Heterosexual Transmission of Human Immunodeficiency Virus Type 1 Subtype C: Macrophage Tropism, Alternative Coreceptor Use, and the Molecular Anatomy of CCR5 Utilization

Abstract

Human immunodeficiency virus type 1 transmission selects for virus variants with genetic characteristics distinct from those of donor quasispecies, but the biological factors favoring their transmission or establishment in new hosts are poorly understood. We compared primary target cell tropisms and entry coreceptor utilizations of donor and recipient subtype C Envs obtained near the time of acute infection from Zambian heterosexual transmission pairs. Both donor and recipient Envs demonstrated only modest macrophage tropism, and there was no overall difference between groups in macrophage or CD4 T-cell infection efficiency. Several individual pairs showed donor/recipient differences in primary cell infection, but these were not consistent between pairs. Envs had surprisingly broad uses of GPR15, CXCR6, and APJ, but little or no use of CCR2b, CCR3, CCR8, GPR1, and CXCR4. Donors overall used GPR15 better than did recipients. However, while several individual pairs showed donor/recipient differences for GPR15 and/or other coreceptors, the direction of the differences was inconsistent, and several pairs had unique alternative coreceptor patterns that were conserved across the transmission barrier. CCR5/CCR2b chimeras revealed that recipients as a group were more sensitive than were donors to replacement of the CCR5 extracellular loops with corresponding regions of CCR2b, but significant differences in this direction were not consistent within pairs. These data show that sexual transmission does not select for enhanced macrophage tropism, nor for preferential use of any alternative coreceptor. Recipient Envs are somewhat more constrained than are donors in flexibility of CCR5 use, but this pattern is not universal for all pairs, indicating that it is not an absolute requirement.

A majority of new human immunodeficiency virus type 1 (HIV-1) infections are initiated by only a single genetic species, although in some, several closely related variants are transmitted (19, 26, 34, 44, 55). Nevertheless, in all cases, a molecular bottleneck occurs during transmission (8, 15, 22, 33, 56). This bottleneck does not appear to be simply a stochastic result of low-efficiency transmission since viral sequences in recipients do not typically reflect the majority sequences in donors, even in genital secretions responsible for transmission (22, 56). Identifying the biological factors that favor particular variants in transmission and/or establishment in new hosts is essential both to understanding the mechanisms of transmission and to developing approaches, including vaccines and microbicides, that might interrupt transmission.

More than 15 years ago, it was recognized that new infections were nearly always initiated by HIV-1 variants that were macrophage tropic and non-syncytium inducing (NSI) in T-cell lines, even though donors often harbored variants that were syncytium inducing (SI) and non-macrophage tropic (55). The molecular basis for these characteristics was subsequently linked to use of the entry coreceptor CCR5 by macrophage-tropic/NSI transmitted variants (R5 strains), with exclusion of T-cell-line-tropic/SI or dual-tropic variants that use CXCR4 (X4 or R5X4 strains). The critical role for CCR5-using strains in transmission is underscored by the fact that individuals who genetically lack CCR5 expression are highly resistant to HIV-1 infection (32, 49). Many potential mechanisms have been offered to explain the requirement for CCR5-mediated transmission, including R5 infection of macrophages at sites of transmission, preferential uptake by dendritic cells at mucosal sites, selective transcytosis by epithelial cells, or greater susceptibility of CCR5-rich memory T lymphocytes that are the main reservoir for viral amplification during acute infection (reviewed in reference 35). However, the virus-cell interactions that underlie this powerful restriction remain to be defined.

Even among transmission pairs in which donors harbored only R5 variants, however, genetic selection at transmission indicates that selective forces beyond just CCR5 use appear to be operative. In a cohort of serodiscordant Zambian couples infected with subtype C HIV-1 followed prospectively, in which transmission subsequently occurred, the gp120 envelope glycoprotein of transmitted variants were typically more compact than those of chronically infected donors, with shorter V1/V2 regions and fewer potential N-linked glycosylation sites (15). Similar genetic and/or serological selections have been identified in several additional, although not all, cohorts (8, 31, 33). One biological feature associated with the transmitted variants is greater sensitivity to neutralization by the infecting partner's antibody (15). However, it is unclear what selective advantage this property would provide in an immune naïve recipient, compared with other variants present in donors, raising the possibility that these gp120 features may confer other biological characteristics favoring transmission and/or establishment of infection.

In this study, we compared the tropism and coreceptor utilization characteristics of donor and recipient Env glycoproteins derived from subtype C heterosexual transmission pairs obtained near the time of acute infection that have previously been linked to this genetic selection pattern (15). Because of the potential role for macrophages in transmission and uncertainty over the importance of macrophage tropism per se in the bottleneck, we assessed these variants’ ability to mediate entry into primary human macrophages, as well as primary CD4⁺ T cells. In addition to CCR5 and CXCR4, a number of other G protein-coupled receptors (GPCRs) can support HIV-1 entry and infection in in vitro systems, although a role for these alternative coreceptors in vivo has yet to be identified. Therefore, to address the possibility that one of these pathways might be involved in sexual transmission and contribute to the molecular bottleneck, we asked if donor and recipient Envs differed in their abilities to use alternative coreceptors for entry. Finally, since HIV-1 gp120 molecules vary in the molecular details of how they interact with CCR5, and since CCR5 may be expressed differently in the context of different target cells (29), we determined whether molecular anatomy of CCR5 use was different between donor and recipient Envs. Finally, in addition to enabling comparison of donors and recipients within transmission pairs, the panel of Envs also enabled us to address the tropism and coreceptor characteristics of “chronic” versus “acute” subtype C Envs.

MATERIALS AND METHODS

Primary cells and cell lines.

U87/CD4, U87/CD4/CCR5, and U87/CD4/CXCR4 cells were provided by D. Littman through the NIH AIDS Reagent Repository and were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS; Gibco, Carlsbad, CA) along with selection agents as recommended (300 μg/ml G418 for CD4 and 1 μg/ml puromyocin for CCR5 and CXCR4) (5). 293T cells were maintained in Dulbecco's modified Eagle's medium with 10% FBS.

Primary human monocytes were isolated from healthy volunteers by elutriation and maintained for 5 to 7 days in RPMI 1640 medium supplemented with 10% FBS and 50 ng/ml macrophage colony-stimulating factor (Peprotech, Rocky Hill, NJ) to enable differentiation into monocyte-derived macrophages (MDM). After maturation, cells were gently scraped and replated at 5 × 10⁴ cells/well in 96-well plates in medium without macrophage colony-stimulating factor. Primary human CD4⁺ T cells were isolated by leukapheresis followed by immunomagnetic positive selection (Miltenyi, Auburn, CA). T cells were maintained in RPMI 1640 medium supplemented with 10% FBS and stimulated with phytohemagglutinin (5 μg/ml) for 2 days, after which cells were replated at 2 × 10⁵ cells/well in 96-well plates in the presence of interleukin-2 (10 U/ml; Novartis, East Hanover, NJ).

Transmission pair envelopes.

These studies utilized subtype C primary isolate envelopes from eight Zambian heterosexual HIV transmission pairs enrolled in a prospective cohort (2). Five of the pairs (no. 53, 55, 106, 109, and 135) have been previously described, and env genes were cloned into the pCR3.1 expression vector (15). Two additional pairs (no. 205 and 221) are described more recently (22) and, along with pair 153, were generated in a similar manner and cloned into pCDNA3.1, except that env genes were derived by single genome amplification prior to cloning. In cases where highly homologous populations were obtained, the env clones used for this study all exhibited distinct sequences. Envelopes were first screened for function based on ability to mediate pseudotype infection of U87/CD4/CCR5 or CXCR4 cells, and approximately five functional envelopes from each donor and each recipient were studied.

Pseudovirus production and infectivity assessment.

Pseudotype virions were produced using the env-deleted vpr⁺ backbone plasmid pNL-Luc-E-R+, which contains the luciferase gene in place of nef, kindly provided by N. Landau (12). 293T cells were cotransfected using the FuGene reagent (Roche, Indianapolis, IN) with the HIV-1 backbone plasmid and an envelope plasmid. Two days later, pseudotype-containing supernatants were clarified by centrifugation for 10 min at 1,500 × g and frozen at −80°C in 5% sucrose. Coreceptor phenotype and virus stock infectivity were then determined on U87/CD4 coreceptor cells that were plated at 1.5 × 10⁴ cells per well in 96-well tissue culture plates and spinoculated for 2 h at 1,200 × g (40) with pseudovirus stock. Three days later, cells were lysed in lysis buffer and assayed for luciferase activity as described previously (52). Pseudotype stocks were normalized to contain 10⁶ relative luciferase units (RLU) based on U87/CD4/CCR5 cells per 50 μl, which was then used as a standard inoculum in subsequent infections. Titration of pseudotype stocks showed that the inoculum of 10⁶ RLU was within a range that gave a linear relationship between input and RLU output (data not shown) so that output from this inoculum would accurately represent differences in infectiousness on target cells. As a positive control and reference point for primary cell infections, pseudotypes containing the HIV-1 prototype Env BAL were generated in the same way, along with pseudotype viruses lacking Env as a negative control.

Use of alternative, mutant, and chimeric coreceptors.

293T cells were cotransfected, using the FuGene reagent, with plasmids encoding CD4 and target chemokine receptors, or with CD4 and an empty plasmid (pCDNA3.1). One day later, cells were detached with EDTA, washed twice with phosphate-buffered saline, and aliquoted into 96-well plates at a density of 2 × 10⁴ cells/well. Chimeric, mutant, and alternative coreceptor chemokine receptor plasmids have been previously described (47, 48), and the identity of each coreceptor expression plasmid tested here was verified by sequence analysis.

Infections.

Primary MDM, CD4⁺ T cells, or 293T targets were infected using equivalent inocula (10⁶ RLU determined in U87/CD4/CCR5 cells) by spinoculation for 2 h at 1,200 × g. Cells were then incubated for 3 days, after which they were lysed and assayed for luciferase expression. Each infection was done in duplicate or triplicate wells. Infections of 293T cells were repeated in at least two independent experiments, and primary MDM and CD4⁺ T-cell targets were assayed in at least three independent experiments, using cells from different donors. For transfected 293T targets, infection was normalized relative to each Env's entry through wild-type CCR5, carried out in parallel within each experiment, and target cells transfected with CD4 alone served as negative controls. In parallel, each coreceptor was also tested with virions lacking Env, and in all experiments, the two types of negative controls gave equivalent levels of background luciferase expression (data not shown). For infection of primary macrophages and CD4⁺ T cells, infection was normalized to that mediated by the HIV-1 BAL Env, carried out in parallel within each experiment, and infection with virions lacking Env served as a negative control. In selected experiments, the CCR5 blocker maraviroc (2 uM) was also used to establish the floor value for luciferase expression in the absence of infection.

Statistical analysis.

Statistical analysis was performed using STATA/MP 10.1 on log_e-transformed outcomes. Analyses reflect comparisons of the underlying replicates from all experiments. Differences in infectivity between Envs from donors as a group and those from recipients as a group were analyzed by analysis of variance (ANOVA) with adjustment for pair and experiment effects. Differences between Envs within each pair were analyzed by two-way ANOVA with adjustment for experiment effects.

RESULTS

Heterosexual transmission pair envelopes.

To dissect the role of primary cell tropism and coreceptor use in heterosexual transmission of HIV-1, we studied subtype C envelopes isolated from epidemiologically linked transmission pairs identified in a cohort of initially HIV-1 serodiscordant couples in Zambia followed prospectively in which transmission from infected (donor) to uninfected (recipient) partners took place (2, 15, 22). Recipient envelopes were isolated generally within 20 to 90 days of transmission and, therefore, are considered acute/early subtype C variants. Since couples were enrolled on the basis of established infection in one partner and genotypic analysis of donor env sequences revealed heterogeneous quasispecies (2, 15, 22), donor envelopes serve not only as the point source for recipients but also are considered to be chronic subtype C infections. Only pairs that were genetically linked were included in this analysis and, as shown in Table 1, included both male-to-female (MTF) and female-to-male (FTM) transmission. Each recipient demonstrated a monophyletic pattern, indicating a single transmitted variant. Envs were cloned by PCR from uncultured plasma RNA or peripheral blood mononuclear cell provirus, and approximately five functional Envs were studied for each subject. Donor Envs used for this analysis were chosen from across each individual's phylogenetic tree in order to represent as many different clusters as possible. Recipient Envs were much more homogenous, and therefore, Envs that each represented a distinct genetic sequence were chosen. Additional details of transmission pairs, env cloning and genetic analysis have been described for pairs 53, 55, 106, 109, and 135 in reference 15 and for pairs 205 and 221 in reference 22.

TABLE 1.

Characteristics of the heterosexual transmission pair envelopes

ID no.a	Typeb	Timing (mo)c	Viral loads (RNA copies/ml)d		Env source(s)e	No. of Envsf
			Donor	Recipient		Donor	Recipient
53	FTM	3.4	150,699	26,643	PB	5	4
55	MTF	3	501,927	88,544	PB	5	3
106	MTF	4.3	267,961	48,442	PB	4	5
109	MTF	3.2	847,759	887,586	PB	5	5
135	FTM	3.6	65,784	202,999	PL	5	5
153	FTM	3	44,251	429,581	PB, PL	5	5
205	MTF	0.6	47,133	6,125	PB, PL	5	5
221	FTM	0.75	62,741	750,000	PB, PL	5	5

^aTransmission pairs are identified by identification (ID) numbers.

^bRefers to FTM or MTF transmission direction.

^cIndicates the period between the recipients’ last seronegative test and collection of samples for analysis (15), except for pairs 205 and 221, for which time of infection was calculated on the basis of the most recent common ancestor (22).

^dPlasma viral loads were determined at the time of sample collection.

^eIndicates whether env genes were amplified from peripheral blood mononuclear cell DNA (PB) and/or plasma RNA (PL).

^fIndicates the number of Envs analyzed from each sample.

To assess envelope function, pseudoviruses carrying individual Envs were generated with an env-deleted lentiviral backbone carrying the luciferase gene in place of nef, employing the vpr⁺ version in order to properly test macrophage infection (12). Pseudoviruses were first evaluated on U87 cells expressing CD4 alone or in combination with CCR5 or CXCR4. All envelopes mediated infection through CD4 and CCR5, while none were infected through CD4 and CXCR4 or CD4 alone (data not shown). The finding that all Envs were R5, including chronic isolates, is consistent with observations that evolution to CXCR4 use occurs less commonly with subtype C than with other subtypes (11). We therefore standardized inocula based on levels of infection on U87/CD4/CCR5 cells to ensure that infections were carried out using comparable infectious units.

Primary macrophage and CD4 T-cell entry.

New infections are almost always associated with R5 strains, but a critical question is whether that finding reflects a transmission advantage for variants that efficiently infect macrophages or a selection for the CCR5-using strains and against the CXCR4-using variants that typically infect macrophages less well. Because all of these subtype C donors exhibited an R5 phenotype, it enabled us to evaluate tropism changes across the transmission barrier separately from coreceptor use. We determined whether transmission led to enhanced macrophage tropism by comparing donor and recipient pseudoviruses from six transmission pairs for infection of primary human monocyte-derived macrophages. Because macrophages from different donors can vary dramatically in permissiveness to HIV-1, we tested each envelope by using cells from at least three different donors and included the macrophage-tropic prototype strain Bal in each experiment as an internal reference. Within each experiment we also included virions lacking Env as a control to establish a threshold value in the absence of infection. In selected experiments, we also used the CCR5 antagonist maraviroc to block infection and found values that were similar to those of virions lacking Env (data not shown). Primary CD4⁺ T cells were infected in parallel by using a similar approach to normalize infection among experiments and establish the threshold value.

We first assessed the ability of the acute/recipient envelopes as a group to infect MDM, compared with that of the chronic/donor envelopes. As shown in Fig. 1, both donor and recipient Envs mediated only modest levels of macrophage infection, which was approximately 100-fold less efficient than Bal (range, 10- to 1,000-fold), despite the use of inocula that were equivalent on CCR5/CD4-expressing U87 indicator cells. Importantly, there was no evidence for macrophage infection capacity being greater among recipient Envs than that among donor Envs, indicating that heterosexual transmission of HIV-1 subtype C does not select for enhanced macrophage tropism. Conversely, there was also no evidence that acute-infection variants were less macrophage tropic than chronic-infection variants, as has been suggested previously (20).

FIG. 1.

Donor/chronic and recipient/acute Env-mediated infection of primary human macrophages and CD4⁺ T cells. MDM and CD4⁺ T cells were infected by pseudotypes carrying donor (circles) and recipient (triangles) envelopes from pairs 53, 109, 135, 153, 205, and 225, using equivalent inocula based on infectivity in U87/CD4/CCR5 cells, in parallel with the macrophage-tropic prototype BAL. Macrophage and T-cell infection was measured by luciferase expression at day 3 and normalized relative to that produced by BAL to account for cell donor-dependent differences in primary cell permissiveness. Each point represents the relative infection by one envelope, showing the average of the results for infections using cells from at least three independent donors. Horizontal bars indicate the geometric mean of the population. The horizontal dashed lines show the negative control values (cells infected with virions lacking Env). Analysis by ANOVA showed no significant difference between donor/chronic and recipient/acute envelopes in either cell type.

Primary CD4⁺ T-cell infection mediated by these Envs was markedly more efficient than that mediated by macrophages (Fig. 1), although still not quite to the level of Bal. However, there was also no overall difference between chronic/donor and acute/recipient Envs in primary lymphocytes. This observation suggests that transmission also does not select for greater T-cell infection capacity, nor is chronic infection characterized by Envs with more-efficient lymphocyte entry. Since inocula were equilibrated based on U87/CD4/CCR5 infectivity, this analysis could not completely exclude differences in overall infectivity between Envs if such differences were reflected in all cell types, including indicator cells. However, there was no evidence to suggest that pseudoviruses carrying donor and recipient Envs differed systematically in infectivity based on infection of U87/CD4/CCR5 cells (RLU) per ng of p24 Gag antigen (data not shown).

We then asked if transmission across the sexual barrier was associated with differences in primary cell infectivity within individual transmission pairs (Fig. 2). Only pairs 53 and 205 showed significant differences in macrophage tropism between donor and recipient Envs. However, the differences were not consistent, since recipient Envs were more macrophage tropic than were donor Envs in pair 53 whereas the opposite was noted for pair 205, and the remaining pairs failed to show any significant differences across the transmission barrier (Fig. 2A). A similar pairwise analysis was done for primary CD4⁺ T-cell infection (Fig. 2B). As with primary macrophages, only two pairs (no. 53 and 109) showed significant donor/recipient differences and, again, differences were not conserved among the pairs. On the other hand, it is apparent that despite standardization of inocula by infectivity on U87/CD4/CCR5 cells, there were striking differences (up to 10-fold) between pairs in ability to infect macrophages and CD4⁺ T cells, at least compared with a standard reference strain (Fig. 2). However, in most cases these differences appeared to be similar within donor and recipient pairs.

FIG. 2.

Primary macrophage and CD4⁺ T-cell infection by donor and recipient envelopes from each transmission pair. MDM (A) and CD4⁺ T cells (B) were infected with donor (circles) and recipient (triangles) envelopes from transmission pairs as described in the legend for Fig. 1. The horizontal dashed lines indicate the negative control values (cells infected with virions lacking Env). Differences in primary cell infection between donors and recipients within each pair were compared by ANOVA. **, P < 0.01; ***, P < 0.001.

Taken together, these data indicate that these subtype C Envs display considerable heterogeneity in primary cell infection capacity but, in most cases, this characteristic is conserved upon transmission and that while transmission may, in some cases, lead to differences in primary cell infectivity, there is no systematic trend in primary cell tropism.

Infection through alternative coreceptors.

HIV-1 can infect cells in vitro by using a number of GPCRs other than CCR5 and CXCR4, but the significance of these alternative coreceptors in pathogenesis remains unclear. We hypothesized that selective use of one or more of these pathways might be involved in preferential transmission of particular variants. Consequently, we looked for differences between donor/chronic and recipient/acute envelopes in efficiency of entry through the alternative coreceptors CCR2b, CCR3, CCR8, GPR1, GPR15, CXCR6 (STRL33), and APJ by transfecting 293T cells with CD4 and a coreceptor followed by pseudovirus infection. In order to compare pathways, for each Env, we normalized alternative coreceptor-mediated infection to that supported by CCR5 carried out in parallel. To ensure that alternative coreceptors were expressed in a functional manner, each experiment included internal controls with at least one HIV-1 or simian immunodeficiency virus (SIV) Env pseudotype known to utilize that alternative coreceptor (HIV-1/89.6 or SIVmac239) (data not shown). While the level of infection that might be of biological significance in vivo is unknown, we arbitrarily set a threshold at 1% of the level of infection relative to that mediated by CCR5.

As shown in Fig. 3, we found that a large majority of these Envs used GPR15, CXCR6, and APJ. In many cases, infection efficiency was 10% or greater than that mediated by CCR5. This result was unexpected, as broad use of these alternative coreceptors by subtype C Envs has not previously been noted. In contrast, few envelopes infected target cells through CCR2b, CCR3, CCR8, or GPR1. When chronic/donor and early/recipient isolates were compared, we noted that the chronic population was slightly but significantly better able to infect target cells through GPR15, whereas no other alternative coreceptor showed differences. These data suggest that subtype C HIV-1 Envs are flexible in their use of some alternative coreceptors in vitro but that, as a group, acute/recipient variants do not exhibit selective use of any particular alternative pathway compared with chronic/donor isolates.

FIG. 3.

Donor/chronic and recipient/acute Env infection of cells expressing alternative entry coreceptors. 293T cells were transfected with CD4 in conjunction with CCR2b, CCR3, CCR8, GPR1, GPR15, CXCR6, or APJ, as well as with CCR5 or with no coreceptor. Two days later, the cells were infected by pseudotype viruses carrying donor/chronic (circles) and recipient/acute (triangles) envelopes from eight transmission pairs. Infection was measured 3 days later via luciferase expression and normalized to infection mediated by CCR5. Each point represents the average of the results for at least two independent experiments, and the horizontal bars indicate the population geometric mean. A dashed line was drawn at 1% infection relative to CCR5 to indicate an arbitrarily selected level of biological significance. Donor/chronic and recipient/acute envelopes were compared by ANOVA, and statistically significant differences are indicated. ***, P < 0.001.

Next, we determined whether systematic selection for alternative coreceptor use across the transmission barrier might be identified within the individual transmission pairs (Fig. 4). Donor isolates were more capable than recipients in using GPR15 for three pairs (no. 53, 109, and 135), and two pairs showed similar use by donors and recipients (no. 205 and 221), while neither donors nor recipients used GPR15 efficiently in three pairs (no. 55, 106, and 153). For CXCR6, donor isolates from three pairs mediated infection significantly better than did recipient isolates (no. 53, 109, and 153). However, the opposite pattern was seen for two pairs (no. 135 and 205), and no difference was seen for two others (no. 106 and 221). Donor isolates used APJ more robustly than did recipient isolates for three pairs (no. 53, 55, and 109); however, the opposite pattern was seen for two pairs (no. 106 and 205), and equivalent use was found for two pairs (no. 153 and 221). For the poorly used alternative coreceptors, one pair (no. 53) showed an increase in CCR3 utilization following transmission, whereas CCR8 was used by both donor and recipient in one other pair (no. 106).

FIG. 4.

Use of alternative coreceptors by donor and recipient envelopes of each transmission pair. 293T cells transfected with CD4 and alternative coreceptors were infected by donor/chronic (circles) and recipient/acute (triangles) envelope pseudotypes as described in the legend for Fig. 3. Donor and recipient envelopes within each donor pair were compared by ANOVA, and significant differences are shown. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Taken together, these data suggest that sexual transmission of HIV-1 can lead to a variety of outcomes in regards to alternative coreceptor utilization, but there is no systematic selection for use of any alternative coreceptor across the transmission barrier. In contrast, there is a tendency for chronic/donor variants to use GPR15 more efficiently than acute/recipient Envs, but there is neither consistent nor systematic selection against GPR15 use with sexual transmission.

Molecular anatomy of CCR5 utilization.

Structural studies of the molecular interactions between gp120 and the CD4/coreceptor complex demonstrate critical contacts on both N-terminal and extracellular loop (ECL) coreceptor regions for the prototype isolates analyzed (25, 45). In addition, studies using modified coreceptor molecules offer further opportunity to refine the relative importance of specific domains and compare different viruses (reviewed in reference 16). Although results are not always comparable because of the different assay systems used (chimeras between human and nonhuman CCR5 or between CCR5 and GPCRs that do not support entry; point mutants; truncation mutants), such studies confirm central roles for the CCR5 N terminus (including several sulfated tyrosines in the N terminus) and interactions among multiple ECL regions. However, those studies demonstrate that marked differences exist between individual R5 strains in specific dependence on domains of CCR5 critical for entry, indicating that there are strain-specific differences in the details of how Envs interact with the coreceptors (4, 48). Given the genetic selection in Env at transmission, we hypothesized that differences in the molecular anatomy of CCR5 utilization might be linked to HIV-1 variants which transmit or that they might differ between acute and chronic infection variants.

To address this possibility, we tested the pseudoviruses on a panel of chimeric coreceptors cotransfected along with CD4 into 293T target cells. For this, we focused on the CCR5 N terminus and on the ECL domains, using chimeras that exchanged these regions with CCR2b. CCR2b was chosen because none of the Envs tested used that coreceptor efficiently for entry (Fig. 4). Staining with antibodies directed at the N terminus (monoclonal antibody 3A9) or ECL2 (monoclonal antibody 2D7) confirmed expression of each molecule, with expression levels (based on mean fluorescence intensity) that were similar for the coreceptors (data not shown). Within each experiment, infection through the mutant coreceptor was normalized as a percentage of that mediated by wild-type CCR5.

As shown in Fig. 5A, entry of both chronic/donor and acute/recipient isolates was markedly impaired when the CCR5 N terminus was replaced with that from CCR2b (chimera 2555). Furthermore, both populations were able to infect target cells expressing the CCR5 in which the ECL2 was replaced with that from CCR2b (chimera 5525), with a moderate reduction in efficiency compared to that of wild-type CCR5. Both sets of Envs were also able to use a coreceptor in which the CCR5 N terminus was grafted onto CCR2b (chimera 5222) although with considerably reduced efficiency. For both chimeras 5525 and 5222, donor/chronic Envs exhibited slightly but significantly greater ability to use the modified molecules. These results indicate that both the CCR5 N terminus and ECL domains play a role in entry, with N-terminal, CCR5-specific sequences being particularly critical, and that chronic/donor Envs display slightly greater flexibility than do acute/recipient Envs in their interactions with CCR5.

FIG. 5.

Donor/chronic and recipient/acute Env infection of cells expressing chimeric and mutant CCR5 molecules. 293T cells were transfected with CD4 along with a panel of CCR5/CCR2b chimeras (A) or N-terminally truncated CCR5 mutants (B), and then infected by pseudotype viruses carrying donor/chronic (circles) and recipient/acute (triangles) envelopes from eight transmission pairs. Infection was measured 3 days later via luciferase and normalized relative to wild-type (WT) CCR5. Each point represents the average of the results for at least two independent experiments. The horizontal bars represent the population geometric mean, and the dashed lines are shown at 1% infection relative to CCR5. Differences between the ability of donor/chronic and recipient/acute envelopes to mediate infection were compared by ANOVA, and statistically significant differences are shown. **, P < 0.01; ***, P < 0.001.

We further addressed the N terminus of CCR5 by testing N-terminally truncated CCR5 molecules in which sequentially greater regions were deleted (Fig. 5b). Both chronic/donor and acute/recipient pseudovirus populations entered through CCR5 lacking the first 4 residues of the N terminus (CCR5 Δ4) at levels generally similar to those of the wild-type CCR5, while deletion of the first 8 N-terminal residues (CCR5 Δ8) led to a moderate reduction in entry for both populations. In contrast, deletion of the 12 (CCR5 Δ12) or 16 (CCR5 Δ16) N-terminal amino acids abrogated target cell infection by the majority of isolates. Interestingly, chronic/donor isolates entered through CCR5 Δ4 with slightly but statistically greater efficiency than did acute/recipient isolates, whereas the opposite was seen for CCR5 Δ8. However, these differences were very small and unlikely to be biologically significant. Together, these results confirm for these subtype C Envs that the CCR5 N terminus plays an essential role, especially truncation of the first 12 amino acids, which is beyond the first critical sulfated tyrosine at residue 10 (17). However, neither the chimeric coreceptors nor the N-terminally truncated molecules suggest substantial differences between chronic/donor and acute/recipient populations.

Even though the majority of Envs failed to use certain mutants, there were some exceptions (e.g., Fig. 5, chimera 2555, CCR5 Δ12, and CCR5 Δ16), as well as variants that could not enter through molecules used by most of the others (e.g., chimeras 5525 and 5222 and CCR5 Δ8). Therefore, we then analyzed chimeric and truncated coreceptor data within transmission pairs (Fig. 6 and 7), reasoning that stratifying our analysis in this manner might reveal systematic differences not otherwise evident. Instead, our investigation showed that donor-to-recipient differences varied both in extent and direction from pair to pair. For three transmission pairs (no. 53, 109, and 135), donor Envs were significantly better able than recipients to use chimeric (5525 and 5222) and truncated (CCR5 Δ8 and/or CCR5 Δ12 or CCR5 Δ16) CCR5 coreceptors. In contrast, three pairs (no. 55, 153, and 205) showed the opposite pattern, in which recipient Envs were more flexible than donors at infecting through altered CCR5 coreceptors, and in two pairs (no. 106 and 121), donor and recipient Envs showed generally similar levels of flexibility in their use of modified CCR5. The extent of variation between the pairs suggests that neither more-flexible nor more-restricted use of the molecular anatomy of CCR5 is a consistent selective factor in HIV-1 subtype C sexual transmission.

FIG. 6.

Use of CCR5/CCR2b chimeric coreceptors by donor and recipient envelopes from each transmission pair. 293T cells were transfected with CD4 along with wild-type (WT) or CCR5/CCR2b chimeric coreceptors and infected by pseudotype viruses carrying donor (circles) and recipient (triangles) envelopes from transmission pairs as described in the legend for Fig. 5. Infections by donor and recipient Envs within each pair were analyzed by ANOVA, and significant differences are shown. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

FIG. 7.

Use of N-terminally truncated CCR5 molecules by donor and recipient envelope transmission pairs. 293T cells were transfected with CD4 and wild-type (WT) or N-terminally truncated CCR5 molecules and then infected by pseudotype viruses carrying donor (circles) and recipient (triangles) envelopes from transmission pairs as described in the legend for Fig. 5. Infections by donor and recipient Envs within each pair were analyzed by ANOVA, and significant differences are shown. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Relationship between use of alternative and modified coreceptors.

We then asked if within transmission pairs, there is a relationship between the patterns of alternative coreceptor use and flexibility in the molecular anatomy of CCR5 utilization. As shown in Fig. 8, for pairs 53, 109, and 135, donor Envs exhibited a fairly broad advantage over recipient Envs in the ability to use multiple alternative and mutant coreceptors. For pairs 106 and 221, envelopes from either side of the transmission barrier were similarly infectious on the different coreceptors. In contrast, recipient isolates from pair 205 mediated infection through these panels of coreceptors better than did their donor counterparts. For pairs 55 and 153, recipient Envs used the chimeric and mutant CCR5 molecules better than did donor Envs, although alternative coreceptor use did not show the same pattern.

FIG. 8.

Directionality of donor/recipient differences in alternative coreceptor and modified CCR5 use. Transmission pairs are identified by identification numbers and types of transmission as defined in Table 1. For each coreceptor, directionality of usage is shown in green (D > R); yellow (D = R); or red (D < R). White boxes indicate no utilization above the arbitrarily determined threshold (1% of CCR5 use) by either donor or recipient. CCR2b and GPR1 are not shown because no Envs reached the 1% threshold.

Thus, there appears to be a general correspondence for most pairs in the directionality of efficiency for different coreceptor molecules, suggesting that both alternative coreceptor and modified CCR5 use may reflect an overall flexibility in particular Envs’ abilities to interact with the coreceptor. In contrast, there was no relationship between use of these molecules and efficiency of macrophage or primary CD4⁺ T-cell entry (data not shown). However, this overall flexibility was clearly not particularly associated with either chronic/donor or acute/recipient Envs, arguing against it playing a role in selection at transmission. Of note, there was no correlation with the mode of transmission either, as the group in which the donor Envs displayed greater flexibility than did the receptor Envs (D > R) included both FTM and MTF transmission pairs, as did the group in which the levels of flexibility were the same amongst donor and receptor Envs (D = R) and the group in which the donor Envs displayed less flexibility than did the receptor Envs (D < R).

DISCUSSION

The AIDS epidemic worldwide is being driven largely by heterosexual transmission. Transmission is a relatively inefficient event, however, so understanding the factors that select for HIV-1 variants capable of breaching the transmission barrier should provide important insight into the mechanisms, cellular pathways, and viral features that must be targeted in order to interrupt transmission. Transmitted variants are distinguished from donor viruses by envelope glycoproteins that display near-universal use of CCR5, more-compact V1/V2 domains, fewer N-linked glycosylation sites, and increased sensitivity to neutralization by autologous partner serum (8, 15, 33, 56). It is unclear why neutralization-sensitive variants would have a selective advantage in transmission or amplification in an immune naïve new host, so we focused here on the functional aspects of viral entry and target cell tropism determined by Env that might distinguish the transmitted variants from the donor variants.

A principal observation of this study is that transmitted variants do not exhibit enhanced macrophage tropism compared with that of donor variants. It is well established, mainly from studies with subtype B, that new infections are characterized almost exclusively by macrophage-tropic/NSI variants, whereas chronic infections often had mixtures of macrophage-tropic/NSI, T-cell-line-tropic/SI, and dual-tropic variants. However, it has been uncertain whether this bottleneck was due to a critical role for macrophage infection, per se, or dependence on CCR5 in transmission apart from macrophages (with the incidental exclusion of CXCR4-using variants that generally infect macrophages less efficiently than R5 strains). Since all of the donor strains tested here used CCR5 and none used CXCR4, as is typical of subtype C, we were able to evaluate target cell tropism independently from coreceptor use. Our results suggests that macrophage infection itself is not responsible for selection at transmission and that the association between acute infection and macrophage tropism is likely an indirect result of a critical role for CCR5 on some target cell other than macrophages and/or selection against CXCR4-using variants. It remains to be determined whether CCR5 dependence is linked to (i) very early events in mucosal tissues, such as epithelial transcytosis, dendritic cell interactions, or intraepithelial lymphocyte infection; (ii) amplification in CCR5-rich gastrointestinal mucosal T cells during acute infection; (iii) greater susceptibility of CXCR4-using variants to inactivation by mechanical or innate immune mechanisms; or (iv) other factors (35, 37, 43).

In contrast to the hypothesis for macrophage tropism and acute infection, an alternative model has been described, in which early variants infected macrophages poorly, while progressive disease was characterized by increased macrophage tropism, possibly in conjunction with waning immune pressures that reduce the necessity for a highly cryptic coreceptor binding site (18). While our Envs were not derived from longitudinal sampling, we did not find greater macrophage infection capacity among chronic-infection Envs than that of acute-infection Envs, which would tend to argue against the notion of enhanced macrophage tropism with later-stage disease.

Finally, in addition to comparing chronic/donor Envs and acute/recipient Envs, these data also allowed us to address the overall macrophage infection capacity of uncultured Envs from these 16 subjects. Despite using inocula with equivalent levels of infectiousness, these Envs mediated infection of macrophages at a level that was approximately 20-fold less efficient than that of primary CD4⁺ T cells and ∼100-fold less efficient in macrophages than the prototype strain BAL (Fig. 1). This finding indicates that, on an absolute scale, neither the acute nor chronic viruses are highly macrophage tropic. This result with subtype C tends to support findings that were reported recently for subtype B that most unamplified HIV-1 strains derived from blood are poorly macrophage tropic and that efficient macrophage tropism is mainly seen among variants from compartments in which infection is particularly macrophage dependent, such as the central nervous system (42).

The second focus of our study was to identify a potential role for alternative entry coreceptors in selection during transmission. Many GPCRs can support HIV-1 entry in conjunction with CD4 in transfected cells, and several are expressed by and support infection in CD4⁺ transformed T-cell lines. Their distribution and coreceptor function on primary CD4⁺ lymphocytes are more restricted, however. CCR2, CCR3, CCR8, and CXCR6 (STRL33) are expressed on small subpopulations of blood CD4⁺ T cells or thymocytes (10, 14, 30, 50, 53), and both CCR8 and CXCR6 have been reported to support primary CD4⁺ cell infection by certain HIV-1 strains (10, 30, 50). Interestingly, CCR2 is also expressed on a small proportion of normal gut mucosal lymphocytes, but becomes highly expressed on a large proportion of intraepithelial CD4⁺ lymphocytes in inflamed intestine (14). This observation raised the possibility that CCR2 might be involved in selective viral amplification in mucosal lymphocytes, the predominant site of replication during acute infection, but our results suggest that is not the case. In contrast, GPR1, GPR15, and APJ are not known to be expressed on CD4⁺ T lymphocytes, although GPR1 is expressed on and reportedly supports in vitro HIV-1 infection of several nonlymphoid target cells (51, 54). Within the genital tract, CCR2 and CCR3 are expressed in both female and male genital mucosa (36, 41) although little is known about mucosal tissue expression of the other GPCRs used by HIV-1. However, no alternative coreceptor was used by recipient viruses more efficiently than by donor viruses, as a group, and while some individual transmission pairs showed significant donor/recipient differences for individual coreceptors, there was no consistent pattern among the pairs. In fact, examination of the data suggests that pair-dependent differences in patterns of alternative coreceptor use were frequently conserved across the transmission barrier (Fig. 4), such as (i) the high GPR15/CXCR6 but low APJ use by both donor and recipient Envs of pair 135; (ii) the high use of each of those coreceptors by both donor and recipient Envs of pair 221; and (iii) the low GPR15/CXCR6 but high APJ use by Envs of pair 55. Together, these data argue against a role for CCR2b, CCR3, CCR8, GPR1, GPR15, CXCR6, or APJ in selection at heterosexual transmission.

Contrary to the initial hypothesis that transmitted variants might use an alternative pathway more efficiently, GPR15 was used significantly less efficiently by recipient viruses than by donor viruses overall. However, this difference was quantitatively modest, and the pattern was not uniform among individual pairs. Nevertheless, since these Envs also reflect “acute” versus “chronic” variants, the greater use of GPR15 by chronic variants than by acute variants might reflect broadened alternative coreceptor use associated with disease progression, as has been noted for subtype B (13), although without acquisition of CXCR4 utilization.

Apart from a potential role in transmission, this study is also the broadest analysis of alternative coreceptor use by subtype C HIV-1 and is the first to test APJ, CCR8, and GPR1. While it is unclear what, if any, level of alternative coreceptor-mediated entry might have biological significance, we arbitrarily selected 1% of that supported by CCR5. Considering donors and recipients independently, and based on the majority of clones within each swarm, we found surprisingly wide use of GPR15 (9 of 16 subjects), CXCR6 (11 of 16), and APJ (12 of 16), with infection efficiencies that were up to or more than 10% of that for CCR5. In contrast, only one set of Envs used CCR3, two used CCR8 (both from the same transmission pair), and none used CCR2b or GPR1. Prior studies have also found little or no CCR2b and CCR3 use by subtype C HIV-1. However, our findings differ somewhat from the previously reported infrequent use of GPR15 and CXCR6 in subtype C (1, 6, 9, 38, 39). Additional isolates will be needed to determine if the use of these molecules is widespread among certain cohorts or if it reflects particular biological properties linked to other aspects of pathogenesis.

The third question we addressed here was whether sexual transmission selected for variants that exhibited distinct patterns in the structural anatomy of how they interact with CCR5. The rationale for this hypothesis was based on the fact that despite common fundamentals in the structural basis for gp120/CD4/CCR5 interactions (25, 45), HIV-1 isolates differ considerably in the details of how they utilize CCR5 (4, 48), and that the structural basis for CCR5 interactions is at least in part regulated by gp120 variable domains. While V3 is most prominent in affecting coreceptor interactions (23, 24), other variable domains are involved as well, and V1/V2, in particular, cooperates with V3 to regulate coreceptor interactions (27). Further informing this rationale, epitope analysis indicates that entry coreceptors may assume distinct configurations on different cell types (3, 7), raising the possibility that selection pressures on Env during transmission might reflect the use of CCR5 on particular target cells. As a group, recipient Envs were somewhat more constrained in their use of CCR5 molecules in which the critical second ECL was replaced with that from CCR2b alone or in conjunction with the first and third loops. While these differences were statistically significant for the donor/recipient viruses overall, the pattern was seen in only some of the pairs. In contrast, both donors and recipients were highly dependent on the CCR5 N terminus, based on chimeric and N-terminal truncation mutants. Taken together, these results do not suggest that transmitted HIV-1 variants exhibit a unique structural basis for interactions with CCR5.

While we selected recipient Envs that were genetically distinct, as a rule they were much more homogenous than donor Envs (15, 22). Biologically, recipient Envs also generally exhibited greater homogeneity than donor Envs, especially for transfection-based studies of alternative coreceptors and CCR5 mutants (Fig. 4, 6, and 7). This was the case in primary cells for most of the pairs as well, particularly in macrophage infection (Fig. 2). However, pair 205 recipient Envs showed considerable diversity in both macrophages and CD4⁺ T cells, even though they are genetically quite similar (22). Because heterogeneity was seen in both macrophages and T cells, it likely reflects different overall infectiousness for primary cells compared to that for indicator cells on which inocula were standardized. While the basis for this diversity remains to be determined, it may be related to the ability to use both CCR5 and CD4 expressed at lower levels, as is the case in primary CD4 T cells and macrophages, respectively, or levels of Env incorporation that might affect interactions at low receptor expression levels. Further studies will be needed to clarify the genetic and functional bases for this diversity. Nevertheless, it shows that in some cases, significant differences in biological properties may result from quite small genetic differences.

It is not clear whether the genetic bottleneck, and presumably biological selection, occurs at the time of actual transmission, or during early outgrowth following local mucosal infection (22, 28, 46). These recipient Envs were obtained 3 to 12 weeks after the transmission event. Those that were isolated earliest (pair no. 205 and 221) are comprised of populations that are highly homologous to the recipient consensus sequence and thus are closely related, if not identical, to “founder” populations (22, 26). Those isolated later or with less precision relative to the actual time of transmission could have been subjected to additional selection pressures during early replication in recipients, including early immune pressures. Nevertheless, the observation of characteristic structural motifs in recipient Envs distinct from those in donors (15) suggests that the relevant biological features responsible for selection are likely still retained, no matter what level at which it occurs, and should be reflected in this analysis.

One consideration here is the considerable heterogeneity among transmitter pairs, which included both MTF and FTM pairs, and Envs derived from both plasma RNA and peripheral blood mononuclear cell provirus. In the analysis presented here, all transmission pairs were evaluated as a group, but separate evaluation of MTF (no. 55, 106, 109, 205) and FTM (no. 53, 135, 153, 221) transmissions did not reveal any more consistent patterns of donor/recipient differences for the parameters tested, nor did separate consideration of plasma or proviral sequences (data not shown). Thus, we think it is unlikely that the heterogeneity obscured important differences, but we cannot completely exclude that possibility.

The factors regulating transmission within discordant couples are highly relevant to the AIDS epidemic worldwide, as it is being increasingly recognized that transmission within discordant partnerships is responsible for a large proportion of new infections (21). At the same time, couples who initially present as discordant may not be representative of all transmission events, in that they might reflect conditions of relatively low transmission efficiency that enabled them to be identified as discordant, and thus, different factors might be operative under other circumstances. Our data indicate that heterosexual transmission of subtype C HIV-1 within discordant partnerships does not select for enhanced macrophage tropism, preferential use of an alternative entry coreceptor pathway, or use of CCR5 in a manner that has major structural differences from variants in donor partners. The biological factors that lead to a viral genetic bottleneck at transmission remain to be identified.