Cloning of full-length HIV-1 env genes.
Full-length env clones (rev/env cassettes) were cloned from short-term-cultured PBMC DNA of 12 primary isolates obtained from patients with acute or early HIV-1 infection (Table ; Fig. ). Corresponding culture supernatants from the infected PBMC were harvested, filtered, and stored in aliquots as reference stocks of the immediate uncloned parental viruses (Fig. ). rev/env cassettes from seven additional acutely infected subjects were cloned directly from plasma viral RNA. Sequence analysis of multiple plasma-derived env clones (9 to 12 clones per patient) revealed a homogeneous virus population for all seven subjects, with average V1V5 nucleotide sequence differences of less than 1.2% (not shown). Multiple clones from both the PBMC viruses and plasma viruses were screened for infectivity in JC53-BL cells as Env-pseudotyped viruses, and representative clones were selected for further analysis. Clones with the highest infectivity were selected for the technical advantage of generating high-titer stocks of Env-pseudotyped viruses for neutralization assays. Multiple clones of each plasma-derived env were screened in neutralization assays with high-reactive and low-reactive plasma pools from HIV-1-infected individuals. Uniform results were obtained with all clones of a corresponding env in most cases, although we did occasionally detect a single outlier clone that was unusually sensitive compared to other clones in the matched set. These outlier clones were not selected for further use in this study. All selected clones used CCR5, but not CXCR4, to gain entry into JC53-BL cells.
Demographic and biologic properties of molecularly cloned, Env-pseudotyped strains of clade B HIV-1
FIG. 1. Generation of pseudovirions containing cloned Env from primary PBMC-cultured isolates. Primary isolates of HIV-1, obtained by standard PBMC coculture, were used to infect fresh cultures of PHA-stimulated PBMC. DNA from the infected PBMC was used for the (more ...)
Acute/early infections occurred between 1994 and 2004 in the United States, Italy, and Trinidad. Thirteen infections arose from male-male transmission, five from female-male transmission, and one from male-female transmission. Plasma viral loads at the time of virus isolation in many subjects were typical of the high levels seen at or near the peak of viremia soon after primary infection (21
). Lower levels of plasma viral RNA in other cases were an indication that some viruses may have been isolated either prior to or after the initial peak in plasma viremia. It was possible to document in 13 cases that viruses were obtained within 6 weeks from the time of seroconversion. No serologic testing was performed for the remaining eight subjects prior to seroconversion. For these remaining subjects, the duration of infection from the time of onset of symptoms of acute retroviral infection (102
) was estimated to be 24 weeks (3988.25), 4 weeks (BG1168.1, PVO.4 and TRO.11), 6 weeks (RHPA4259.7), and < 3 months (CAAN.A2). Subject AC10 was seropositive at the time of virus isolation, but Western blot reactivity indicated a very early stage of seroconversion (73
). Subject SS1196 had no symptoms of acute infection, and therefore the duration of infection at the time of virus isolation (<4 months) was taken as the midpoint between the last negative serologic test and the first positive serologic test.
Sequence analysis of full-length HIV-1 env genes.
Phylogenetic analysis of full-length gp160 nucleotide sequences confirmed that all 19 functional clones grouped within clade B (Fig. ). A wide spectrum of genetic diversity was readily apparent between most clones, as would be expected for independent infections. Only two clones, both from Trinidad (SC422661.8 and QH0515.1), clustered with a high bootstrap value. This linkage is consistent with a previous report on these and other isolates from Trinidad (22
). Deduced amino acid sequences showed an uninterrupted env
open reading frame for 18 of the 19 functional gp160 clones (Fig. ). Clone 6101.10 contained a premature stop codon that is predicted to truncate the terminal 100 amino acids of the cytoplasmic domain of gp41.
FIG. 2. Phylogenetic relationships of newly derived HIV-1 env sequences with subtype B reference strains. The newly characterized sequences are indicated by boxes, and the 12 env clones that are recommended as standardized reference reagents are highlighted by (more ...)
FIG. 3. Alignment of deduced amino acid sequences from acute/early HIV-1 env genes. Nucleotide sequences of newly derived env genes were translated, aligned, and compared with a consensus sequence generated by MASE. Numbering of amino acid residues begins with (more ...)
Natural variation in the native structure of Env trimeric spikes on the HIV-1 virion is thought to influence the neutralization phenotype of the virus. In particular, epitope exposure may be influenced in part by the size and structure of variable loops (11
) and the position of N-linked glycosylation sites (67
). Cysteine residues that form the V1/V2, V3, and V4 loops of gp120 were conserved in all 19 clones (Fig. ). Clone BG1168.1 contained two additional cysteine residues in V1. Another clone (PVO.4) contained an extra cysteine pair in the central portion of V4. Most size variation was seen in the V1/V2 and V5 loops and was most pronounced in the central region of V1 and the C-terminal region of V2 (Table ). Little size variation was seen in V3 and V4.
Moderate variation was seen in the number and position of potential N-linked glycosylation sites in gp120, ranging in number from 22 to 27 sites (Table ) and most heavily clustered in V1/V2 and V4 (Fig. ). Five sites were conserved on all 19 clones, including one site at the N terminus of the V3 loop that has been shown to mask adjacent neutralization epitopes (2
). All clones also contained four highly conserved potential N-linked glycosylation sites in the gp41 ectodomain. Two clones contained a fifth potential N-linked glycosylation site in the membrane-proximal ectodomain of gp41.
Numbers of potential N-linked glycosylation sites in gp120 and gp41
One of our goals in devising a standard panel of clade B env reference clones was to select clones that are not unusually sensitive or resistant to neutralization while at the same time preserving an ability to detect multiple antibody specificities that are not biased toward a particular antibody repertoire. Because the antibody specificities that confer protection against HIV-1 are unknown, a wide spectrum of serologic reagents was used to characterize neutralization phenotypes. Individual serum samples and pooled plasma samples from HIV-1-infected individuals were used as a means to evaluate epitopes targeted during infection. Sensitivity to inhibition by sCD4 was used as a relative measure of epitope exposure in and around the CD4 binding site of gp120. Four human MAbs with broadly cross-reactive neutralizing activity were used, because many novel immunogens aim to generate antibodies with similar specificities. Three of these MAbs were used in combination (TriMab) as a possible broad-spectrum positive control reagent. Additional MAbs were used to probe neutralization epitopes in the V3 loop.
The results of a comprehensive analysis of the neutralization phenotypes of the 19 pseudoviruses containing cloned acute/early infection Env are shown in Table . We included MN, IIIB, and SF162.LS in this analysis to allow comparison to strains that are known to be highly sensitive to neutralization. The acute/early Env-pseudotyped viruses exhibited variable patterns of neutralization by individual patient serum samples and pooled plasma samples. Most of these Env-pseudotyped viruses were relatively insensitive to neutralization by individual serum samples from four HIV-1-infected donors compared to the high sensitivity exhibited by MN, IIIB, and SF162.LS. A fifth HIV-1-positive serum sample, DUMC-1, exhibited potent neutralizing activity against all viruses, which is an unusual property for serum from a single donor. We have found that this serum sample has potent neutralizing activity against a broad spectrum of additional HIV-1 strains regardless of genetic subtype (unpublished data).
Neutralization phenotypes of pseudoviruses containing cloned Env from acute or early HIV-1 infections as determined with serum from HIV-1-infected individuals, sCD4, and MAbs.
Pseudoviruses containing acute/early Env were broadly sensitive to neutralization by HIV-1-positive plasma pools representing clades A through D. The overall potencies of the plasma pools ranked in order clade C > clade B > clade A > clade D. Differences in potency were significant for the clade C pool compared to clade A (P = < 0.001) and D (P = < 0.001) pools. Differences in potency also were significant for the clade B pool compared to the clade A (P = 0.004) and D (P = < 0.001) pools. No significant differences in potency were seen between the clade B and C pools (P = 0.06) and the clade A and D pools (P = 1.0). These results suggest that clade B viruses are more sensitive to neutralization by plasma from individuals who are infected with clade B and C viruses compared to infection with clade A and D viruses.
The geometric mean titer (GMT) of neutralizing activity exhibited by HIV-1-positive serum samples and plasma pools was calculated for each Env-pseudotyped virus and used to rank the clones in order of neutralization sensitivity (Fig. ). Acute/early Env-containing pseudoviruses were considerably less sensitive to neutralization than MN, SF162.LS, and, to a lesser extent, IIIB in a majority of cases. Specifically, these pseudoviruses were 50- to 500-fold less sensitive to neutralization than were MN and SF162.LS and were 5- to 50-fold less sensitive than IIIB. The neutralization phenotypes of two Env-pseudotyped viruses (SS1196.1 and TRJO4551.58) closely resembled that of IIIB.
FIG. 4. Rank order of the neutralization sensitivities of acute/early HIV-1 Env clones as determined with serum and plasma samples from HIV-1-infected individuals. The bar height represents the GMT of the HIV-1-positive serum and plasma samples shown in Table (more ...)
All acute/early Env-pseudotyped viruses were sensitive to sCD4, but, in general, they were much less sensitive than MN, IIIB, and SF162.LS. Specifically, they were 30-fold to 1,770-fold less sensitive than HIV-1 MN and IIIB (mean, 370-fold) and 6-fold to 354-fold less sensitive than SF162.LS (mean, 74-fold).
As expected, all pseudoviruses carrying acute/early Env were neutralized by one or more human MAbs. The broadest sensitivity was seen with IgG1b12 and/or 2G12. IgG1b12 recognizes a complex epitope in the CD4-binding domain of gp120 that is sensitive to mutations in V2 and C3 (15
). 2G12 recognizes a mannose cluster on gp120 that can involve N-linked glycans at multiple sites, including residues 295, 332, and 392, with possible contributions from residues 339 and 386 (9
). Sequence motifs predicting N-linked glycosylation at all of these residues were conserved in 11 Env clones, only one of which (clone TRJO4551.58) was resistant to 2G12 (Fig. ). Two additional clones lacked predicted N-linked glycosylation sites at residues 339 (clone SS1196.1) and 386 (clone 6535.3) but remained sensitive to 2G12, confirming that these two residues are not always essential for 2G12 recognition. The remaining six clones lacked one or more predicted N-linked glycosylation sites at residues 295, 332, 339, 386, and 392 and were resistant to 2G12. In two cases, 2G12 resistance was associated with a loss of predicted N-linked glycosylation at either residue 295 (clone AC10.0.29) or 332 (clone THRO4156.18).
Thirteen acute/early Env-containing pseudoviruses were sensitive to neutralization by 2F5. The epitope for this MAb resides in the membrane-proximal ectodomain of gp41 and involves residues ELDKWAS (3
), with possible contributions from flanking residues (83
). In a recent study of 90 HIV-1 strains belonging to multiple genetic subtypes, the DKW motif proved to be the minimum requirement for 2F5 recognition (9
). Our results are in general agreement with this observation. Thus, 11 of 13 Env clones containing the DKW motif were sensitive to neutralization by 2F5. Two additional Env clones (PVO.4 and THRO4156.18) contained this motif and were not sensitive to 2F5; however, both of these Env clones contained sequence changes elsewhere in the 2F5 epitope. The remaining 2F5-resistant Env clones contained a mutation in the DXW motif.
All 19 acute/early Env-pseudotyped viruses were highly sensitive to neutralization by 4E10. This MAb targets an epitope (NWFDIT) located immediately downstream from the 2F5 epitope in the membrane-proximal ectodomain of gp41 (110
). Recent results suggest that the WFXI motif is the minimum requirement for 4E10 neutralization (9
). In agreement with this observation, D→S was the most common sequence change in the 4E10 epitope of the Env clones examined here (Fig. ). Notably, two clones (AC10.0.29 and WITO4160.33) contained a potential N-linked glycosylation site (D→N) that would be expected to shield the virus from 4E10 recognition. The fact that both Env clones were sensitive to 4E10 suggests an absence of glycans at this site. Biochemical evidence for a lack of glycosylation at this site has been reported previously (55
TriMab exhibited potent neutralizing activity (ID50
of <6 μg/ml) against all acute/early Env-pseudotyped viruses except TRJO4551.58, which resisted neutralization (ID50
of >25 μg/ml). Part of our rationale for evaluating TriMab as a broad-spectrum positive control reagent was that certain combinations of these MAbs have been shown to act synergistically (58
), although others have reported a lack of synergism (9
). Our results support the utility of TriMab as a broad-spectrum positive control reagent, but it is not clear that the extended range of neutralizing activity was due to synergistic effects. In most cases, the potency of TriMab approximated the potency of the single most effective component in the trivalent mixture. For example, all three components of the TriMab mixture neutralized Env clone 6535.3 as single MAbs. Of these, 2G12 was most potent (ID50
of 0.6 μg/ml), whereas TriMab had an ID50
of 1.2 μg/ml (equivalent to 0.4 μg 2G12/ml). These results indicate that strong synergy between the three MAbs in TriMab was rare, at least against clade B viruses. We do not rule out the possibility that weak synergy was present in some cases and went unnoticed in our assays.
Additional assessments were made with six MAbs that target epitopes in the V3 loop of gp120 (42
). Relatively few cases of positive neutralization were detected with these MAbs. SF162.LS and SS1196.1 were notable exceptions in that both were highly sensitive to neutralization by all six MAbs (ID50
of <0.04 to 2). Other exceptions included 6535.3, which was neutralized by five V3-specific MAbs. Also, clones 7165.18 and REJO4541.67 were neutralized by three and two MAbs, respectively. Four additional Env clones (QH0692.42, REJO4541.67, TRJO4551.58, and WITO4160.33) were sensitive to a single V3-specific MAb, while the remaining 10 Env clones were resistant to all V3-specific MAbs tested.
It was important to determine whether the early/acute Env clones were sensitive to neutralization by serum samples from gp120-immunized human subjects. Antibodies elicited by gp120 immunogens often neutralize TCLA strains but exhibit poor neutralizing activity against primary isolates (5
). We aimed to determine whether the Env clones selected for study here would yield similar findings. Sera from 14 recipients of a gp120 immunogen and six control subjects in a recent phase I clinical trial were screened at a 1:10 dilution for neutralizing activity against all 19 early/acute Env-pseudotyped viruses (Table ). Titers of neutralizing Abs against the Env vaccine strain of virus, W61D-TCLA, also were measured. ID50
titers against the vaccine strain ranged from 188 to 9,582, including seven samples that had titers of >1,500. Despite this potent neutralization of W61D-TCLA, little or no neutralizing activity against most acute/early Env-pseudotyped viruses was detected. Occasional weak positive neutralizing activity was detected, but this activity was not significant compared to occasional weak false-positive neutralizing activity (51 to 59% reduction in RLU) detected in sera from control subjects. SS1196.1 was an exception in that ≥90% neutralization of this clone was seen with sera from three gp120-immunized subjects, possibly due to the heightened sensitivity of this virus to neutralization by V3-specific antibodies. This Env clone was more sensitive to neutralization by V3-specific MAbs than all other acute/early infection Env clones examined (Table ) and is highly sensitive to neutralization by V3 loop-specific antibodies generated by a variety of HIV-1 Env immunogens (D. C. Montefiori, unpublished observations). Interestingly, Env clone 6535.3, which was also broadly sensitive to V3-specific MAbs (Table ), was relatively insensitive to the anti-gp120 serum samples.
Neutralization phenotypes of pseudoviruses containing cloned Env from acute or early HIV-1 infections as determined with serum from HIV-Igp120-vaccinated subjects
Sensitivity of Env clones to neutralization by V3-specific MAbs
Neutralization sensitivities of Env-pseudotyped viruses produced in 293T cells compared to uncloned parental viruses produced in PBMC.
Past measurements of HIV-1-specific neutralizing Abs have relied on uncloned virus stocks produced and assayed in human PBMC. Here we used acute/early infection Env clones in pseudoviruses that were produced by transfection in 293T cells and assayed in JC53-BL cells; both of these cell lines are epithelial in origin (40
). These new technologies afford significant technical advantages for assay standardization, validation, and high throughput, making them an attractive alternative to PBMC assays. However, the type of cells used to prepare virus stocks has been shown to influence the neutralization phenotype of the virus by either genetic selection (99
) or the impact of one or more additional host cell factors (32
). Because PBMC are a more natural cell substrate for HIV-1, it was of interest to determine how the neutralization phenotypes of 293T pseudovirions compared to those of PBMC viruses.
The immediate uncloned parental viruses of all 12 molecularly cloned Envs derived originally from PBMC cocultures (Table ) were assayed with sCD4, MAbs, and a subset of the serum samples used for Table . A comparison of ID50
values revealed that the 293T Env-pseudotyped viruses were more sensitive to neutralization by a wide spectrum of antibody specificities than were their PBMC parental viruses (Fig. ). This dichotomy in neutralization sensitivity was best illustrated with serum DUMC-1, which was ~10-fold more potent against 293T pseudoviruses than against PBMC parental viruses. In addition, 293T pseudoviruses were ~22-fold more sensitive to inhibition by sCD4. The greatest discrepancy was seen with 4E10, which was very potent against 293T pseudoviruses (ID50
range of 0.1 to 8.5 μg/ml) (Table ) but had little or no detectable neutralizing activity against PBMC parental viruses (ID50
range of 17 to >50 μg/ml) (Fig. ). Similar results with 4E10 have been reported previously (9
). We observed no cases where a 293T pseudovirus was more resistant to neutralization than its PBMC parental virus. These results indicate that Env-pseudotyped viruses produced in 293T cells are more sensitive for detecting neutralizing Abs in JC53-BL cells than are PBMC-grown viruses.
FIG. 5. Neutralization phenotypes of acute/early HIV-1 Env clones compared to the phenotypes of the immediate uncloned parental viruses. Uncloned parental PBMC-grown viruses (gray bars) and the corresponding Env-pseudotyped viruses produced in 293T cells (black (more ...)