic cells using Northern blot analysis of samples from peripheral blood leukocytes, bone marrow (see Fig.  2A), and the cell lines HL-60, Raji, K562, and THP-1 (data not shown). However, PCR has demonstrated hD6 expression in RNA prepared from leukocytes derived from umbilical cord blood and in the primitive erythromyeloid leukemic cells lines, K562, and THP-1 monocytic cells. Raji and HL-60 cells remain negative in this more sensitive assay (Fig. 2B). The expression of hD6 in K562 cells correlates well with the documented expression of the murine counterpart of hD6 in primitive bone marrow cells (30).
The general weak hemopoietic expression of this gene suggests that the solid organ expression of hD6 may be due to nonhemopoietic cells or to the activation status and type of resident leukocytes. In situ hybridization studies should distinguish between these two possibilities.
Ligand Binding Profiles of hD6 -Ligand displacement analysis was performed on CHO cells stably expressing hD6, and the data from these studies are outlined in Fig. 3, A and B and summarized in Table I. The data indicate that hD6 is a highly promiscuous ␤-chemokine receptor that displays relatively high affinity binding for the majority of members of the ␤chemokine family. The highest affinity ligands for hD6 appear to include murine MIP-1␣ and MIP-1␤ and murine JE and human MCP2. Human MCP1 has approximately a 30-fold lower affinity for hD6 than the presumed murine homologue (JE), a similar difference in affinity to that seen with the mouse D6 receptor, which again binds human MCP1 with 30-fold lower affinity than murine JE (30). More recently, it has been suggested that murine MCP5 is the closer murine homologue of MCP1 than JE (39) and indeed this is borne out by our binding analysis that shows murine MCP5 and human MCP1 to have very similar dissociation constants for hD6. Binding to hD6 is also seen with the human chemokines RANTES, MCP3, MCP4, and HCC1, and binding of both murine and human eotaxin is detectable, although this is a relatively lower affinity interaction than that seen with many of the other ␤-chemokines, and the physiological relevance of this interaction remains to be determined. No binding of the ␤-chemokine C10 is detectable, and in addition, I309 appears not to be a ligand for hD6. The specificity of hD6 for the ␤-chemokines is confirmed by the inability of various ␣-chemokines (IL8, Gro␥, IP10, and MIP2) and lymphotactin to displace 125 I-labeled mMIP-1␣ from the receptor. It is noteworthy that in our hands K562 cells display a receptor that exhibits similar properties with respect to affinity and specificity as does hD6 (not shown).
Human and murine D6 exhibit many similarities with respect to ligand binding, and it appears from our recent studies using ␤-chemokines tested here that were not commercially available at the time of publication of the murine receptor (e.g. MCP4 and 5) that murine D6 displays a similar ligand promiscuity to that seen with hD6 (data not shown). Nevertheless, a number of discrepancies do exist between D6 receptors from these two species. For instance, human MCP2 is a high affinity ligand for hD6 but does not bind to murine D6 (30). However, it remains possible that the murine homologue of MCP2 may exhibit an affinity for murine D6. Also, surprisingly, human MIP-1␣ has markedly reduced affinity for hD6 compared with murine MIP-1␣ for mD6; hMIP-1␣ may not be able to bind to hD6 at physiological concentrations. However, recent results from our laboratory strongly suggest that the human and murine MIP-1␣ molecules used in this study cannot be considered to be functionally homologous. 2 Thus, we suggest that despite these anomalies it is highly likely that human and murine D6 are functionally analogous.
The basis for the remarkable promiscuity of hD6 is not immediately obvious. It is likely that this is a reflection of the usage of elements common to the majority of ␤-chemokines in binding such as the clusters of highly conserved residues between cysteines 3 and 4. This assertion is further substantiated by the observations that alterations to the cluster of basic residues in this area substantially affect binding. Thus, weakly binding chemokines such as eotaxin and MCP1 have nonconservative alterations in this area. In addition, the nonproteoglycan binding variant of murine MIP-1␣, which we have previously generated by neutralization of two of the basic residues in this cluster (32), shows a 30-fold reduction in affinity for hD6 compared with the wild-type mMIP-1␣ protein (data not shown). It is important to highlight however that while this region may well be of importance in hD6 binding, it is not a singular requirement as this site is conserved in human MIP-1␣, which shows only limited binding to hD6. Therefore it is likely that other regions of the ligands are also important in hD6 binding.
The relatively unique nature of the amino terminus of murine and human D6 suggests that this region may be involved to some extent in defining the ligand promiscuity of this receptor. We are in the process of generating receptor chimeras to test this hypothesis.
We have been interested in determining the signaling competence of hD6 in response to the various binding ␤-chemokines, however we have been unable to elicit any detectable calcium flux in hD6 expressing HEK293, human osteosarcoma (see below), or CHO cells. This is in contrast to the murine D6 receptor and to human CCR1 and CCR5, which in our hands do signal in response to ligand binding in 293 cells, and may be indicative of alternative G-protein coupling by the hD6. The divergent nature of the hD6 carboxyl terminus, the altered DRYLAIVHA motif, and the absence of the aspartic acid residue in the second transmembrane domain in hD6 (described above) may conspire with other sequence peculiarities of hD6 to explain this phenomenon, This matter is currently under investigation in our laboratory.
It is unclear at present what likely roles a promiscuous ␤-chemokine receptor may play in either normal or pathological situations. If this receptor is involved in chemokine-mediated chemoattraction then, assuming a relationship between binding and signaling, it appears to be fairly indiscriminate in the ligands to which it will respond. The expression of the murine receptor on a range of cell types including primitive hemopoietic cells, and the expression of the human homologue on the primitive K562 cells and on U937 cells, suggest that it may be a general chemoattractant receptor perhaps involved in the general migration of cells around the body rather than in the direct recruitment to inflammatory sites. However, the high expression in solid organs such as the liver may indicate that hD6 acts as a chemokine receptor for nonhemopoietic cells. Evidence for these roles awaits the generation of hD6 -/mice, which we are in the process of making.
hD6 as an HIV Co-receptor-The recent characterization of the chemokine receptors CCR5 and CXCR4 as major co-receptors with CD4 for entry of macrophage-tropic or T cell-tropic strains of HIV-1, respectively, into host cells (see the Introduction), has prompted us to examine the ability of hD6 to act in this manner. To this end we have generated a stably transfected cell line Hos.CD4.hD6 that co-expresses the CD4 and hD6 molecules on the cell surface. Expression of functional hD6 on these cells has been confirmed by Scatchard analysis, which calculated that these cells express ϳ3 ϫ 10 6 receptors/cell (data not shown). Neither macrophage-tropic (JRFL) nor T cell-tropic (HXB2) HIV-1 isolates, or the SIV-1A11 SIV strain, were able to infect human osteosarcoma cells expressing either CD4 alone, or CD4 and hD6, whereas virus pseudotyped with amphotropic leukemia virus (A-MLV) env showed a high level of FIG. 3. Displacement analysis of chemokine binding to hD6-expressing CHO cells. hD6-expressing CHO cells were plated at 2 ϫ 10 5 cells/well in 24-well plates and left overnight at 37°C. After washing, cells were incubated in binding buffer with 600 pM 125 I-labeled murine MIP-1␣ in the presence or absence of increasing amounts of unlabeled chemokines for 90 min at room temperature. Cells were washed 3 times with icecold phosphate-buffered saline prior to lysis in 1% SDS and counting on a gamma counter. Data was analyzed using the Scahot and Scafit program in LIGAND software. A, displacement of 125 I-labeled murine MIP-1␣ by a range of human chemokines as indicated. B, displacement of 125 I-labeled murine MIP-1␣ by a range of murine chemokines as indicated. infectivity in both cell lines (data not shown). As expected, human osteosarcoma cells co-expressing CD4 and hCCR5 showed capacity for entry of the macrophage-tropic variant of HIV-1 and in addition displayed co-factor activity toward SIV-1A11. This data suggests that using the strains described, hD6 is unable to act as a co-factor with CD4 for entry of common laboratory strains of M-or T-tropic isolates of HIV-1 or SIV into human cells. However, it remains possible that other less common strains of HIV or SIV could use hD6 as an entry coreceptor, and we are currently investigating this possibility.