INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The human and simian immunodeficiency viruses (HIV-1, HIV-2, and SIV) normally require the presence of both CD4 and a chemokine receptor at the cell surface for entry into a target cell. Different viral strains use distinct members of the chemokine receptor family as coreceptors (for reviews, see references 5 and 44). The chemokine receptors CCR5 and CXCR4, in particular, function prominently in HIV-1 infection. Viral strains are classified as R5, X4, or X4R5 according to whether they use CCR5, CXCR4, or both as coreceptors. While macrophage-tropic primary isolates preferentially use CCR5 and many T-cell-tropic isolates use both CXCR4 and CCR5, viruses adapted to growth in T-cell lines preferentially use CXCR4 (6, 66).

Enveloped viruses enter cells by fusion with the plasma membrane or with the endosomal membrane after endocytosis (for a recent review, see reference 31). In HIV infection of model cell lines, the viral envelope fuses with the plasma membrane of the target cell (38, 46, 49, 61). Although the molecular mechanism of HIV fusion is not well understood, some of the interactions that precede it have been described in great detail. The outer envelope glycoprotein, gp120 or SU (surface unit), of HIV-1 binds to CD4 with high affinity. The critical residues in both molecules have been identified by mutagenesis and crystallography (10, 11, 32, 34, 40). The SU-binding site on CD4 is centered on the CDR2-like region of the N-terminal immunoglobulin (Ig)-like domain (D1), whereas residues in SU that make contact with CD4 are located in six distinct regions of the polypeptide (32). The binding of the envelope glycoprotein, Env, to CD4 induces conformational changes in the Env-CD4 complex (42, 43, 55) that appear to facilitate a subsequent interaction with the cognate chemokine receptor (32, 62, 65). The recruitment of a chemokine receptor could be a limiting step in the fusion process. It may promote fusion merely by placing the Env-CD4 complex in close proximity to the target cell membrane, or it may trigger a final fusogenic conformational change in the Env-CD4 complex.

In order to explore the interactions between Env, CD4, and chemokine receptors in more detail we designed a series of CD4-chemokine receptor hybrids. When the orientation of CD4 to the rest of the hybrids is appropriate, such constructs might be predicted to enhance the functional affinity of SU for the chemokine receptor moieties by allowing two-point interactions on a single molecule. The hybrid constructs might also circumvent the potentially limiting step of coreceptor recruitment. The two most N-terminal Ig-like domains of CD4 (D1D2) were linked to the N termini of CXCR4, CCR5, CCR2b, and CXCR2. The latter two chemokine receptors are known to have only weak or no coreceptor function for HIV-1 (5, 44), although they can function as coreceptors with CD4 in cell-to-cell fusion induced by an HIV-2 envelope glycoprotein (9). The hybrids between D1D2 of CD4 and CXCR4 were constructed with and without a Gly- and Asn-rich 39-residue spacer between the CD4 moiety and the N terminus of the chemokine receptor. In addition, the entire extracellular four-domain fragment of CD4 was linked to the N terminus of CXCR4 (Fig. 1).

The physiological ligand for CXCR4 is the CXC chemokine stromal cell-derived factor 1 (SDF-1) (7, 45), a peptide with a molecular mass of approximately 8 kDa. SDF-1 blocks X4 virus infection by two mechanisms: down-modulation of CXCR4 from the cell surface and competition with SU for binding to CXCR4 (2, 59). It could therefore be predicted that enhancement of SU-CXCR4 interactions will diminish the antiviral potency of SDF-1. A synthetic nonpeptide molecule, the bicyclam AMD3100, with a molecular mass of only 1 kDa, competes with both SDF-1 and the CXCR4-specific monoclonal antibody (MAb) 12G5 for binding to CXCR4. It also blocks X4 virus infection (15, 57, 58) by a mechanism that, as we demonstrate, does not involve down-modulation of CXCR4. In these and other regards we illustrate how hybrids between receptors and coreceptors can be useful tools in analyzing interactions with envelope glycoproteins and the antiviral mechanisms of coreceptor ligands.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells. The cell line Mv-1-lu, derived from fetal mink lung epithelium, and Mv-1-lu cells stably expressing human CD4 (12), were obtained from the AIDS Reagent Project (ARP) of the United Kingdom Medical Research Council (Potters Bar, United Kingdom). The Mv-1-lu cells expressing wild-type CXCR4 and CD4 were previously described (59). All tissue culture media and reagents were from Gibco Ltd. (Paisley, United Kingdom). The Mv-1-lu cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, penicillin (10 IU/ml), and streptomycin (10 mg/ml). In addition, 1 mg of G418 per ml and 0.5 mg of hygromycin per ml were included for cell lines stably transfected with plasmids carrying the respective resistance markers.

Virus. A stock of HIV-1_LAI was prepared from chronically infected H9 cells provided by the ARP. The infectivity of this stock on Mv-1-lu cells expressing wild-type CXCR4 and CD4 was approximately 10³ focus-forming units per ml. The X4R5 strain HIV-1_89.6 (17) and the R5 strain HIV-1_SF162 (13), which had been cultured in peripheral blood mononuclear cells and yielded a 50% tissue culture infective dose on peripheral blood mononuclear cells of 10⁶/ml, were donated by G. Simmons (The Institute of Cancer Research, London, United Kingdom).

MAbs, recombinant proteins, and CXCR4 ligands. The MAb Q4120, which reacts with an epitope overlapping the SU-binding site in the N-terminal Ig-like domain, domain 1 (D1), of CD4 (24), was obtained from the ARP. The MAb 12G5 (20), directed to a discontinuous epitope, which involves the second extracellular loop of CXCR4 (8), was provided by J. Hoxie (University of Pennsylvania, Philadelphia, Pa.). The MAbs to HIV-1 Gag proteins 38:96K and EF7 (27) were obtained from the ARP. Q4120 and 12G5 were labeled with ¹²⁵I as previously described (59).

Construction of CD4-chemokine receptor hybrids. The signal sequence and the first two Ig-like domains (D1 and D2) of human CD4, from Met¹ to Phe²⁰⁴ (54, 63), were fused to the N terminus of human CXCR4, yielding the hybrid designated CD4(2D)CXCR4. The corresponding linkages were made for CXCR2, CCR5, and CCR2b. CD4 fragments with overlaps to the N-terminal end of the respective chemokine receptors and the entire chemokine receptor genes with overlaps to the C-terminal end of the CD4 fragment were generated separately in initial PCRs with wild-type CD4 and the corresponding chemokine receptor cDNAs as templates. The primers were designed to include appropriate sites for restriction enzymes. The primer for the N-terminal end of CD4 was TAG AAG CTT ACC ATG AAC CGG GGA GTC CCT (sense). The primers for the C-terminal ends of the chemokine receptors were as follows: for CXCR4, ACC GAA TTC TTA GCT GGA GTG AAA ACT TGA (antisense); for CXCR2, CAC GAA TTC CTA TTA GAG AGT AGT GGA AGT (antisense); for CCR5, CAC GGA TCC TCA CAA GCC CAC AGA TAT TTC (antisense); and for CCR2b, CAC GGA TCC CTA TTA TAA ACC AGC CGA GAC TTC (antisense). The primers used to create the overlaps for fusing the CD4 and chemokine receptor moieties were as follows: for CD4(2D)CXCR4, GTG CTA GCT TTC ATG GAG GGG ATC AGT ATA TAC (sense) and GTA TAT ACT GAT CCC CTC CAT GAA AGC TAG CAC CAC GAT GTC (antisense); for CD4(2D)CXCR2, GTG CTA GCT TTC ATG GAA GAT TTT AAC ATG (sense) and CAT GTT AAA ATC TTC CAT GAA AGC TAG CAC CAC GAT GTC (antisense); for CD4(2D)CCR5, GTG CTA GCT TTC ATG GAT TAT CAA GTG TCA AGT (sense) and TCA TGA CAC TTG ATA ATC CAT GAA AGC TAG CAC CAC GAT GTC (antisense); for CD4(2D)CCR2b, GTG CTA GCT TTC ATG CTG TCC ACA TCT CGT TCT CGG (sense) and AGA TGT GGA CAG CAT GAA AGC TAG CAC CAC GAT GTC (antisense). The CD4 and chemokine receptor fragments were joined in a PCR ligation. The fusion constructs were then inserted into the HindIII-EcoRI-BamHI sites of the pTEJ8 vector (28), and the DNA sequence was confirmed.

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Schematic representations of CD4-CXCR4 hybrids. Wild-type CD4, including its four Ig-like domains, single transmembrane segment, and C-terminal cytoplasmic tail, is shown in schematic outline next to CXCR4, which has seven transmembrane segments. The CD4-CXCR4 hybrids are illustrated below. The two N-terminal Ig-like domains of CD4 were linked directly to the N terminus of CXCR4 [CD4(2D)CXCR4]. Similar CD4(2D) hybrids were also constructed with CCR2b, CCR5, or CXCR2 as the chemokine receptor moiety. As shown, the two N-terminal Ig-like domains of CD4 were also linked via a spacer to CXCR4 [CD4(2D)-Sp-CXCR4]. In addition, all four extracellular CD4 domains were linked to CXCR4 [CD4(4D)CXCR4].

Transfection of CD4-chemokine receptor hybrid genes into Mv-1-lu cells. Mv-1-lu cells (2 × 10⁶) were washed in HEBS buffer (20 mM HEPES, 137 mM NaCl, 5 mM KCl, 0.7 mM Na₂HPO₄, and 6 mM D-glucose [adjusted to pH 7.0]) and transferred to a 0.4-cm-path-length electroporation cuvette containing 10 µg of DNA in 250 µl of HEBS buffer. The cells were electroporated with a single pulse in a Gene Pulser (Bio-Rad, Hercules, Calif.) set to 250 µF, 400 V, and infinite resistance. Clones expressing the hybrids were selected by limiting dilution in medium containing 1 mg of G418 per ml.

Flow cytometric analysis of antibody and SU binding. Cells grown to confluence were detached in 10 mM EDTA in phosphate-buffered saline (PBS) and transferred to U-bottomed 96-well plates (Costar, Cambridge, Mass.) at approximately 10⁶ cells per well. All subsequent steps were performed at 4°C. Cells were spun and resuspended in 200 µl of fluorescence-activated cell sorter (FACS) wash buffer (FWB; 2% FCS and 0.02% NaN₃ in PBS). The cells were incubated with primary antibody at various concentrations in 100 µl of FWB for 2 h on a shaker. Subsequently the cells were washed twice and then incubated for 45 min with fluorescein isothiocyanate-conjugated goat anti-mouse antibodies (Pierce and Warriner UK Ltd., Chester, United Kingdom) diluted 1/200 in FWB. After three more washes, the cell-bound fluorescence was detected with a FACSCalibur flow cytometer (Becton Dickinson, San Jose, Calif.). The relative MAb binding was calculated as the mean fluorescence intensity (mfi) for each concentration of MAb (MAb[XnM] in the formula below) divided by the mfi at the plateau of binding [MAb(plateau)], after subtraction from both of the background mfi for antigen-negative cells (bd), i.e., [mfi_MAb[XnM]  mfi_bd]/[mfi_MAb(plateau)  mfi_bd].

Electron microscopy. The ultrastructural localization of hybrid receptors was determined by the use of colloidal gold labeling of cryosections essentially as described previously (35). Two days after passage, Mv-1-lu cells stably expressing CD4(2D)CXCR4 or CD4(4D)CXCR4 were fixed for 100 min in 4% paraformaldehyde, washed, embedded in 10% gelatin, infiltrated with 2.3 M sucrose, and frozen in liquid nitrogen. Cryosections (approximately 60 nm thick) were labeled with the MAb Q4120 at 5 µg/ml and a goat anti-mouse antibody conjugated to 10-nm-diameter gold particles (British Biocell International, Cardiff, United Kingdom), or with the latter alone as a control, and then examined with a transmission electron microscope (EM 420; Phillips, Eindhoven, The Netherlands).

Western blotting. Mv-1-lu cells stably expressing hybrids or wild-type CD4 and CXCR4 were lysed in 20 mM Tris-HCl buffer, pH 8, containing 1% Triton X-100, 150 mM NaCl, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 10 µg each of chymostatin, leupeptin, antipain, and pepstatin per ml. The nuclei were spun down at 3,000 rpm for 10 min in a benchtop centrifuge. Approximately 20 µg of protein was loaded per lane in 10% sodium dodecyl sulfate-polyacrylamide gels. Electrophoresis and blotting under nonreducing conditions without preheating were carried out as previously described (48). The MAb Q4120 was used at a concentration of 7 µg/ml for detecting CD4 and hybrids, and anti-mouse IgG-peroxidase conjugate (Pierce and Warriner) was diluted 1/2,000. The blots were developed with enhanced chemiluminescence (Super Signal; Pierce and Warriner).

Quantitative infectivity assay. Mv-1-lu cells stably expressing CD4, CXCR4, or hybrids were seeded in 96-well plates at 8 × 10³ cells/well 24 h before challenge. The cells were incubated with 50 µl of a 1/5 dilution of HIV-1 for 3 h at 37°C. The inoculum was then aspirated and replaced with medium. After two more days of culture, the cells were fixed in cold methanol/acetone (1/1) and the viral Gag protein was detected (9) with the MAbs 38:96K and EF7 (27) followed by -galactosidase-conjugated sheep anti-mouse antibody (Genosys, The Woodlands, Tex.). Then the substrate chlorophenol red -D-galactopyranoside (Boehringer) at a concentration of 2 mM was incubated with shaking at 37°C for 3 h, and the absorbance was measured in a spectrophotometer at 562 nm. With this modification of the readout, the experiments to measure SDF-1 inhibition of infection were carried out as previously described (59). Similar experiments were performed to measure the inhibition of infection by the bicyclam AMD3100 (58).

SDF-1 binding to hybrids and wild-type CXCR4 transiently expressed in COS-7 cells. SDF-1 was labeled with ¹²⁵I by the oxidative iodination procedure and purified by high-performance liquid chromatography (52). Binding of ¹²⁵I-labeled SDF-1 to COS-7 cells transiently transfected by the calcium phosphate method (23) with CD4-CXCR4 hybrids and wild-type CXCR4 was measured as previously described (52, 60). Briefly, the transfected COS-7 cells were transferred to 24-well culture plates 1 day after transfection, i.e., 1 day before the binding experiments. The numbers of cells per well were adjusted to give 5 to 10% binding of the added radiolabeled ligand. Binding was performed for 3 h at 4°C in binding buffer (50 mM HEPES, 5 mM MgCl₂, 1 mM CaCl₂, and 0.5% bovine serum albumin) with increasing concentrations of cold ligand. The reactions were terminated by washing the wells four times in binding buffer supplemented with 0.5 M NaCl. All determinations were performed in duplicate, and the nonspecific binding was determined in the presence of 1 µM SDF-1. The data were analyzed with Inplot (GraphPad Software, San Diego, Calif.).

Internalization of CD4-CXCR4 hybrids. Endocytosis assays on adherent cells were performed essentially as described previously (59). Briefly, cells were seeded in 16-mm-diameter wells in 24-well plates and grown for 2 days to a final density of 1 × 10⁵ to 2 × 10⁵ cells per well. The cells were cooled on ice, washed with binding medium (BM; see reference 59), and incubated for 2 h at 4°C with 250 µl of 0.3 nM ¹²⁵I-labeled Q4120 antibody in BM. Subsequently, the cells were washed in BM to remove free antibody and then warmed by addition of 250 µl of BM (37°C) with or without SDF-1 (125 nM). At the indicated times, the cells were returned to 4°C and washed with cold BM. For each time point six wells were used. From half of the wells, the cells were collected directly in 400 µl of NaOH (0.2 M) and transferred to tubes for -counting (as a measure of total cell-associated activity). To determine the intracellular activity, the remaining wells were rinsed twice with 0.5 ml of BM at 4°C and adjusted to pH 2.0 and then incubated twice for 3 min each time with 1 ml of the same medium to remove cell surface-bound antibody. The cells were harvested in NaOH as described above. The proportion of internalized antibody at each time point was determined by dividing the acid-resistant activity by the total cell-associated activity.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Expression of hybrids of CD4 and chemokine receptors in Mv-1-lu cells. Mv-1-lu cells were used for these studies because they are resistant to infection by HIV-1 when CD4 (12) or CXCR4 is expressed alone but show strong susceptibility when these two molecules are expressed together (59). Furthermore, the rate and extent of SDF-1- and phorbol ester-induced endocytosis of CXCR4 in Mv-1-lu cells are similar to those seen in T cell lines (59).

View larger version (29K): [in this window] [in a new window]   FIG. 2.   CD4-CXCR4 hybrid expression in Mv-1-lu cells. (A) Flow cytometric detection of stably expressed CD4 and CD4-CXCR4 hybrids on intact Mv-1-lu cells. The fluorescence intensity is depicted on the x axis, and the cell counts are on the y axis. The binding of the anti-CD4-D1 MAb Q4120 at a saturating concentration of 10 nM is illustrated. The labels 1 through 5 above the peaks indicate the following: 1, parental Mv-1-lu cells (mfi = 12.6); 2, wild-type CD4 (mfi = 882); 3, Mv-1-lu stably expressing CD4(2D)-Sp-CXCR4 (mfi = 159); 4, CD4(2D)CXCR4 (clone B; mfi = 580); and 5, CD4(4D)CXCR4 (mfi = 1340). (B) Western blot of lysates of Mv-1-lu cells expressing wild-type and hybrid receptors. Two CD4(2D)CXCR4-positive clones (clones B and H) were included. The CD4 moiety was detected with the MAb Q4120. Clone H had a similar level of cell-surface expression in flow cytometry to that of the CD4(2D)-Sp-CXCR4-expressing cells. Molecular standards (the masses are given in kilodaltons) migrated as indicated in the right-hand margin. The mass of the CXCR4 moiety predicted from the amino acid sequence is approximately 40 kDa; for two and four domains of CD4 the masses are 20 and 40 kDa, respectively.

View larger version (92K): [in this window] [in a new window]   FIG. 3.   Ultrastructural localization of CD4-CXCR4 hybrids. The MAb Q4120, bound to its epitope on the first Ig-like domain (D1) of CD4 in sections of Mv-1-lu cells stably expressing CD4(2D)CXCR4 (A) or CD4(4D)CXCR4 (B), was detected with anti-mouse antibody conjugated to 10-nm-diameter gold particles. (C) Lack of nonspecific binding of gold-conjugated antibody to Mv-1-lu cells expressing CD4(4D)CXCR4. The section was incubated without primary antibody and only with the gold-conjugated second antibody. The scale bar represents 100 nm.

View larger version (16K): [in this window] [in a new window]   FIG. 4.   MAb binding to CD4 and CXCR4 epitopes on receptor hybrids. The relative MAb binding (see Materials and Methods) determined from flow cytometric analyses is plotted on the y axis as a function of MAb concentration, which is plotted on the x axis. The value of x corresponding to y = 0.5 is an approximation of the Kd for MAb binding. (A) Titration curves for the binding of the CD4 MAb Q4120 to Mv-1-lu cells expressing wild-type CD4 (), wild-type CD4 together with wild-type CXCR4 (), or the hybrid CD4(2D)CXCR4 (), CD4(2D)-Sp-CXCR4 (), or CD4(4D)CXCR4 (). (B) Titration curves for the binding of anti-CXCR4 MAb 12G5 to Mv-1-lu cells expressing wild-type CXCR4 () or other receptors (symbols are as described for panel A).

CD4-CXCR4 hybrids mediate HIV-1_LAI infection. In order to determine whether the CD4-chemokine receptor hybrids can function as virus receptors, we challenged the hybrid-expressing Mv-1-lu cells with different strains of HIV-1. Infection was determined by immunodetection of newly synthesized viral Gag protein and quantitated spectrophotometrically (see Materials and Methods). In each experiment the mean optical density (OD) of at least three wells was determined for each cellular clone and virus combination; in some experiments Mv-1-lu clones expressing different levels of the hybrid receptors were included, as indicated.

View larger version (31K): [in this window] [in a new window]   FIG. 5.   HIV infection of Mv-1-lu cells expressing CD4-chemokine receptor hybrids. (A) On the horizontal line the hybrids or receptors stably expressed by clones of Mv-1-lu cells are indicated. Cells were incubated with HIV-1LAI (), HIV-1SF162 (), or HIV-189.6 () for 3 h at 37°C. Bars represent averages ± standard errors of means of OD at 562 nm (OD562) in the immunostaining assay from two experiments with at least three replicates in each. (B) The bars represent susceptibility to infection by HIV-1LAI. The hybrids and receptors, CD4 and CXCR4 (one clone), CD4(2D)CXCR4 (four clones), CD4(2D)-Sp-CXCR4 (two clones), CD4(4D)CXCR4 (eight clones), and CXCR4 (one clone), expressed in Mv-1-lu cells, are indicated under each bar. The bars show the means ± standard errors of means from six to eight experiments with at least three replicates for each clone in each. (C) As in panel B, the susceptibility of Mv-1-lu cells to infection by HIV-1LAI is represented by blocks above the labels on the horizontal line. Only clones with similar expression levels have been selected (mfi = 100 to 200); three CD4(2D)CXCR4, two CD4(2D)-Sp-CXCR4, and four CD4(4D) CXCR4 clones were included. The relative OD shown on the y axis is the OD for the respective CD4-CXCR4 hybrid divided by the OD for CD4+ CXCR4+ after subtraction from both of the background OD for CXCR4. The bars represent the means ± standard errors of means from four to five experiments.

SU competition with anti-CD4 and -CXCR4 MAbs for binding to CD4-CXCR4 hybrids. As the CD4-CXCR4 hybrids exhibited different abilities to mediate infection, we examined their interactions with recombinant SU. Initially, we detected bound SU with an antiserum to its C-terminal region. The concentration of SU_LAI that yielded half-maximal binding to cells expressing CD4, both CD4 and CXCR4, and CD4 (2D)CXCR4 or CD4(2D)-Sp-CXCR4 was in the range from 1 to 3 nM (data not shown), demonstrating that the recombinant SU had the requisite conformation for high-affinity binding at least to CD4. However, since the C-terminal part of SU may influence the interaction with chemokine receptors (41), we used a competition assay in which the ability of X4 or R5 SU to impede the subsequent binding of MAbs to either the CD4 or CXCR4 epitopes was detected. In Fig. 6A, SU_LAI (X4) and SU_JR-FL (R5) are shown to compete potently for binding to the Q4120 epitope on wild-type CD4 expressed together with CXCR4 as well as on the CD4-CXCR4 hybrid constructs. Similar curves were obtained for CD4 expressed alone or together with CCR5 (data not shown). Whereas SU_LAI and SU_JR-FL proteins showed comparable binding to wild-type CD4, SU_LAI was marginally more potent than SU_JR-FL in competing with Q4120 for binding to the hybrid receptors. The half-maximal inhibitory concentration, IC₅₀, of SU_LAI competition with Q4120 for binding to wild-type CD4 was 8 nM; for the 2D hybrid constructs CD4(2D)CXCR4 and CD4(2D)-Sp-CXCR4 the IC₅₀s were 20 and 15 nM, respectively; for binding to CD4(4D)CXCR4 it was 4 nM. Thus the affinity of monomeric recombinant SU_LAI for binding to the CD4 moiety was slightly increased for the 4D hybrid and somewhat reduced for the 2D constructs compared with wild-type CD4. Since HIV-1 infection was mediated more efficiently by the 2D than the 4D hybrid constructs (Fig. 5C), the SU affinity for CD4 as determined here does not appear to be a limiting factor in that process.

View larger version (25K): [in this window] [in a new window]   FIG. 6.   SU competition with the MAbs Q4120 and 12G5 for binding to CD4, CXCR4, and CD4-CXCR4 hybrids. After preincubation of cells with the various concentrations of SU, the binding of the MAb Q4120 to CD4 (A) or 12G5 to CXCR4 (B) was measured by FACS. The mfi in the flow cytometric analyses is indicated as a proportion of binding relative to that at the plateau of binding after subtraction of a background for the parental Mv-1-lu cells. The relative MAb binding expressed as a percent is plotted on the y axis, as a function of the SU concentration (shown on the x axis). , X4 SULAI; , R5 SUJR-FL. The receptors and hybrids are indicated above the diagrams.

Inhibition of infection of hybrid-expressing cells by SDF-1 and AMD3100. Both the chemokine SDF-1 and the nonpeptide compound AMD3100 are known to inhibit HIV infection mediated by CXCR4. However, low concentrations of SDF-1, which blocked HIV-1_LAI infection of CD4+ CXCR4+ Mv-1-lu cells, slightly enhanced the infection of cells expressing the 2D hybrids (Fig. 7A). At higher concentrations, SDF-1 blocked the infection mediated by the 2D hybrids but less efficiently than for wild-type receptors. In contrast, the effects of the nonpeptide AMD3100 were very similar on the CD4+ CXCR4+ cells and those expressing the 2D hybrids, whereas it was markedly more potent in blocking infection mediated by the 4D construct (Fig. 7B). In conclusion, the clear differences in the ability of SDF-1 and AMD3100 in blocking HIV-1_LAI infection mediated by wild-type and hybrid receptors indicate that these two compounds block by distinct mechanisms. However, these results raise the questions of whether SDF-1 binds with the same affinity to hybrids as to wild-type CXCR4 and whether the hybrids are internalized as efficiently in response to SDF-1 binding as wild-type CXCR4.

View larger version (24K): [in this window] [in a new window]   FIG. 7.   Inhibition of HIV-1 infection of Mv-1-lu cells expressing CD4 and CXCR4 or CD4-CXCR4 hybrids by SDF-1 and AMD3100. SDF-1 and AMD3100 inhibition of HIV-1LAI infection was assayed as described in Materials and Methods. The OD562 determined in the immunostaining assay is expressed as a proportion of that seen for cells infected in the absence of inhibitor after subtraction of the background obtained with nonpermissive cells (y axis). The bars are plotted above the concentrations of SDF-1 (A) and AMD3100 (B) and represent data for Mv-1-lu cells expressing CD4 and CXCR4, CD4(2D)CXCR4, CD4(2D)-Sp-CXCR4, and CD4(4D)CXCR4. The means ± standard errors of means from two to four experiments are shown.

SDF-1 binding to CXCR4 and hybrid receptors. Using ¹²⁵I-labeled SDF-1 as a tracer, we found that wild-type CXCR4 as well as the CD4-CXCR4 hybrid receptors bound SDF-1 with dissociation constants in the nanomolar range (Fig. 8A), similar to (14, 60) or somewhat lower than (25) what has previously been reported for CXCR4. The CD4(4D) CXCR4 hybrid and wild-type CXCR4 bound SDF-1 very similarly in a monocomponent fashion. In contrast, both the CD4(2D)CXCR4 and the CD4(2D)-Sp-CXCR4 hybrid constructs gave competition curves with Hill coefficients of magnitude less than unity, indicating more than one binding mode for the ligand (Table 1). Two-component analysis showed that SDF-1 bound to the 2D hybrids with affinities corresponding to that observed for the wild-type receptor, i.e., with IC₅₀s of 1.1 and 2.6 nM (compared with 1.3 nM for the wild-type receptor), as well as with higher affinities with IC₅₀s of 0.10 and 0.11 nM (Table 1).

View this table: [in this window] [in a new window]   TABLE 1.   Binding constants for SDF-1 and the anti-CXCR4 MAb 12G5 determined by radioligand competition on COS-7 cells expressing CXCR4 and CD4-CXCR4 hybrid receptorsa

View larger version (25K): [in this window] [in a new window]   FIG. 8.   SDF-1 binding to wild-type CXCR4 and CD4-CXCR4 hybrid receptors. 125I-SDF-1 binding displaced by SDF-1 (A) or the MAb 12G5 (B) from CXCR4 wild-type receptor (), CD4(2D)CXCR4 (), CD4(2D)-Sp-CXCR4 (), and CD4(4D)CXCR4 () expressed transiently in COS-7 cells. For CD4(2D)CXCR4 and CD4(2D)-Sp-CXCR4 the curves are calculated from two-site competition binding analysis.

SDF-1-induced down-modulation of CD4-CXCR4 hybrids. Although the affinity of SDF-1 for the hybrids was at least as high as that for the wild type, it remained a possibility that they differed from the wild type in their capacity to undergo SDF-1-induced down-modulation. Fig. 9A shows the efficient and rapid down-modulation from the cell surface of the CD4(2D)CXCR4, CD4(2D)-Sp-CXCR4, and CD4(4D)CXCR4 hybrids as a result of incubation with SDF-1 in a range of concentrations similar to those used for SDF-1 inhibition of infection. In this experiment the MAb 12G5 was used in order to allow comparison with wild-type CXCR4. Since this antibody and SDF-1 partially compete for binding to CXCR4, SDF-1 was first allowed to bind and receptor internalization proceeded at 37°C. Then the amount of CXCR4 remaining at the cell surface was determined after acid stripping of the cell surface-bound SDF-1 (59). This method showed a similar down-modulation of hybrid and wild-type CXCR4 both as a function of SDF-1 concentration and kinetically (Fig. 9A and B). The presence of the Q4120 epitope on the CD4 moiety of the hybrids offered an alternative means of measuring the internalization of CXCR4 in response to SDF-1: Q4120 can be prebound to hybrid-expressing cells which are subsequently incubated with SDF-1, because the binding of ¹²⁵I-labeled Q4120 does not interfere with the SDF-1 binding. Thus endocytosis rates can be measured directly as percentages of hybrid molecules at the cell surface that become internalized over time (47). This approach also showed rapid induction of endocytosis of the three CD4-CXCR4 hybrids by SDF-1 (Fig. 9D), which reached steady state when 80 to 90% of the molecules were internal. The rates of endocytosis from two experiments with 250 nM SDF-1 were 7.0 to 7.2%/min for CD4(2D)CXCR4, 7.2 to 8.2%/min for CD4(2D)-Sp-CXCR4, and 3.8 to 4.7%/min for CD4(4D)CXCR4. All results shown in Fig. 8 were obtained with the same clones as used for the experiment presented in Fig. 2A. Hence, the somewhat slower internalization of CD4(4D)CXCR4 than of the other two hybrids could be related to its higher expression level and may not translate into smaller numbers of molecules internalized per minute.

View larger version (31K): [in this window] [in a new window]   FIG. 9.   SDF-1-induced internalization of CD4-CXCR4 hybrids. Mv-1-lu cells expressing wild-type CXCR4 () or the hybrids CD4(2D)CXCR4 (), CD4(2D)-Sp-CXCR4 (), and CD4(4D)CXCR4 () were treated with increasing concentrations of SDF-1 for 30 min (A) or with 125 nM SDF-1 for up to 60 min (B through D). The means of triplicate values from one experiment are shown. The error bars, each representing 1 standard deviation, are sometimes hidden within the symbols. (A and B) Comparisons of the SDF-1-induced down-modulation of CXCR4 and hybrids from the cell surface. Panel A shows the dependence of down-modulation on SDF-1 concentration, and panel B shows down-modulation as a function of time. After incubation in BM with SDF-1 at 37°C, the cells were cooled down on ice, acid washed, and then labeled with 125I-12G5 at 4°C. The graphs show the cell-associated 12G5 binding for SDF-1-treated cells as a proportion of 12G5 binding on untreated cells for the indicated concentrations and time points. (C) SDF-1-induced down-modulation of CD4-CXCR4 hybrids detected with the anti-CD4 MAb Q4120. The cells were labeled with 125I-Q4120 at 4°C. The results are expressed as described for panels A and B. (D) Direct measurement of the endocytosis of CD4-CXCR4 hybrids detected with Q4120. Cells were incubated first on ice with 125I-Q4120 for 2 h and then in BM with SDF-1 at 37°C for the indicated period. Each time point indicates the acid-resistant (internalized) radioactivity as a proportion of the total cell-associated activity (y axis).

AMD3100 binding to hybrid receptors and effect on their cell surface expression. We assessed the AMD3100 binding to Mv-1-lu cell-expressed CXCR4 and hybrid receptors in a ¹²⁵I-labeled 12G5 competition assay. AMD3100 was titrated from 10² to 10² nM and allowed to bind to Mv-1-lu cells expressing wild-type CXCR4, CD4(2D)CXCR4, or CD4(4D)CXCR4 for 1 h at 4°C before the addition of ¹²⁵I-labeled 12G5. The AMD3100 concentration yielding half-maximal 12G5 competition was approximately 5 nM for wild-type CXCR4 and 2 nM for CD4(2D)CXCR4 and CD4(4D)CXCR4 (Fig. 10A). Hence, the similar apparent AMD3100 affinities for the latter two receptor hybrids and the only marginally lower affinity for wild-type CXCR4 do not explain the strong sensitivity to inhibition of infection of the CD4(4D)CXCR4-expressing cells.

View larger version (17K): [in this window] [in a new window]   FIG. 10.   AMD3100 binding to CXCR4 and CD4-CXCR4 hybrids. (A) Binding of AMD3100 to Mv-1-lu cells expressing CXCR4 (), CD4(2D)CXCR4 (), and CD4(4D)CXCR4 () detected by competition with 125I-12G5 binding. The 12G5 binding, expressed as a percentage of binding in the absence of AMD3100 (y axis), is shown as a function of the AMD3100 concentration (x axis). (B) The effect of AMD3100 binding on the cell surface expression of CD4-CXCR4 hybrids. Mv-1-lu cells expressing CD4(2D)CXCR4 ( and ) or CD4(4D) CXCR4 ( and ) were incubated with 100 nM AMD3100 in BM for up to 1 h. The cells were then cooled on ice, washed, and labeled on ice with 125I-12G5 ( and ) or 125I-Q4120 ( and ). The y axis represents the amount of cell-associated iodinated antibody for AMD3100-treated cells expressed as a proportion of antibody bound to untreated cells at the indicated time points. The data in panels A and B are from one of two similar experiments and represent the means ± standard deviations of three replicates.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Evidence is accumulating that direct physical interactions between the viral outer Env protein, SU, CD4, and specific chemokine receptors are required in the fusion and entry of most HIVs (4, 25, 41, 62, 65). CD4 binding is known to induce conformational changes in the Env oligomeric complex (55). Furthermore, an epitope on SU that is induced by CD4 binding is implicated in interactions with the chemokine receptor (32, 62, 65). This suggests a chronology of events in which CD4 binding precedes interactions with the chemokine receptor. The recognition of the chemokine receptor by SU may serve as a trigger for further conformational changes in Env that are conducive to fusion. Alternatively, the chemokine receptor may be an additional point of contact for SU that brings it closer to the membrane. The distinct properties that we identified for three different CD4-CXCR4 hybrids shed light on several functions of the receptor-coreceptor complexes in HIV-1 entry.

We found that hybrids between CD4 and CXCR4 could be expressed efficiently on the cell surface and that both the CD4 and chemokine receptor components of these hybrid molecules adopted conformations similar to those of their parent proteins. These hybrids were able to mediate infection by X4 HIV-1; moreover, their capacities to do so correlated with the efficiency of SU interactions with their CXCR4 moieties. The fusion activity of these hybrids may occur through an ability to act in trans, i.e., by an intermolecular mechanism in which one hybrid molecule provides the CD4 function and another provides the CXCR4 function. Complementation of CXCR4 with CD4(2D)CCR5 gave some susceptibility to X4 virus infection, indicating that a CD4 moiety on a noncognate hybrid can function together with a cognate nonhybrid chemokine receptor (data not shown). Although it is thus plausible that the hybrids can function in trans, several lines of evidence argue against this as a dominant mechanism. First, in an intermolecular mechanism, the CD4(4D)CXCR4 and CD4(2D)-Sp-CXCR4 hybrids might be expected to function as efficiently as the CD4(2D)CXCR4 hybrid, whereas our results demonstrated differential efficacy. Second, the level of hybrid expression at the cell surface correlated with infection mediated by CD4(4D)CXCR4 but not with CD4(2D)CXCR4-mediated infection. This suggests that a dominant intermolecular mechanism may only occur for the 4D hybrid. Third, the competition by X4 SU_LAI with the MAb 12G5 for binding to the hybrids suggests that one SU molecule can interact with both the CD4 and the CXCR4 moieties on one and the same 2D hybrid molecule. Fourth, the less potent antiviral effect of SDF-1 on the 2D hybrids than on wild-type CXCR4 may be suggestive of a mechanism in which both components of a single hybrid molecule are used (see below). Fifth, the R5 and X4R5 viruses did not infect CD4(2D)CCR5-expressing cells, indicating that an intermolecular mechanism was not active in those cases.

However, the CXCR4 hybrids that functioned in HIV-1 entry did not enhance susceptibility to infection. Why these molecules were not more efficient than the wild-type receptors is unclear. One possibility is that the recruitment of CXCR4 into fusion complexes is not a limiting step. Alternatively, even the most permissive hybrids may be geometrically disadvantaged for interactions with the Env protein. The relative susceptibilities to infection suggest that such disadvantages would be less pronounced for the 2D hybrids, in particular, CD4(2D)CXCR4, than for CD4(4D)CXCR4. However, the two- to fourfold less potent SU competition with the anti-CD4 MAb on the 2D hybrids than on the wild-type CD4 and CD4(4D)CXCR4 (Fig. 6A) may indicate that conditions favoring simultaneous CD4 and CXCR4 interactions on the 2D hybrids are suboptimal.

The CD4 hybrids with chemokine receptors other than CXCR4 did not provide receptor-coreceptor function for the X4, X4R5, or R5 strains tested (Fig. 5A): expressing CD4 in juxtaposition to a chemokine receptor did not overcome the specific chemokine receptor requirements for infection. We did observe some R5 SU_JR-FL competition with 12G5 for binding to CD4(2D)CXCR4, which although it was weaker than that of X4 SU_LAI, suggests physical contact of the R5 SU with CXCR4. However, this R5-SU association with CXCR4 was not sufficient to allow infection by the R5 strain HIV-1_SF162. The chemokine receptor binding site on SU is likely to contain conserved elements from the interdomain bridging sheet (32, 51), which may be the basis for low-affinity, nonspecific chemokine receptor interactions. Other more variable adjacent segments, e.g., the V3 region (41, 62, 65), may determine receptor preference; however, the exposure of these sites on monomeric, recombinant SU may differ from that on virion-associated Env oligomers.

The segments of CXCR4 and CCR5 that are necessary for their respective coreceptor functions are being mapped (3, 8, 16, 18, 19, 22, 36, 53). Whereas the second extracellular loop is most strongly implicated in the coreceptor function of CXCR4 in HIV-1_LAI infection, other strains have additional requirements, such as for the N-terminal segment. This may explain why the CXCR4 hybrids functioned more efficiently for HIV-1_LAI than for HIV-1_89.6 in relation to the infection mediated by the wild-type molecules. Similarly, the lack of infection via the CCR5 hybrid even for the R5 virus HIV-1_SF162 and the X4R5 virus HIV-1_89.6 may be attributed to differences in the importance of the N-terminal segment of the coreceptor. Since the capacity of HIV-1_89.6 to use CCR2b as a coreceptor is weak (17), we cannot answer the question of whether CD4(2D) CCR2b may be used with similar efficiency. It has recently been reported that a hybrid of all four domains of CD4 linked to CCR5 can function in infection of virus pseudotyped with Env of HIV-1_JR-FL, HIV-2, and SIV (26), albeit at a 10-fold lower efficiency than wild-type molecules. Although this cannot be directly compared with the lack of detectable function of the CD4(2D)CCR5 hybrid for HIV-1_SF-162, the possibility remains that a CD4(2D)CCR5 hybrid might give relatively efficient infection with specific R5 viruses.

We and others have previously postulated two mechanisms for chemokine inhibition of HIV-1 infection: direct competition with SU for binding and down-modulation of the chemokine receptor from the cell surface (1, 2, 37, 59, 64). Here we found that SDF-1 inhibition of HIV-1_LAI infection via the CD4-CXCR4 hybrids was weak compared with that via wild-type CD4 and CXCR4. The average affinity of SDF-1 for the 2D hybrids, which showed the least sensitivity to SDF-1, was two- to threefold higher than those for wild-type CXCR4 and CD4(4D)CXCR4. This affinity difference was in the opposite direction to a change that would most readily explain the relative insensitivity of the 2D hybrids to SDF-1, although it cannot be ruled out that the higher-affinity SDF-1 binding to these hybrids may have qualitatively different effects; for example, it might explain the weak enhancement of infection via CD4(2D)CXCR4 by SDF-1 at a low concentration (Fig. 7A).

In contrast to the differences in inhibition of infection, the 2D hybrids were similar to wild-type CXCR4 and CD4(4D) CXCR4 in their susceptibilities to SDF-1-induced endocytosis (Fig. 9). As the hybrids contain both the CD4 and coreceptor moieties within a single molecule, the possibility of down-modulating both the receptor and coreceptor from the surface of hybrid-expressing cells might have been expected to augment the inhibition of infection. On the contrary, we found weaker SDF-1 inhibition of infection via the hybrids than wild-type receptors; infection via CD4(2D)CXCR4, which was enhanced at the lowest SDF-1 concentration, was the most insensitive. However, we also found that infection did not correlate with the level of CD4(2D)CXCR4 cell surface expression, whereas it did for CD4(4D)CXCR4. Thus differences in the amounts of receptor at the cell surface that are required for infection may partly explain the insensitivity to SDF-1. The direct block of SU binding to CXCR4 may also differ between hybrids and wild-type CXCR4. SDF-1 may compete less efficiently the greater the capacity of SU to establish two-point binding. In fact, the 12G5 competition experiments suggest that SU-CXCR4 binding ranks similarly to the relative SDF-1 insensitivity for wild-type CXCR4 and the three CD4-CXCR4 hybrids. An additional possibility is that, for wild-type molecules, SDF-1 may block the formation of CD4-CXCR4 complexes. Such a mechanism may be hampered when the CD4 moiety is covalently linked to CXCR4.

Unlike the larger molecule SDF-1, AMD3100 might not interfere with CD4-CXCR4 interactions. Indeed, AMD3100 gave similar inhibition of infection via wild-type receptor, CD4(2D)CXCR4, and CD4(2D)-Sp-CXCR4. However, infection via CD4(4D)CXCR4 showed significantly greater sensitivity to AMD3100, which could not be attributed to any detectable differences in the binding of the bicyclam to this hybrid. One possible explanation is that AMD3100 blocks CXCR4 function irreversibly. It has been suggested that AMD3100 interacts with residues in transmembrane segment 4 of CXCR4 that might contribute to irreversible binding (33); furthermore, we found that AMD3100 binding to CXCR4 at the surface of Mv-1-lu cells was resistant to acid washes, suggesting irreversibility. CD4(4D)CXCR4 might be more sensitive to an irreversible block, because efficient infection via this hybrid required higher levels of cell-surface expression than for the other hybrids. The stronger AMD3100 inhibition of infection via CD4(4D)CXCR4 than via wild-type molecules or the 2D hybrids is also in keeping with these hybrids' SU interactions. For, unlike the SDF-1 effect on infection, the antiviral potency of an irreversible blocker added to cells before exposure to virus is not likely to depend on the strength of the SU-receptor interactions. In conclusion, the distinct sensitivities of the wild-type and hybrid receptors to SDF-1 and AMD3100 inhibition of infection indicate that the modes of action of these two CXCR4 ligands differ in more than just their capacities to down-modulate CXCR4.

From a practical point of view, the most efficiently functioning CD4-CXCR4 hybrids may prove useful in the construction of receptor-coreceptor-bearing virus-like particles in order to target HIV-1 Env-expressing cells, as has been achieved for CD4 and chemokine receptors (21, 39, 56). By varying the CD4 linkage it may be possible to overcome the nonfunctionality of the hybrids for R5 virus (26). However, the hybrids also offer experimental advantages: the covalent linkage of receptor to coreceptor allows measurement of SU-chemokine receptor interactions that are otherwise barely detectable. Furthermore, the coexpression of receptor and coreceptor as one polypeptide provides a means of varying the expression level of the two moieties at a constant stoichiometric ratio, which may prove valuable in the study of the dependence of infectivity on cell surface concentrations of the two receptor components (29, 50). Another possible use of the hybrids is in facilitating the cocrystallization of Env, CD4, and chemokine receptor for structural studies.

We conclude that a simulation of complexes between receptors and coreceptors is feasible. The study of these molecules covalently linked in different manners may help elucidate their interactions with viral envelope proteins.