(Received for publication, March 17, 1997, and in revised form, May 23, 1997)
From the Laboratory of Viral Diseases and the
¶ Laboratory of Host Defenses, NIAID, National Institutes of
Health, Bethesda, Maryland 20892 and the ** Krebs Institute, Department
of Molecular Biology and Biotechnology, University of Sheffield,
Western Bank, Sheffield, S10 2TN, United Kingdom
The chemokine receptors CXCR4, CCR2b, CCR3, and CCR5 are cell entry coreceptors for HIV-1. Using an HIV-1 envelope (Env)-dependent cell-cell fusion model of entry, we show that CCR3 can interact with Envs from certain macrophage (M)-tropic strains (which also use CCR5), T cell line (TCL)-tropic laboratory-adapted strains (which also use CXCR4), and a dual-tropic primary isolate (which also uses CCR2b, CCR5, and CXCR4). Paradoxically, CCR1 is the closest homologue to CCR3 (63% amino acid identity), but lacked HIV-1 coreceptor activity. These results confirm and extend previous reports. Replacing the N-terminal segment of CCR3 with that of CCR1 abolished activity of the resulting chimera for M-tropic and TCL-tropic Envs, but not for the dual-tropic Env. Replacing extracellular loop 2 of CCR3 with that of CCR1 abolished activity for TCL-tropic Envs, but not for M- and dual-tropic Envs. A chimera containing all four extracellular regions of CCR3 on a backbone of CCR1 lacked any activity. Env-CCR3 interactions were strongly inhibited by the major CCR3 ligand eotaxin, but weakly or not at all by other CCR3 ligands. With primary macrophages, eotaxin induced transient calcium flux and partially inhibited fusion with cells expressing M-tropic Envs. We conclude that specificity determinants for different Envs are located in shared and distinct extracellular regions of CCR3, the transmembrane/cytoplasmic domains make major contributions to coreceptor function, and CCR3 may be used by certain HIV-1 strains as a cell fusion factor on macrophages.
All HIV-1 strains infect peripheral blood mononuclear cells, but can differ in their ability to infect primary macrophages (M)1 and transformed T cell lines (TCL) (1, 2). Viruses recovered from individuals before AIDS develops are usually M-tropic and do not induce syncytia in cultured cells, whereas many isolates recovered from patients with AIDS are TCL-tropic and syncytium-inducing; the latter phenotype is also observed with laboratory-adapted strains. Some primary isolates are dual-tropic, able to infect both M and TCL.
The viral determinants of cytotropism include the V3 loop of the envelope glycoprotein gp120 (Env) (3). The host cell determinants of cytotropism are specific chemokine receptors which act with Env and CD4 to promote fusion of viral envelope with target cell membrane (4). Chemokine receptors are 7-transmembrane domain G protein-coupled receptors that normally regulate leukocyte trafficking (5). Four of them (CXCR4, CCR5, CCR3, and CCR2b) have been shown to have HIV-1 coreceptor activity.
CXCR4 was originally identified as a coreceptor for laboratory-adapted TCL-tropic strains (6), whereas CCR5 functions preferentially for M-tropic primary isolates (7-11). Numerous dual-tropic primary strains can use both recombinant coreceptors (10-14). The importance of endogenous CCR5 in HIV-1 pathogenesis is solidly established by the ability of its chemokine ligands to block fusion/infection of primary leukocytes with M-tropic strains in vitro (7-10, 15), by the high level resistance to natural HIV-1 infection found in individuals lacking CCR5 because of homozygous inheritance of an inactive mutant CCR5 allele (16-20), and by the resistance of primary leukocytes from such individuals to infection with M-tropic strains (16, 17, 21, 22). Involvement of endogenous CXCR4 in HIV-1 infection is supported by the ability of its natural ligand SDF-1 (23, 24) and anti-receptor antibodies (6, 25) to block infection of primary leukocytes with TCL-tropic laboratory-adapted strains in vitro; however proof of its importance in people is currently lacking. Only one HIV-1 strain has been reported to interact with recombinant CCR2b so far (11), and there is currently no evidence that endogenous CCR2b is used for virus entry.
Recombinant CCR3 has been shown to interact with several primary HIV-1 isolates including dual-tropic and some M-tropic strains (10, 11, 14, 26). Among TCL-tropic strains, CCR3 usage was observed for a primary isolate, but not for the laboratory-adapted strains examined (10, 11). Recently, several M-tropic HIV-1 strains were shown to use endogenous CCR3 for infection of microglial cells in vitro (26). CCR3 is also strongly expressed on primary eosinophils and at lower levels on monocytes and a small subset of T cells (27-30), although its role as an HIV-1 coreceptor on these cells has not yet been shown.
Interestingly, the amino acid sequence of CCR3 is much less related to CCR5 than to CCR1 (51 versus 63% identity) (27, 31-33), yet CCR1 has little if any HIV-1 coreceptor activity (7, 8, 10, 12). Similarly, CCR5's sequence is much less related to CCR3 than to CCR2b (51 versus 75% identity) (27, 31-34), yet CCR2b has little HIV-1 coreceptor activity. In previous reports, CCR2b/CCR5 chimeric receptors were used to map Env specificity determinants on CCR5 (35-37). Here we identify additional Envs from M-tropic and some TCL-tropic laboratory-adapted strains that can interact with CCR3, and we use CCR1/CCR3 chimeras to map Env specificity determinants. We also provide evidence that CCR3 may function as an HIV-1 coreceptor in blood-derived macrophages.
Human HeLa cells and murine NIH 3T3 cells were obtained from the American Type Culture Collection (Rockville, MD). For preparation of primary macrophages (38), elutriated monocytes were obtained from healthy blood donors under an approved protocol by the Dept. of Transfusion Medicine, Clinical Center, National Institutes of Health. Monocytes were washed and resuspended in Dulbecco's modified Eagle's medium plus 10% human AB serum (Advanced Biotechnologies, Columbia, MD) and then incubated on bacteriological plates for 2 weeks, with fresh medium added every 7 days. The cells were recovered by incubation at 4 °C for 15 min, then cryopreserved in 90% fetal bovine serum, 10% Me2SO in liquid nitrogen until use.
[Ca2+]i MeasurementsMacrophages (107/ml) were loaded with Fura-2 (Molecular Probes, Eugene, OR) as described previously (27). 2 × 106 cells in 2 ml of Hanks' balanced salt solution were placed in a continuously stirred cuvette at 37 °C in a fluorimeter (Photon Technology Inc., South Brunswick, NJ) and stimulated with recombinant chemokines purchased from PeproTech (Rocky Hill, NJ). The data are presented as the relative ratio of fluorescence excited alternately at 340 and 380 nm every 0.5 s, monitored at 510 nm.
Construction of Epitope-tagged CCR1 and CCR3Wild type CCR1 and CCR3 were epitope-tagged by polymerase chain reaction using: 1) the p4 cDNA encoding CCR1 (31) and the clone 3 cDNA encoding CCR3 (27) as templates; 2) specific sense primers containing a 33-base pair sequence encoding the FLAG sequence MDYKDDDDK immediately downstream of the Kozak sequence, and specific antisense primers ending at the corresponding stop codons; and 3) Pfu polymerase (Stratagene, La Jolla, CA). Inserts were blunt-ligated into the StuI site of plasmid pSC59,2 which contains a vaccinia virus synthetic strong early/strong late promoter.
Construction of Chimeric ReceptorsChimeric receptors were constructed by overlap extension, as described previously (39), using the appropriate flagged wild type receptor cDNA as template and specific primers flanking the extracellular domains. Switch sites for the chimeras were chosen so as to independently exchange each of the four extracellular domains of CCR1 and -3. Correct sequence and orientation in the vaccinia expression plasmid pSC59 were verified for all constructs.
Cell Fusion Assay of HIV-1 Coreceptor ActivityFusion between Env-expressing effector cells and CD4/coreceptor-expressing target cells was monitored by a vaccinia-based reporter gene assay for cell-cell fusion (40). Recombinant plasmids containing coreceptor constructs downstream of the vaccinia synthetic promoter were transfected into target NIH 3T3 cells by lipofection using DOTAP (Boehringer Mannheim). The vaccinia promoter was activated by coinfection of the transfected cells with recombinant vaccinia viruses vCB-3 encoding human CD4 (41) and vTF7-3 encoding T7 RNA polymerase (42). Surface expression of the flag epitope was verified by FACS using the M2 monoclonal antibody according to the recommendations of the manufacturer (Eastman Kodak Co., Rochester, NY).
Effector HeLa cells were co-infected with vCB-21R containing the
Escherichia coli lacZ gene under control of the T7 promoter (43) plus one of the following Env-encoding vaccinia viruses (44):
vCB-41, LAV; vCB-36, RF; vCB-43, Ba-L; vCB-28, JR-FL; vCB-32, SF-162;
vCB-39, ADA; vCB-16, a nonfusogenic uncleavable (Unc) Env. The IIIB
(BH8) Env was expressed using vSC60.2 The 89.6 Env (cloned
into pSC59) was expressed by lipofection of HeLa cells and infection
with vCB-21R. Where indicated, infections of HeLa cells were performed
in the presence of 40 µg/ml cytosine arabinoside (araC). Effector
HeLa cells and target NIH 3T3 cells were incubated separately overnight
at 31 °C to allow synthesis of vaccinia virus-encoded proteins, then
washed, and resuspended at 106 per ml. Mixtures (duplicate)
were prepared containing equal numbers (1 × 105) of
effector and target cells and incubated for 2.5 h at 37 °C in
the presence or absence of recombinant chemokines. Reactions were
terminated by addition of 0.5% Nonidet P-40. Fusion was scored by
colorimetric assay of lysates for -galactosidase activity. When
primary macrophages were used as targets, macrophages were infected
with vTF7-3, and fusion with Env-expressing HeLa effector cells was
analyzed as described above.
CCR3 exhibited strong
fusion activity with all M-tropic Envs tested (Ba-L, JR-FL, SF-162, and
ADA), two of three laboratory-adapted TCL-tropic Envs tested (the
closely related IIIB and LAV, but not RF), and the dual-tropic primary
isolate 89.6 Env (Fig. 1). In marked
contrast, CCR1 was inactive with all Envs tested despite similar levels
of surface expression, as determined by FACS with the M2 antibody
recognizing the epitope tag (see Fig. 1 legend).
Env Specificity Determinants on CCR3
To map Env specificity
determinants on CCR3, we analyzed the HIV-1 coreceptor activity of
eight CCR1/CCR3 chimeras for the LAV, IIIB, 89.6, Ba-L, JR-FL, and ADA
Envs (Figs. 2 and
3). Since the extracellular regions of
CCR1 and CCR3 contain the majority of amino acid differences between
the two molecules (Fig. 2) and are the most likely regions to directly
contact Env, we switched only these regions, singly and in
combinations. The boundaries for each exchange were based on hydropathy
analysis of the primary sequences, and were limited precisely to the
predicted extracellular hydrophilic domains by appropriate design of
polymerase chain reaction amplimers (Figs. 2 and 3). All chimeras were
expressed on the cell surface at levels similar to those for CCR1 and
CCR3 as determined with the M2 anti-FLAG monoclonal antibody (see Fig. 3 legend).
In the first set of chimeras, we replaced extracellular domains of CCR3 with the corresponding domains of CCR1 one at a time (chimeras CHI 1 through CHI 4). We observed that the chimera containing the N-terminal segment of CCR1 (chimera CHI 1) had <20% of the fusion activity of wild type CCR3 for Envs from the laboratory-adapted TCL-tropic strains (IIIB and LAV), as well as for Envs from the M-tropic strains (JR-FL, ADA, and Ba-L). This suggests that the N-terminal segment of CCR3 is necessary for interaction with the laboratory-adapted TCL-tropic and the M-tropic Envs. However, it is not sufficient since chimera CHI 4, which contains only the third extracellular loop of CCR1, also exhibited a marked reduction in fusion activity for the same Envs.
Compared with the M- and TCL-tropic Envs tested, Env from the dual-tropic primary isolate 89.6 was exceptional in exhibiting a high level (80%) of the fusion activity of wild type CCR3 when tested with chimera CHI 1, although, like the other Envs, it also interacted poorly with chimera CHI 4 (20% of wild type CCR3). Thus, different HIV-1 strains can finely discriminate not only among different chemokine receptors, but also among different domains of the same receptor. This point is further supported by the phenotype of chimera CHI 3, which contains the second extracellular loop of CCR1. While this chimera exhibited fusion activity at levels comparable to those of wild type CCR3 for all M-tropic Envs tested, it did not exhibit any activity with the laboratory-adapted TCL-tropic Envs. Together with the phenotypes of chimeras CHI 1 and CHI 4, this indicates that the N-terminal segment and the second and third extracellular loops of CCR3 are necessary, but that none is sufficient, for interaction with laboratory-adapted TCL-tropic Envs. CHI 3 supported intermediate activity when tested with the 89.6 Env (40% of wild type CCR3 activity). The first extracellular loop of CCR3 appeared to make a modest contribution to fusion activity and specificity (Fig. 3, CHI 2); fusion activity for all Envs was reduced, but in most cases less than 50%.
We next determined the activities of a second set of chimeras (Fig. 3), in which we replaced individual or multiple extracellular domains of CCR1 with the corresponding domains of CCR3 (chimeras CHI 5, CHI 6, CHI 7, and CHI 8). Thus, all of these chimeras contain the transmembrane and cytoplasmic domains of CCR1. FACS analyses confirmed cell surface expression at levels similar to CCR1 and CCR3 (see Fig. 3 legend). We expected that gain of function might occur with one or more combinations; however, no fusion activity was observed for any Env with any of these chimeras.
Relationship of Env and Chemokine Interaction Sites on CCR3To assess whether the Env specificity determinants on CCR3
overlap with its chemokine specificity determinants, we tested coreceptor activity in the presence and absence of the CCR3 ligands eotaxin, RANTES, and MCP-3. Eotaxin inhibited fusion mediated by
recombinant wild type CCR3 in a dose-dependent manner
(Figs. 4 and
5). The extent of inhibition ranged from
55-95% when 1 µM eotaxin was tested. RANTES and MCP-3
were much less effective inhibitors (Fig. 4).
Analysis of fusigenic CCR3/CCR1 chimeras indicated that the
determinants of eotaxin and Env specificity were overlapping but not
identical (Fig. 6). In particular, fusion
of cells expressing chimera CHI 1, which contains the N-terminal
segment of CCR1, with cells expressing the 89.6 Env was highly
sensitive to eotaxin, whereas fusion of cells expressing other chimeras
permissive for interaction with the same Env was eotaxin-resistant.
Evidence That Endogenous CCR3 Interacts with HIV-1 Envs on Primary Macrophages
We have previously reported that chemokine ligands
specific for CCR5 can inhibit fusion of primary macrophages with cells expressing M-tropic Envs (9). However, inhibition was typically incomplete suggesting the possibility that other HIV-1 coreceptors may
be used on these cells. We have previously reported that CCR3 mRNA
is present in primary macrophages, albeit at very low levels compared
with primary eosinophils (27). To test whether macrophages express
functional eotaxin receptors, we measured calcium flux responses in
Fura-2 loaded cells. Calcium flux is highly associated with chemokine
receptor signal transduction and can be used as a convenient assay of
receptor activation in real time. As shown in Fig.
7, primary macrophages exhibited a
calcium flux in response to multiple CC chemokines including eotaxin.
To test whether endogenous CCR3 might account for Env interactions with
primary macrophages, we carried out cell fusion assays with HeLa cells
expressing a variety of HIV-1 Envs as effectors and primary macrophages
as targets, in the presence and absence of eotaxin. As reported
previously (9, 44), cells expressing M-tropic Envs fused with primary macrophages, and RANTES inhibited M-tropic Env-dependent
fusion, but MCP-3 did not. However, we also observed that fusion
supported by several of the M-tropic Envs was inhibited by eotaxin in a dose-dependent manner (Fig.
8).
Our results add to a growing body of literature demonstrating that chemokine receptors can finely discriminate among different HIV-1 Envs. CCR3 has combined Env specificities found separately in CCR5 and CXCR4, displaying coreceptor function with M-tropic strains (which can also use CCR5), some TCL-tropic laboratory-adapted strains (which can also use CXCR4), and dual-tropic strains (which can use both CCR5 and CXCR4). With the demonstration of functional CCR3 on cells believed to be natural targets for HIV-1 including microglia in the central nervous system (26) and blood-derived macrophages (this report), the coreceptor activity of this molecule assumes added significance.
Our results agree with previous reports using various experimental systems (7, 10, 11, 26) with respect to CCR3 usage by some M-tropic strains and the dual-tropic 89.6 isolate. However, we have also detected CCR3 function for some (but not all) TCL-tropic laboratory-adapted isolates not observed in other systems. In particular, recombinant CCR3 was reported to function as a coreceptor for primary syncytium-inducing strains (10, 14) including the TCL-tropic ELI isolate, but several reports (7, 8, 10-12, 14) found no CCR3 coreceptor activity for the closely related laboratory-adapted strains HXB2, IIIB, and NL4-3 (in contrast with our positive results with IIIB and LAV). Our finding that Env from the TCL-tropic laboratory-adapted RF strain does not function with CCR3 adds further confidence to the positive results we obtained with the other Envs and demonstrates that the coreceptor function of CCR3 does not extend to all TCL-tropic laboratory-adapted strains. Differences in the expression systems used by various groups may account for these apparent discrepancies.
It is important to point out that CCR3 has been difficult to express in heterologous cell lines (28).3 For this reason, it is critical that surface expression be shown. Only one previous report on CCR3 interactions with HIV-1 (10) verified expression of recombinant CCR3. In this case, the expression level was only 1.5-fold over background, as determined using an anti-FLAG antibody directed against the same epitope tag used in our study; in contrast, expression of other chemokine receptors was much more robust. In our study, we clearly detected CCR3 expression, with signals typically 10-fold over background (see legends to Figs. 1 and 3). It is possible that at very low coreceptor surface densities, threshold effects influence the observed coreceptor specificities of different Envs, accounting for the negative results with laboratory-adapted strains in other studies.
A potentially significant difference between our expression system and others previously used in HIV-1 coreceptor studies is the nature of the promoter used. The vaccinia promoter we used drives high level transcription in the cytoplasm, whereas other vectors used for CCR3 all employ nuclear promoters. However, since nuclear promoters allowed high level expression of other chemokine receptors in these previous studies, the specific problem with CCR3 is more likely to be post-translational. We also note that the vaccinia assay system isolates the fusion reaction from other HIV-mediated processes and is thus ideally suited for the structure-function analyses that form the basis for this report.
Specificity Determinants of CCR3Our data indicate that different HIV-1 strains discriminate among different regions of the same receptor, strengthening a concept that has been proposed based on the fusion specificities of CCR5/CCR2b and human/mouse CCR5 chimeric coreceptors (35, 45), as well as on the inhibitory effects of coreceptor-derived synthetic peptides.4 The N-terminal segment and third extracellular loop of CCR3 appear to be critical for interaction with Envs from both M-tropic strains and TCL-tropic laboratory-adapted strains, whereas the second extracellular loop is critical for TCL-tropic but not M-tropic Envs. However, unlike the case for CCR5 where it has been shown that the N-terminal segment can function when substituted for the corresponding region of a nonfunctional chemokine receptor (35, 36), this segment of CCR3 was not sufficient for fusion by TCL-tropic or M-tropic Envs when substituted for the corresponding region of CCR1. Thus, as for CCR5 (35, 36), multiple extracellular regions of CCR3 are involved in Env recognition.
An intriguing finding in the present study is that the transmembrane/cytoplasmic domains of CCR1 and CCR3, which have highly related sequences (Fig. 2), are not interchangeable scaffolds for the extracellular domains. Thus, chimera CHI 8, which contains all the extracellular regions of CCR3 and the transmembrane/cytoplasmic domains of CCR1, lacked coreceptor activity for any of the Envs tested; similarly the fusion activity displayed by CHI 3 was not observed with CHI 7, despite the fact that these molecules differed only in the transmembrane/cytoplasmic domains. FACS analyses indicated that these differences were not attributable to impaired surface expression. Additional work is needed to learn whether the transmembrane/cytoplasmic domains interact directly with Env, or whether they act indirectly by affecting folding of the extracellular domains which contact Env.
We have confirmed previous reports (10, 26) that eotaxin can inhibit
the interaction of Envs with CCR3, although the efficiency of
inhibition was lower than in the earlier studies. Differences in
experimental parameters, including coreceptor density, could contribute to these variables. We observed substantial variability in
eotaxin inhibition of fusion depending on the specific Env tested
(55-95%), again suggesting that there may be subtle differences in
the CCR3 interaction sites for various Envs. Two other CCR3 ligands,
RANTES and MCP-3, were only weak inhibitors of
CCR3-dependent fusion reactions. This is consistent with
the potency hierarchy for CC chemokines acting at CCR3, defined either
with calcium flux or chemotaxis assays: eotaxin RANTES > MCP-3, and this correlates well with the relative binding affinities of
these chemokines for CCR3 (28, 30). This implies that eotaxin may be a
better lead structure than RANTES or MCP-3 for developing antagonists
of the CCR3-Env interaction.
Previous work with CCR5 and CXCR4 has suggested that Env interacts with the chemokine receptors and CD4 to form a physical complex (46-48). Thus the simplest interpretation of the eotaxin inhibition results is that the chemokine and the Envs bind to shared sites on CCR3. However, the finding that chemokines can induce surface down-modulation of their receptors (49, 50) suggests another mode by which they may inhibit coreceptor activity.
Role of Endogenous CCR3 in HIV-1 InfectionOur results on the Env specificity of recombinant CCR3 are consistent with the ability of eotaxin to partially inhibit fusion of macrophages with effector cells expressing M-tropic Envs (Fig. 8) and with the ability of eotaxin and an anti-CCR3 antibody to inhibit infection of cultured microglial cells with M-tropic viruses (26). Together the results suggest the importance of examining other HIV-1 target cells, such as dendritic cells and CD4+ lymphocytes for CCR3 expression and CCR3-dependent coreceptor function. It will also be important to test the role of CCR3 in HIV-1 infection of eosinophils which also express CD4.
However, the results are difficult to reconcile with the relative refractoriness of primary macrophages to productive infection by TCL-tropic viruses. The same inconsistency holds for CXCR4, which is also expressed in primary macrophages (51). Macrophages fuse at low levels with effector cells expressing Envs from TCL-tropic laboratory-adapted strains (44), and this fusion can be blocked by the CXCR4 ligand SDF-1 and anti-CXCR4 antibodies,4 suggesting a cell type- and HIV-1 strain-specific block after the entry step.
The HIV-1-resistant phenotype of individuals genetically deficient in
CCR5 (16-20) implies that CCR5 is necessary for efficient initial
infection in vivo, but does not exclude a role for other chemokine receptors at this step or in subsequent events underlying disease progression. Moreover, recent case reports indicate that CCR5(/
) individuals can become infected with HIV-1 and progress to
AIDS (52, 53). Based on its activity with the major classes of Envs and
its broad expression on blood leukocytes, CCR3 may very well function
as an alternative coreceptor for initial HIV-1 infection and disease
progression. In addition, our demonstration that CCR3 can interact with
TCL-tropic Envs suggests that it may also support viral expansion after
clinical progression to AIDS, where these strains are found more
frequently (1, 2). If so, targeting the Env-CCR3 interaction could be
therapeutically efficacious. Our data suggest that efforts to block the
N-terminal segment and third extracellular loop may have the broadest
efficacy.
We thank P. Brown and A. Moir for DNA sequencing, B. Moss for the pSC59 vector, and G. Russ, S. A. Prasad, R. Brutkiewicz, and J. Yewdell for help in FACS analysis.