We describe the expression and regulation of the HIV-1 coreceptor CXCR4/fusin. Using anti-CXCR4 mAb, we demonstrate that this chemokine receptor is highly expressed on neutrophils, monocytes, B cells, and naive T cells among peripheral blood cells. In secondary lymphoid organs CXCR4 was found to be expressed on B cells. However, individual variations with regard to surface expression could be observed on T cells. Expression of the receptor is not confined to the cell surface, as large amounts of intracellular stores can be found on various leukocytes. Upon activation with phorbol esters the amount of cell surface-expressed CXCR4 on lymphocytes increases twofold within 30 s before it is completely down-regulated within the next 2 min. Incubation of leukocytes with stroma derived factor-1α, the natural ligand for CXCR4, induces down-regulation of up to 60% of surface-expressed receptors in a pertussis toxin-insensitive manner. Interestingly, receptor cross-linking caused by incubation of cells with anti-CXCR4 mAb triggers receptor trafficking, in that the receptor is rapidly internalized and recycled to the cell surface. Therefore, receptor internalization and recycling may regulate the functional interaction of the receptor with envelope proteins during an initial step of HIV-1 infection.
Chemokines represent a family of closely related chemotactic cytokines known to be potent attractors for various leukocyte subsets such as neutrophils, monocytes, or lymphocytes (recently reviewed in 1 . Based on the locations of four distinct cystein residues that form disulfide bonds, these small (6–14 kDa) basic substances are divided into two subgroups. In the CXC family the first two cysteines are separated by a nonconserved amino acid, whereas in members of the CC family the two cysteines are adjacent to each another. It was recently shown that apart from their chemotactic activity, the CC chemokines RANTES, macrophage inflammatory protein-1α, and macrophage inflammatory protein-1β suppress HIV-1 infection, particularly of macrophage-tropic (M-tropic)2 isolates (2), indicating that chemokines or chemokine receptors might play a role in HIV-1 infection. Recently, Feng and co-workers demonstrated that the chemokine receptor CXCR4 (also termed fusin, LESTR, and HUMSTR) is used by lymphotropic strains of HIV-1 as a cofactor for viral entry (3), and others succeeded in identifying the CC chemokine receptors CCR5 and CCR3 as cofactors necessary for infection with M-tropic isolates (4, 5, 6, 7, 8). These findings shed light on previous observations that expression of CD4 in nonhuman cell lines is required for effective virus binding via envelope glycoproteins (Env) to the cell surfaces but not sufficient for productive virus infection (9, 10). Although little is known about the interaction of HIV-1 and chemokine receptors, it has been suggested that the V3 loop of gp120 may participate in this process. Exchange of only a few amino acids of this loop, which is not engaged in CD4 interaction, alters the ability of the virus to infect macrophages or transformed T cells (11, 12). Therefore, available evidence suggests a direct interaction of gp120 with CXCR4 or with CCR5, which may be strong and efficient only in the precence of CD4 (13, 14, 15).
Chemokine receptors belong to the family of G protein-coupled receptors that are characterized by seven transmembrane-spanning domains and are coupled to heterotrimeric GTP binding proteins (reviewed in 16 . Sequence analysis of leukocyte chemoattractant receptors reveals a subgroup of primarily lymphocyte-expressed chemokine receptors, now consisting of BLR1, BLR2/EBI1, and CXCR4 (16). The ligand for BLR1 is unknown, and both stroma-derived factor-1 (SDF-1) and ECL were only recently identified as ligands for CXCR4 (17, 18) and BLR2/EBI1 (19), respectively. Therefore, little is known about the physiologic function of these receptors. Using gene-targeted mice lacking the chemokine receptor BLR1, we recently succeeded in identifying BLR1 as the first chemokine receptor involved in B cell migration and microenvironmental homing of recirculating B cells to the spleen and Peyer’s patches (20). To further characterize the role of this chemokine receptor subfamily in lymphocyte migration and viral pathogenesis, we generated mAbs against CXCR4. We demonstrate that CXCR4 is expressed on various leukocyte subpopulations, including monocytes, but not on T memory cells. Interestingly, the cellular localization of this molecule is not restricted to the surface, as large intracellular stores of the receptor can be identified in some cell types. Upon activation by PMA, these storage vesicles are translocated to the cell surface within 30 s, leading to a twofold increase in the concentration of surface-exposed receptor. Subsequently, all surface CXCR4 is internalized rapidly (2–4 min). The anti-CXCR4 mAb isolated by our group are able to inhibit HIV-1-mediated syncytia formation. Moreover, they were found to induce receptor internalization. While it is still controversial whether HIV uptake into coated vesicles is required for a productive infection, it is interesting that the HIV-1 virion docks to a receptor prone to endocytosis under receptor cross-linking conditions.
Materials and Methods
Cloning of CXCR4 and production of specific mAbs
RNA was isolated from human REH cells, and reverse transcriptase-PCR was performed to amplify a fragment coding for the complete cDNA of human CXCR4. A 1/20th volume of the reverse transcriptase reaction served as a template for PCR amplification, which was performed using vent polymerase (Biolabs, Schwalbach, Germany) and 50 pmol of each primer (5′ primer, 5′-CAT GCC ATG GAG GGG ATC AGT ATA TAC AC-3′; 3′ primer, 5′-CCG GGC CCG CTG GAG TGA AAA CTT GAA GAC-3′) under the following conditions: 35 cycles of 94°C for 60 s, 55°C for 60 s, and 72°C for 90 s. After size fractionation, fragments were isolated and cloned into pGem4zMYCc vector, then transferred into the expression vector Rc/CMV, which allows the expression of a myc-tagged-CXCR4 fusion protein in accordance to the procedure described previously for human BLR1 (21, 22). A BstEII/BamHI fragment was cut from pGem4zCXCR4-MYCc containing the first N-terminal 190 bp of the cDNA of CXCR4. The fragment was cloned in the pGex-1 vector (Pharmacia, Piscataway, NJ). Using this p3x-NCXCR4 plasmid coding for a GST fusion protein containing the first 63 amino-terminal amino acids of CXCR4 (GST-NCXCR4), Escherichia coli were transformed and induced with isopropylthiogalactoside according to standard procedures. Cells were harvested and subjected to French press lysis, and the GST fusion protein was purified using glutathione-agarose and affinity chromatography. Lou/C rats were immunized with 100 μg of GST-NCXCR4 in CFA and were boosted once with the same amount of Ag. Rat splenocytes were fused with X63Ag8.653 cells as described previously (23), and 7 days later hybridomas were tested for the production of specific Abs by differential ELISA using GST-NCXCR4 and 10 irrelevant GST fusion proteins as controls, including those for the N-termini of BLR1 and BLR2/CCR7. The specificity of the mAb obtained was further tested using murine 3T3 cells transfected with expression plasmids coding for CXCR4, CXCR1, BLR1, and BLR2/CCR7 in accordance to the procedure described previously for human BLR1 (22). Only mAb binding to GST-NCXCR4- and CXCR4-transfected 3T3 cells, but not those binding to the other GST fusion proteins or to 3T3 cells transfected with the other plasmids, were considered to be specific for CXCR4.
The following Abs were used in this study: anti-CD4-FITC, anti-CD8-PE, anti-CD14-PE, anti-CD19-FITC, and anti-CD62L-PE (all purchased from Becton Dickinson, Mountain View, CA). Anti-CD45RA-FITC, anti-CD45R0-PE, and mouse anti-rat F(ab′)2-FITC were purchased from Dianova (Hamburg, Germany). The anti-BLR1 mAb RF8B2 (rat IgG2b) was produced at our laboratory and was characterized previously (24). Cy3 and Cy5 conjugates of Abs were prepared as recommended by the manufacturer (Amersham, Arlington Heights, IL).
PBL were isolated from venous blood of healthy donors or of HIV+ individuals at various stages of the disease. Mononuclear cells were separated by centrifugation on Ficoll gradients (Biochrom, Berlin, Germany) for 30 min at 1200 × g. The interphase was harvested, and cells were washed twice in staining buffer (PBS, 4% FCS, 5 mM EDTA, and 0.1% NaN3, pH 7.4) at 4°C and kept on ice until used. In some experiments leukocytes were isolated by lysis of erythrocytes using the NH4Cl method. Normal tonsil tissues were obtained from tonsillectomies of children 4 to 14 yr of age. Tonsil tissues were minced and strained through a stainless steel mesh. To remove debris, cells were centrifuged over a Ficoll cushion as described above. Thymic tissue was obtained from children 3 to 5 yr of age undergoing cardiovascular surgery at the Deutsches Herzzentrum (Munich, Germany). Suspensions were made as described for tonsillar lymphocytes, omitting the Ficoll separation step.
Cell culture and Western blotting
B and T cell lines used were cultured in RPMI/10% FCS in a humidified incubator at 37°C 5% CO2. Western blotting was performed as described previously (21).
Immunohistology and flow cytometry
For immunohistochemical analysis of lymphoid tissue, tonsils were embedded in Tissue-Tek (Miles, Elkhart, IN), snap-frozen in liquid nitrogen, and stored at −80°C. Eight-micron cryostat sections on gelatin-covered slides were dried overnight and fixed with acetone. Slides were rehydrated and preincubated for 1 h in staining buffer (0.1 M Tris, pH 8.0, and 0.1% Tween-20) containing 5% heat-inactivated rat serum. Sections were incubated with conjugated primary Abs in a humidified chamber for 1 h at room temperature. After three washes in staining buffer, binding of biotinylated Abs was revealed with streptavidin-TRITC (tetramethyl rhodamine isothiocyanate) (The Jackson Laboratory, Bar Harbor, ME; 1/150) or streptavidin-Cy5 (Amersham; 1/150). After three final washes the slides were mounted in Moviol (Calbiochem, La Jolla, CA) and analyzed by confocal microscopy (Leica, Heidelberg, Germany). Digitized images were printed on a Sony video printer (UP-D8800; Sony, Berlin, Germany). Flow cytometric analysis of lymphoid cells was performed as described previously (24).
PBL were adjusted to 2 × 106/ml in RPMI/5% FCS, incubated at 37°C, and then stimulated with PMA (50 ng/ml) and ionomycin (800 ng/ml). After various incubation periods at 37°C, aliquots were added to 10-fold volumes of ice-cold FACS buffer (PBS, 2% FCS, 10 mM EDTA, and 0.1% NaN3, pH 7.2) and placed on ice. Cells were stained with saturating amounts of FITC-labeled anti-CXCR4 and various PE-labeled mAb (CD4, CD8, CD14, and CD19) at 4°C for 30 min and analyzed by flow cytometry. Linearized mean fluorescence values of cell populations were calculated using Winlist 3.0 software (Verity Software House, Inc., Topsham, ME) and used as a parameter to monitor relative changes in the amount of cell surface-expressed CXCR4 and CD4 (see below). To quantitate the uptake of anti-CXCR4 mAb and BLR1 mAb, cells were incubated with saturating amounts of fluorochrome (FITC or CY3)-conjugated mAb for 30 min on ice. Some aliquots were transferred to 37°C, while others were kept on ice for various periods of time. Uptake of labeled Ab was analyzed by flow cytometry or confocal microscopy. To determine the amount of surface CXCR4 remaining on the cell surface after various incubation periods with anti-CXCR4, cells were incubated with unlabeled anti-CXCR4 mAb for 30 min on ice before some samples were transferred to 37°C. After various incubation periods, aliquots were removed, washed, and stained with FITC-labeled anti-species conjugate. In some experiments cells were coincubated with FITC-labeled transferrin (Sigma Chemical Co., St. Louis, MO) and anti-CXCR4-Cy3 mAb to check for colocalization of transferrin-containing endosomes and anti-CXCR4-containing endosomes by confocal microscopy. To assess the influence of pertussis toxin (PTX) on CXCR4 internalization, cells were preincubated with various amounts of PTX for 2 or 18 h before FITC-labeled anti-CXCR4 was added. To test whether SDF-1α induces down-regulation of surface CXCR4, cells were incubated with recombinant SDF-1α (Peprotech, London, U.K.) at 4 or 37°C. After 1 h, cells were washed extensively and stained with anti-CXCR4 mAb (30 min on ice) to determine the amount of surface-expressed CXCR4.
Inhibition of HIV-1/RF-induced cytotoxicity by anti-CXCR4 mAb
The c8166 cells (2 × 104) were incubated with 100 μl of serially diluted mAb in RPMI 1640/10% FCS. After 2 h at 37°C, cells were infected by addition of 100 μl of virus stock (HIV-1/RF; multiplicity of infection = 0.05) and subsequently cultivated for 3 to 5 days at 37°C. Virus-induced cytotoxic syncytia formation was quantified by measuring metabolization of MTT dye as described previously (25). Culture medium in hexplicates (SD = <12%) and rat isotype mAb in serial dilutions served as controls. The OD values were compared with the mean OD value of the culture medium controls (virus without mAb), which is given as 100%.
Generation of mAbs
Rats were immunized with a GST fusion protein containing the first 63 amino-terminal amino acids of CXC4 (GST-NCXCR4), and hybridoma supernatants were tested for the presence of CXCR4-specific mAb by differential ELISA using either GST-NCXCR4 or various irrelevant GST fusion proteins (see Materials and Methods). This procedure enabled us to identify 44 clones binding to GST-NCXCR4 but not to the control GST fusion proteins (data not shown). The specificity of the hybridomas obtained was further tested by using murine 3T3 cells transiently transfected with a myc-tagged expression plasmid encoding for the entire sequence of CXCR4 or other chemokine receptors. Using these cells and flow cytometry, three mAb (1D9, 2B11, and 3D2) were found to bind to CXCR4-MYC transfected 3T3 cells, but not to BLR1-MYC transfected 3T3 cells (Fig. 1⇓A) or to 3T3 cells transfected with plasmids coding for the chemokine receptors CXCR1 or BLR2/CCR7 (data not shown). These Abs were applied in protein immunoblot analysis and were found to identify a major protein band with an apparent molecular mass of 45 kDa, which is close to the predicted molecular mass derived from the amino acid sequence (40 kDa; see Fig. 1⇓B). All mAb reacted with the same specificity, and all additional experiments were performed using anti-CXCR4 2B11.
Expression of CXCR4 on lymphoid cells
Using anti-CXCR4 mAb and three-color flow cytometry we found CXCR4 to be expressed in a variety of lymphoid cells. On peripheral blood leukocytes, expression of CXCR4 was observed on all PMN (Fig. 2⇓A), on about 75% of monocytes (Fig. 2⇓B), and on about 90% of B cells (Fig. 2⇓C). The expression pattern of CXCR4 on T cells was analyzed in more detail. Seventy percent of CD4+ lymphocytes and 40% of CD8+ T cells were associated with this receptor. Using mAb against L-selectin (CD62L) or CD45RA, CXCR4-positive T cells were primarily identified as naive, Ag-inexperienced T cells (reviewed in 26 , as 90% of the CD4+CXCR4+ cells and about 75% of the CD8+CXCR4+ cells coexpressed L-selectin (Fig. 2⇓, D and E) or CD45RA (data not shown). In contrast, only 20 to 40% of CD45RO+ T memory cells were associated with this chemokine receptor (data not shown). As CXCR4 was expressed strongly on naive T cells, we also looked for CXCR4 expression at earlier stages of T cell development. In total, CXCR4 is expressed on the majority (>90%) of T cells isolated from juvenile thymi. Interestingly, a similar expression pattern was observed for both immature (CD4+CD8+) and mature (CD4+CD8− or CD4−CD8+) T cells, and no difference in CXCR4 expression during these stages of T cell development was observed (Fig. 3⇓, A–C). In contrast to high CXCR4 expression levels on T cells isolated from peripheral blood or thymus, CXCR4 was barely detectable on T cells isolated from three (of four) different tonsils (data not shown and compare Fig. 4⇓A, T zone). In one case, however, tonsillar T cells were found to express CXCR4 at levels similar to those expressed by peripheral blood T cells (data not shown). In contrast, almost all CD19+ B cells (>80%) isolated from tonsils expressed CXCR4 at a high rate. Interestingly, counterstaining with anti-CD10 or anti-CD38 mAb, which both stain defined germinal center B cells, marked a subpopulation of germinal center B cells not expressing CXCR4 (Fig. 3⇓, Dand E).
Confocal microscopy on tonsillar sections confirmed the absence or low levels of CXCR4 on T cells, whereas all areas of B cell follicles were readily stained by anti-CXCR4 mAb. Although this mAb does not allow the identification of follicular substructures (Fig. 4⇑A), some cells within the germinal center could be identified that were not stained with anti-CXCR4 mAb. However, extrafollicular cells could be identified that were stained brightly with this mAb. These cells were in close proximity to the tonsillar epithelium (Fig. 4⇑B, arrowheads) and along tonsillar septae (Fig. 4⇑B, long arrows). They were classified as B cells because they coexpress CD19 (data not shown). Further functional and flow cytometric experiments must be performed to further delineate the phenotype of these cells, but the data presented here are in accordance with the localization and phenotype recently described for subendothelial B cells (27, 28). Surprisingly, in those cells not only was CXCR4 expressed on the cell surface as expected for members of this family, but they accumulated in large intracellular compartments (Fig. 4⇑C).
Rapid up-regulation and subsequent complete internalization of surface CXCR4 upon activation with phorbol esters
As we had identified preformed cytoplasmic CXCR4 in some cell types (compare Fig. 4⇑C), we tested whether activation influences the distribution between pools of cytoplasmic and surface-expressed CXCR4. PBL were therefore stimulated with PMA and the calcium ionophore ionomycin. The amount of surface CXCR4 on different leukocyte subpopulations was determined by flow cytometry using FITC-labeled anti-CXCR4 mAb and various PE-labeled mAb specific for leukocyte subpopulations. As shown in Figure 5⇓, activation of PBL with PMA/ionomycin rapidly releases cytoplasmic CXCR4 to the cell surface. Thirty seconds poststimulation, the mean fluorescent intensities of the lymphocyte, monocyte, and neutrophil (not shown) subpopulations within the sample increased approximately twofold compared with those at the zero time point. This peak in expression was followed by a steep decline in CXCR4 surface concentration within <10 min. However, receptor down-regulation occurred with different kinetics on different leukocyte subpopulations. Four to six minutes after activation, surface CXCR4 was completely undetectable on the lymphocyte surface. On monocytes, down-regulation was significantly slower; after the same incubation period, approximately identical levels of surface CXCR4 could be detected as at the beginning of the experiment. Here, incubation periods had to be extended to 15 min until surface CXCR4 was completely down-regulated on these cells (Fig. 5⇓B). As PMA is also known to induce down-regulation of CD4, we compared the down-regulation kinetics of CD4 and surface CXCR4 on peripheral blood Th cells. In contrast to the pattern observed for CXCR4, no initial increase in surface-bound CD4 could be observed, and even 20 min poststimulation, >50% of the surface CD4 was still detectable. To address the question of whether CXCR4 is internalized or shed from the cell upon activation, we stained PBL with FITC-labeled anti-CXCR4 mAb. Cells were washed and activated with PMA/ionomycin as described above. After different incubation periods aliquots were removed and subjected to flow cytometry. During a 30-min period there was no change in fluorescence activity, indicating that the receptor/mAb moiety remains cell associated. This observation was confirmed by confocal microscopy demonstrating that the receptor/mAb complex is transferred from the cell surface (Fig. 5⇓C) into intracellular vesicles following activation of cells with PMA/ionomycin (Fig. 5⇓D).
SDF-1α induces receptor down-regulation
As receptor down-regulation was observed upon activation with PMA, we asked whether SDF-1α, the natural ligand for CXCR4 would induce this effect as well. White blood cells (WBC) were therefore incubated with different concentrations of this chemokine for 1 h at either 4 or 37°C before cells were washed, stained with anti-CXCR4 mAb, and subjected to flow cytometry. The different forward/side scatter properties of lymphocytes, monocytes, and neutrophils were used to set electronic gates on these cell populations, and changes in the relative amount of surface-expressed CXCR4 were determined using the mean fluorescence intensity of fluorochrome-labeled anti-CXCR4 mAb. As demonstrated in Figure 6⇓A, we did not observe any difference in surface-expressed CXCR4 on any population regardless of the presence or the absence of SDF1α when cells were incubated at 4°C (hatched bars in Fig. 6⇓A). This observation indicates that binding of SDF-1α to CXCR4 does not interfere with binding of mAb 2B11 to the receptor. However, incubating cells with SDF-1α at 37°C induced down-regulation of up to 60% of surface-expressed CXCR4 on all WBC populations (Fig. 6⇓A, filled bars). Interestingly, we constantly observed a marked increase in CXCR4 on lymphocytes and monocytes after a 1-h incubation period in RPMI/5% FCS at 37°C but not at 4°C (compare hatched and closed bars in Fig. 6⇓A at 0 ng/ml SDF-1α), indicating that internal stores of the receptor are translocated to the cell surface under this condition (see also below). As both the SDF-1α-induced increase in free intracellular Ca2+ and cell migration are sensitive to PTX (29), we tested whether inhibitory G proteins (Giα) are also involved in receptor internalization. Interestingly, preincubation of WBC with 100 ng/ml PTX for 2 h at 37°C could not prevent ligand-induced internalization of CXCR4 on PMN, lymphocytes, or monocytes (Fig. 6⇓B, data shown for lymphocytes only). Together with the observations made by others, these data demonstrate that PTX-sensitive G proteins are involved in defined cellular signaling pathways but not in receptor internalization, thus resembling the situation described by others for the angiotensin I receptor (30).
Anti-CXCR4 mAb induce receptor internalization and recycling
We then asked whether anti-CXCR4 mAb could induce receptor activation or internalization in a similar fashion. Therefore, various T and B cell lines were incubated with FITC-labeled anti-CXCR4 mAb for various periods at 4 or 37°C. Aliquots were removed, washed, and subjected to flow cytometry. Within the incubation period observed, the fluorescence intensity of CEM T cells continuously increased (Fig. 7⇓A), reaching a plateau at approximately 90 min. A similar effect was observed for other T cell lines when incubated at 37°C (Fig. 7⇓B, closed symbols). In contrast, in none of the cell lines tested was increased fluorescence intensity observed when cells were incubated at 4°C (Fig. 7⇓B, open symbols). Neither CEM cells nor ESIII cells (see below) possessed noticeable amounts of internal CXCR4 as observed for primary lymphocytes. Thus, the release of preformed cytoplasmic CXCR4 to the cell surface could not account for the increased cell-bound fluorescent label. Likewise, de novo synthesized and mAb-stained receptor did not add to the increased fluorescence intensity, since the uptake kinetics did not change when cells were treated with cycloheximide, which is known to prevent de novo protein synthesis (data not shown). Therefore, we suggest that incubation with anti-CXCR4 mAb may induce receptor recycling in a way that internalized receptor becomes redistributed to the cell surface, initiating a new round of Ab transport inside the cell. To test this more vigorously, we first asked whether Ab-induced receptor internalization is characteristic of other chemokine receptors as well. ESIII Burkitt’s lymphoma cells that express BLR1, BLR2, and CXCR4 were incubated with specific mAbs. As observed for T cell lines, all anti-CXCR4 mAb used generated continuously induced receptor internalization over a period of 2 h before reaching a plateau indicative of saturation (Fig. 7⇓, C and E, data shown for clone 2B11). In contrast, neither anti-BLR1 mAb nor anti-BLR2 mAb induced receptor internalization at all (Fig. 7⇓, D and E, data for BLR2 not shown). To address the question of whether receptor internalization is a continuous process independent of Ab-mediated cross-linking, we used FITC-labeled Fab of anti-CXCR4 mAb in the experiment outlined in Figure 7⇓E with ESIII cells. Although these Fab bound CXCR4, they did not induce internalization of CXCR4, suggesting that receptor cross-linking is required for receptor uptake (data not shown). As mentioned before (see also Fig. 7⇓E), receptor uptake was saturated 2 h after the start of incubation with mAb. It is conceivable that previous Ab incubation cleared all CXCR4 molecules from the cell surface, preventing ongoing Ab uptake. Therefore, ESIII cells were incubated with a large excess of unlabeled anti-CXCR4 mAb. Samples were withdrawn after various incubation periods, and levels of surface-bound anti-CXCR4 mAb were determined by counterstaining with FITC-conjugated anti-rat Ab. As demonstrated in Figure 7⇓F, expression levels of sCXC4 did not depend on either the incubation temperature (4 or 37°C) or the incubation time, indicating that the number of surface CXCR4 remains constant during the process of Ab internalization. As incubating cells in hypertonic medium has been shown to prevent the formation of clathrin-coated pits, we incubated CEM T cells in medium containing 0.6 M sucrose to determine whether mAb-mediated receptor internalization and recycling use this pathway. As shown in Figure 7⇓G, incubating cells in hypertonic medium completely inhibited receptor internalization, indicating that clathrin-coated pits are involved in receptor internalization and recycling. In summary the data presented provide evidence that CXCR4 cycles from the surface to the cell interior and back upon cross-linking mediated by anti-CXCR4 mAb. To address the question of whether inhibitory G proteins are involved in receptor internalization, ESIII cells were incubated with pertussis toxin before labeled Ab was added. As described above for SDF-1α (compare Fig. 6⇑B), at no concentration tested did PTX exert any influence on receptor internalization, indicating that this process is also independent of Giα (Fig. 7⇓H).
To further characterize the pathway of CXCR4 internalization, we fixed and permeabilized ESIII cells and subsequently stained them with anti-CXCR4 mAb. As demonstrated in Figure 8⇓A, most of the Ab bound to the cell surface, indicating the absence of large quantities of cytoplasmic CXCR4 in these cells. A similar staining pattern was observed analyzing unfixed cells when incubated with anti-CXCR4 at 4°C for 1 h (Fig. 8⇓B). As expected from flow cytometry data, following incubation of cells for the same time at 37°C, large cytoplasmic vesicles stained by the Ab could be observed (Fig. 8⇓C). Next, ESIII cells were incubated with Cy3-labeled anti-CXCR4 and FITC-labeled transferrin. Confocal microscopy (Fig. 8⇓D) demonstrated partially overlapping endocytotic pathways for both transferrin and the anti-CXCR4 complex, as some vesicles contained transferrin (green) or anti-CXCR4 (red) only, whereas others contained both molecules (yellow).
Anti-CXCR4 mAb neutralize infection with HIV-1
As CXCR4 has been identified as a coreceptor for HIV-1, we used HIV-1/RF to test CXCR4 mAb for their ability to prevent HIV-1 infection. In the presence of anti-CXCR4 mAb, c8166 cells were infected with HIV-1/RF, and virus-induced syncytia formation was quantitated indirectly analyzing metabolic activity of the cells. As demonstrated in Figure 9⇓A, anti-CXCR4 mAb reduces HIV-1 cytotoxicity in a dose-dependent manner, demonstrating that this Ab could inhibit virus infection. Since receptor internalization was observed upon addition of anti-CXCR4 Ab (see Fig. 7⇑), it was possible that inhibition of virus infection was due to Ab-induced down-regulation of surface CXCR4. However, when tested, slightly increased levels of sCXCR4 could be detected after prolonged incubation periods (24 h) with the Ab (data not shown). As CXCR4 is expressed on various lymphocyte subpopulations, we investigated whether altered expression patterns of this receptor can be observed in HIV-1+ individuals. Analyzing PBL from 30 patients at different stages of the disease, we found an overall significantly reduced percentage of both CD4+ or CD8+ cells coexpressing this HIV coreceptor (Fig. 9⇓B).
It has been suggested that HIV-1 gp120 simultaneously interacts with both CD4 and CXCR4 (15). Therefore, we stained CEM T cells with anti-CD4-Cy5 and CXCR4-Cy3 mAb. Large areas could be identified on the cell surface where both molecules colocalized (Fig. 9⇑C). Interestingly, when incubating CEM cells with gp120-coated latex beads, surface-expressed CXCR4 was primarily found at those areas of the cell where latex beads had bound (Fig. 9⇑D). As we did not observe binding of gp120 latex beads to ESIII cells that do not express CD4 but high levels of CXCR4 (data not shown), these observations indicate that CD4 and CXCR4 are closely associated at the cell surface, and that CD4 and CXCR4 are cocapped upon binding of latex-bound gp120.
It has been known for >10 yr that CD4 acts as a receptor for HIV-1, HIV-2, and SIV, permitting firm attachment of these particles to the cell surface (31, 32, 33, 34). However, it soon became clear that expression of this molecule is not sufficient to allow productive infection with any of these viruses, and it has been postulated that other (co)-receptors must be existing. Recently, the G protein-coupled chemokine receptor CXCR4 has been identified to function as a coreceptor for T cell-tropic (T-tropic), but not for M-tropic, strains of HIV-1 (3) and to allow CD4-independent infection with some strains of HIV-2 (35). As the CC chemokine receptors CCR5 and CCR3 have been demonstrated to act as coreceptors for M-tropic strains (4, 5, 6, 7, 8), it has been suggested that cell tropism of virus strains is defined by both molecular evolution of HIV Env and the expression pattern of the corresponding chemokine receptors on the surface of the target cell. Furthermore, as T-tropic strains cannot infect macrophages, it has been postulated that CXCR4 is expressed on the surface of T cell lines and primary T cells, but not on macrophages or monocytes (36). However, earlier experiments using Northern blot analysis demonstrated expression of CXCR4-specific transcripts in myeloid cell lines as well as in primary monocytes and neutrophils (37), suggesting expression of CXCR4 in these cells. These findings were corroborated in the present study using an mAb specific for human CXCR4 and by SDF-1α-induced receptor internalization observed on lymphocytes, monocytes, and neutrophils. As both monocytes and naive T cells express similar levels of sCXCR4, it is unlikely that the number of surface receptors expressed on monocytes or macrophages is too low to allow infection with T-tropic strains. Interestingly, mutation analysis of CCR5 demonstrated that different M-tropic strains require different residues of the first 10 N-terminal amino acids or the first extracellular loop of the receptor for membrane fusion and infection (38). Although there are currently no data available, it seems plausible that 1) due to differential splicing or post-translational modifications, different forms of CXCR4 are expressed on macrophages and T cells that cannot be distinguished with any of the three mAb but can be distinguished by different HIV-1 strains. 2) In addition to CXCR4 (and other chemokine receptors identified to date), further cofactors are required to determine cell tropism, i.e., to allow for productive infection.
Secondary lymphatic organs are known as a major source of HIV replication and virus spread. Various cell populations such as macrophages, dendritic cells, and T cells have been identified as major players in this process (39, 40, 41). Interestingly, we found that expression of CXCR4 on T cells is primarily restricted to naive T cells, as thymocytes and L-selectin+ or CD45RA+ peripheral blood T cells were stained with the mAb. This is of particular interest as on T cells, expression of most of the known chemokine receptors is restricted to activated or memory (CD45R0+) cell types (24, 42, 43). CXCR4 is therefore the first chemokine receptor showing preferential expression on naive T cells. Another remarkable feature regarding expression of this receptor is the increase in surface expression following incubation of WBC in culture medium. After a 1-h incubation period we usually observed a twofold increase in the amount of surface CXCR4 on lymphocytes and monocytes (see Fig. 6⇑A). It seems possible that storage temperature of blood samples before FACS analysis is a crucial parameter when analyzing the expression pattern of this receptor.
We have shown that the mAb generated can inhibit HIV-1/RF-induced cytotoxicity in c8166 cells. However, rather high amounts of mAb (50–80 μg/ml) were needed to completely inhibit infection with this strain. Recently, others demonstrated that their mAb against CXCR4 is able to inhibit infection with some strains of HIV-2 at lower concentration (10 μg/ml), but also reported that inhibition of HIV-1 infection with their Ab depends on the viral isolate and the target cell used (35), indicating that viruses might contact different epitopes of CXCR4. Cellular receptors for enveloped viruses are characteristically removed from the cell surface to prevent superinfection (44). Our data indicate that receptor down-regulation is not a mechanism underlying the neutralizing effect of anti-CXCR4 mAb, as even after prolonged incubation periods (<24 h), this Ab does not induce receptor clearance from the cell surface. It is therefore more likely that the Ab block receptor epitopes necessary for infection with individual strains of HIV-1. Although we frequently observed reduced proportions of CXCR4+ T cells in HIV+ individuals, it seems unlikely that this finding directly reflects virus-induced down-regulation of the chemokine receptor as both CD4+ and CD8+ cells were equally affected (see Fig. 7⇑B).
CXCR4 is a member of the G protein-coupled receptor family that is characterized by seven hydrophobic transmembrane-spanning domains. Some members of this family are known to be internalized upon binding with their natural ligand (45, 46). However, this report describes for the first time the existence of lymphocytes that contain large intracellular stores of a preformed chemokine receptor. One of the most interesting features of this receptor was the rapid translocation upon activation with PMA (Fig. 5⇑). Similar phenomena have been described for various molecules, including CD4 and CD40L (47, 48). Preformed CD40L has been identified in a subpopulation of Th memory cells, and activation with PMA resulted in a permanent translocation to the cell surface, reaching the maximum after 15 min (47). In contrast, activation of resting PBL with PMA immediately transferred preformed CXCR4 to the cell surface, reaching a maximum within 30 s. Furthermore, in contrast to the observation made for CD40L, CXCR4 did not remain at the cell surface, but was completely internalized within 2 to 5 min. Interestingly, receptor internalization upon activation with PMA has also been observed for CD4, the primary HIV receptor (48). In contrast to the observation made for CXCR4, we never observed an increase in surface CD4 upon stimulation, and receptor internalization occurred considerably slower, as after 20 min >50% of the CD4 molecules were still found at the cell surface (Fig. 5⇑).
Internalization of CXCR4 could also be induced upon incubation with any of the three mAb specific for the receptor. Under these conditions, receptor internalization occurred at a constant rate for the first 40 to 60 min and reached saturation after approximately 2 h. After this incubation period, the amount of mAb found in the cell was about 3 times higher than mAb bound to the cell surface, demonstrating a high frequency of receptor trafficking. Interestingly, although internalization was observed at high frequency, the amount of surface CXCR4 remained constant over that period. This finding together with the observation that intracellular stores of CXCR4 were not detectable in ESIII cells indicates that CXCR4 is internalized upon binding by mAb and is then completely recycled to the cell surface once the Ab is released to intracellular vesicles. In control experiments we could not observe an increase in cell-associated Fab derived from parental 2B11 mAb with time. This indicates that cross-linking of CXCR4 by bivalent mAb is required for uptake, and that unlike the situation observed for CTLA-4 (49), continuous circulation of unstimulated CXCR4 between intracellular vesicles and the cell surface is not responsible for Ab uptake. As observed for the mAb, incubating cells with SDF-1α also resulted in receptor internalization. In contrast, binding of SDF-1 to CXCR4 resulted in an internalization of the receptor without significant concomitant receptor recycling. Therefore, we assume that CXCR4 enters different pathways for receptor internalization upon stimulation with either ligand or mAb. It is likely, however, that the described processes, CXCR4 internalization caused by PMA stimulation, Ab cross-linking, or binding to SDF-1α, trigger different secondary signaling events. Several lines of evidence would support this idea: 1) the different time kinetics of receptor internalization (compare Figs. 5–7⇑⇑⇑ and 2) complete receptor clearance from the cell surface upon PMA treatment as opposed to unchanged steady state levels of surface-exposed receptor upon Ab cross-linking and partially down-regulated expression after SDF-1α binding. However, it remains to be determined if activation events caused by either mechanism merge into identical signaling pathways. Our observations that both SDF-1α- and mAb-triggered receptor trafficking are insensitive to PTX treatment, whereas cell migration toward SDF-1α is sensitive to PTX (29), would argue against this.
Recent evidence suggests cell surface association between CXCR4 and CD4 upon binding with HIV gp120 (15). The interaction between gp120 and CXCR4 is apparently weak, as exemplified by the lack of detectable binding of gp120-coated latex beads to CXCR4 expressing ESIII cells (see Results). Therefore, in most cases CD4 remains the key determinant for initial attachment of virus to the cell surface. It is remarkable that this seven transmembrane receptor whose affinity toward the virus is low in most cases is involved in viral entry. However, the CD4 gp120 interaction itself causes conformational changes in the gp120 molecule (50, 51). This, in turn, may not only strengthen the binding of gp120 toward CD4, but may also lead to the presentation of binding epitopes for CXCR4. We showed that CXCR4 is internalized readily and quickly upon activation by PMA, mAb cross-linking, and binding to SDF-1α. Assuming that both Ab cross-linking and cross-linking caused by interaction with the virus trigger similar events, it is tempting to speculate that HIV exploits the property of this activated coreceptor to promote membrane rearrangements and connections to the cytoskeleton of the cell. These represent processes necessary to imitate receptor uptake, but may be required as well to render a large patch of membrane competent for fusion (a step in viral entry probably triggered by viral gp41). Indeed, we found that CXCR4 cocaps with a gp120/CD4 complex (see Fig. 9⇑D). As chemokine receptors have been repeatedly postulated to activate adhesion molecules during cell migration (52), the cross-linking of CXCR4 by attached virus may also activate adhesion molecules on the cell surface via the CXCR4 signaling pathway. This could lead to intimate and broad contact areas between membranes of the virus and the cell.
The authors are very grateful to Dr. Sabine Johann (Medical School, Munich, Germany) for helpful discussions and to Brigitte Scherz for critically reading the manuscript.
↵1 Address correspondence and reprint requests to Dr. Reinhold Förster, Max Delbrück-Center, Robert Rössle Str. 10, 13122 Berlin-Buch, Germany. E-mail address:
↵2 Abbreviations used in this paper: M-tropic, macrophage-tropic; SDF-1, stroma-derived factor-1; GST, glutathione S-transferase; PE, phycoerythrin; PTX, pertussis toxin; MTT, 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide; WBC, white blood cells; T-tropic, T cell tropic; CD40L, CD40 ligand.
- Received January 21, 1997.
- Accepted October 22, 1997.
- Copyright © 1998 by The American Association of Immunologists