Primary Human Immunodeficiency Virus Type 2 (HIV-2) Isolates Infect CD4-Negative Cells via CCR5 and CXCR4: Comparison with HIV-1 and Simian Immunodeficiency Virus and Relevance to Cell Tropism In Vivo

Abstract

Cell surface receptors exploited by human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV) for infection are major determinants of tropism. HIV-1 usually requires two receptors to infect cells. Gp120 on HIV-1 virions binds CD4 on the cell surface, triggering conformational rearrangements that create or expose a binding site for a seven-transmembrane (7TM) coreceptor. Although HIV-2 and SIV strains also use CD4, several laboratory-adapted HIV-2 strains infect cells without CD4, via an interaction with the coreceptor CXCR4. Moreover, the envelope glycoproteins of SIV of macaques (SIV_MAC) can bind to and initiate infection of CD4⁻ cells via CCR5. Here, we show that most primary HIV-2 isolates can infect either CCR5⁺ or CXCR4⁺ cells without CD4. The efficiency of CD4-independent infection by HIV-2 was comparable to that of SIV, but markedly higher than that of HIV-1. CD4-independent HIV-2 strains that could use both CCR5 and CXCR4 to infect CD4⁺ cells were only able to use one of these receptors in the absence of CD4. Our observations therefore indicate (i) that HIV-2 and SIV envelope glycoproteins form a distinct conformation that enables contact with a 7TM receptor without CD4, and (ii) the use of CD4 enables a wider range of 7TM receptors to be exploited for infection and may assist adaptation or switching to new coreceptors in vivo. Primary CD4⁻ fetal astrocyte cultures expressed CXCR4 and supported replication by the T-cell-line-adapted ROD/B strain. Productive infection by primary X4 strains was only triggered upon treatment of virus with soluble CD4. Thus, many primary HIV-2 strains infect CCR5⁺ or CXCR4⁺ cell lines without CD4 in vitro. CD4⁻ cells that express these coreceptors in vivo, however, may still resist HIV-2 entry due to insufficient coreceptor concentration on the cell surface to trigger fusion or their expression in a conformation nonfunctional as a coreceptor. Our study, however, emphasizes that primary HIV-2 strains carry the potential to infect CD4⁻ cells expressing CCR5 or CXCR4 in vivo.

Human immunodeficiency virus type 2 (HIV-2) is endemic in West Africa and has spread in the last decade to the west coast of India (3, 43, 67), as well as causing numerous infections in Europe. The mortality rate following HIV-2 infection is estimated to be a third lower than that for HIV-1 (84). HIV-2 is closely related to simian immunodeficiency virus of sooty mangabeys (SIV_SM) and SIV of macaques (SIV_MAC). SIV_SM is endemic and nonpathogenic in West African sooty mangabey monkeys, even though high viral loads can sometimes be detected in plasma (65). The HIV-2 epidemic is likely to have resulted from several zoonoses from wild SIV_SM-infected sooty mangabeys, and, consequently, primary HIV-2 strains are closely related by sequence to SIV_SM strains (30).

HIV and SIV are viruses with a lipid membrane that must fuse with the cell membrane to allow the virus core and RNA genome access to the cell cytoplasm. Glycoprotein spikes on the surface of virus particles attach to specific receptors at the cell surface and induce fusion of viral and cellular membranes. HIV-1, HIV-2, and SIV strains interact with cell surface CD4 and seven-transmembrane (7TM) coreceptors to infect cells. An interaction with CD4 triggers conformational changes in gp120 allowing a secondary interaction with a 7TM molecule to occur. The crystal structure of an HIV-1 gp120 core, complexed with soluble CD4 (sCD4 [domains 1 and 2]) and a Fab fragment of an antibody to a CD4-induced epitope, has been solved (45). The 7TM receptor binding site is predicted to be composed of conserved regions encompassing a bridging sheet domain and residues within V3 (66, 88). CCR5 and CXCR4 are major coreceptors for HIV-1; however, there are marked differences in coreceptor use between SIV and HIV-1. In particular, SIV_MAC strains use CCR5 but not CXCR4, while other coreceptors, including GPR15/BOB, STRL33/BONZO, and GPR1, are more likely to be used (2, 15, 22, 28, 48, 49). Previously we and others have shown that many primary and laboratory-adapted HIV-2 strains can exploit a broad range of coreceptors for infection of CD4⁺ cell lines, including CCR5 and CXCR4 (9, 32, 51, 58, 78), while some primary HIV-2 strains from asymptomatic individuals predominantly use CCR5 (32, 58, 78).

HIV-1 infection of CD4⁻ cell cultures in vitro has been extensively reported (for reviews, see references 12 and 13); however, this is usually much less efficient than infection of cells that express CD4. The relevance of CD4-independent entry in vivo and its influence on pathogenesis are therefore unclear. There is, however, evidence that CD4⁻ brain astrocytes become infected by HIV-1 in vivo, particularly in pediatric AIDS patients (68, 74). A CD4-independent variant of HIV-1/IIIB selected by multiple passage in a CD4⁻ T-cell line was recently described. This virus utilized CXCR4 to infect CD4⁻ cells (36), yet substitution of the V3 loop with that from the R5 BaL strain resulted in a virus capable of CD4-independent infection via CCR5 (35). In contrast to HIV-1, T-cell-line-adapted (TCLA) strains of HIV-2 can be readily adapted to infect a subset of CD4⁻ human cell lines (14). This CD4-independent infection occurs predominantly via CXCR4, most likely reflecting the passage of these viruses through CXCR4⁺ T-cell lines (27, 63). Low-level CD4-independent infection has been reported for a single R5 HIV-2 isolate (11). For HIV-2 strains that are CD4 dependent, infection of CD4⁻ cells is often potently induced by prior treatment of virus particles with sCD4 (14). Interestingly, recombinant envelope proteins derived from some SIV_MAC strains have been shown to interact directly with CCR5 (26, 50), and infection of primary CCR5⁺ CD4⁻ brain endothelial cultures has also been reported (26).

Receptor use has profound implications for the cell tropism and pathogenesis of HIV-2 strains in vivo. For instance, if CD4-independent viruses occur or evolve in an infected individual, then such strains are likely to be able to infect a broader range of cell types at different sites in vivo. Moreover, the conformation of the envelope glycoproteins that confer a direct interaction with coreceptors may expose antigenic epitopes to neutralizing and other antibodies and thus influence the capacity of the host to control viral replication. Here, we show that many primary HIV-2 strains can infect CD4⁻ cell lines expressing either CCR5 or CXCR4. Primary cultures of CD4⁻ astrocytes were susceptible to infection by the TCLA HIV-2 variant ROD/B. Intriguingly, however, primary X4 strains only infected astrocytes if virus was treated with sCD4. These results indicate that CD4-independent infection of cell lines observed in vitro may not reflect infection of CD4⁻ cell types in vivo or may require high levels of CXCR4 expression.

MATERIALS AND METHODS

Cells.

Peripheral blood mononuclear cells (PBMCs), cultured in RPMI 1640 medium (GIBCO) supplemented with 20% fetal calf serum (FCS), 60 μg of penicillin and 100 μg of streptomycin per ml (pen/strep), were stimulated for 2 to 3 days with phytohemagglutinin (PHA; 0.5 μg/ml) and then cultured with interleukin-2 (IL-2; 20 U/ml) for 2 to 3 days prior to infection. T-cell lines H9 and Molt 4 were cultured in RPMI 1640 medium supplemented with 10% FCS and pen/strep. The human glioma cell lines U87, U87/CXCR4, U87/CD4, and U87/CD4 cells stably expressing chemokine receptors CCR1, CCR2b, CCR3, CCR5, and CXCR4 (6, 22, 86), as well as NP2, NP2/CCR5, NP2/CD4, and NP2/CD4/CCR5 (89) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; GIBCO) supplemented with 5% FCS and pen/strep. The CD4⁻ human rhabdomyosarcoma cell line RD/TE671 (79); feline kidney cell lines CCC, CCC/CXCR4, CCC/CD4, and CCC/CD4/CXCR4 (19, 77); and the human osteosarcoma cell line GHOST and the GHOST-derived lines expressing CCR1, CCR2b, CCR3, CCR5, and CXCR4 (10) were also cultured in DMEM supplemented with 5% FCS and pen/strep. Primary astrocytes prepared from fetal brain (83) were cultured in DMEM supplemented with 10% FCS, 20 mM l-glutamine per ml, pen/strep, and 17.5 μg of neomycin per ml. Astrocyte cultures were positive for the astrocytic marker glial fibrillary acidic protein (GFAP), but negative for CD4 expression and the macrophage/microglial marker CD68. The use of fetal brain samples was approved by the Royal Marsden NHS Trust Research Ethics Committee and complied with institutional and ethical regulations. Fetal brains were obtained from the Medical Research Council Tissue Bank (Hammersmith Hospital, London, United Kingdom).

Viruses.

Primary HIV strains were isolated from PHA- and IL-2-stimulated PBMCs derived from the peripheral blood of infected individuals. Isolates were minimally passaged in PBMCs from HIV-negative donors to prepare virus stocks. Stocks of TCLA viruses were produced from the CD4⁺ T-cell lines H9 for HIV-1 and HIV-2 or Molt 4 for SIV strains. Table 1 lists all of the HIV and SIV strains used in this study and provides current information on the coreceptors used by each strain to infect CD4⁺ cell lines. The coreceptors used by primary HIV-2 strains ALI, MLC, TER, ETP, JAU, MIL, and SAB were characterized in this study.

TABLE 1.

HIV-1, HIV-2, and SIV strains^a

Virus strain	Coreceptor(s) used to infect CD4+ cells (where known)b
	R5/X4	Other(s)
HIV-2
Primary
MLC	R5	(R1), (R3), (U87)
TER	R5	R1, R3, U87
ALI	R5, (X4)	(R1)
ETP	R5, X4	R1, R2b, R3
JAU	R5, X4	R1, R2b, R3
MIR	(R5), X4	(R1), (R2b), R3
prCBL-20	(R5), X4	R1, R2b, R3
prCBL-23	(R5), X4	(R1), R2b, R3
A-ND	X4	R1, R2b, R3
MIL	X4
SAB	X4	(R3)
TCLA
CBL-23	R5, X4
ROD/A (pACR23)	X4	R2b, R3, (CX3CR1)
ROD/B	X4	R2b, R3, (CX3CR1)
CBL-20	X4
SIV
SIVAGMTYO-2	R5
SIVMAC251	R5
SIVMAC32H	R5	R1, GPR15, STRL-33
SIVSMB670	R5
SIVSMg1010.2c	R5	GPR15
SIVSMswg497c
HIV-1
Primary
BR49	R5
BR92	R5
SL-2	R5
2028	R5, X4	(R3), (R8), (STRL-33), (GPR-15)
2076	R5, X4	R3, R8, STRL-33, (GPR-15)
ACH320.3.1	R5, X4
HAN-2	R5, X4
HAN-2-2mc	R5, X4
SL-12	R5, X4
2005	X4
2044	X4
TCLA
GUN-1	R5, X4
HXB2	X4
RF	X4

^aHIV and SIV strains used throughout this study are listed along with current information on coreceptors used to infect CD4⁺ cell lines (9, 23, 51, 63, 76, 77). 

^bParentheses denote inefficient use of a coreceptor, where the infectivity titer (FFU per milliliter) via this receptor is at least 2 logs lower than that of infection via either CXCR4 or CCR5 or less than 1 log above infection of parental cells. “U87” indicates that the parental U87/CD4 cells were susceptible without expression of additional coreceptors. 

^cReisolation of SIV_SMB670 following infection of rhesus macaques. 

Primary HIV-1 isolates.

The CCR5 tropic (R5) primary HIV-1 isolates used include BR49 and BR92, from Brazilian patients (23), and SL-2 from a patient from Thailand (77), all of which were subtype B viruses isolated from asymptomatic individuals. The CXCR4 tropic (X4) viruses used include strains 2005 and 2044 (77); both subtype B viruses were isolated from patients registered in England with CD4 blood cell counts of <190 cells mm⁻³. R5/X4 viruses included subtype B strains, 2028 and 2076, from English patients with CD4 counts of <190 cells mm⁻³ (77). ACH-320.3.1.mc is a molecular clone of an isolate originating from Amsterdam (31). HAN2 and HAN2-2mc (69) are a primary isolate and the corresponding molecular clone from Germany. SL-12 is a subtype E virus isolated from an asymptomatic individual from Thailand, at St. Mary’s Hospital, London, England.

TCLA HIV-1 viruses.

The TCLA viruses included X4 HXB2 (62) and RF (60), as well as R5/X4 GUN-1 (80).

Primary HIV-2 isolates.

Primary HIV-2 isolates described previously, including MIR (17), prCBL-20, prCBL23, and A-ND (51), use a broad range of coreceptors, including CCR1-3, CCR5, and CXCR4, to infect CD4⁺ cells. MIR and prCBL-20 were isolated from Guinea-Bissau and Gambian AIDS patients respectively, while prCBL-23 and A-ND were from Gambian and Portuguese symptomatic individuals, respectively. Additional primary HIV-2 strains, characterized for receptor use in this study, were all isolated from Portuguese patients with CD4 counts of <200 cells mm⁻³. ALI was isolated from a patient with AIDS-related complex. TER, JAU, MIL, and SAB were from AIDS patients, and MLC and ETP were from symptomatic patients.

TCLA HIV-2 viruses.

The TCLA X4 HIV-2 strains included ROD/A, which was generated from the CD4-dependent, infectious proviral clone of ROD, pACR23 (38). ROD plasmid DNA was transfected into RD cells, and virus progeny were seeded onto H9 cells to produce virus stocks (64). ROD was the first reported isolate of HIV-2 which originated from the Cape Verde Islands, Senegal (16). ROD/B is a CD4-independent variant derived from ROD/A following passage through C8166 cells (14). Other TCLA HIV-2 strains included CBL-20 and CBL-23 (73), which are derived from the primary isolates prCBL-20 and prCBL-23, respectively (described above). Stocks of CBL-20 and CBL-23 were produced from H9 cells.

SIV strains.

The R5 TCLA SIV strains used were SIV_MAC251 (21), SIV_MAC32H (18), SIV_SMB670 (56), and SIV_AGMTYO-2 (4). G1010.2 and swg497 are reisolations of SIV_SMB670 following infection of rhesus macaques (kindly provided by M. Murphey Corb).

Infectivity assays.

Cells were seeded into 48-well trays on the day prior to infection, at 1 × 10⁴ cells/well for U87, NP2 cells, and derivatives and 4 × 10⁴ cells/well for astrocytes, CCC cells, and derivatives. Infections were performed in duplicate, or with serial dilutions of 100 μl of cell-free virus supernatant in the absence or presence of 5 μg of baculovirus-derived sCD4 per ml. Virus was incubated with cell lines for 3 h before addition of 500 μl of growth medium. Cells were immunostained for virus expression 4 days postinfection. Astrocytes were challenged with 5 × 10³ focus-forming units (FFU) of each virus (as measured on U87/CD4/CCR5 or U87/CD4/CXCR4 cells). Viral supernatant was removed from infected astrocytes 16 h postinfection, and cells were washed four times before addition of 500 μl of culture medium. Supernatants, sampled over 26 days, were assayed for reverse transcriptase (RT) activity by an enzyme-linked immunosorbent assay (Retrosys RT activity kit; Cavidi Tech, Uppsala, Sweden). Following the final harvest of supernatant for RT analysis, astrocytes were immunostained for viral antigen expression.

Receptor ligands, tested for their ability to inhibit HIV-2 ROD/B infection of primary astrocytes, included CXCR4 ligand SDF-1α (7, 57) and the CXCR4-specific monoclonal antibody (MAb) 12G5 (27, 53) and CCR5 ligand RANTES (70) and the CD4-specific MAb Q4120, which binds domain 1 and interferes with gp120 binding (33). Briefly, primary fetal astrocytes were preincubated with ligands at 2× final concentration for 1 h before an equal volume of virus was added for 3 h. Cells were then washed three times in growth medium, and 500 μl of medium was replaced. Cultures were incubated for 3 days, fixed in methanol-acetone (1:1), and immunostained for viral antigens.

Immunostaining.

HIV-1-infected cells were immunostained for p24 antigen as previously described (14). HIV-2-infected cells were fixed for 10 min in methanol-acetone (1:1). Cells were then immunostained with serum pooled from six HIV-2⁺ individuals (World Health Organization panel C) at a dilution of 1:4,000. SIV-infected cells were immunostained with HIV-2 serum (as described above) or with a mixture of SIV envelope MAbs, KK7a and KK41 (39, 40). β-Galactosidase conjugates of antihuman or antimouse antibodies (Southern Biotechnology Associates, Inc. [dilution 1:400]) were used to detect first-layer antibodies, as appropriate. Infected cells were immunostained blue with addition of 5-bromo-4-chloro-3-indolyl-β-galactopyranoside (X-Gal; 0.5 mg/ml in phosphate-buffered saline [PBS] containing 3 mM potassium ferricyanide, 3 mM potassium ferrocyanide, and 1 mM magnesium chloride) as previously described (14). Individual or groups of blue-stained cells were regarded as foci of infection, and virus infectivity was estimated as FFU per milliliter.

Comparison of CD4-independent infection by HIV-1, HIV-2, and SIV.

Infectivity titers (FFU per milliliter) for 14 HIV-1, 15 HIV-2, and 6 SIV strains were determined on CD4⁻ cells in the presence and absence of sCD4 and compared to infectivity titers on CD4⁺ cells. Ratios of infectivity for CD4⁻ and CD4⁺ cells were calculated from virus titrations of CCC and CCC/CD4 cells transfected with either CCR5 or CXCR4 expression vectors. Ratios for some strains were determined from titers on the stable cell lines NP2/CCR5 and NP2/CD4/CCR5 and on CCC/CXCR4 and CCC/CD4/CXCR4. Background infectivity on corresponding coreceptor-negative cells was subtracted. A ratio of 1 indicates equivalent infection on CD4⁻ and CD4⁺ cells, while a ratio of 0.1 implies a 10-fold-less-efficient infection of CD4⁻ compared to that of CD4⁺ cells.

Determination of cell surface receptor expression.

Primary astrocyte cultures were analyzed for cell surface expression of CD4, CXCR4, and CCR5 by flow cytometry. CXCR4 expression on astrocytes was compared to that on CCC/CXCR4 and U87/CXCR4 cells. Cells (3 × 10⁵; preincubated in PBS–1% FCS–0.05% sodium azide for 30 min) were incubated with MAb 12G5 (27, 53) (2 μg/ml) to detect CXCR4 expression, MAb 2D7 (2 μg/ml) (87) to detect CCR5 expression, or an anti-CD4 domain 4 R-phycoerythrin conjugate (CD4 D4 R-PE; 1:10 dilution [Becton Dickinson]) to detect CD4 expression, as well as the appropriate isotype controls diluted in 100 μl of PBS–1% FCS–0.05% sodium azide for 1 h at room temperature. Cells incubated with 12G5, 2D7, and isotypes were washed twice in PBS–1% FCS–0.05% sodium azide before resuspension in 100 μl of antimouse immunoglobulin G (IgG) conjugated with fluorescein isothiocyanate (FITC; 1:40 dilution [DAKO]) for 30 min. All cells were then washed once in PBS–1% FCS–0.05% sodium azide, twice in PBS–0.05% sodium azide, resuspended in 100 μl of PBS–0.05% sodium azide, and added to 300 μl of formal saline (4% of formaldehyde in 0.5% NaCl and 1.5% Na₂SO₄) before analysis by flow cytometry.

Immunostaining for envelope expression in ROD/B-infected astrocytes.

Astrocytes were seeded, the day prior to infection, onto 8-well chamber slides (Nunc) at 4 × 10⁴ cells/well. Cells were infected with 100 μl of virus supernatant for 4 h before addition of 500 μl of culture media. Seven days postinfection, cells were fixed in methanol-acetone (1:1; 5 min), washed in PBS–1% FCS, and incubated with antibodies to HIV-2 envelope and GFAP (rat MAb 44.2g [52]; rabbit anti-cow GFAP [DAKO], 1:100). Bound primary antibodies were detected with anti-rat IgG R-PE (1:120 [Harlan Sera-Labs]) or swine F(ab′)₂ anti-rabbit FITC (1:15 [DAKO]). Immunostained cells were visualized with a fluorescent microscope.

RESULTS

HIV-2 coreceptor use on CD4⁺ cells.

Table 1 lists the HIV-1, HIV-2, and SIV strains used in this study and known coreceptors used for infection of CD4⁺ cell lines. We showed previously that several primary HIV-2 isolates use a broad range of coreceptors, including CCR5 and CXCR4 (51). A further seven primary HIV-2 isolates were analyzed for coreceptor use. Two of these (ETP and JAU) could utilize a range of receptors, including CCR1, CCR2b, CCR3, CCR5, and CXCR4. A further isolate (TER) could infect CD4⁺ cells expressing CCR1, CCR3, and CCR5. Two strains were identified that predominantly use CCR5 (ALI and MLC), and two were identified that predominantly or exclusively use CXCR4 (MIL and SAB). Coreceptor use was assessed by testing infection of a set of U87/CD4 cell lines that individually express CCR1, CCR2b, CCR3, CCR5, and CXCR4 (6, 22). Where possible, results were confirmed by using the panel of GHOST/CD4/coreceptor cell lines (10) or CCC/CD4 cells transfected with and transiently expressing each coreceptor. Assessment of coreceptor use by some HIV-2 strains is complicated by use of unidentified coreceptors expressed naturally on U87/CD4 and CCC/CD4 cells, while GHOST/CD4 cells express low levels of CXCR4 that result in background infectivity. For instance, TER infects the parental U87/CD4 cell line without expression of exogenous coreceptors; however, infection was increased 30- to 100-fold on U87/CD4 cells expressing CCR5, CCR3, and CCR1 (Table 1).

Primary HIV-2 infection of CD4⁻ cells expressing either CCR5 or CXCR4.

We tested CD4-independent infection for seven primary HIV-2 isolates, including two X4 strains (SAB and MIL), two R5 strains (ALI and MLC), and TER, ETP, and JAU, which use a broad range of coreceptors for infection of CD4⁺ cells (Table 1). Infection was compared to that of the R5 SIV_MAC32H strain and two X4 TCLA strains of the HIV-2 ROD isolate: ROD/A, which is mainly CD4 dependent; and ROD/B, which efficiently infects CXCR4⁺ CD4⁻ cell lines (63).

The CD4⁻ cell lines currently available that stably express high levels of either recombinant CXCR4 or CCR5 are CCC/CXCR4 (feline kidney), U87/CXCR4, and NP2/CCR5 (human gliomas). Figure 1A shows infectivity for CCC/CXCR4 compared to infection of the counterpart CCC/CD4/CXCR4 cell line, while Fig. 1B shows infectivity titers for NP2/CCR5 and NP2/CD4/CCR5. We also assessed the effect of sCD4 on infectivity for CD4⁻ CCC/CXCR4 and NP2/CCR5 cells. Substantial CD4-independent infection was recorded for ETP, MIL, and SAB on CCC/CXCR4 cells (Fig. 1A). Infection by these three strains was enhanced between 10- and 100-fold by sCD4. As expected, ROD/B infected CD4⁻ CCC/CXCR4 cells as efficiently as the equivalent CD4⁺ cells. ROD/A infectivity was about 1,000-fold less efficient without CD4 (accounting for background infection), but was enhanced 100-fold following sCD4 treatment. Other strains either infected only CD4⁺ cells or failed to use CXCR4 as a coreceptor and were unaffected by sCD4. As expected, no infection of the CCC/CXCR4 cells by R5 strains was detected. Similar results were obtained following infection of CD4⁻ U87/CXCR4 cells (data not shown).

FIG. 1.

Infection of CD4⁻ and CD4⁺ cells stably expressing either CCR5 or CXCR4 by primary isolates of HIV-2. Infectivity titers of primary HIV-2 isolates on CD4⁻ cells (+sCD4 or − sCD4) and CD4⁺ cells are compared to those of TCLA HIV-2 (ROD/A and ROD/B) and SIV (MAC32H). Results are representative of at least three experiments with titers expressed as FFU per milliliter + standard deviation. (A) Infection of CCC/CXCR4 and CCC/CD4/CXCR4 cells, showing substantial CD4-independent infection by MIL, SAB, ETP, and TCLA ROD/B on CXCR4⁺ cells. (B) Infection of NP2/CCR5 and NP2/CD4/CCR5 cells showing CCR5-mediated CD4-independent infection by JAU, TER, and ALI as well as SIV_MAC32H. Asterisks denote background infection in which virus infects CD4⁺ parental cells in the absence of additional coreceptor expression.

Primary HIV-2 strains that use CCR5 on CD4⁺ cells ranged from ALI (predominantly CCR5) to strains that used multiple coreceptors (Table 1). On CD4⁻ NP2/CCR5 cells, infection by three strains, ALI, JAU, and TER, was observed (Fig. 1B). Infection was comparable with that for SIV_MAC32H. sCD4 enhanced infection of NP2/CCR5 by these four strains as well as inducing infection by MLC.

Inhibition experiments with receptor ligands (AMD3100, specific for CXCR4 [24, 71, 72]); AOP-RANTES, a potent inhibitor of infection via CCR5 (75); and Q4120 and 5A8, MAbs specific for CD4 (33, 55) confirmed that CXCR4 and CCR5 were used for CD4-independent infection, but not CD4 (data not shown).

Comparison of CD4-independent infection by HIV-1, HIV-2, and SIV.

Infectivity titers of 14 HIV-1, 15 HIV-2, and 6 SIV strains were compared for CD4⁻ and CD4⁺ cells expressing either CCR5 or CXCR4. These strains included R5 and X4 viruses as well as viruses that used a range of different coreceptors, including both CCR5 and CXCR4 (Table 1). Figure 2 shows infectivity ratios for each strain calculated as the infectivity titer (FFU per milliliter) for CD4⁻ cells divided by the infectivity titer for the equivalent CD4⁺ cells. Although several viral strains could use both CCR5 and CXCR4 to infect CD4⁺ cells, CD4-independent infection occurred only via one of these receptors. Ratios for either CCR5⁺ or CXCR4⁺ cells are therefore shown (detailed as R5 or X4), as is infection with or without sCD4 treatment. Ratios for HIV-2 strains were of the same order as those for SIV strains, while HIV-1 ratios were substantially lower, with few strains able to infect CD4⁻ CCC cells that expressed CXCR4 and none able to infect CD4⁻ CCR5⁺ CCC cells. Only one primary HIV-1 strain, 2005, could infect CXCR4⁺ CCC cells, and infection was enhanced by sCD4. Another primary strain (HAN-2) and the TCLA strains RF and GUN-1 also infected CXCR4⁺ CCC cells, but only if treated with sCD4. Interestingly, RF infected CD4⁻ U87/CXCR4 cells without sCD4 (not shown), highlighting the cell type specificity of receptor dependence for this virus. These results show that primary HIV-2 strains, like SIVs, are substantially less reliant on CD4 for infection than HIV-1 strains.

FIG. 2.

Comparison of CD4-independent and CD4-dependent infection by HIV-1, HIV-2, and SIV. Infectivity plotted as ratios between titers on CD4⁻ cells (−sCD4 or +sCD4) and CD4⁺ cells expressing either CCR5 or CXCR4 (as described in Materials and Methods). The coreceptor used by R5X4 viruses for CD4-independent infection is denoted in parentheses after the strain designation. The results represent at least two experiments for each virus.

Seven HIV-2 strains (JAU, MIR, ETP, ALI, prCBL-23, CBL-23, and prCBL-20) used both CCR5 and CXCR4 to infect CD4⁺ cells (Table 1). In the absence of CD4, however, ETP and prCBL-23 could only use CXCR4 and JAU and ALI could only use CCR5, while MIR, prCBL-20, and CBL-23 used neither CCR5 nor CXCR4 efficiently. CD4-independent infection was mediated by the receptor preferentially used on CD4⁺ cells.

Susceptibility of CXCR4⁺ primary fetal astrocyte cultures to CD4-independent infection by HIV-2.

Primary cultures of CD4⁻ fetal astrocytes were analyzed for CXCR4 and CCR5 expression. Figure 3 shows flow cytometric analysis of astrocytes immunostained with 12G5, a CXCR4-specific MAb, and indicates astrocytes expressed CXCR4, but at lower concentrations compared to U87/CXCR4 and CCC/CXCR4 cells. Astrocytes were negative for CD4 and CCR5 expression, as assessed by Q4120 and 2D7 immunostaining (data not shown).

FIG. 3.

Surface expression of CXCR4 on primary fetal astrocytes and CCC/CXCR4 and U87/CXCR4 cells. Flow cytometric analysis using 12G5 (CXCR4-specific MAb) and isotype (IgG2a MAb) shows a low level of cell surface CXCR4 on astrocytes from two fetal brains (top), compared to those of CCC/CXCR4 and U87/CXCR4 cells (bottom).

The susceptibility of astrocytes to CD4-independent infection was determined. Figure 4 shows syncytia in an astrocyte culture infected with ROD/B. Costaining for the astrocyte marker, GFAP, and for gp120 indicated that late gene expression is evident in ROD/B-infected astrocytes. ROD/B infection of astrocytes was inhibited by the CXCR4 ligands MAb 12G5 and SDF-1, but not by the CD4-specific MAb Q4120 or the CCR5 ligand RANTES (Fig. 5). These results confirm that CD4-independent infection was mediated via CXCR4. Astrocytes, purified from two independent fetal brain samples, were then challenged with a panel of HIV strains by using equivalent infectivity doses for each, in the presence and absence of sCD4 (Fig. 6). Infectivity doses for astrocytes were assessed as FFU on U87/CD4/CXCR4 or U87/CD4/CCR5 cells. The HIV-2 strains tested included the two primary HIV-2 X4 strains, SAB and MIL; the R5 strain, TER (which uses CCR5, CCR1, CCR3 and other coreceptors); and the TCLA viruses ROD/A and ROD/B. We also included the primary HIV-1 X4 strain, 2005, which infected CD4⁻ CCC/CXCR4 cells, albeit inefficiently. Replication and virus production were assessed by testing supernatants for RT activity over 26 days, after which cells were fixed and immunostained for viral antigens by using HIV-2⁺ human serum. Figure 6 shows that ROD/B productively infected astrocytes from two fetal brains, albeit rather modestly. The two primary X4 strains, MIL and SAB, only showed positive replication if first treated with sCD4, indicating that astrocytes support postentry replication by primary HIV-2 strains. Astrocyte culture 2 expressed slightly higher levels of CXCR4 (Fig. 3) and when infected produced higher levels of supernatant RT compared to astrocyte culture 1 (Fig. 6B and A, respectively). These results suggests that susceptibility to infection may correlate with CXCR4 concentration on the cell surface. ROD/A replication was detected if astrocytes were challenged with a 20-fold-higher dose of virus infectivity, but only following sCD4 treatment, while 2005 required 1,000-fold more virus to initiate productive infection (data not shown). No replication was observed for the R5 HIV-2 isolate TER. Immunostaining of fixed astrocytes for viral antigens after 26 days of culture showed that the presence of infected cells correlated with detection of RT activity in the cell supernatant. Interestingly, although similar levels of RT activity were detected for ROD/B (without sCD4) as for MIL and SAB (with sCD4), many more ROD/B-infected astrocytes were observed by immunostaining at day 26 postinfection (Fig. 6C). The fewer SAB- or MIL-infected cells must produce more progeny virions than cells infected by ROD/B.

FIG. 4.

Envelope expression and syncytium formation in astrocytes infected by ROD/B. Immunofluorescence of HIV-2 ROD/B-infected fetal astrocyte cultures. (A) Expression of the astrocyte marker GFAP (green). (B) Expression of ROD/B envelope in astrocyte syncytia (red). (C) Overlay of GFAP and envelope antigen staining.

FIG. 5.

CD4-independent infection of astrocytes is inhibited by CXCR4 ligands. (A) Inhibition of ROD/B infection by the CXCR4 ligand SDF, but not the CCR5 ligand RANTES. (B) Inhibition of ROD/B infection by the CXCR4-specific MAb 12G5, but not the CD4-specific MAb Q4120. These results are consistent with CD4-independent infection of astrocytes via CXCR4. Results are representative of at least three experiments, with titers expressed as FFU per milliliter.

FIG. 6.

Susceptibility of CXCR4⁺ primary astrocytes cultures to CD4-independent infection. (A and B) Infection of two independent astrocyte cultures by primary X4 HIV-2 strains MIL and SAB, primary R5 HIV-2 strain TER, TCLA X4 HIV-2 strains ROD/A and ROD/B, and primary X4 HIV-1 strain 2005. Infections were performed in the presence and absence of sCD4. Replication was assessed by RT activity in supernatants determined on duplicate wells. (C) Astrocytes infected with ROD/B, MIL, and SAB (−sCD4 or + sCD4) and immunostained for HIV-2 antigens.

We determined if virus released from astrocytes was infectious by plating supernatants on U87/CD4 cells expressing either CXCR4 or CCR5. MIL and SAB, rescued from sCD4-induced astrocyte infections, and ROD/B, rescued from astrocytes infected in the absence or presence of sCD4, were fully infectious for U87/CD4/CXCR4 cells. No infectious virus was rescued from TER-infected astrocytes or from astrocytes infected with 2005 or ROD/A at doses equivalent to that of ROD/B (data not shown).

DISCUSSION

Some T-cell-line-passaged HIV-2 strains infect CXCR4⁺ cells without CD4 (27, 63). The first CD4-independent variant we identified (ROD/B) emerged spontaneously from a T-cell line chronically infected with the prototype HIV-2_ROD strain (14). The ROD/B envelope retains the capacity to interact with CD4, but can efficiently utilize CXCR4 alone for infection of CD4⁻ cells (63). Only two amino acid substitutions in the envelope (one at the base of the V4 loop and one near the leucine zipper-like domain in the transmembrane) were required to confer CD4 independence on ROD, although further changes (in the V3 loop and at the base of V4) increased the efficiency of infection without CD4 (64). Whether ROD/B-like strains evolve or exist in vivo and whether CD4-independent infection influences HIV-2 tropism or pathogenesis has been unclear. We show here, however, that many primary HIV-2 isolates can infect CD4⁻ cells via human CCR5 or CXCR4. CD4-independent infection via CCR5 was at levels similar to those of the CD4-independent SIV strains. Rhesus CCR5 has been shown to function more efficiently than human CCR5 as a primary receptor for SIV (25), but was not utilized in these studies. No HIV-1 strains were found to use CCR5 in the absence of CD4 (Fig. 2), although one primary HIV-1 X4 isolate (2005) infected CD4⁻ CCC cells via CXCR4. To assess whether CD4-independent and sCD4-induced infection of CD4⁻ cell lines was relevant for in vivo replication, we tested if primary HIV-2 isolates infected primary CXCR4⁺ fetal astrocyte cultures. Only the TCLA HIV-2 strain, ROD/B, infected primary astrocytes. The primary X4 strains, MIL and SAB, both of which efficiently infect CXCR4⁺ CCC and U87 cells without CD4, did not infect the astrocyte cultures, although infection was induced by sCD4. These results demonstrate that the capacity of HIV-2 strains to infect CD4⁻ cells is profoundly influenced by cell type and determined by the concentration/presentation of cell surface coreceptors and/or by currently unidentified cell surface factors.

For HIV-1, non-syncytium-inducing/R5 strains are usually transmitted. Syncytium-inducing (SI) strains that use CXCR4 can be isolated from about 50% of AIDS patients, and their emergence correlates with a more rapid decline in numbers of CD4⁺ T-cells (42). Such SI viruses either can use a range of coreceptors, including CCR5 and CXCR4, or alternatively seem to be specific for CXCR4 (76). Similarly, HIV-2 isolates that use mainly CCR5 and not CXCR4 have been identified (32, 58, 78); however, the majority of isolates use a broad range of coreceptors, including CCR5 and CXCR4 as well as coreceptors rarely used by HIV-1 (e.g., CCR1) (9, 32, 51, 58, 78). Two primary HIV-2 isolates studied here used CXCR4 only or predominantly, yet few such HIV-2 strains have been reported previously (32). These two X4 viruses were proficient for infection of CD4⁻ CXCR4⁺ cell lines.

HIV-1 strains that use both CCR5 and CXCR4 (R5X4) interact differently with CCR5 compared to R5 viruses. CCR5-dependent infection by R5X4 strains is especially sensitive both to CCR5 amino acid substitutions (5, 59) and to inhibition by the β-chemokine RANTES (41). Thus, evolution of HIV-1 from R5 to R5X4 seems to compromise the interaction of the viral envelope with CCR5. Here, CD4-independent infection by HIV-2 R5X4 strains indicated a spectrum of phenotypes, none of which were able to use both CCR5 and CXCR4. Of seven R5X4 strains, one used CCR5 only and three used CXCR4 only, while the three others used neither CCR5 nor CXCR4 efficiently for CD4-independent infection. It is possible therefore that CD4-independent infection by these R5X4 strains reflects an evolution from high-CCR5–low-CXCR4, low-CCR5–low-CXCR4 to low-CCR5–high-CXCR4 affinity. For these strains, interaction with CD4 presumably overrides lower env-7TM interactions and increases the range of coreceptors available for infection.

It is uncertain why HIV usually needs two coreceptors to enter cells, nor is it clear whether other lentiviruses or retroviruses use one or two receptors. Single receptors have been identified for murine leukemia virus (MLV); the 14-transmembrane cation transporter for ecotropic MLV (1) and the 10-transmembrane phosphate transporter, Pit-2, for amphotropic MLV (54). Gibbon ape leukemia virus and feline leukemia virus both use the related phosphate transporter, Pit-1 (37, 81). Avian leukosis subgroup A (ALV-A) viruses use a receptor related to the low-density lipoprotein receptor (90), while subgroups B and D share a tumor necrosis factor receptor-like molecule (8). Although it seems likely that these receptors are sufficient to trigger virus entry and replication, only for ALV-A is there direct evidence that the identified receptor alone is needed (20, 34). Willett et al. (85, 86) showed that cell-line-adapted strains of feline immunodeficiency virus use CXCR4 (either feline or human CXCR4) for entry, but so far no other receptor equivalent to CD4 has been identified.

We speculate that the viral ancestors of HIV and SIV originally used a 7TM receptor alone. Acquisition of a second receptor such as CD4 may have provided selective advantages to a virus that persistently replicates in the face of a vigorous host immune response. Variation in the envelope must help the virus to escape from neutralizing antibodies, but too much divergence will inevitably weaken the envelope-7TM interaction and reduce the efficiency of infection. On the HIV-1 envelope, the gp120 site for binding the 7TM receptor is exposed only after CD4 is contacted. This mechanism may enable potential neutralizing epitopes on or around the 7TM binding site to be hidden until the fusion reaction is triggered, and perhaps even then. Our results suggest that for HIV-2 and SIV, the envelope glycoproteins form a subtly different conformation compared to HIV-1, where the 7TM binding site on gp120 is at least partially exposed or formed, enabling direct contact without CD4. The role of CD4 binding for these strains is currently unclear but may (i) modify the 7TM binding site to increase the affinity of the env-7TM interaction, or (ii) contribute extra energy or a “kick” to the env-7TM contact needed to trigger fusion of viral and cell membranes. Either or both of these roles would provide HIV-2 with the capacity to exploit coreceptors that otherwise do not interact with gp120 strongly enough to trigger fusion.

Astrocytes do not express CD4 yet become infected in vivo, at least in pediatric HIV-1 AIDS cases (68, 74). Such infection is relatively unproductive, with structural gag and env genes poorly expressed. Coreceptors used for infection of astrocytes have not been identified, although glial cell lines, e.g., U87, NP2 and U373, do not usually express CXCR4 or CCR5. The primary fetal astrocytes used in this study, however, were positive for CXCR4 and supported replication by ROD/B, thus demonstrating the potential of such cells to support replication in vivo. The lack of astrocyte infection by primary X4 HIV-2 strains, in the absence of sCD4, may be due to the relatively low level of CXCR4 expression on astrocytes compared to that CXCR4⁺ CCC and U87 cell lines, although Edinger et al. recently reported that CD4-independent infection by SIV strains required only a low level of CCR5 (25). Alternatively, CXCR4 may be present on astrocytes in a different conformation than that found on the cell lines examined in this paper, as recently shown for CCR5 on different cell types (47). It has also been shown that CXCR4 may exist mainly as oligomers in macrophages, compared to monomers in monocytes, which may influence coreceptor activity (46). Additionally, posttranslational modifications such as glycosylation or sulfation may affect the efficiency of coreceptor utilization (29, 61, 82). Our results showing CD4-independent infection by primary HIV-2 strains on cell lines in vitro should therefore be interpreted with care until further studies are done to elucidate the cell types that are infected by HIV-2 in vivo. We cannot rule out a very low level of infection of astrocytes by primary HIV-2 isolates, because PCR detection or coculture with susceptible cell types was not attempted. Whether mechanisms analogous to sCD4-induced infection occur in vivo is unknown, although soluble forms of CD4 have been detected in serum (44). Astrocytes represent only one cell type that is a potential target for HIV-2 infection in vivo. Other CD4⁻ cell types expressing either CCR5 or CXCR4 may behave more like the CCC, U87, or NP2 cell lines shown here to be susceptible to HIV-2 infection without CD4. Our observations, however, show clearly that primary HIV-2 isolates (as for SIV strains) carry the potential to infect CD4⁻ cells in vivo via an interaction with CCR5 or CXCR4 that bypasses CD4.