Spontaneous Mutations in the env Gene of the Human Immunodeficiency Virus Type 1 NDK Isolate Are Associated with a CD4-Independent Entry Phenotype

Abstract

Human immunodeficiency virus type 1 (HIV-1) entry into target cells is a multistep process initiated by envelope protein gp120 binding to cell surface CD4. The conformational changes induced by this interaction likely favor a second-step interaction between gp120 and a coreceptor such as CXCR4 or CCR5. Here, we report a spontaneous and stable CD4-independent entry phenotype for the HIV-1 NDK isolate. This mutant strain, which emerged from a population of chronically infected CD4-positive CEM cells, can replicate in CD4-negative human cell lines. The presence of CXCR4 alone renders cells susceptible to infection by the mutant NDK, and infection can be blocked by the CXCR4 natural ligand SDF-1. Furthermore, we have correlated the CD4-independent phenotype with seven mutations in the C2 and C3 regions and the V3 loop. We propose that the mutant gp120 spontaneously acquires a conformation allowing it to interact directly with CXCR4. This virus provides us with a powerful tool to study directly gp120-CXCR4 interactions.

Human immunodeficiency virus (HIV) entry into target cells is mediated by CD4 and a coreceptor belonging to the chemokine receptor family (1, 10, 11, 14, 18, 22). HIV type 1 strains are able to infect macrophages and CD4 lymphocytes or T-cell lines, reflecting differences in coreceptor usage (8). Macrophage-tropic isolates use the CC chemokine receptor CCR5 as a coreceptor, whereas T-cell line-adapted (TCLA) isolates use the CXC chemokine receptor CXCR4 as a coreceptor (1, 11, 14, 18). Dual-tropic viruses constitute the majority of primary isolates and can use both coreceptors. Whatever the coreceptor specificity of an HIV isolate may be, an interaction with CD4 is always the first step in a chain of events leading to fusion of the viral envelope with the plasma membrane. Envelope glycoprotein interactions with CD4 and the coreceptor are probably aimed at bringing the viral envelope membrane close to the plasma membrane, allowing transmembrane protein gp41 to initiate the fusion process.

gp120 binds directly to CD4 (23, 28), inducing conformational changes in the envelope glycoproteins thought to expose a binding domain for the coreceptors (25, 31). A direct interaction between gp120 and CCR5 or CXCR4 can be detected only in the presence of soluble CD4 (12a). Recent data suggest an important role of the V3 loop in coreceptor binding (36, 37), though no direct interaction with the coreceptors has ever been formally demonstrated. This region of gp120 is a major determinant of viral tropism, since a portion of the V3 loop from a macrophage-tropic isolate is sufficient to confer macrophage tropism on a TCLA isolate, without affecting binding of gp120 to CD4 (3, 21, 27, 32, 33).

We report the first isolation of an HIV-1 strain that no longer requires the presence of CD4 to enter target cells. This spontaneous and stable entry phenotype occurred after long-term culture of HIV-1 strain NDK in CEM cells (15). We have obtained molecular clones of the mutant virus and show that it is capable of infecting different CD4-negative human cells after a direct interaction with CXCR4. We were able to correlate this new phenotype with mutations in critical regions of the env gene, including the V3 loop. CD4-independent infection has so far been reported for only one HIV-2 isolate and has been correlated with mutations dispersed throughout the env gene (5, 16, 30).

MATERIALS AND METHODS

Cells and viruses.

Nonadherent cells were grown in RPMI 1640 medium (Gibco-BRL) supplemented with 10% fetal calf serum, antibiotics, and glutamine. CD4-positive human lymphoid T-cell lines CEM and H9 were gifts from J.-L. Virelizier (Institut Pasteur, Paris, France). The human T-cell leukemia virus type 3b (HTLV3b) isolate has been described previously (29) and was cultured on H9 cells. The NDK isolate has been described elsewhere (15) and was maintained in CEM cells. It was a gift from F. Barre-Sinoussi (Institut Pasteur). NDK isolate mutants m5 and m7 have been obtained after long-term culture followed by limiting-dilution cloning of the chronically infected CEM cell line. Infections were performed by incubation of target cells with 4 ml of filtered infectious supernatants for 4 h. Virus supernatants were generally titrated the same day on HeLa CD4_LacZ indicator cells. All adherent cell lines were grown in Dulbecco modified Eagle medium (Gibco-BRL) supplemented with 10% fetal calf serum, antibiotics, and glutamine. HeLa_LTRLacZ, HeLaCD4_LTRLacZ, Cos7_LTRLacZ, and Cos7CD4_LTRLacZ indicator cells have been described previously (13) and were a gift from M. Alizon (Institut Cochin de Génétique Moléculaire [ICGM], Paris, France). CaCo2 and HT29 cells were obtained from F. Russo-Marie (ICGM), and the Wish, SW480, and U373MG cell lines were obtained from J.-L. Virelizier (Institut Pasteur). Stable cotransfections were performed by the calcium phosphate coprecipitation technique. Stable clones or populations were functionally tested by coculture experiments. CCR5 expression was monitored by both reverse transcription-PCR and transient transfection of a CD4 expression vector followed by cocultures with cells expressing macrophage-tropic isolate Env proteins.

Cell fusion assays.

SDF-1 (stroma cell-derived factor 1) was obtained by a procedure previously described (6). RANTES was purchased from R&D Systems. Syncytium formation assays were performed with adherent or nonadherent HIV-1-infected cells and adherent indicator target cells. Indicator target cells contained a transiently or stably expressed long terminal repeat (LTR)-luciferase or stably expressed LTR-lacZ reporter gene cassette. Cell ratios were usually 1:1 for adherent cell cocultures and 1:2 to 1:5 for adherent-nonadherent cocultures. Depending on cell fusion efficiency, samples were analyzed between 8 and 16 h later. Too-efficient fusions were analyzed visually by May-Grumwald-Giemsa staining and photographed by using a phase-contrast microscope under ×40 or ×80 magnification. Depending on the indicator cells used, analyses using two indicator reporter gene systems (lacZ or luciferase gene) were performed. Reporter gene assays allowed quantitative or in situ analysis. For in situ analysis using β-galactosidase activity, cells were fixed in 0.5% glutaraldehyde, and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) assays were performed for 4 h at 37°C or overnight at 4°C, depending on the experiments, as previously described (13). Blue-stained syncytia were scored under ×40 binocular magnification. Quantitative chlorophenol red-β-d-galactopyranoside (CPRG) (Boehringer Mannheim) assays were done after recovery of cell extracts following incubation in a lysis buffer (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 10 mM MgSO₄, 2.5 mM EDTA, 50 mM β-mercaptoethanol, 0.125% Nonidet P-40). The cell extracts were then incubated in a reaction buffer (0.9 M phosphate buffer [pH 7.4], 9 mM MgCl₂, 11 mM β-mercaptoethanol, 7 mM CPRG) for 90 min. The resulting activities were measured with an LP400 (Becton Dickinson) plate reader as optical density at 570 nm. Experiments were performed in triplicate. Luciferase tests were performed on a LB9501 luminometer (EGG Berthold) as previously described (34).

Constructs.

The NDK proviral genomic clone was a gift from B. Spire (35). The CXCR4 expression vector was a gift from B. Moser (University of Bern, Bern, Switzerland) (24), and the CCR5 expression vector was obtained from J. Moore and T. Dragic (Aaron Diamond AIDS Research Center, New York, N.Y.) (14). The pLTR-Luc plasmid used in transfection experiments has been described elsewhere (34). The env cassette expression vector contained a PCR-amplified, blunted HIV-1 NDK LTR cloned in the blunted SacI site of pBS_KS+ (Stratagene). The primers used were 5′8602 (5′TAATTTGGTCAAAGAAAAGACAAGAG3′) and 3′313 (5′ATCTCTCTCCTTCTAGCCTC-CGC3′), and the 877-bp amplified fragment contained the whole promoter, the trans-activating region (TAR), and the common splice donor (SD) site of the virus. The XhoI-BglII (blunted) fragment from the poly(A) of the murine PGK gene was inserted in the XhoI-Asp718 (blunted) sites of the polylinker. env regions were PCR amplified from genomic DNA from infected cells by using the primers 5′5236 (5′AAACTTATGGGGATACCTGGGCAGG3′) and NDK6 (5′ATTGCCCCATGTTTTTCCAGG3′). The resulting 3,193-bp fragment was digested with EcoRI and XhoI enzymes and cloned in the same sites of the polylinker of the expression vector. The ligated fragment contained env, tat, rev, and vpu gene open reading frames, and the proteins are expressed after alternative splices between the common SD site from the promoter region and gene-specific acceptor sites. Chimeric constructs between derived and nonderived cloned env genes were made by using unique restriction sites in the env gene. EcoRV and HindIII sites were used for amino acids located in the V3-C3 region, EcoRI and HindIII sites were used for amino acids situated in the C2, V3, and C3 regions, and EcoRI and EcoRV sites were used for the C2 region. Regions were exchanged as indicated below (see Fig. 7).

FIG. 7.

Chimeric constructs between derived or nonderived cloned env genes. Regions were exchanged as indicated, with EcoRV and HindIII sites used for the V3-C3 region and EcoRI and HindIII used for the C2, V3, and C3 regions. The fusion phenotypes of the resulting chimeric env expression vectors were analyzed by coculture between transiently transfected HeLa cells and either CD4-positive or CD4-negative HeLa lacZ indicator cells. An in situ X-Gal assay was performed after 16 h of coculture. Results represent schematic means of at least three to five independent experiments. Scoring of syncytia was done as follows: −, <5; ++, 50 to 200; +++, 200 to 1,000; and ++++, >1,000. wt, wild type.

Sequence analysis.

Relevant mutations of env genes were analyzed either by direct sequencing of PCR-amplified fragments or by sequencing after cloning in the expression vector. In the latter case, only transfected vectors showing expected cell fusion tropism were sequenced. The sequencing method used the ABI-Prism kit (Perkin-Elmer), allowing PCR sequencing with fluorescent nucleotides and automated gel reading. Results were directly computerized for performance of sequence analyses and comparisons. The sequences were established twice in both senses by the use of the following oligonucleotide primers dispatched approximately every 200 to 300 bases: NDK1 (5′AATAGC-AATAGTTGTGTGGACC3′), NDK2 (5′TTGTAGCTACCTGTTGTAAAGC3′), NDK3 (5′AAGCACATTGTAAAATTAGCA3′), NDK4 (5′AAATAATCCGTTCACCAATCG3′), NDK5 (5′TCACTTCTGTCATTTCAGACC3′), NDK6 (5′ATTGCCCCATGTTTTTGGAGG3′), 5′5236 (5′AAACTTATGGGGATACCTGGGCAGG3′), 5′5979 (5′TACCCACGGACCCCAACCC3′), 5′6379 (5′TTTCCAATTCTATTTCTTGTGGG3′), 5′7058 (5′AAATGTTCATCAAATATTACAGGG3′), 5′7573 (5′TAATAGATCTCTAGATGAGATTTGG3′), 5′8069 (5′ATTGTGGAACTTCTGGGACGC3′), 3′5995 (5′TTTCCAATTCTATTTCTTGTGGG3′), 3′6377 (5′AAAAATGTATGGGAATTGG3′), 3′7155 (5′ATAATTCACTTCTCCAATTGTCCC3′), 3′7617 (5′AATTGTCAATTTCTCTTTCCC3′), and 3′8379 (5′ATACTGCTCCTACCCCATCTGC3′). The relevance of mutations was ensured by use of the PCR method to determine consensus sequences.

RESULTS

Characterization of a mutant NDK tropism for CD4-negative cells.

Cocultures of LTR-lacZ indicator cells and CEM cells chronically infected with the HIV-1-mutated NDK isolate revealed syncytium formation with both HeLaCD4 and HeLa cells (Fig. 1A, panels a and b). By contrast, H9 cells infected with the HIV-1 HTLV3b isolate efficiently formed syncytia only with HeLaCD4 indicator cells (Fig. 1A, panel c) and, as expected, did not fuse with HeLa cells (Fig. 1A, panel d). Usually, 10- to 15-fold-more syncytia were observed when the indicator cells expressed the CD4 antigen, showing that although it is no longer necessary, CD4 optimizes viral entry of mutant NDK (mNDK). The infected CEM population therefore contains a viral population which has undergone a phenotypic switch of its cell tropism. We refer to this uncloned population as CEM_mNDK. Different human adherent cell lines were analyzed for their ability to allow CD4-independent mNDK entry. We transiently transfected an LTR-luciferase vector into different cell lines and cocultured them with CEM_mNDK. Detection of luciferase activity revealed that CD4-negative cell lines such as HeLa, Wish, and SW480 were able to fuse with CEM_mNDK cells, whereas human U373MG, CaCo2, and HT29 cells, which do not express CXCR4 (data not shown), were not (Fig. 1B). This also demonstrated that galactosylceramide, expressed on U373MG, CaCo2, and HT29 cells, was not involved in this phenomenon (9, 17, 19).

FIG. 1.

Coculture experiments with HIV-1-infected cells and different human target cells. (A) CD4-positive (panels a and c) or CD4-negative (panels b and d). HeLa indicator cells stably expressing an HIV LTR-lacZ gene were cocultured with either CEM_MNDK cells (panels a and b) or HTLV3b-infected H9 cells (panels c and d). An in situ β-galactosidase test was performed after 16 h of culture to score and analyze specific fusion events. Cells were photographed under ×40 magnification. (B) Different human cell lines were transiently transfected with an HIV-1 LTR-luciferase gene and trypsinized 16 h thereafter to be used for cocultures with different CEM cell lines infected with either NDK or mNDK virus. The experiment whose results are presented is representative of at least three independent manipulations. Fold induction (indicated above the bars) was standardized in comparison to luciferase activities obtained for cocultures with the nonderived virus.

Subcloning of mutant viruses.

The CEM_mNDK infected-cell population was cloned by limiting dilution. Sixty-four independent clones were analyzed for their fusion ability by coculture with indicator cells (Table 1), and we used two different indicator gene systems (luciferase [Fig. 2A] and lacZ [Fig. 2B and C]). Only eight clones exhibited a CD4-independent tropism. The others were either nonfusogenic or strictly CD4 dependent (Table 1). We selected three clones (CEM clones 15, 27, and 29) for further analysis. The two reporter systems strictly correlated and clearly demonstrated that CEM clone 15 presented a strict CD4-dependent tropism, whereas CEM clone 29 presented a CD4-independent cell tropism (Fig. 2). Results illustrated in Table 1 strongly suggest that a progressive phenotype shift occurred, since intermediate fusion phenotypes were characterized in the infected-cell population. Fusion efficiencies varied from one clone to another. Only 12.5% of the cellular clones were CD4 independent (Table 1). This may signify that the CD4-independent tropism represents a dominant phenotype. Our results clearly demonstrate that the infected CEM population contains phenotypically heterogeneous viruses.

TABLE 1.

Cellular cloning of the mNDK-infected CEM population: coculture analysis of cell clone entry phenotype

Syncytium formationa		No. of clonesb
HeLaCD4LTRLacZ	HeLaLTRLacZ
−	−	14
++	−	36
+++	−	4c
+++	±	2
++++	+++	8d

^aSyncytia were scored as follows: −, <5; ±, 5 to 10; +, 10 to 50; ++, 50 to 200; +++, 200 to 1,000; and ++++, >1,000. 

^bSixty-four clones in total (mean clone dilution, 1/8 per well). 

^cFour clones, from which clone 15 was selected for further studies. 

^dEight clones, from which clones 27 and 29 were used for further studies. 

FIG. 2.

Fusion phenotype of infected CEM cell clones. (A) An HIV LTR-luciferase plasmid was transiently transfected in HeLa cells, and coculture experiments with CEM clone (Cl.) 15 or 29 were then performed. Fold induction (indicated above the bars) was standardized in comparison to luciferase (Luc.) activities obtained for cocultures with CEM clone 15. (B) Coculture experiments between CD4-negative indicator HeLa lacZ cells were performed with CEM cell clones 15 and 29. Blue-stained syncytia were scored, and fold induction was standardized in comparison to the number of syncytia scored for cocultures with CEM clone 15. (C) Coculture experiments between CD4-negative (panels b and d) or CD4-positive (panels a and c) indicator HeLa lacZ cells were performed with CEM clones 15 (panels a and b) and 29 (panels c and d). An in situ β-galactosidase test was performed after 16 h of coculture to analyze and photograph specific fusion events (magnification, ×40).

Stability of the phenotype shift.

We infected SW480 and HeLa cells with filtered supernatant from the CEM_mNDK infected-cell population. Chronically infected HeLa_mNDK and SW480_mNDK cell populations were obtained, filtered supernatants from which were used to infect the CD4-positive human T-cell line H9. Chronically infected cell populations were designated H9_H and H9_S, respectively (Fig. 3A). The chronically infected populations were cocultured with human LTR-lacZ indicator CD4⁺ or CD4⁻ HeLa cells (Fig. 3B, panels a, c, e, and g, and B, panels b, d, f, and h, respectively). In all cases, the same entry phenotype as for the original mNDK-infected CEM cell population was observed (Fig. 3B). Infection of the CD4⁺ human T-cell line H9 revealed no reversion of the fusion phenotype.

FIG. 3.

Experimental strategy followed to analyze the stability of the new entry phenotype and to localize mutations of the derived virus. (A) CEM_mNDK infected-cell supernatants were used to infect either SW480 or HeLa cells. Supernatants of the resulting chronically infected cell lines were used to infect the H9 CD4-positive T-lymphocytic cell line. Chronically infected cell lines H9_H and H9_S were obtained for further syncytium formation analysis. Cellular clones were also obtained as indicated. Fusion phenotype analysis of different infected-cell populations, molecular cloning, and sequencing of the env gene were performed at each step. (B) Coculture experiments between CD4-negative (panels b, d, f, and h) or CD4-positive (panels a, c, e, and g) indicator HeLa lacZ cells were performed with HeLa_mNDK (panels a and b), SW480_mNDK (panels c and d), H9_H (panels e and f), and H9_S (panels g and h) cells. An in situ β-galactosidase test was performed after 16 h to analyze and photograph (magnification, ×40) specific fusion events.

Chronically infected HeLa and SW480 cells were also cloned. We analyzed 116 independent cell clones from HeLa_mNDK and 16 from SW480_mNDK. Cocultures were performed with indicator cells, and we observed that 80% of the HeLa clones were infected, 100% of which had the mutant tropism. Around the same percentage of SW480 cell clones were infected (75%), of which 100% presented the mutant entry phenotype (Table 2). All infected CD4-negative clones expressed viruses which exhibited the mutant tropism, although viral supernatants from the CEM_mNDK cell population contained a mixed population (compare Tables 1 and 2). Nonetheless, all the cloned mutant viruses were able to enter HeLaCD4 cells 5 to 15 times more efficiently than HeLa cells (depending on the clone [Tables 1 and 2]).

TABLE 2.

Cellular cloning of the mNDK-infected HeLa and SW480 populations: coculture analysis of cell clone entry phenotype

Syncytium formationa		No. of clonesb
HeLaCD4LTRLacZ	HeLaLTRLacZ	HeLa	SW480
−	−	45	4
+	±	28	11
++	±	15	0
+++	++	22c	0
++++	+++	6c	1

^aSyncytia were scored as follows: −, <5; ±, 5 to 10; +, 10 to 50; ++, 50 to 200; +++, 200 to 1,000; and ++++, >1,000. 

^bIndependent clones were recovered by using glass cylinders. 

^cTwenty-eight clones, from which clones 2, 14, and 48 were selected. 

We concluded that the derived viral tropism is stable once acquired and is associated with a viral genetic shift rather than depending on a particular cell type or cell clone.

CXCR4 is sufficient for viral entry.

U373MG cells were stably cotransfected with a luciferase vector under the control of the HIV-1 LTR and an expression vector for CXCR4 or CCR5. As shown in the representative experiment results in Fig. 4A, an 11-fold transactivation of the luciferase reporter gene was observed after coculture of U373MG_CXCR4 cells with mNDK-infected CEM cells. A greater fusion efficiency was observed with U373_CXCR4 clone 31 (Fig. 4A), which expressed a higher level of CXCR4 protein (not shown). These results show that CXCR4 acts as the receptor for mNDK virus. This was confirmed by a May-Grunwald-Giemsa (MGG) in situ analysis of cocultures of U373MG_CXCR4 positive clone (clone 23) with either CEM_mNDK cells (Fig. 4B, panel b) or HTLV3b-infected H9 cells (Fig. 4B, panel a), resulting in the occurrence of large syncytia only with the mNDK isolate (Fig. 4B). In contrast, expression of CCR5 in U373MG cells did not allow syncytium formation with CEM_mNDK cells (Fig. 4A).

FIG. 4.

CXCR4 acts as the receptor for the derived virus. (A) CXCR4 or CCR5 was transiently or stably cotransfected with the HIV LTR-luciferase vector in the human U373MG cell line. Coculture experiments were then performed with HTLV3b- or mNDK-infected H9 cells, and Tat-directed transactivation was measured. Fold induction was standardized in comparison to luciferase (Luc.) activities obtained for cocultures with the nonderived virus. The experiments whose results are presented are representative of at least three independent manipulations. Pop., population; Cl., clone. (B) An in situ MGG test was performed 8 h after cocultures between CXCR4-stabilized U373MG clone 23 with HTLV3b-infected (a) or mNDK-infected (b) H9 cells. Specific fusion events were analyzed and photographed under an original magnification of ×80.

Next, we attempted cell-cell fusion inhibition using different chemokines. The CXCR4 ligand SDF-1 (2, 26) or the CCR5 ligand RANTES (1) was added to HeLa CD4⁺ or CD4⁻ indicator HIV-1_LTR-lacZ cells. Fusions between those cells and chronically HTLV3b- or mNDK-infected H9 cells were analyzed 8 h later by a CPRG β-galactosidase test (Fig. 5). Specific inhibition of syncytium formation was observed when SDF-1 was used with either mNDK-infected cells (98.7%) or HTLV3b-infected cells (95.4%) when HeLa CD4⁺ indicator cells were used (Fig. 5A). Syncytia were observed only with mNDK virus when HeLa CD4⁻ indicator cells were used and were specifically inhibited (97%) with the CXC chemokine SDF-1 (Fig. 5B). The CC chemokine RANTES had no effect. This suggests that mNDK uses CXCR4 directly as a receptor and can efficiently be competed with the specific receptor ligand.

FIG. 5.

SDF-1 specifically inhibits cell-cell fusion. SDF-1 (300 nM) or RANTES (300 nM) was added to 2 × 10⁴ indicator cells per well in a 96-well plate 30 min prior to addition of HTLV3b- or mNDK-infected H9 cells or noninfected H9 cells (3 × 10⁴ cells/well). Cocultures were performed in triplicate. Analysis of fusion efficiency was performed 8 h after coculture by a quantitative β-galactosidase test using CPRG substrate. β-Galactosidase activities were quantified after 90 min at an optical density of 570 nm (OD₅₇₀). (A) Cell-cell fusion with HeLaCD4 lacZ indicator cells. (B) Cell-cell fusion with HeLa lacZ indicator cells. Untr., untreated cells.

We performed cocultures of mNDK-infected cells and CD4-positive, CXCR4-negative U373MG cells. As no fusion could be detected, we concluded that CD4 alone could not allow cell fusion even though the mutated virus was used (not shown).

Cloning and functional analysis of env genes.

PCR-amplified env genes from different cell populations infected with either derived or nonderived viruses (CEM_NDK, CEM_mNDK, HeLa_mNDK, SW480_mNDK, H9_H, and H9_S) were cloned in a eucaryotic expression vector, as were the PCR-amplified env gene fragments from CEM clones 15, 27, and 29 and HeLa clones 2, 14, and 48. The different envelope (Env) proteins expressed from those plasmids were designated according to the cell populations or cell clones they were derived from.

env expression vectors were transfected into HeLa cells, and cocultures with human CD4-positive or CD4-negative indicator cells were performed. All of the env expression vectors from derived viruses (either from cell populations or from cell clones) were able to form syncytia with CD4⁺ and CD4⁻ HeLa cells (not shown). As previously observed for infected cells, an increase in fusion efficiency (ranging from 2- to 18-fold, depending on the cell clone) could be detected in the presence of the CD4 molecule on the cell surface. As expected, Env₁₅ (from CEM clone 15), Env_Ori (from the cloned NDK provirus [35]), and Env_NDK (from nonderived NDK-infected CEM cells) formed syncytia only with HeLaCD4 indicator cells. This shows that the env gene is responsible for the CD4-independent phenotype and that the CD4 molecule is not necessary for viral entry, although it increases fusion efficiency.

Molecular analysis of the env genes from derived viruses.

We performed consensus sequence analysis on PCR-amplified fragments from env genes of genomic DNA in HeLa_mNDK, SW480_mNDK, and CEM_mNDK cells. We sequenced whole env genes which we aligned and compared with two reference sequences, from the cloned NDK provirus (35) and the env gene from proviral nonderived NDK virus from the infected CEM cell population. Silent point mutations were observed throughout the env coding sequence. However, no mutations were found in gp41. The critical gp120 region V1, V2, V4, and V5 loops and the CD4-binding domains also do not contain mutations. In contrast, compared with the wild-type env gene, seven amino acid changes were observed in all sequenced env genes from CEM_mNDK, HeLa_mNDK, and SW480_mNDK cells. Two of them were located in the C2 region (D¹⁹²→N¹⁹² and T¹⁹⁵→S¹⁹⁵), four were in the V3 loop (K²⁹⁶Y²⁹⁷T²⁹⁸→N²⁹⁶N²⁹⁷I²⁹⁸ and R³⁰⁷→G³⁰⁷), and one was in the C3 region (A³³³→V³³³). The presence of these seven specific mutations was confirmed after sequencing of the C2, V3, and C3 regions from the env genes from H9_S and H9_H cells (Fig. 6A).

FIG. 6.

Consensus sequence of env genes. The sequencing strategy used (see Materials and Methods) allowed detection of only relevant mutations. The different origins of sequenced genes and the panel of primers used gave the sequences twice on both strands. (A) Schematic map of relevant mutations observed after consensus sequencing performed directly on PCR-amplified env genes from infected-cell populations of different origins. Env_Ori, sequences obtained from nonderived CEM_NDK cell population and from the cloned provirus (35); Env_mNDK sequences obtained from different derived cell populations: CEM_mNDK, H9_H, H9_S, HeLa_mNDK, and SW480_mNDK. (B) Schematic map of relevant mutations observed after consensus sequencing performed on cloned env genes from different origins ligated in the expression vector. Their ability to direct cell fusion when expressed in HeLa cells has previously been tested. Env_NDK and Env_NDKsb, clonotypic sequences from the env gene from the nonderived CEM_NDK cell population; Env_CEM15, clonotypic sequence of the cellular clone 15 presenting a strict CD4-dependent entry phenotype; Env_CEM29,27 and Env_HeLa2,14,48 clonotypic sequences of cellular clones presenting a CD4-independent entry phenotype. wt, wild type.

Some of the cloned env genes were also sequenced to correlate the fusion phenotype with the seven mutations. Cloned env genes from CEM clone 29 and HeLa clone 14 were sequenced. As expected, few silent mutations were found, and only the seven mutations previously observed were common to these two clones.

We have also isolated two env genes carrying mutations that are not associated with a CD4-independent entry phenotype. In CEM clone 15, five mutations localized in the V3 and C3 regions and identical to those observed in derived env genes were observed (K²⁹⁶Y²⁹⁷T²⁹⁸→N²⁹⁶N²⁹⁷I²⁹⁸, R³⁰⁷→G³⁰⁷, and A³³³→V³³³ [Fig. 6B]). In the molecular subclone Env_NDKsb, two mutations in the V3 loop which were not present in Env_wt (K²⁹⁶Y²⁹⁷→N²⁹⁶N²⁹⁷) were observed. These mutations correspond to two of the four mutations observed in all V3 loops from derived viruses and probably represent intermediate genotypes.

Virus names are based on the number of mutations in the env gene. Therefore, CD4-independent virus was designated m7 NDK, and CD4-dependent mutant viruses were designated m2 or m5 NDK.

V3 and C3 mutations associated with C2 changes are responsible for the CD4 independence.

To link the onset of the derived phenotype to the appearance of specific mutations, we constructed chimeras between m5, m7, and wild-type NDK env genes which were cloned into a eucaryotic expression vector. Transient transfections into HeLa cells and coculture experiments with indicator cells were performed. We observed that the C2 region from Env_m7 cloned in Env_wt did not confer a CD4-independent phenotype. The reverse chimera (C2 from Env_wt cloned into Env_m7) revealed a loss of fusion with CD4-negative cells (Fig. 7). When C2, V3, and C3 regions from Env_m7 were cloned in Env_m5, the resulting chimeric Env protein allowed a CD4-independent cell fusion. The amino acid changes in the C2 domain thus correlate with the onset of the mutant tropism in the context of modified V3-C3 regions. Since there is no difference between those Env proteins other than these five or seven mutations, it can be concluded that the appearance of CD4-independent tropism requires seven mutations and that five of the seven mutations in the V3-C3 region are not sufficient to allow CD4 independence. Work is in progress to determine if all mutations are required to allow this particular phenotype.

DISCUSSION

We report here a spontaneous shift in cell tropism for a TCLA laboratory HIV-1 isolate. This NDK derivative is able to infect some human CD4-negative cell lines, such as HeLa, SW480, and Wish. This adaptation is genetically stable, since chronically infected CD4-negative HeLa and SW480 cells could be obtained and since recurrent infection of human CD4-positive T-cell line H9 with supernatants from those adherent infected cell lines revealed a stable entry phenotype. The host cell range of this virus proved that the alternate receptor galactosylceramide is not implicated in the phenomenon, since U373MG, HT29, and CaCo2 cells were not efficient in syncytium formation (9, 17, 19). As expected for a long-term laboratory TCLA isolate, complementation of the U373MG cell line with a CXCR4 expression vector and inhibition experiments using SDF-1 clearly demonstrate that the mNDK virus uses CXCR4 as receptor, no longer needing CD4. Our results also shown that mNDK is not able to use CCR5 (Fig. 4). Our virus is thus the first HIV-1 laboratory-adapted isolate for which the CD4 molecule is not necessary for infection and which interacts with its specific coreceptor, CXCR4.

We suggest the occurrence of a progressive genetic switch in HIV-1 NDK cell tropism related to specific mutations in critical regions C2, V3, and C3 of the env gene. The characterization of intermediate genotypes with two amino acid changes in the V3 loop or five amino acid changes in the C3-V3 regions (from Env_m2 and Env_m5, respectively) supports the idea of a progressive evolution towards the seven mutations leading to a CD4-independent entry phenotype. These seven mutations could be responsible for a spontaneous exposure and better flexibility of the V3 loop. In all cases, either infected-cell populations, infected-cell clones, or transiently expressed Env_m7 proteins always formed syncytia 5 to 10 times more efficiently with HeLaCD4 than with HeLa cells (Tables 1 and 2 and Fig. 1 to 3). This indicates that although no longer necessary, the CD4 molecule can still interact with the mutant gp120 and may optimize its fusogenic conformation. To support this, it must be noted that this spontaneous change occurred in the human CD4-positive T-cell line CEM without any selective pressure and that once acquired, this phenotype was genetically stable. This clearly indicates that this phenotype is favored in a genetically mixed viral population (Table 1) and suggests a progressive adaptation of the Env proteins in vitro to gain in infection efficiency and/or viral replication.

The recent characterization of several members of the chemokine receptors as the long-sought HIV entry cofactors (1, 4, 11, 12, 14, 18) raised the question of a direct interaction between viral glycoproteins and the coreceptor. Some evidence suggests that CD4-dependent isolates undergo a direct but weak interaction between gp120 and CXCR4 (20) or CCR5 (37). A critical role for the V3 loop in this interaction has been reported (7, 36). This region of the viral gp120 influences cellular tropism and participates in coreceptor binding. Perhaps the function of CD4 is limited to inducing changes in gp120 conformation allowing efficient binding to the coreceptor and subsequent viral entry. This would explain why although a direct interaction with CXCR4 might be possible, the native gp120 conformation does not allow efficient interaction with the coreceptor for membrane fusion to occur. This would prevent viral entry in CXCR4-expressing CD4-negative cells and restrict the cellular host range. A progressive increase in coreceptor binding affinity would correlate with the possibility for gp120 to efficiently interact with the coreceptor. Studies of CD4-independent HIV-2 isolates confirm that an interaction with CXCR4 is sufficient to allow subsequent membrane fusion (16). Among the mutations characterized for those HIV-2 isolates, those found in V1, V2, and V3 loops could not be associated with the CD4-independent phenotype (30). In contrast, for the m7 NDK virus, we were able to correlate phenotype changes with genetic modifications. There is no mutation in the V1, V2, V4, and V5 regions and CD4-binding domains of gp120 or in the fusion peptide of gp41. Nevertheless, four mutations in the V3 loop and three mutations in adjacent regions C2 and C3 were observed. We have demonstrated that these seven mutations are directly implicated in the CD4-independent phenotype that was confirmed by fusion analysis of chimeric env gene expression vectors. Mutations in the C2 and C3 regions could change the conformation of the V3 loop, which could then interact directly with CXCR4.

The absence of fusion between m7 NDK and CD4-expressing, CXCR4-negative U373MG cells suggests that CXCR4 is essential for viral entry, whereas in some cases CD4 is dispensable. The availability of these modified Env proteins thus represents a powerful tool with which to directly analyze gp120 interaction with HIV-1 coreceptors.