Cutting Edge: CXCR4 Is a Functional Coreceptor for Infection of Human Macrophages by CXCR4-Dependent Primary HIV-1 Isolates¹

Human immunodeficiency virus-1 strains have been traditionally divided into two categories based on their cellular tropism, replication kinetics, and ability to induce syncytia formation (1, 2, 3). Isolates that are commonly referred to as macrophage (Mφ)³-tropic (M-tropic), slow/low, or non-syncytia-inducing (NSI) infect both monocyte-derived Mφ (MDMs) and primary CD4⁺ T lymphocytes but do not usually infect established T cell lines such as MT-2. By contrast, T cell line-tropic (T-tropic), rapid/high, or syncytia-inducing (SI) strains grow in T cell lines and form syncytia in MT-2 cells and in PBMCs. The ability of SI isolates to infect Mφ productively is controversial. Indeed, while T cell line-adapted (TCLA) strains usually fail to replicate in MDMs (4), conflicting results have been reported when primary isolates were used (5, 6, 7, 8, 9, 10).

The recent identification of several chemokine receptors as HIV coreceptors has provided a molecular basis for the difference in tropism of different HIV-1 strains. In particular, C-C chemokine receptor (CCR)5, which is the RANTES, MIP-1α, and MIP-1β receptor, has been shown to serve as the main coreceptor for NSI viruses (11, 12, 13, 14); CXC chemokine receptor-4 (CXCR4)/fusin, which is the natural receptor for stem cell differentiation factor (SDF)-1 (15, 16), mediates the entry of both primary and TCLA SI HIV-1 strains (17). CCR2b and CCR3 can also serve as entry cofactors for certain virus strains (14, 18). More recently, Bob and Bonzo (19), which are two orphan seven-transmembrane domain G protein-coupled receptors that are expressed in T cells but weakly, if at all, in Mφ, were reported as potential new coreceptors for fusion by M-tropic and T-tropic HIV-1 strains as well as by SIV.

While the lack of CCR5 expression on most T cell lines (20) has provided a rationale for the inability of NSI strains to infect these cells, the issue of MDM infection by HIV-1 strains with an SI phenotype remains unresolved. Although Mφ express significant levels of CXCR4 on their membranes (21), this coreceptor is assumed to be nonfunctional for infection (22). Because of the critical role of Mφ in the pathogenesis of HIV-1 infection, we addressed this apparent paradox by infecting normal human MDMs in vitro with a panel of primary HIV-1 isolates and TCLA strains that had been rigorously characterized for coreceptor usage. Furthermore, we added the natural CXCR4 ligand, SDF-1, to specifically block CXCR4-mediated viral entry. Our results show that primary HIV-1 isolates can productively infect human MDMs using CXCR4 as a coreceptor. By contrast, productive infection was not observed with CXCR4-dependent TCLA HIV-1 strains, even though viral entry occurred with variable efficiency.

mAbs that were specific for human CXCR4 (12G5) (21) and CCR5 (2D7) were kindly provided by J. Hoxie (University of Pennsylvania, Philadelphia, PA) and C. Mackay (Leukosite, Cambridge, MA), respectively. Anti-CD14 mAb P9, FITC-conjugated goat anti-mouse IgG, and isotype controls were purchased from Becton Dickinson (Mountain View, CA). SDF-1 and RANTES were obtained from Upstate Biotechnology (Lake Placid, NY) and R&D Systems (Minneapolis, MN), respectively. The endotoxin content of the cell culture reagents was assessed by the Limulus amebocyte lysate assay (BioWhittaker, Walkersville, MD) and was <0.125 Eu/ml.

Cell migration was assayed in 48-well transwell chambers using a 5-mm pore size polycarbonate membrane (Costar, Cambridge, MA). Chemokines diluted in RPMI 1640/0.3% human serum albumin were added to the lower chamber. Cells (3 × 10⁶/ml, 100 μl) were added to the upper chamber. The filter was removed after a 2-h incubation at 37°C, and the cells that migrated in the lower chamber were counted using a FACScan (Becton Dickinson) at 60 μl/min for 30 s. Specific cell types were selected by gating on the appropriate forward and side scatter. Cell migration was assessed in duplicate and expressed as a chemotactic index (the ratio between the number of cells that migrated in the presence of chemokines and spontaneous migration).

MDMs were isolated as described previously (23). MDM preparations contained ≥90% CD14⁺ cells as assessed by immunofluorescence. To obtain monocytes, nonadherent cells were removed after 1 h, and the remaining adherent cells were cultured for 24 h.

MDMs were infected with three primary CCR5-dependent HIV-1 isolates (HIV-1₅₅₀₈, HIV-1₆₀₈₈, and HIV-1₁₀₀₀₅), one primary isolate (HIV-1₅₇) that uses both CCR5 and CXCR4, three primary CXCR4-dependent HIV-1 isolates (HIV-1₂₇, HIV-1₃₄, and HIV-1₁₃₀), and the TCLA strains HIV-1_IIIB and HIV-1_MN that were continuously grown in MOLT-3 and PM1 cells, respectively. All isolates were characterized for coreceptor usage by infecting U87.CD4 glioma cells that coexpressed CCR1, CCR2B, CCR3, CCR5, and CXCR4 (9) and osteosarcoma GHOST34.CD4 cells that had been transfected with the Bob or Bonzo genes (kindly provided by D.R. Littman, Skirball Institute, New York, NY). MDMs were infected with DNase-treated virus (tissue culture ID₅₀: 50/10⁶ cells). The p24 Ag concentrations in the culture supernatants were determined by ELISA (23).

DNA was extracted from MDMs at 14 h postinfection by salting-out. PCR was performed using primers 1 and 2II (24) that amplify a 218-base pair (bp) fragment from the HIV-1 gag gene. Samples were subjected to 50 cycles of amplification (95°C for 1 min, 63°C for 1 min, and 72°C for 1 min). The PCR products were separated on a 1.8% agarose gel, transferred to a nylon membrane, and hybridized with a gag-specific, ³²P-labeled oligonucleotide (5′-AGGCGACTGGTGAGTACGCCAAAA). To normalize for the quantity of DNA in each sample, a 441-bp region of the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene was amplified using primers 5′-GGGAAGGTGAAGGTCGGAGTC and 5′-GCTGATGATCTTGAGGCTGTTGTC. The results are expressed as the ratio between the intensities of the HIV-1 and GAPDH bands as assessed by scanning densitometry. Each sample was amplified in duplicate or triplicate.

As a first step in assessing the role of CXCR4 in MDM infection by SI HIV-1 isolates, we analyzed CXCR4 expression on monocytes and MDMs from normal donors. Figure 1 shows that immunofluorescence with mAb 12G5 detected variable but significant levels of CXCR4 protein on monocytes at 1 day postisolation as assessed by the percentage of positive cells and by mean fluorescence intensity (MFI). Although culture-induced differentiation resulted in a progressive decrease of CXCR4, the receptor was expressed at a comparable intensity on MDMs at the time of infection (day 5 of culture, MFI = 20–29) and on MOLT-3 cells, which are a T cell line that is widely used to expand SI HIV-1 strains (MFI = 24). The expression of the other major HIV-1 coreceptor, CCR5, followed a similar pattern in all of the donors examined (n = 3).

Notably, the levels of CXCR4 that were available on MDMs at the time of in vitro infection were sufficient to support a brisk chemotactic response to rSDF-1 (Fig. 2). RANTES-induced chemotaxis was in a comparable range. Taken together, these results show that, at the time of exposure to the virus, CXCR4 is expressed on MDMs at substantial levels and is functional.

To define the role of CXCR4 in MDM infection by HIV-1, we selected from a panel of 33 primary HIV-1 isolates that had been rigorously characterized for coreceptor usage (9), 3 isolates that exclusively use CXCR4, together with a control group of 3 CCR5-dependent isolates. As expected, MDMs were efficiently infected by primary NSI HIV-1 isolates. In a representative experiment, p24 Ag levels at 9 days after the infection of cultures with HIV-1₅₅₀₈, HIV-1₆₀₈₈, and HIV-1₁₀₀₀₅ were 5.6, 2.3, and 7.2 ng/ml, respectively. Notably, MDMs were also efficiently infected by all of the CXCR4-dependent primary SI strains, with p24 Ag release rapidly reaching substantial levels (Fig. 3 A). The source of HIV in our cultures was most likely MDMs rather than contaminating T cells. Indeed, no p24 Ag secretion was ever detected despite an intense surface expression of CXCR4 when nonadherent CD3⁺/CD14⁻ cells were infected with the same isolates at 5 days after purification (data not shown).

It has recently been shown that SDF-1 is the natural ligand for CXCR4 but not for the other chemokine receptors that mediate HIV-1 fusion and entry (15, 16). To formally prove that CXCR4 acts as a coreceptor for MDM infection by primary SI HIV-1 isolates, we infected MDMs in the presence or absence of rSDF-1 (2 μg/ml). Figure 3,B shows that the addition of rSDF-1 blocked the replication of all of the CXCR4-dependent primary HIV-1 strains. The SDF-1-dependent inhibition of MDM infection was even more efficient than that observed previously with human PBMCs (9). A semiquantitative PCR analysis (Fig. 3,C) revealed an SDF-1-induced decrease in proviral DNA at 14 h postinfection that pointed to a block at the level of viral entry. By contrast, SDF-1 had no inhibitory effect on infection by CCR5-dependent HIV-1 isolates (data not shown). Consistent with the selective use of CXCR4 as a coreceptor for entry, the addition of the CCR5 ligand RANTES at a concentration (100 ng/ml) that completely inhibits infection with NSI isolates (23) did not affect the entry or the replication of CXCR4-dependent HIV-1 isolates (Fig. 3, B and C). These data demonstrate that CXCR4 is a functional coreceptor for the entry of primary HIV-1 SI isolates in MDMs.

Further supporting this conclusion, Table I shows that MDMs from a ccr5Δ32-homozygous individual were infected by two HIV-1 primary isolates (HIV-1₃₄ and HIV-1₁₃₀) that selectively use CXCR4, as well as by a primary isolate (HIV-1₅₇) that uses both CXCR4 and CCR5. The addition of SDF-1 efficiently blocked HIV infection by all viruses. By contrast, RANTES had no significant effect (data not shown). Taken together, these results show that CXCR4 supports CCR5-independent HIV-1 entry in Mφ.

We then compared the ability of CXCR4-dependent TCLA strains and primary isolates to productively infect MDMs. Proviral DNA was assessed at 14 h postinfection with two TCLA strains (HIV-1_IIIB and HIV-1_MN) and three primary isolates (HIV-1₂₇, HIV-1₃₄, and HIV-1₁₃₀). Table II shows that the level of viral entry was variable but comparable overall for TCLA strains and primary isolates. However, productive infection could not be detected with TCLA strains, even when entry had occurred with substantial efficiency (e.g. HIV-1_IIIB for donor 1 and HIV-1_MN for donor 2). These results suggest that the low or absent viral replication in MDMs that had been infected with TCLA HIV-1 strains was due to both entry and postentry defects.

The current availability of assays that determine HIV-1 coreceptor usage and of ligands that selectively block HIV entry provides a rational way out of the existing maze of viral phenotypes and nomenclatures, prompting us to readdress the issue of Mφ infectability by primary HIV-1 strains with different biologic properties. Our data show that human MDMs can be efficiently infected by primary HIV-1 isolates that selectively use CXCR4 as a coreceptor. This notion is supported by a rigorous characterization of all of the relevant viral isolates as selective CXCR4 users, by the demonstration that CXCR4 is functional in an independent assay (i.e., chemotaxis), and most importantly, by the ability of SDF-1, the natural ligand of CXCR4, to prevent HIV-1 infection. We conclude that MDMs support the entry and replication not only of CCR5-dependent but also of CXCR4-dependent primary HIV-1 isolates. Our conclusion is consistent with the recent demonstration that CXCR4 supports the infection of Mφ by a dual-tropic primary isolate (10). Thus, in addition to their well-established role in the early stages of disease and in viral transmission (25), Mφ are both a source of HIV during the opportunistic infections that mark the progression of HIV-1 disease (26) and a target for the CXCR4-dependent HIV-1 strains that emerge in the late stages of HIV infection (9).

Our findings cast some doubt on the traditional definition of HIV-1 tropism based on the infection of cells that have been manipulated by culture conditions, and more generally, on the usefulness of thinking about HIV isolates as M- vs T-tropic. In particular, the results obtained by us and others with CXCR4-dependent TCLA HIV-1 strains are conflicting, and underline how the cellular tropism of HIV isolates is determined by multiple virus/host cell interactions. Blocks have been observed at the entry step and have been ascribed to limited coreceptor availability (10) and/or to the intrinsic fusogenic properties of env proteins (27). Postentry defects have also been shown, implicating the cellular factors required to activate viral replication (28, 29). In this respect, the transcription factors NF-ATc (30) and GATA-3 (31) activate HIV-1 transcription and replication in T cells, whereas the binding of CCAAT/enhancer-binding proteins to the HIV-1 long terminal repeat is required for HIV-1 replication in MDMs (32). By the same token, the HIV-1-encoded protein vpr is important for efficient viral replication in primary MDMs but not in activated T cells (33). It is tempting to speculate that HIV-1 strains that are continuously grown in T cell lines might become highly dependent upon T cell-specific transcription factors for their replication and/or develop mutations in the genomic regions that are critical for replication in Mφ. Such events would remain functionally silent as long as the virus is passaged in T cells but would be likely to undermine replication in Mφ.