Use of Inhibitors To Evaluate Coreceptor Usage by Simian and Simian/Human Immunodeficiency Viruses and Human Immunodeficiency Virus Type 2 in Primary Cells

Abstract

We have used coreceptor-targeted inhibitors to investigate which coreceptors are used by human immunodeficiency virus type 1 (HIV-1), simian immunodeficiency viruses (SIV), and human immunodeficiency virus type 2 (HIV-2) to enter peripheral blood mononuclear cells (PBMC). The inhibitors are TAK-779, which is specific for CCR5 and CCR2, aminooxypentane-RANTES, which blocks entry via CCR5 and CCR3, and AMD3100, which targets CXCR4. We found that for all the HIV-1 isolates and all but one of the HIV-2 isolates tested, the only relevant coreceptors were CCR5 and CXCR4. However, one HIV-2 isolate replicated in human PBMC even in the presence of TAK-779 and AMD3100, suggesting that it might use an undefined, alternative coreceptor that is expressed in the cells of some individuals. SIV_mac239 and SIV_mac251 (from macaques) were also able to use an alternative coreceptor to enter PBMC from some, but not all, human and macaque donors. The replication in human PBMC of SIV_rcm (from a red-capped mangabey), a virus which uses CCR2 but not CCR5 for entry, was blocked by TAK-779, suggesting that CCR2 is indeed the paramount coreceptor for this virus in primary cells.

Like human immunodeficiency virus type 1 (HIV-1), simian immunodeficiency viruses (SIV) and human immunodeficiency virus type 2 (HIV-2) use seven-transmembrane receptors as coreceptors during the process of virus-cell fusion (reviewed in references 6, 8, 9, 25, 74, and 87). Usually, this process requires an initial interaction of the viral envelope glycoproteins with CD4 (48, 107, 114). However, rare examples of CD4-independent HIV-1 isolates have been described (29, 47), and several HIV-2 and SIV strains can interact with coreceptors quite efficiently in the absence of CD4 (14, 30, 33, 35, 65, 89, 97). The first coreceptor described for HIV-1 was CXCR4, which serves to mediate the entry of the so-called T-cell line-tropic or syncytium-inducing (SI) viruses (39), now designated X4 isolates (7). The major coreceptor for the macrophage-tropic, non-syncytium-inducing (NSI) viruses, now designated R5 isolates, was found to be CCR5 (2, 20, 22, 27, 28). A plethora of other coreceptors has also been described to function for HIV-1 entry, especially for SI viruses, at least in the context of coreceptor-transfected cells in vitro (5, 19, 20, 22, 27, 31, 32, 38, 40, 50, 60, 62, 84, 85, 92, 93, 96, 102, 104, 105, 116).

HIV-2 isolates can also use CCR5 and CXCR4 for entry into coreceptor-transfected cells in vitro. In general, coreceptor usage by HIV-2 is broader than that for HIV-1, in that many seven-transmembrane receptors have been reported to support HIV-2 entry when transfected into cell lines (10, 14, 22, 35, 45, 48, 52, 68, 77, 80, 89, 106).

The first coreceptor to be identified as supporting SIV entry was CCR5, which was shown to function with SIV_mac isolates soon after it was found to be an HIV-1 coreceptor (12, 18, 38). The use of CCR5 by several other SIV strains, including primary isolates from the natural host, has since been documented (14, 22, 30, 31, 33, 48, 53, 54, 64, 65, 93). CXCR4 usage by SIV strains is, however, very rare, although an example of SIV entry via CXCR4 is known, albeit for an isolate obtained from mandrills (SIV_mnd) (98). However, the initial reports of SIV_mac entry via CCR5 concluded that additional coreceptors used by SIV may be substitutes for CXCR4 (12, 18, 38). Since then, several seven-transmembrane receptors have been reported to support SIV entry in vitro, often with an efficiency comparable to that of CCR5 (3, 22, 31, 32, 38, 93, 96). Arguably, the most efficient among these SIV coreceptors are the ones variously designated BOB/GPR15 and Bonzo/STRL33/TYMSTR (3, 22, 31, 62).

The question arises, however, as to whether these other coreceptors are as important as CCR5 and (for HIV-1 and HIV-2) CXCR4 for viral replication in vivo. There is mounting evidence that many, if not all, of the other coreceptors have only limited, if any, relevance to viral replication in primary cells and hence in vivo, except perhaps in specialized tissues and cell types (13, 31, 43, 70, 81, 86, 101, 102, 117). Care must always be taken when evaluating viral entry mediated via transfected seven-transmembrane receptors (66). We have continued to address this issue here using inhibitors targeted at CCR5 and CXCR4 (117). Our conclusion is that CCR5 is the most important SIV_mac coreceptor in primary CD4⁺ T cells but that an alternative coreceptor(s) may indeed be relevant, at least in cells from some macaques. SIV_rcm, however, uses CCR2 and not CCR5. These observations may be useful in studies of SIV-infected nonhuman primates as model systems for the development of HIV-1 vaccines (23, 57, 73). For the same reason, we have also studied the coreceptor usage of selected simian/human immunodeficiency viruses (SHIV).

MATERIALS AND METHODS

Coreceptor inhibitors.

The bicyclam AMD3100, a small-molecule inhibitor of HIV-1 entry via CXCR4 (26, 37, 56, 100), and TAK-779, a small-molecule inhibitor of HIV-1 entry via CCR5 (4), were both gifts from Annette Bauer, Michael Miller, Susan Vice, Bahige Baroudy, and Stuart McCombie (Schering Plough Research Institute, Bloomfield, N.J.). Aminooxypentane-RANTES (AOP-RANTES), a derivatized CC-chemokine that interacts with CCR5, was provided by Amanda Proudfoot, Serono Pharmaceutical Research Institute, Geneva, Switzerland (34, 63, 103, 113). The human chemokines monocyte chemotactic peptide (MCP) 1 (MCP-1), MCP-3, and stromal cell-derived factor 1α (SDF-1α) were purchased from Peprotech Inc. (Norwood, Mass.).

Viral isolates and preparation of virus stocks.

The HIV-1 primary isolates 5160 and 5073, derived from individuals with AIDS, have been described previously (115), as have two other primary isolates, M6-v3 and P6-v3, obtained from an HIV-1-infected mother-child transmission pair (116, 117). All these viruses have the SI phenotype, except for P6-v3. Six HIV-2 primary isolates have also been described elsewhere (41, 80). Three of these (7924A, 77618, and GB122) were isolated from individuals with AIDS, one (7312A) was isolated from an individual with lymphadenopathy, and two (310340 and 310342) were isolated from blood donors whose clinical conditions were unrecorded (80). The origins of HIV-1 SF162, DH123, and NL4-3 have been described elsewhere, as have their coreceptor usage profiles (116, 117). All HIV-1 and HIV-2 isolates were propagated and titrated in phytohemagglutinin-activated human peripheral blood mononuclear cells (PBMC) before use.

The SIV strains SIV_mac251, SIV_mac239, SIV_mac251/1390, SIV_mac239/5501, SIV_sm (variant SIV_smpbj), and SIV_rcm were all provided by Preston Marx and Zhiwei Chen (14, 15). SIV_mac251/1390 and SIV_mac239/5501 were isolated from macaques which progressed to AIDS after infection with SIV_mac251 and SIV_mac239, respectively (14, 67). SIV_rcm was originally isolated from a red-capped mangabey by cocultivation with human PBMC (15). All SIV strains were propagated and titrated in rhesus macaque PBMC, except for SIV_rcm, for which human PBMC were used (15).

SHIV strains 89.6, 89.6P, and 89.6PD were obtained from David Montefiori (90, 91). SHIV strain SF33A was obtained from Cecilia Cheng-Mayer (46), and SHIV strain KU-2 was obtained from Opendra Narayan (51). All SHIV stocks were prepared in macaque PBMC, except for a second stock of 89.6PD, which was grown in human PBMC for comparison (89.6PD-hu).

Virus replication in PBMC.

Human PBMC were isolated from various healthy blood donors by Ficoll-Hypaque separation and stimulated for 3 days with phytohemagglutinin (5 μg/ml) and interleukin-2 (IL-2; 100 U/ml) (a gift from Hofmann-La Roche, Inc., Nutley, N.J.). These donors were all homozygous for the CCR5 wild-type allele. PBMC from three individuals known to be homozygous for the CCR5 Δ32 allele (Δ32-CCR5) were also used. Activated PBMC (2 × 10⁵/well) were cultured in 96-well plates with 150 μl of RPMI 1640 medium containing 10% fetal calf serum and IL-2. Virus inocula (100 or 1,000 50% tissue culture infective doses [TCID₅₀] in 75 μl) were added to duplicate or triplicate wells. Three wells lacked cells to provide a control for the viral antigen input.

Rhesus macaque PBMC were prepared by similar procedures, except that they were stimulated for 3 days with staphylococcal enterotoxin B (Sigma Chemical Co., St. Louis, Mo.) at 5 μg/ml in RPMI 1640 growth medium containing IL-2 (46).

CEMx174 cells in RPMI 1640 growth medium were used at concentrations of 4 × 10⁴/well. Culture supernatants were harvested on days 7 and 11 postinfection, and fresh medium was added to replenish the cultures.

Viral antigen detection.

Virus production was measured using a Gag antigen capture enzyme-linked immunosorbent assay. A commercial diagnostic kit (Cellular Products Inc., Buffalo, N.Y.) was used, with modifications, to quantitate HIV-2 and SIV p27 antigen. Briefly, p27 antigen in a 100-μl volume was captured onto wells of a 96-well plate by the adsorbed anti-p27 monoclonal antibody provided with the kit. The captured p27 antigen was then detected using the biotin-labeled anti-SIV Gag polyclonal antibodies provided with the kit. To increase the sensitivity of antigen detection, we used a modified protocol that involved streptavidin-conjugated alkaline phosphatase (DAKO, Carpinteria, Calif.) and a chemiluminescent alkaline phosphatase substrate (ELISA-Light; Tropix Inc., Bedford, Mass.). The plates were read with a microtiter plate luminometer (Dynex Technologies Inc.), and the amount of antigen detected was calculated using a standard antigen curve prepared in each assay. The use of the chemiluminescent detection system increased the sensitivity of HIV-2 or SIV p27 detection by more than 100-fold. HIV-1 p24 antigen was detected as described previously (109, 111), except that the chemiluminescent detection system was used.

Determination of coreceptor usage by viral isolates using GHOST cells expressing CD4 and coreceptors.

Coreceptor usage was determined essentially as described previously (109, 116, 117). Human osteosarcoma (GHOST) cells expressing CD4 and one of the following coreceptors were obtained from Dan Littman and Vineet KewalRamani (Skirball Institute, New York University School of Medicine, New York, N.Y.): CCR1, CCR2, CCR3, CCR4, CCR5, CCR8, CXCR4, BOB, Bonzo, GPR1, APJ, V28, and US28. These cells were cultured in complete Dulbecco's minimal essential medium containing G418 (5 μg/ml), hygromycin (1 μg/ml), and puromycin (1 μg/ml). GHOST cells expressing only CD4 (GHOST-CD4 cells) served as controls; they were cultured in the same medium, except that puromycin was omitted.

GHOST cells (10⁵/ml; 500 μl per well) were maintained in 24-well plates for 24 h. The medium was then removed, and 200 μl of fresh medium was added, along with a viral inoculum of 1,000 TCID₅₀. On the next day, residual virus was removed and the cells were washed once with 1 ml of medium. A 750-μl aliquot of fresh complete medium containing the selection antibiotics was then added. At approximately day 5 postinfection, Gag antigen production in 100 μl of harvested culture supernatant was measured. For a few slowly replicating SIV isolates, it was necessary to replenish the cultures and repeat the antigen assay on day 7 or 10 postinfection. In all cases, the amount of antigen produced in control GHOST-CD4 cells was subtracted from the amount produced in coreceptor-transfected GHOST-CD4 cells. Whether this is a sufficient correction for use by some isolates of the low level of endogenous CXCR4 in GHOST-CD4 cells is discussed in Results. Attempts were made to quantify CXCR4 expression on the various coreceptor-transfected GHOST-CD4 cell lines. All the lines do express CXCR4, but at very low levels that are difficult to quantify accurately by fluorescence-activated cell sorting (FACS). Thus, we could not accurately quantitate the extent to which CXCR4 expression varied among the various lines. This situation is consistent with the experience of others (Dan Littman, personal communication).

Effect of coreceptor-targeted inhibitors on viral replication.

Human PBMC were used with HIV-1, HIV-2, and SIV_rcm, rhesus macaque PBMC were used with other SIV isolates, and both human and macaque PBMC were used with SHIV. Stimulated PBMC (75 μl) were cultured in 96-well plates at 2 × 10⁵ per well for human cells and 1 × 10⁵ per well for macaque cells. A range of concentrations of inhibitors (75 μl) was incubated with the cells, in duplicate or triplicate wells, for 1 h at 37°C before addition of the viral inoculum (100 TCID₅₀ in 75 μl). The final inhibitor concentrations used, unless otherwise specified, were as follows: AMD3100, 400, 40, and 4 nM; AOP-RANTES, 40, 4, and 0.4 nM; TAK-779, 3.3 μM, 330 nM, and 33 nM; and MCP-1 and MCP-3, 400, 40, and 4 nM. For each virus tested, five wells without drugs and five wells containing only virus served as positive and negative controls for virus production, respectively. Culture supernatants (200 μl) were harvested for measurement of Gag antigen content (in 100 μl) by an enzyme-linked immunosorbent assay on days 4, 7, and 10. Inhibitors were added back each time. Only when sufficient antigen had been produced was the effect of the inhibitors on virus production calculated.

To determine the specificity of the inhibitors, GHOST-CD4 cells and a coreceptor were used. The cells were cultured as described above. Briefly, 24 h after the cells were plated, inhibitors in a total volume of 200 μl were added to each well of a 24-well plate. AMD3100 was used at 1.2 μM, AOP-RANTES was used at 120 nM, and TAK-779 was used at 10 μM. After incubation for 1 h at 37°C, a viral inoculum of 1,000 TCID₅₀ was added for overnight incubation. The cells were then washed, and 750 μl of fresh medium was added. The production of p24 antigen and the effect of the inhibitors were determined as for the PBMC cultures, except that the supernatants were harvested on days 3, 6, and 10.

RESULTS

Coreceptor usage by HIV-1, HIV-2, SHIV, and SIV in transfected cells.

We assembled a panel of HIV-1, HIV-2, SHIV, and SIV isolates to study their replication in primary cells. We first determined which coreceptors these viruses could use, at least under artificial conditions, by measuring their replication in human GHOST-CD4 cell lines stably transfected with one of several seven-transmembrane receptors (Table 1).

TABLE 1.

Coreceptor usage by HIV-1, SHIV, HIV-2, and SIV isolates in GHOST-CD4 cells expressing a transfected seven-transmembrane, G-protein-coupled receptor

Viral isolate	Inoculum (TCID50)	Replication in the presence of the following receptora:
		CCR1	CCR2	CCR3	CCR4	CCR5	CCR8	CXCR4	BOB	Bonzo	GPR1	V28	APJ	US28
HIV-1 P6-v3	1,000	−	−	−	−	+++	−	−	−	+++	−	−	−	−
HIV-1 P6-v3	1,000		−	−	−	+++	+	+++	−	+++	−	+	+	−
HIV-1 5073	1,000	+	−	−	−	−	−	++	−	−	−	+	+	+
HIV-1 5160	1,000	++	−	−	−	−	−	++	−	−	−	+	+	+
HIV-1 NL4-3	1,000	−	−	−	−	−	−	+++	−	−	−	−	−	−
SHIV 89.6PD	500	−	−	+++	−	+++++	+++	+++++	−	−	−	++++	++	+++
SHIV 89.6PD-hu	500	−	−	++	−	+++++	++	+++++	−	−	−	+++	++	++
SHIV KU-2	500	−	−	−	−	−	−	++++	−	−	−	+++	++	++
SHIV SF33A	500	−	−	−	−	−	−	+++	−	−	−	++	+	−
HIV-2 310340	1,000		−	−	−	+++++	−	−	−	−	−	−	−	−
HIV-2 310342	1,000		−	−	−	+++	−	−			−	−	−	−
HIV-2 7312A	1,000		−	−	−	+++	−	−	+	+	−	−	−	−
HIV-2 GB122	1,000	++	−	−	−	−	−	++++	−	−	−	++	++	++
HIV-2 77618	1,000	++	−	−	−	−	−	+++	−	−	−	++	++	+
HIV-2 7924A	1,000	++	−	−	−	−	−	+++++	−	−	−	++	++++	+++
SIVrcm	100	−	+++++	−	−	−	−	−	−	+++	−	++++	−	+++++
SIVrcm	500	−	+++++	−	−	−	−	−	−	+++++	−	+++++	−	+++++
SIVmac239	500	−	−	−	−	++++	−	−	++	+++	+	−	++	−
SIVmac251	500	−	−	−	−	+++	−	−	+	+	−	−	−	−
SIVmac239/5501	500	−	−	−	−	++++	−	−	++++	++++	+++	−	+++	−
SIVmac251/1390	500	−	−	−	−	+++	−	−	+	+	−	−	−	−
SIVsmpbj	500	−	−	−	−	++++	−	−	−	−	−	−	−	−

^aAbility to replicate in GHOST-CD4 cells expressing the seven-transmembrane, G-protein-coupled receptor indicated. The extent of replication (Gag antigen production) is recorded as follows: −, <0.1 ng/ml; +, 0.1 to 1 ng/ml; ++, 1 to 5 ng/ml; +++, 5 to 20 ng/ml; ++++, 20 to 100 ng/ml; and +++++, >100 ng/ml. For each CXCR4-utilizing virus, the amount of p24 antigen produced in the parental GHOST-CD4 cells was subtracted from the amount produced in the coreceptor-expressing GHOST-CD4 cells. Whether this is always a sufficient correction for the use of the CXCR4 that is endogenous to GHOST-CD4 cells is discussed in the text. 

CCR5 and CXCR4 were clearly the coreceptors most widely and efficiently used by HIV-1, HIV-2, and SHIV isolates. None of the SIV used CXCR4, a feature that distinguishes SIV from HIV-1 and HIV-2 (3, 10, 22, 31, 32, 34, 38, 45, 48, 68, 77, 80, 89, 93, 96, 106), but all the SIV except for SIV_rcm used CCR5 (Table 1). SIV_rcm was originally isolated from a red-capped mangabey, a monkey species with a high frequency of a mutated, inactive CCR5 gene, the Δ24-CCR5 allele (15). SIV_rcm has a unique pattern of coreceptor usage in that it uses CCR2 and not CCR5 as its major coreceptor (15). We confirmed this fact and found that SIV_rcm can also use Bonzo/STRL33, V28, and US28 efficiently (Table 1).

Consistent with previous reports, some HIV-2 and SIV isolates were able to enter cells expressing several other coreceptors (3, 10, 22, 31, 32, 34, 38, 45, 48, 68, 77, 80, 89, 93, 96, 106). For instance, some SIV isolates were able to use BOB/GPR15 and Bonzo/STRL33 efficiently—notably, SIV_mac239/5501 (Table 1). HIV-1 and SHIV isolates of the SI phenotype, i.e., viruses that could use CXCR4 efficiently, were usually able to replicate in GHOST-CD4 cells expressing various coreceptors (Table 1). Any differences in coreceptor usage patterns between this and previous reports (80, 115) probably arises from the use of different GHOST-CD4 cell clones and/or isolates with a different passage history.

The broad tropism of SI viruses in coreceptor-transfected cell lines is well known (5, 19, 20, 22, 27, 31, 32, 38, 40, 50, 60, 62, 84, 85, 92, 93, 96, 104, 105, 116). However, the growth of HIV-1, HIV-2, and SHIV isolates in GHOST-CD4 cells transfected with other coreceptors was only rarely comparable to the replication of the same viruses in CCR5- or CXCR4-expressing cells. Examples of relatively efficient replication include that of HIV-1 P6-v3 and M6-v3 in Bonzo-transfected cells, HIV-2 7924A in APJ- or US28-transfected cells, and SHIV 89.6PD in V28-transfected cells (Table 1). Whether virus entry into the various GHOST-CD4 cell lines actually occurs via the transfected coreceptor is discussed below.

Replication of HIV-1, HIV-2, SHIV, and SIV isolates in PBMC from donors expressing or not expressing CCR5 and in CEMx174 cells.

The above experiments showed that many of the test isolates can apparently use multiple coreceptors to enter transfected human cell lines. To gain insights into the importance of CCR5 for viral replication in primary cells, we compared the abilities of the isolates to replicate in human PBMC from either donors who had wild-type CCR5 alleles or donors who were homozygous for the Δ32-CCR5 mutation and so did not express functional CCR5 proteins (21, 61, 95). We also used the CEMx174 human B/T-hybrid line because these cells can support high-level SIV replication. CEMx174 cells are CXCR4⁺ but CCR5⁻ (18, 58, 110) and strongly express the SIV_mac251 and SIV_mac239 coreceptor BOB/GPR15 (22, 31, 86).

Among the six HIV-1 isolates tested, the R5, NSI viruses SF162 and P6-v3 were unable to replicate in the Δ32-CCR5 PBMC from donor 1 (Table 2). Similar results were obtained with PBMC from two other Δ32-CCR5 donors (data not shown; see also Table 5). These observations are consistent with the known dependence of these viruses on CCR5, so they validate the use of the Δ32-CCR5 cells for subsequent studies of HIV-2 and SIV replication. HIV-1 SF162 and P6-v3 also failed to replicate in CEMx174 cells (Table 2). In contrast, the X4 HIV-1 clone NL4-3 and the multitropic HIV-1 isolates DH123, M6-v3, and 5073 all replicated in both wild-type and Δ32-CCR5 PBMC (donor 1) as well as in CEMx174 cells. This finding was also true of the three SHIV tested, 89.6PD, KU-2, and SF33A (Table 2). Hence, all seven of these HIV-1 and SHIV isolates can use a coreceptor other than CCR5 to enter PBMC and CEMx174 cells, consistent with their replication patterns in the various GHOST-CD4 cell lines (Table 1).

TABLE 2.

Replication of HIV-1 and SHIV isolates in PBMC from wild-type and Δ32-CCR5 donors and in CEMx174 cells

HIV-1 or SHIV isolate	Virus (ng of p24 or p27 antigen/ml) produced in the following cells on the indicated day postinfectiona:
	Wild-type PBMC		Δ32-CCR5 PBMC		CEMx174
	7	11	7	11	7	11
SF162	15.2	13.4	0	0	0	0
P6-v3	10.2	14.4	0	0	0	0
NL4-3	2.6	10.1	5.7	14.3	8	17.3
DH123	14.1	11.6	13.2	6.2	7.3	15
M6-v3	12.6	13.9	6.5	15.5	4.1	18.2
5073	7	13.2	5.6	13.8	2.9	17.3
89.6PD	85.2	134.2	53.3	154.3	41.5	181.9
KU-2	37	102	15.7	70	94	134
SF33A	73	131	72	107	51.4	155

^aThe inoculum was 1,000 TCID₅₀, but an identical pattern of data was found at 100 TCID₅₀ (data not shown). PBMC were from donors A (wild type) and 1 (Δ32-CCR5). 

TABLE 5.

Replication of SIV and HIV-1 isolates in PBMC from wild-type and Δ32-CCR5 donors and in CEMx174 cells in two different experiments

SIV or HIV-1 isolate	TCID50	Virus (ng of p24 or p27 antigen/ml) produced in the following cells on the indicated day postinfectiona:
		Wild-type PBMC (D)		Δ32-CCR5 PBMC (1)		Δ32-CCR5 PBMC (2)		Wild-type PBMC (E)		Δ32-CCR5 PBMC (3)
		7	11	7	11	7	11	7	11	7	11
SIVmac239	100	10.9	87.7	0.2	0.5	3	44.2	34	383	8.7	201
	500	27.4	112	0.5	0.8	7.7	53.4	134	460	112	298
	1000	37.6	105	1.6	3.4	7.6	52	ND	ND	ND	ND
SIVmac251	100	4.4	146	0.5	1.4	0.6	9.9	85	481	46	309
SIVrcm	100	154	707	65	710	55	637	ND	ND	294	358
SF162	100	18.1	16.8	0	0	0	0	0	0	0	0

^aTwo experiments are recorded: one that compared donors D, 1, and 2 and the other that compared donors E and 3 (donor designations in parentheses after cell types). ND, not done. 

One of the five HIV-2 isolates tested, 310340, failed to replicate in Δ32-CCR5 PBMC from donor 1 and in CEMx174 cells (Table 3). This virus was also unable to use any coreceptor other than CCR5 to enter GHOST-CD4 cells (Table 1). Another HIV-2 isolate, 7312A, grew very poorly, but detectably, in Δ32-CCR5 PBMC and CEMx174 cells; the extent of 7312A production in Δ32-CCR5 PBMC was 5 to 10% that in wild-type PBMC (Table 3). Of note is that HIV-2 7312A could use BOB/GPR15 and Bonzo/STRL33 inefficiently; the amount of p24 produced from GHOST-CD4 cells expressing BOB or Bonzo was approximately 5% that derived from GHOST-CD4 cells expressing CCR5 (Table 1; also data not shown). The remaining three HIV-2 isolates, GB122, 77618, and 7924A, all replicated to comparable extents in the wild-type and Δ32-CCR5 PBMC and replicated efficiently in CEMx174 cells (Table 3). These results are consistent with the ability of these three isolates to use CXCR4 and other coreceptors (Table 1).

TABLE 3.

Replication of HIV-2 isolates in PBMC from wild-type and Δ32-CCR5 donors and in CEMx174 cells

HIV-2 isolate	TCID50	Virus (ng of p27 antigen/ml) produced in the following cells on the indicated day postinfectiona:
		Wild-type PBMC		Δ32-CCR5 PBMC		CEMx174
		7	11	7	11	7	11
310340	100	1,250	1,509	0	0	0	0
	1,000	1,329	1,717	0	0	0	0
7312A	100	2.3	10.3	0	0.7	0.4	0.8
	1,000	15	69.8	0.2	5.5	1.5	8.9
GB122	100	58	87	85	130	121	114
	1,000	82.9	105	108	133	116	135
77618	100	37.7	58.2	42.6	159	167	186
	1,000	125	182	113	195	168	237
7924A	100	88	141	78	141	29.9	89
	1,000	84	68	91	99	59	161

^aPBMC were from donors B (wild type) and 1 (Δ32-CCR5). 

The replication of SIV_mac239 and SIV_mac251 in human PBMC from an individual homozygous for the Δ32-CCR5 allele has been taken as strong evidence that these viruses can use a coreceptor other than CCR5 to enter primary, CD4⁺ cells (18). We sought to confirm this. In the first experiment, the extent of SIV_mac239, SIV_mac251, SIV_mac239/5501, and SIV_mac251/1390 replication in Δ32-CCR5 PBMC from donor 1 was never more than 5% and usually was less than 1% the replication of the same viruses in wild-type PBMC (Table 4). A second experiment also included Δ32-CCR5 PBMC from two more donors, 2 and 3. There was, again, little or no production of SIV_mac251 and SIV_mac239 in Δ32-CCR5 PBMC from donor 1 (Table 5). However, both isolates replicated well in PBMC from Δ32-CCR5 donors 2 and 3, although antigen production from SIV_mac251 in cells from donor 2 was lower than that from typical CCR5 wild-type donors (Table 5). Thus, PBMC from some, but not all, human donors must express a coreceptor other than CCR5 that can be used with reasonable efficiency by members of the SIV_mac group of viruses.

TABLE 4.

Replication of SIV isolates in PBMC from wild-type and Δ32-CCR5 donors and in CEMx174 cells

SIV isolate	TCID50	Virus (ng of p24 or p27 antigen/ml) produced in the following cells on the indicated day postinfectiona:
		Wild-type PBMC		Δ32-CCR5 PBMC		CEMx174
		7	11	7	11	7	11
SIVmac239	100	69.9	414.8	0.4	4.1	146	444
	500	373	289	5.2	1.9	209	555
	1,000	303	381	8	5.6	755	470
SIVmac239/5501	100	394	366	4.5	7	923	455
SIVmac251	100	121	606	6.8	6.8	47	402
SIVmac251/1390	100	22	299	0.9	0.5	20.9	442
SIVrcm	100	2,536	2,413	266	678	12	4
	1,000	7,550	2,975	5,016	2,707	79	46

^aPMBC were from donors C (wild type) and 1 (Δ32-CCR5). 

SIV_rcm replicated efficiently in wild-type and Δ32-CCR5 PBMC (Tables 4 and 5), consistent with its lack of dependence on CCR5 for entry into PBMC (15). SIV_mac239 and SIV_mac251 also replicated efficiently in CEMx174 cells, as found previously (18), but SIV_rcm replication was inefficient in these cells (Table 4). Thus, neither the major coreceptor for SIV_rcm, CCR2, nor the minor ones Bonzo, US28, and V28 are expressed in CEMx174 cells.

Evaluation of the specificity of coreceptor-targeted inhibitors.

Coreceptor-targeted inhibitors are useful for evaluating which coreceptors are relevant for viral entry into PBMC. One suitable inhibitor of entry via CXCR4 is the bicyclam AMD3100 (26, 37, 56, 100). Inhibitors of entry via CCR5 are the TAK-779 molecule (4) or the CC-chemokine derivative AOP-RANTES (62, 103, 113). The specificity of these agents is an important issue. Previous studies have found that AMD3100 is specific for CXCR4 (26, 37, 56, 100) and that TAK-779 can interact with both CCR5 and CCR2 (4). Although RANTES fully activates all of its receptors, AOP-RANTES is able to do this only for CCR5; it has half the activity of RANTES for CCR3 and is very inefficient at activating CCR1 (79, 88). AOP-RANTES is therefore a moderate inhibitor of CCR3-mediated HIV-1 infection, compared to its effect on entry mediated by CCR5 (34).

To confirm these specificities, we determined whether AMD3100, TAK-779, and AOP-RANTES could inhibit viral entry into GHOST-CD4 cells transfected with other coreceptors by using viruses that were broadly tropic in these cells. For each test virus, AMD3100 was used at 1.2 μM, AOP-RANTES was used at 120 nM, and TAK-779 was used at 10 μM (Fig. 1).

FIG. 1.

Testing of the specificity of coreceptor-targeted inhibitors. The replication of the test viruses in GHOST-CD4 cells expressing the coreceptor indicated in the presence and absence of AMD3100 (1.2 μM), AOP-RANTES (AOP-R) (120 nM), TAK-779 (10 μM), or SDF-1α (500 nM) was evaluated. The extent to which replication was inhibited by each agent was recorded.

SHIV 89.6PD replication in GHOST-CD4 cells expressing CCR5 was sensitive to both TAK-779 and AOP-RANTES but not to AMD3100, as expected (Fig. 1a). We also found that AOP-RANTES, but not TAK-779, inhibited SHIV 89.6PD entry into GHOST-CD4 cells expressing CCR3, consistent with an interaction between AOP-RANTES and CCR3, a known RANTES receptor (data not shown). However, neither TAK-779 nor AOP-RANTES had any significant effect on SHIV 89.6PD replication in GHOST-CD4 cells expressing CXCR4, CCR8, V28, US28, or APJ (Fig. 1a; also data not shown). Both TAK-779 and AOP-RANTES inhibited SIV_mac239 entry into GHOST-CD4 cells expressing CCR5, but the entry of this virus into GHOST-CD4 cells expressing either BOB, Bonzo, GPR1, or APJ was unaffected by TAK-779 or AOP-RANTES (Fig. 1a). The entry of SIV_rcm into GHOST-CD4 cells expressing CCR2 was completely inhibited by TAK-779, whereas AOP-RANTES had only a marginal effect on entry via CCR2 (Fig. 1a). SIV_rcm replication in GHOST-CD4 cells expressing Bonzo, V28, or US28 was, however, insensitive to TAK-779 or AOP-RANTES (Fig. 1a), as was HIV-2 7924A replication in cells expressing V28, APJ, or US28 (data not shown). Neither TAK-779 nor AOP-RANTES inhibited the replication of HIV-1 P6-v3 in GHOST-CD4 cells expressing Bonzo (data not shown).

Taken together, these data suggest that AOP-RANTES can block viral entry via CCR5 and CCR3 and that TAK-779 inhibits entry via CCR5 and CCR2. The latter result is consistent with the report that TAK-779 binds to both CCR2 and CCR5 but not to CCR1, CCR3, or CCR4 (4). TAK-779 and AOP-RANTES have no effect on viral replication in GHOST-CD4 cells expressing any one of the eight coreceptors that we were able to evaluate: CXCR4, CCR8, V28, US28, APJ, BOB, Bonzo, or GPR1.

A less clear-cut pattern of inhibition was observed with AMD3100. As expected, the efficient replication of SHIV 89.6PD in GHOST-CD4 cells expressing CXCR4 was blocked by AMD3100 (Fig. 1a). However, AMD3100 also prevented the inefficient replication of SHIV 89.6PD in GHOST-CD4 cells expressing either CCR3, V28, APJ, US28, or CCR8 and significantly inhibited the limited replication of HIV-2 7924A in GHOST-CD4 cells expressing V28, APJ, or US28 (Fig. 1b; also data not shown). However, AMD3100 had no detectable effect on SIV_mac239 entry into GHOST-CD4 cells expressing BOB, Bonzo, GPR1, or APJ or on SIV_rcm entry into GHOST-CD4 cells expressing CCR2, Bonzo, V28, or US28 (Fig. 1a). The replication of HIV-1 P6-v3 in GHOST-CD4 cells expressing Bonzo was also unaffected by AMD3100 (data not shown). Thus, entry via V28, US28, and APJ in GHOST-CD4 cell lines can apparently be either sensitive or insensitive to AMD3100, depending upon the test virus.

There are two possible explanations for the unusual pattern of inhibition shown by AMD3100. One is that AMD3100 is broadly reactive with multiple coreceptors but that certain viruses, particularly SIV, can still interact with some of these coreceptors even in the presence of AMD3100. The other is that the apparent cross-reactivity of AMD3100 is an artifact of the presence of low levels of endogenous CXCR4 in coreceptor-transfected GHOST-CD4 cells (109, 110). To address this possibility, we tested the sensitivity of SHIV 89.6PD and HIV-2 7924A replication in several GHOST-CD4 cell lines to SDF-1α. In all cases, whenever AMD3100 inhibited the replication of the test viruses, so did SDF-1α (Fig. 1b; also data not shown). Since SDF-1α is specific for CXCR4 (6, 8, 9, 71, 84), these findings strongly suggest that the entry of SHIV 89.6PD and HIV-2 7924A into several coreceptor-transfected GHOST-CD4 cell lines occurs via endogenous CXCR4. This coreceptor may well be expressed to different levels in different individual GHOST-CD4 cell lines, although we were unable to accurately quantitate this expression by FACS.

The inhibitory effect of AMD3100 in coreceptor-transfected GHOST-CD4 cell lines is, therefore, most probably explained by its antagonism of viral entry via endogenous CXCR4. The coreceptor usage information presented in Table 1 should be interpreted with this caveat in mind. Overall, we can find no evidence that AMD3100 is anything other than specific for CXCR4.

Effect of coreceptor-targeted inhibitors on HIV-1, SHIV, and HIV-2 replication in PBMC.

The replication of each test virus in mitogen-stimulated PBMC in the presence and absence of AMD3100, TAK-779, or AOP-RANTES was evaluated. Combinations of AMD3100 with TAK-779 and AMD3100 with AOP-RANTES were also tested. Each inhibitor, alone and in combination, was used at three different concentrations: 400, 40, and 4 nM for AMD3100; 3.3 μM, 330 nM, and 33 nM for TAK-779; and 40, 4, and 0.4 nM for AOP-RANTES. Preliminary experiments had indicated that the effects of the inhibitors usually titrated out over these ranges. Human PBMC from CCR5 wild-type donors were used in experiments with HIV-1 and HIV-2 isolates and SIV_rcm; rhesus macaque PBMC were used with other SIV; and both human and macaque PBMC were used with SHIV.

Four HIV-1 primary isolates that could use multiple coreceptors, as determined by the GHOST-CD4 cell assays (Table 1), were evaluated with human PBMC (Fig. 2). P6-v3, a virus able to use CCR5 and Bonzo, was completely inhibited by both TAK-779 and AOP-RANTES but not by AMD3100 (Fig. 2a). The more broadly tropic virus M6-v3 was partially sensitive to each of the three inhibitors, but its replication was fully blocked by combinations of either TAK-779 or AOP-RANTES with AMD3100 (Fig. 2a). Isolates 5073 and 5060 were able to replicate in several different coreceptor-expressing GHOST-CD4 cell lines, including GHOST-CD4 cells expressing CXCR4, but their replication was completely inhibited in PBMC by AMD3100 (Fig. 2b). Thus, none of the tested HIV-1 isolates appeared to enter PBMC from the donors included in these studies via a coreceptor other than CCR5 or CXCR4.

FIG. 2.

Effects of coreceptor-targeted inhibitors on HIV-1 replication in human PBMC. The replication of the HIV-1 isolates P6-v3 and M6-v3 (a) and 5073 and 5160 (b) in human PBMC in the presence and absence of AOP-RANTES (AOP-R) (40 nM [left bar], 4 nM [middle bar], and 0.4 nM [right bar]), TAK-779 (3.3 μM, 330 nM, and 33 nM), or AMD3100 (400 nM, 40 nM, and 4 nM) or with combinations of AMD3100 and either AOP-RANTES or TAK-779 was evaluated. When combinations were used, the concentration of each agent was the same as when the agents were used alone. The extent to which replication was inhibited by each agent or combination was recorded. The coreceptors that can be used by each isolate in GHOST-CD4 cells are indicated below the isolate designation in parentheses.

Results similar to those obtained with the broadly tropic HIV-1 isolates were found when SHIV were evaluated (Fig. 3). Thus, SHIV 89.6PD replication in either macaque or human PBMC was fully inhibited by AMD3100, while TAK-779 and AOP-RANTES had no effect (Fig. 3a). The same was true of SHIV KU-2 and SHIV SF33A in human PBMC (Fig. 3b) and also of SHIV 89.6 and SHIV 89.6P (data not shown). The paramount, and most probably exclusive, coreceptor for all of these SHIV in PBMC therefore appears to be CXCR4. This finding was unexpected for SHIV 89.6, 89.6P, and 89.6PD, considering that these viruses efficiently use CCR5 in transfected GHOST-CD4 cells (Table 1 and Fig. 1a; also data not shown).

FIG. 3.

Effects of coreceptor-targeted inhibitors on SHIV replication in PBMC. The experimental design was like that described in the legend to Fig. 2. The SHIV isolates evaluated were 89.6PD in macaque and human PBMC (a) and KU-2 and SF33A in human PBMC (b).

Among the HIV-2 isolates tested, 310342 and 7312A were both completely inhibited by TAK-779 and AOP-RANTES but were insensitive to AMD3100 (Fig. 4a). Although HIV-2 7312A can use BOB and Bonzo, to a limited extent, in GHOST-CD4 cells (Table 1), this property does not allow the virus to evade CCR5-directed inhibitors in PBMC (Fig. 4a). HIV-2 77618 and GB122 were almost completely (>95%) blocked by AMD3100, whereas TAK-779 and AOP-RANTES had no effect on these viruses (Fig. 4b; also data not shown). All of these HIV-2 isolates probably use only CCR5 or CXCR4 to enter PBMC.

FIG. 4.

Effects of coreceptor-targeted inhibitors on HIV-2 replication in human PBMC. The experimental design was like that described in the legend to Fig. 2. The HIV-2 isolates evaluated were 310342 and 7312A (a) and 77618 and 7924A (b).

An exception was, however, noted with HIV-2 7924A. This virus was partially sensitive to AMD3100, but the extent of inhibition did not exceed 30% even at the highest AMD3100 concentration, 400 nM (Fig. 4b). HIV-2 7924A was completely insensitive to TAK-779 or AOP-RANTES, and combining these agents with AMD3100 did not increase the extent of inhibition caused by AMD3100 alone (Fig. 4b).

HIV-2 isolate 7924A has an unusual pattern of sensitivity to coreceptor-targeted inhibitors.

The insensitivity of HIV-2 7924A to AMD3100 is unusual, since this virus can use CXCR4, and perhaps only CXCR4, to enter GHOST-CD4 cells (Table 1 and Fig. 1b). Usually, 50% inhibitory concentrations (IC₅₀s) of AMD3100 against viruses that use CXCR4 in PBMC are 4 to 40 nM (Fig. 2b, 3a and b, and 4b; also data not shown). To evaluate whether the insensitivity of HIV-2 7924A to AMD3100 in PBMC was donor dependent, we tested much higher AMD3100 concentrations in cells from four CCR5 wild-type donors (Fig. 5a). Donor-to-donor variation in the potency of AMD3100 was significant, with IC₅₀s ranging from 2.1 μM (donor 1) to 34 μM (donor 2). However, if sufficient AMD3100 (40 μM) was used, inhibition of HIV-2 7924A was complete in cells from three of the four donors. Whether at a concentration as high as 40 μM AMD3100 remains specific for CXCR4 is not known, although no overt toxicity was observed.

FIG. 5.

Effects of coreceptor-targeted inhibitors on HIV-2 isolate 7924A in PBMC from different donors. The replication of HIV-2 7924A in PBMC from four different human donors in the presence of AMD3100 at 40 μM, 4 μM, 400 nM, and 40 nM (a) and SDF-1α at 400 nM, 40 nM, 4 nM, and 0.4 nM (b) was evaluated. HIV-1 NL4-3 was also tested with SDF-1α. In each case, replication was measured after 7 and 10 days. IC₅₀s of AMD3100 were calculated and are shown in panel a. The data shown were obtained on day 10, but values from day 7 were similar.

We also tested AMD3100 (4 μM) against HIV-2 7924A in PBMC from a Δ32-CCR5 homozygous donor (donor 1). For the first 4 days of culturing, AMD3100 at 4 μM completely suppressed HIV-2 7924A replication; however, by day 7, the virus had broken through, and the extent of inhibition was negligible thereafter. In contrast, HIV-1 5160 was completely inhibited by 4 μM AMD3100 throughout the duration of culturing (data not shown).

To gain more insight into whether HIV-2 7924A could use CXCR4 for entry into PBMC, we determined its sensitivity to SDF-1α in cells from the same four CCR5 wild-type donors as those used in the AMD3100 experiment. Even at the highest concentration tested (400 nM), SDF-1α did not inhibit HIV-2 7924A replication in PBMC from any of the four donors, whereas HIV-1 NL4-3 replication was efficiently blocked (Fig. 5b). Taken together with the insensitivity of HIV-2 7924A to TAK-779 and AOP-RANTES (Fig. 4b), the limited or nonexistent effect of AMD3100 and SDF-1α on HIV-2 7924A replication suggests that this virus uses an undefined coreceptor other than CXCR4 to enter PBMC. An alternative explanation is that HIV-2 7924A uses CXCR4 in a highly unusual, inhibitor-insensitive manner. If this is so, how this virus uses CXCR4 must be cell type dependent, since we determined that the IC₅₀ of AMD3100 for this virus in GHOST-CD4 cells was 0.47 μM. This value contrasts markedly with the IC₅₀s of 2.1 to 34 μM for the same virus in PBMC.

Effect of coreceptor inhibitors on SIV replication in macaque PBMC.

To evaluate the inhibitor sensitivities of SIV isolates, we used macaque PBMC. In cells from the first donor macaque tested, SIV_mac251, SIV_mac239, SIV_mac251/1390, and SIV_mac239/5501 were all inhibited by both TAK-779 and AOP-RANTES to an extent that was complete, or virtually so (>95%), whereas AMD3100 had no effect (Fig. 6a and b; also data not shown). Thus, these SIV isolates all use CCR5, and only CCR5, to enter PBMC from this macaque donor. However, there are issues of donor cell dependency to consider (see below).

FIG. 6.

Effects of coreceptor-targeted inhibitors on SIV replication in PBMC. The experimental design was like that described in the legend to Fig. 2. The SIV isolates evaluated were SIV_mac251 and SIV_mac239 in macaque PBMC (a), SIV_mac251/1390 and SIV_mac239/5501 in macaque PBMC (b), and SIV_rcm in human PBMC (c). MCP-1 and MCP-3 were used at 400, 40, and 4 nM (left to right); TAK-779 was used at 3.3 μM, 330 nM, and 33 nM.

Because SIV_rcm uses CCR2 but neither CCR5 nor CXCR4 for entry (Table 1), we tested chemokine ligands of CCR2 for their abilities to inhibit SIV_rcm replication in human PBMC. Of these, MCP-1 almost completely inhibited SIV_rcm replication, whereas MCP-3 had only a limited effect (Fig. 6c). TAK-779 was also an effective inhibitor of SIV_rcm replication in human PBMC (Fig. 6c), just as it was in GHOST-CD4 cells expressing CCR2 (Fig. 1a). However, AOP-RANTES had no effect on SIV_rcm replication in human PBMC (data not shown). This virus appears to make truly exclusive use of CCR2 as a coreceptor in primary human PBMC.

The effect of TAK-779 on SIV_mac239 replication in macaque PBMC is donor dependent.

We showed above that there is a donor dependency in the ability of SIV_mac239 and SIV_mac251 to replicate in human PBMC from Δ32-CCR5 homozygous individuals (Table 4). There is also a donor dependency in the potency with which CCR5-targeted inhibitors inhibit SIV_mac239 replication in macaque PBMC. Thus, the extent to which TAK-779, at 3.3 μM, inhibited SIV_mac239 replication varied from >99% to <50% in PBMC from four different macaques (Fig. 7a). The IC₅₀s of TAK-779 ranged from 240 nM (macaque 3) to 12.6 μM (macaque 1), a 60-fold variation. However, at the very high concentration of 33 μM, TAK-779 completely inhibited SIV_mac239 replication in all four donors (Fig. 7a). Similar results were obtained with SIV_mac251 in the two donors tested; the IC₅₀s were 0.18 μM (donor 3) and 20 μM (donor 1) (Fig. 7a).

FIG. 7.

Donor-dependent variation in the effects of coreceptor-targeted inhibitors in PBMC. (a) SIV_mac239 replication in PBMC from four different macaques was evaluated in the presence of TAK-779 at 33 μM, 3.3 μM, 330 nM, and 33 nM. SIV_mac251 was similarly evaluated with cells from two donors. (b) HIV-1 P6-v3 replication in PBMC from four different human donors was evaluated in the presence of TAK-779 at 3.3 μM, 330 nM, 33 nM, and 3 nM. In each case, replication was measured after 7 and 10 days, and IC₅₀s of the inhibitor were calculated. The data shown were obtained on day 10, but values from day 7 were similar.

There was less variation in the potency of TAK-779 against HIV-1 replication in human PBMC. For instance, HIV-1 P6-v3 was inhibited by TAK-779 in PBMC from four donors at IC₅₀s ranging from 15 nM to 24 nM (Fig. 7b). This result suggests that major variations in inhibition potency are not an inherent feature of TAK-779.

When the inhibitor sensitivities of SIV_mac239 and SIV_mac251 were evaluated with CEMx174 cells, both viruses were insensitive (<5% inhibition) to AMD3100 (400 nM), TAK-779 (3.3 μM), or AOP-RANTES (40 nM), alone or in combination (data not shown). In contrast, HIV-1 NL4-3 replication in these cells was completely blocked by AMD3100 but not by TAK-779 or AOP-RANTES (data not shown). Thus, whatever coreceptor(s) SIV_mac239 and SIV_mac251 use to enter CEMx174 cells, it is not CCR2, CCR3, CCR5, or CXCR4. Whether this is the same coreceptor that these viruses can use to enter human or macaque PBMC from some donors is not yet known.

DISCUSSION

Primate lentiviruses can use about 12 different seven-transmembrane receptors as coreceptors in transfected cell lines. However, questions have been raised as to whether coreceptors other than CCR5 and CXCR4 are relevant for viral entry into primary cells and, hence, for viral replication in vivo (31, 43, 70, 86, 101, 116, 117). This issue affects the development of antiviral drugs aimed at coreceptors. Must multiple coreceptors be targeted, or just CCR5 and CXCR4 (117)? Does the ability of SIV and SHIV to use multiple coreceptors in vitro influence the interpretation of vaccine experiments with primates (73)?

We addressed these issues by using coreceptor-targeted inhibitors to block viral replication in primary PBMC, focusing here on HIV-2 and SIV isolates. As inhibitors, we used AMD3100 for CXCR4 and TAK-779 and AOP-RANTES for CCR5. These agents are not completely specific: TAK-779 and AOP-RANTES also inhibit viral entry via CCR2 and via CCR3, respectively. However, we could find no evidence that AMD3100 is anything other than specific for CXCR4, as found previously with other assay systems and test viruses (26, 37, 56, 100).

The low-level entry of viruses such as SHIV 89.6PD and HIV-2 7924A into GHOST-CD4 cells expressing V28, US28, APJ, and others actually occurs via endogenous CXCR4 and not via the transfected coreceptor, since it is inhibited by both SDF-1α and AMD3100. Of note is that SHIV 89.6PD and HIV-2 7924A use CXCR4 very efficiently, so they may be able to enter coreceptor-transfected GHOST-CD4 cells that express very low levels of CXCR4; the levels of expression of this coreceptor may also vary slightly among different GHOST-CD4 clones, making some transfected cell lines particularly susceptible to viruses that use CXCR4, although we could not accurately quantitate such variation by FACS. Whatever the explanation, ambiguities can arise when the coreceptor usage of CXCR4-tropic viruses is determined with transfected GHOST-CD4 cell lines. Many, if not all, of the “positives” for use of coreceptors other than CCR5 and CXCR4 by CXCR4-tropic viruses in GHOST-CD4 cell lines (Table 1) may simply reflect entry via endogenous CXCR4 and not the transfected coreceptor. This caveat may also apply to other studies that have used these cell lines. A similar conclusion was recently reached by others (59).

We previously concluded from an inhibitor-based study that coreceptors other than CCR5 and CXCR4 made, at most, only a limited contribution to HIV-1 replication in PBMC (117). Our inhibitor studies are now strengthened by the recent availability of TAK-779 (4). This CCR5-targeted inhibitor does not inhibit viral entry via CCR3, whereas AOP-RANTES can do so, albeit inefficiently compared to its effect on CCR5-mediated entry (34). Since TAK-779, by itself, is able to block the replication in PBMC of all of the R5, NSI HIV-1 and HIV-2 isolates that we tested, CCR3 is not relevant to their entry. Any possible use of CCR2 that might be masked by TAK-779 is not supported by the complete inhibition of the same isolates by AOP-RANTES. This chemokine derivative does not block viral entry via CCR2, at least for SIV_rcm, which is the only truly CCR2-tropic virus yet identified (15). Another advantage of TAK-779 is that it avoids the potential complications of AOP-RANTES-induced enhancement of attachment and entry of X4 HIV-1 isolates (44, 108). However, we did not observe infectivity enhancement with human or macaque PBMC at the AOP-RANTES concentrations tested in this study. Overall, the use of TAK-779 reinforces our previous conclusion about the paramount role of CCR5 and CXCR4 in HIV-1 replication in PBMC (116). This is not to say that other coreceptors are completely irrelevant; Bonzo/STRL33 can be used by rare HIV-1 isolates for entry into a minor subset of PBMC in a donor-dependent manner (102), and CCR3 and CCR8 are potential coreceptors expressed on some T-cell subsets (94, 118).

The SHIV isolates that we evaluated—89.6, 89.6P, 89.6PD, SF33A, and KU-2—all exclusively used CXCR4 in human and macaque PBMC; AMD3100 was sufficient to completely inhibit their replication, while neither TAK-779 nor AOP-RANTES had any effect. Thus, although HIV-1 89.6 can enter transfected cells via several coreceptors, including CCR5 (27), the SHIV derived from it use only CXCR4 to enter PBMC. Of note is that SHIV 89.6, SHIV 89.6P, and SHIV 89.6PD very efficiently enter GHOST-CD4 cells expressing CCR5 (Table 1; also data not shown). Thus, these viruses can use CCR5 for entry, at least in CCR5-transfected cells, but CXCR4 is preferred in primary cells. Why this should be the case and whether it matters for transmission and pathogenesis studies with these viruses in macaques are open questions.

We also conclude that, for most HIV-2 strains, CCR5 and/or CXCR4 are the principal coreceptors relevant to the replication of these strains in PBMC. Thus, TAK-779, AOP-RANTES, and AMD3100, alone or in combination, completely or very substantially inhibited the replication of almost all of our test viruses. HIV-2 isolate 7924A is an apparent exception. The replication of this broadly tropic virus in PBMC was inhibited only by very high concentrations of AMD3100 and was completely insensitive to SDF-1α, TAK-779, or AOP-RANTES. One possibility is that HIV-2 7924A is able to use an alternative coreceptor to enter human PBMC, perhaps the CXCR5 receptor reported recently to function with some HIV-2 isolates but not with HIV-1 or SIV isolates (52). Alternatively, HIV-2 7924A may use CXCR4 in a manner that is relatively insensitive to AMD3100. The latter explanation would be consistent with the observation that very high concentrations of AMD3100 do completely inhibit the replication of HIV-2 7924A, although there may be concerns about the specificity of AMD3100 for CXCR4 at such concentrations. Escape mutants of HIV-1 NL4-3 that continue to use CXCR4, but in a drug-insensitive manner, are known to emerge in response to selection pressure from AMD3100 and SDF-1α (24, 99). It has been suggested that CXCR4 can exist in different isoforms on different cell types (69); this property might be one explanation for why AMD3100 is a potent inhibitor of HIV-2 7924A in GHOST-CD4 cells expressing CXCR4 (IC₅₀ = 0.47 μM) but can be such a weak one in PBMC (IC₅₀ = 2.1 to 34 μM, depending upon the donor). Additional studies of HIV-2 7924A are warranted.

Our conclusions for SIV_mac isolates are more complicated. The CCR5 proteins from multiple primate species can function as viral coreceptors (55, 78), and our inhibitor studies are consistent with an important role of CCR5 in SIV_mac entry into primary cells. One aspect of coreceptor usage that distinguishes SIV from HIV-2 isolates is the inability of almost all SIV to use CXCR4. This property contrasts with the efficient use of CXCR4 by many HIV-2 isolates. In this sense, HIV-2 more closely resembles HIV-1 than it does SIV, an unexpected finding given the genetic relationships among these virus families and the evolution of HIV-2 from SIV_sm (16, 17, 41, 42, 44, 49). The minimal use of CXCR4 by SIV strains is mirrored by that of HIV-1 isolates from genetic subtype C (1, 11, 82, 83, 112), although SI primary viruses from this subtype are known (111).

Although CCR5 is important and CXCR4 is unimportant for SIV_mac entry, we found indications that SIV_mac239 could use a coreceptor other than CCR5 to enter PBMC from some human and macaque donors. Thus, in PBMC from one Δ32-CCR5 homozygous human donor, SIV_mac239 replication was negligible. However, in cells from a second such individual, the virus replicated fairly efficiently, as observed previously (18). Furthermore, there was considerable variation in the potency with which TAK-779 inhibited the replication of SIV_mac239 in PBMC from different macaques. This result might be accounted for by the use of an additional coreceptor that is expressed in PBMC from only a subset of macaques or that is expressed in cells from all macaques but at different levels that are sometimes below a threshold needed for infection. The expression of both CCR5 and Bonzo/STRL33 varies from donor to donor, in both humans and macaques, to an extent that can affect infection efficiency (102, 107). The ability of SIV_mac239 to use a coreceptor other than CCR5, perhaps Bonzo/STRL33, in an animal-dependent manner might influence the highly variable rates at which different infected macaques progress to disease and death (23, 57, 73). However, at least for SIV_mne, CCR5 usage is maintained throughout the course of disease progression in infected macaques (53). This is also true of SIV_mac239 and SIV_mac251 (14).

One coreceptor used efficiently by SIV_mac239 in vitro is BOB/GPR15 (22, 38). Pöhlmann et al. have, however, shown that this coreceptor has no relevance to SIV_mac239 replication in vivo, at least in some macaques (86). An unknown, alternative coreceptor(s) also mediates the AMD3100-, TAK-779-, and AOP-RANTES-insensitive entry of SIV_mac239 into CEMx174 cells; this coreceptor cannot, therefore, be CCR2, CCR3, CCR5, or CXCR4. It is not known whether this is the same coreceptor as the one used by SIV_mac239 to enter human or macaque PBMC from some donors.

We could not distinguish SIV_mac239 from the closely related SIV_mac251 in terms of their sensitivity to coreceptor inhibitors. Although SIV_mac251 but not SIV_mac239 replicates efficiently in macrophages, there is no correlation between the coreceptor usage profiles of these viruses in transfected cells and their tropism for primary cells (53, 75, 76, 86). There is also no relationship between the in vitro tropisms of SIV_mac strains and their abilities to be transmitted to uninfected animals (36, 71, 72). We have not yet performed coreceptor inhibitor studies with these viruses and purified macrophages and CD4⁺ T cells from macaques, as opposed to unfractionated PBMC.

SIV_rcm clearly uses CCR2 as its primary coreceptor (15), in a manner that we have shown is sensitive to TAK-779. The ability of SIV_rcm to enter GHOST-CD4 cells expressing US28 and V28 in vitro is likely to be of limited relevance to the replication of this virus in red-capped mangabeys.

Overall, we conclude that there is a greater complexity to coreceptor usage by SIV strains in PBMC than there is for HIV-1 and HIV-2, for which CCR5 and CXCR4 are usually the paramount coreceptors. An unknown coreceptor(s) can perhaps be used by SIV_mac239 and HIV-2 7924A to enter PBMC, at least from some macaque and human donors. Involvement of the same coreceptor in the entry of both SIV_mac239 and HIV-2 7924A might conceivably have relevance to cross-species viral transmission and the evolution of HIV-2 from SIV_sm (16, 17, 41, 42, 44, 49).