Coreceptor Phenotype of Natural Human Immunodeficiency Virus with Nef Deleted Evolves In Vivo, Leading to Increased Virulence

Preparation of viral stocks.

NL4-3 was a gift from Malcolm Martin via the AIDS Research and Reference Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health. The molecular clone 49-5 was a gift from Bruce Chesebro. Infectious virus stocks were prepared by transfecting 293T cells with proviral DNA as described previously (2). The primary isolates D36/95, D36/99, 7/86, and 1/85 were expanded by infection of heterologous peripheral blood mononuclear cells (PBMC). Isolates 7/86 and 1/85 were gifts from Ruth Connor (14). The p24^gag concentrations of viral stocks were assessed by enzyme-linked immunosorbent assay (ELISA) (NEN Life Sciences, Boston, Mass.).

Culture and infection of human lymphoid tissues ex vivo.

Human noninflammatory tonsil tissue removed during tonsillectomy (provided by the National Disease Research Interchange, Philadelphia, Pa., and the Kaiser hospitals in San Francisco, South San Francisco, and San Rafael) was prepared for lymphoid aggregate culture as previously described (23, 37). In brief, tonsil tissue was mechanically dispersed, and isolated cells were transferred to 96-well U-bottomed plates at a concentration of 10⁷ cells per ml, 200 μl per well. Cells were allowed to aggregate at the bottom of the well and were not dispersed for the remainder of the culture period. Ex vivo human lymphoid cell cultures were inoculated within 24 h of preparation with HIV-1 at 80 50% tissue culture infective doses, as determined by terminal dilution of the virus stocks in quadruplicate on heterologous phytohemagglutin-activated PBMC as described previously (45).

Assessment of CD4⁺ T-cell depletion by FACS analysis.

At the indicated time points, cells from infected and uninfected lymphoid cell cultures were stained for cell surface markers CD3, CD4, CD8, and CCR5 as described previously (45, 55) with the following monoclonal antibodies: anti-CD3 (clone SK7, phycoerythrin [PE] conjugated), anti-CD4 (clone SK3, fluorescein isothiocyanate [FITC] conjugated), anti-CD8 (clone SK1, peridimin chlorophyll protein [PerCP] conjugated) (Becton Dickinson), and anti-CCR5 (clone 2D7, allophycocyanin conjugated) (PharMingen). Then, 10,000 lymphocytes positive for CD3 surface marker were counted by fluorescence-activated cell sorting (FACS), and the data were analyzed with Cellquest software (Becton Dickinson). To facilitate comparison among experiments, CD4⁺ T-cell depletion was assessed by measuring the ratio of CD4⁺ to CD8⁺ T cells. This value was normalized to the CD4/CD8 ratio of control (uninfected) samples.

Measurement of apoptosis.

At the indicated time points, cells from infected and uninfected lymphoid cell cultures were washed with phosphate-buffered saline-2% fetal bovine serum-2.5 mM CaCl₂, stained for cell surface markers CD3, CD4, and either annexin V-phycoerythrin (Alexis) or 200 nM tetramethylrhodamine methyl ester (TMRM; Molecular Probes) for 30 min at room temperature, washed again, and subjected to flow cytometric analysis. To determine activation of caspase-3, cells were incubated with 10 μM PhiPhiLux-G₁D₂ (Alexis) for 1 h at 37°C, washed, stained for CD4 and CD3, and subjected to FACS analysis.

GHOST cell and HeLa infection assays.

GHOST cell assays were performed as reported previously (11). Briefly, 20,000 CXCR4⁺ CD4⁺ (GHOST-X4) or CCR5⁺ CD4⁺ GHOST (GHOST-R5) cells were plated in 12-well plates and infected at a multiplicity of infection (MOI) of 0.1. At 72 h after infection, infected cells were identified by flow cytometric analysis. CXCR4⁺or CCR5⁺ HeLa CD4 cells were pretreated for at least 12 h with the indicated concentrations of AMD3100 and inoculated with HIV at an MOI of 0.01. At 72 h after infection, the concentration of p24^gag in the culture supernatant was assessed by anti-p24 ELISA.

Sequence analysis of the gp120 V1/V2 and V3 regions of D36/99.

CXCR4⁺ CCR5⁺ CD4⁺ HeLa cells were infected with D36/99 at an MOI of 0.5. At 12 days after infection, genomic DNA was isolated with the Qiagen DNeasy Tissue kit. The gp120 gene was PCR amplified with the following primers at an annealing temperature of 51°C: 5′ primer D36-5680, GTTGGTCACAGTCTATTATGG; 3′ primer D36-7089rev, TGGGTGCTACTCCTAATGGTT. Specific bands were purified from an agarose gel with the Qiagen gel extraction kit. PCR products were cloned into the pCR2.1 vector (Invitrogen). Inserts of two clones were sequenced with the M13 reverse sequencing primer (CAGGAAACAGCTATGAC) and the M13 sequencing (−20) primer (GTAAAACGACGGCCAGT).

Nucleotide sequence accession number.

The sequence of the D36/95 isolate is available under accession number AF042105.

Characterization of the virulence of D36/95 and D36/99.

First, the clinical isolates from patient D36 were tested for their potential to deplete CD4⁺ T cells in ex vivo human lymphoid cultures. The kinetics of HIV-induced CD4⁺ T-lymphocyte depletion was measured by staining cultures with antibodies to CD3, CD4, and CD8 and displayed as a ratio of CD4⁺ to CD8⁺ T cells as described previously (23, 37, 45, 54, 55). We compared the two D36 isolates with two other primary isolates from a different cohort. These isolates, called 7/86, an R5X4 strain, and 1/85, an R5 strain (14), are wild-type nef strains. As expected, the number of CD4⁺ T cells in 1/85-infected cultures decreased only slightly over time, since R5 strains can infect and deplete only the small subset of CCR5-expressing CD4⁺ T cells (28, 45, 54, 55). In contrast, the dualtropic virus 7/86 depleted CD4⁺ T cells markedly due to its expanded target cell range (Fig. 1A) (55).

Overall, CD4 depletion was less severe in cultures infected with either of the two D36 isolates than in cultures infected with the R5X4 reference strain 7/86 (Fig. 1A). However, the later D36/99 isolate was measurably more cytopathic than the earlier D36/95 isolate, with more severe CD4⁺ T-cell depletion despite comparable inoculum size (Fig. 1A). The increased virulence of D36/99 was also reflected in faster replication kinetics. D36/99 replicated with a profile similar to that of the two reference strains, whereas replication of D36/95 was significantly delayed (Fig. 1B).

The elevated potential of D36/99 to deplete CD4⁺ T lymphocytes was paralleled by its ability to induce apoptosis in CD4⁺ T cells. Apoptosis was measured independently by several assays: annexin V binding to phosphatidylserine as a marker for the loss of cell membrane asymmetry (Fig. 2A); activation of caspase-3, a key enzyme of the apoptotic signal transduction pathway (Fig. 2B); and depolarization of the mitochondrial membrane as a marker of the mitochondrial branch of apoptotic signaling (Fig. 2C). To this end, cells were stained with antibodies to CD4 and CD3 along with annexin V-FITC, the membrane-permeable caspase-3 substrate PhiPhiLux-G₁D₂, or TMRM and subjected to FACS analysis (6, 29). Loss of TMRM binding to mitochondria is an indicator of the breakdown of the mitochondrial membrane potential (6). Examination by all three methods showed that cultures infected with D36/99 had much higher levels of apoptosis in CD4⁺ T cells than those infected with the earlier D36/95 isolate (Fig. 2). However, apoptosis in D36/99-infected cultures was still slightly lower than that observed in cultures infected with the wild-type strain 7/86.

Depletion and apoptosis of CCR5⁺ and CCR5⁻ CD4⁺ T cells.

Disease progression in HIV-infected patients has been shown to correlate with a change in coreceptor usage in many, though not all, individuals (14, 56, 65). Therefore, we addressed the question of whether the increased potential of D36/99 to induce apoptosis and hence deplete CD4⁺ T cells was caused by a broadened coreceptor usage. We used flow cytometry to distinguish CD4⁺ T cells into CCR5⁺ and CCR5⁻ subsets as described previously (28, 37, 45, 55). Interestingly, D36/95 depleted the CCR5⁺ subset of CD4⁺ T cells to an extent comparable to that of both D36/99 and the R5X4 reference strain 7/86 (Fig. 3A). In contrast, while D36/99 and 7/86 also caused marked depletion of the CCR5⁻ subset of CD4⁺ T cells, the earlier D36/95 isolate and the R5 reference strain 1/85 did not affect this subset significantly (Fig. 3B). Similarly, D36/95 and 1/85 induced apoptosis mainly in CCR5⁺ CD4⁺ T cells, whereas D36/99 and 7/86 promoted apoptosis equally in both the CCR5⁺ and CCR5⁻ cellular subsets (Fig. 3C and D). Thus, the effects of D36/95 appeared to be restricted to CCR5⁺ CD4⁺ T cells, while cytopathic effects of the later D36/99 isolate were broadened to include both CCR5⁺ and CCR5⁻ CD4⁺ T cells.

Coreceptor phenotype of the D36 isolates.

To define the coreceptor usage by the two D36 isolates more directly and quantitatively, GHOST cell infection assays were performed. CCR5-expressing GHOST (GHOST-R5) cells and CXCR4-expressing GHOST (GHOST-X4) cells were infected with D36/95 and D36/99 at an MOI of 0.1 (11). The prototypical strains 49-5 (R5) and NL4-3 (X4) were used as positive controls for CCR5 and CXCR4 usage, respectively (62). As expected, 49-5 exclusively infected GHOST-R5 cells, as indicated by LTR-mediated expression of the fluorescent marker, whereas NL4-3 infected GHOST-X4 cells (Fig. 4A). The residual infection of GHOST-R5 cells by the X4 strain NL4-3 was due to the endogenous low-level expression of CXCR4 in parental GHOST cells. Importantly, both D36/95 and D36/99 infected GHOST-R5 cells to an extent comparable to that of 49-5. In contrast, D36/95 infected GHOST-X4 cells but with a fourfold lower efficiency than D36/99. These results indicate that D36/99 has a significantly higher potential to infect CXCR4-expressing cells than D36/95 (Fig. 4A).

We tested the differential CXCR4 utilization by these viruses further by analyzing their sensitivity to the CXCR4 antagonist AMD3100 (22, 53). Dose-response curves were determined with AMD3100 to inhibit infection of CXCR4⁺ CCR5⁺ HeLa CD4 cells (7, 33) by D36/95, D36/99, and the two reference viruses 49-5 (R5) and NL4-3 (X4). Infection by either 49-5 or D36/95 was not significantly inhibited by AMD3100, confirming their independence from CXCR4 (Fig. 4B). In contrast, infection by NL4-3 or D36/99 was inhibited effectively by AMD3100, with a 50% inhibitory concentration (IC₅₀) of 38 nM and 85 nM, respectively (Fig. 4B). Notably, even at AMD3100 concentrations as high as 2,500 nM, infection with D36/99 could not be inhibited completely, while that by NL4-3 was nearly fully abolished (Fig. 4B). This suggests that D36/99 can use both CCR5 and CXCR4 as a coreceptor. Taken together, these results demonstrate that both D36 isolates can utilize CCR5 as a coreceptor and that D36/99 can also use CXCR4 efficiently.

Next, we analyzed the coreceptor usage of the D36 isolates within the biologically relevant ex vivo lymphoid cell culture system. Lymphoid cultures were pretreated with AMD3100 (250 nM) and infected with equal doses of D36/95 or D36/99. Depletion and apoptosis of CD4⁺ T cells overall as well as specific depletion of the CCR5⁻ and CCR5⁺ subsets of CD4⁺ T cells were determined 12 days after the infection. As expected, the modest depletion (Fig. 5A) and apoptosis (Fig. 5B) of CD4⁺ T cells induced by D36/95 were not significantly affected by AMD3100. In contrast, both CD4 depletion (Fig. 5A) and apoptosis (Fig. 5B) induced by D36/99 were strongly inhibited by AMD3100 (Fig. 5A and B). Furthermore, analysis of the CCR5⁻ and CCR5⁺ subsets showed that AMD3100 prevented depletion of CCR5⁻ CD4⁺ T cells by D36/99, but not depletion of CCR5⁺ CD4⁺ T cells (Fig. 5C). Again, depletion of the CCR5⁺ subset by D36/95 was unaffected by AMD3100 (Fig. 5C). These results confirm that D36/99 has a broadened coreceptor usage that allows it to infect both CCR5⁺ cells and CCR5⁻CXCR4⁺ CD4⁺ T cells.

Sequence analysis of the gp120 V1/V2 and V3 regions of the D36 isolates.

Coreceptor usage is mostly determined by specific amino acid residues in the V1/V2 and V3 loops of the viral envelope protein gp120 (5, 9, 18, 25, 30, 31, 58). In particular, the high charge and low N-glycosylation status of the V3 loop as well as the high N-glycosylation status in V1/V2 have been linked to X4 usage (12, 18, 34, 39, 47, 63). Therefore, we compared the V1/V2 and V3 regions of the two different D36 isolates to analyze whether specific amino acid changes in these domains could be the molecular determinants for the broadened coreceptor usage of D36/99. We did not detect any amino acid mutations in the V3 loop that would be expected to alter its charge or glycosylation status (Fig. 6). However, we found that the later D36/99 isolate contains an additional potential N-glycosylation site at position 161 in the V1/V2 region (Fig. 6). Thus, it appears that the more efficient CXCR4 usage of the later D36/99 isolate may be attributable to an additional N-glycosylation site in the V1/V2 region.