Human Immunodeficiency Virus Type 1 Subtype B Ancestral Envelope Protein Is Functional and Elicits Neutralizing Antibodies in Rabbits Similar to Those Elicited by a Circulating Subtype B Envelope

Ancestral state simulations.

To assess the accuracy of the ancestral state reconstruction method, we performed the following in silico experiment. Using the program Seq-Gen (55), the evolution of HIV-1B env sequences was simulated, given an ancestral sequence and following the phylogenetic histories depicted in the trees in Fig. 1a, below, and the model of evolution provided in Table S1 in the supplemental material. One hundred replicate data sets were generated for both tree shapes. We then used PAUP* (63) to generate a maximum-likelihood tree and an ancestral sequence for all 200 data sets (100 simulations on both the star- and caterpillar-shaped phylogenies). To be conservative, the simulations were performed on trees with very long external branches (having a genetic distance of 20%), and the caterpillar trees had internal branch lengths of 1% genetic distance.

Ancestral state reconstruction of the SIVmacBK28 env sequence.

SIVmac sequences obtained from a series of experimental infections of rhesus macaques (15) were used to reconstruct an ancestral sequence to compare to the plasmid-derived simian immunodeficiency virus (SIV) clone used to inoculate these monkeys. A region of env from position 8265 to 8827 of the SIVmacBK28 genome was amplified by nested PCR from uncultured peripheral blood mononuclear cell (PBMC) DNA from the SIV-infected macaques. First-round primers were UP-3, positions (SIVMM239 coordinates [8]) 8117 to 8130, AGACTGCAGATGTGAAGAGGTACAC, and PEXTM6 (8977 to 8953), GGATCTGGTATGCTCATAGCAA. Second-round primers were PEXTM7 (8265 to 8286), GATACTGCAGCAACAGCAACAGCTG, and UP-5 (8827 to 8810), GCAAAGCTTCTCTGGTTGGCAGTG. Amplified products were then cloned and sequenced as described previously (14). The GenBank accession numbers for these sequences are AY169007 to AY169163. Methods for SIVmacBK28 ancestor reconstruction were identical to those outlined below. The model of evolution (Table S1) and the reconstructed sequence (Fig. S1) are provided in material available at the website http://ubik.mullins.microbiol.washington.edu/HIV/Doria-Rose2005/.

Ancestral state reconstruction of an HIV-1 B env sequence.

Thirty-eight HIV-1 subtype B env genes were selected from GenBank, with three additional sequences from clade D used as an outgroup for rooting (35) (see Fig. 1c, below, and also Table S2 in the supplemental material). Sequences were aligned using CLUSTALW (64), and the alignment was refined using GDE (58). Inferred amino acid sequences were used to guide the introduction of alignment gaps such that they were inserted between codons. This alignment was then modified for phylogenetic analysis such that regions that could not be unambiguously aligned were removed (38).

An appropriate evolutionary model for phylogeny and ancestral state reconstructions was selected using the Akaike information criterion as implemented in Modeltest 3.0 (53) (Table S1). Evolutionary trees were inferred using maximum-likelihood estimation methods as implemented in PAUP* version 4.0b10 (63). Ten different subtree-pruning-regrafting heuristic searches were performed using a different random-addition order each time and using a neighbor-joining tree of the subtype B sequences as a backbone constraint. The ancestral nucleotide sequence was inferred as the sequence at the basal node of the clade using the inferred phylogenies. The nucleotide at each site of the ancestral sequence had the highest likelihood, integrated over all state assignments at other nodes. The nucleotide sequence is provided in material posted on the website http://ubik.mullins.microbiol.washington.edu/HIV/Doria-Rose2005/ (Fig. S2).

To predict the amino acid sequences for the complete gp160, the inferred ancestral sequences were visually aligned to the complete alignment prior to gap stripping and then translated using GDE (58). Since the highly variable regions were deleted as complete codon triplets, the translations into amino acids were in the correct reading frame and codons were properly maintained. The ancestral amino acid sequences for the regions deleted from the gap-stripped alignment were predicted visually and refined using parsimony-based sequence reconstruction for these sites using the program MacClade (42). The complete gp160 amino acid sequences were converted to DNA sequences optimized for expression in human cells using the BACKTRANSLATE program of the Wisconsin Sequence Analysis Package (GCG), version 10, and a human gene codon usage table from the Codon Usage database (http://www.kazusa.or.jp/codon/cgi-bin/showcodon.cgi?species = Homo+sapiens+[gbpri]) (48).

The AN1-envB gene was chemically synthesized commercially by Midland Certified Reagent Company (Midland, TX). It was sequence verified and inserted into a mammalian expression vector, pVR1012 (27), to produce the plasmid pVR1012-AN1env. To generate a gp140 expression plasmid, pVR1012-AN1env was subjected to site-directed mutagenesis (QuikChange kit; Stratagene) to introduce two stop codons after the codon for amino acid lysine 711 (nucleotides 2134 to 2139). Mutagenesis was verified by sequencing.

Fusion assay.

GHOST-CCR5 or GHOST-CXCR4 cells (gift of S. Zolla-Pazner) (10) were seeded in two-chamber culture slides (Falcon, Franklin Lakes, NJ) at 5 × 10⁴ cells/well and transfected the next day with 0.5 μg env plasmid using Fugene-6 (Roche, Indianapolis, IN). Transfected cells were allowed to fuse for 2 days and then were fixed and stained for immunofluorescence using the monoclonal antibody b12 (gift of D. Burton) and 1:2,500 fluorescein isothiocyanate-conjugated goat anti-human immunoglobulin G (IgG; Cappel). The presence of large multinucleated cells indicated that cell-cell fusion had occurred.

Pseudotype assay.

293T cells were transfected in 75-cm flasks with 4 μg env plasmid, 4 μg pNL43lucR-E- (28), and 12 μl Fugene-6. Two days later, medium was collected and spun at 3,000 × g, and the resulting cell-free pseudovirus was stored at −80°C. Pseudovirus was used to infect TZM-bl cells (12) in the presence of 7.5 μg/ml DEAE-dextran. Two days later, infection was measured by luciferase expression using a commercial kit (Bright-Glo; Promega, Madison, WI) and an EG&G Berthold MicroLumat Plus luminometer. Alternatively, 293T cells were transfected with env plasmids, and Q23Δenv (40) (gift of J. Overbaugh) and pseudovirus were used to infect cMAGI cells (11) (gift of J. Overbaugh) in the presence of 10 μg/ml DEAE-dextran. After 2 days, cells were fixed and stained for β-galactosidase expression (indicating infection).

Western blot analysis.

293T cells were transfected in six-well dishes with 1 μg env plasmid and 6 μl Fugene-6, with medium and cell lysates collected 48 h later. Twenty-one microliters of each sample was used for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to a polyvinylidene difluoride membrane, probed with heat-inactivated pooled human HIV⁺ serum (gift of L. M. Frenkel), and horseradish peroxidase-conjugated goat anti-human IgG (Sigma, St. Louis, MO) and then developed with the color substrate 3-amino-9-ethylcarbazole.

Recombinant protein production.

Recombinant gp120 from HIV-1 SF162 was prepared as described elsewhere (3). AN1-EnvB rgp120 was derived from the synthetic gp160 gene by site-directed mutagenesis to place double termination codons immediately after the gp120 reading frame. The BstEII-NotI fragment of the gene, containing most of the gp120 and all of the gp41 reading frame, was excised and cloned into the cognate region of plasmid pSF128, which consists of the pcDNA3.1+/Zeo expression vector (Invitrogen, Carlsbad, CA) and a synthetic gene for HIV (Ba-L) gp160, which in HIV-1 signal peptide was replaced with that for human CD5. Expression of this plasmid in eukaryotic cells is regulated by the cytomegalovirus early promoter and the bovine growth hormone polyadenylation site. The resultant clone, pSF182, encodes a protein with the CD5 signal peptide, which was removed during expression, and a gp120 in which the first 10 amino acid residues are derived from HIV (Ba-L) and the remainder are from AN1-Env B.

pSF128 was purified using an Endofree pDNA kit (QIAGEN) and transfected into A293 cells using Fugene-6 (Roche) at a ratio of 6 μl of Fugene to 2 μg DNA. Cells were selected with zeocin at 200 μg/ml and 10% fetal bovine serum for 5 days and then transferred to T-75 flasks. When foci appeared, weaning to serum-free medium was initiated using an increasing proportion of 293-SFM II medium from Invitrogen and gradually reducing the zeocin concentration. Cells were considered weaned when they were able to grow for three passages in 100% 293-SFM II-20 μg/ml zeocin. Large quantities of 293 cells expressing AN1-Env B gp120 were then grown in roller bottles in 100% 293-SFM II-20 μg/ml of zeocin. The culture supernatant was subjected to lectin affinity chromatography using a plant-derived lectin isolated from snow drop (Galanthus nivalis) bulb. The affinity matrix, in which the lectin was covalently linked to agarose beads, was washed thoroughly to remove nonspecifically bound proteins. The bound protein was then eluted with alpha-methyl-d-mannopyranoside and dialyzed against phosphate-buffered saline (PBS). Production of gp120 was monitored by enzyme-linked immunosorbent assay (ELISA) at all stages of production, and purity was verified by SDS-PAGE. The SF162 protein was prepared as previously described (60).

Rabbit immunizations.

New Zealand White rabbits were vaccinated by Gene Gun (Bio-Rad, Hercules, CA) (41). Eighteen immunizations of 2 μg DNA each were given for each dose in clusters of three nonoverlapping positions at six sites (back, legs, and abdomen); rabbits were shaved in these areas prior to immunization. Animals were DNA vaccinated at weeks 1, 5, 10, 19, 23, and 46. Protein boosts, consisting of 44 μg recombinant gp120 mixed with an equal volume of incomplete Freund's adjuvant, were injected intramuscularly at weeks 68 and 78. Blood was collected the day before and 2 weeks after each immunization. Animals were housed at R+R Rabbitry, Marysville, WA, and procedures followed IACUC-approved protocols.

IgG purification.

Individual samples (800 μl) of rabbit sera, collected at weeks 5 and 80 of vaccination, were heat inactivated at 56°C for 1 hour prior to purification. The samples were each mixed with 800 μl of 4 Fast Flow Protein A Sepharose (Pharmacia) slurry and incubated for 1 hour at room temperature. Resin was collected and washed three times in PBS using two Handee spin cups (Pierce, Rockford, IL). IgG was eluted in 400 μl 0.1 M glycine-HCl, pH 2.7, for 5 minutes; the eluate was collected immediately by centrifugation and neutralized with 40 μl of 3 M Tris, pH 9.5. Samples were then equilibrated to PBS by dialysis using Slide-a-lyzer cassettes, with a molecular mass cutoff of 10,000 Da (Pierce, Rockford, IL). Thirty-five to 60% of the binding activity was recovered as determined by ELISA. Purity was ∼90%, as estimated by Coomassie staining of SDS-PAGE gels.

Antibody assays.

Binding antibody responses to HIV-1 envelope antigens were measured by ELISA as described previously (13). Briefly, Immunosorp plates (Nalge Nunc, Rochester, NY) were coated with 2 μg/ml HIV-1 SF162 rgp120. Diluted plasma was incubated for 1 hour on the plates and detected with horseradish peroxidase-conjugated anti-rabbit IgG (Kirkegaard and Perry, Gaithersburg, MD). The endpoint titer was defined as the reciprocal of the highest dilution that gave an absorbance value more than twice that of preimmune serum at the corresponding dilution. Values were standardized to a positive control rabbit serum sample (kind gift of L. Stamatatos) that was included with every assay.

Neutralization assays: cMAGI assay.

Assays were performed as described in reference 13. Briefly, serial dilutions of sera were incubated with virus for 1 hour and then added to duplicate wells of cMAGI cells (11). After 2 days, cells were fixed and stained for β-galactosidase expression (indicating infection). The percent neutralization at a given titer was calculated by using the equation [(Vo − Vn)/Vo] × 100, where Vn is the number of infected cells in the virus plus antibody wells and Vo is the number of positive cells in virus-alone wells. We graphed percent neutralization versus dilution for each sample and matched prebleed sera. Sera that failed to achieve 80% neutralization or were not distinguishable from prebleed samples were scored as negative. Subtraction of prebleed activity at each dilution resulted in the same value assignments for titers. Titers were normalized to that of a standard HIV⁺ human serum pool (gift of L. M. Frenkel) that was included on each assay plate. Viruses HIV-1 SF162, 92US657 (also called 301657), 92TH014, 92HT593, 91US056, and 93IN101 were obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health, and were propagated in anonymous donor human PBMCs.

Neutralization assays: luciferase reduction assay (M7-luc assay).

Neutralization was measured as a function of the reduction in luciferase reporter gene expression after multiple rounds of virus replication in 5.25.EGFP.Luc.M7 cells (43). Briefly, 5,000 50% tissue culture infective doses of virus was mixed with serial dilutions of serum in triplicate and incubated for 1 h at 37°C in 96-well microculture plates. A total of 5 × 10⁴ 5.25.EGFP.Luc.M7 cells (gift of N. Landau), suspended in RPMI with 12% heat-inactivated fetal bovine serum and 10 μg/ml DEAE-dextran, were added to each well and incubated for 3 to 4 days. These cells possess Tat-responsive reporter genes for luciferase (Luc) and green fluorescence protein (GFP). Luciferase was measured using a commercial kit (Bright-Glo; Promega, Madison, WI) and a Victor-2 luminometer.

Neutralization titers were calculated as the dilution at which the relative luminescence units (RLU) were 50% that of virus-only wells. For the large panel of viruses, sera from week 80 (post-eighth vaccination) were assayed at a 1:10 dilution in triplicate, and the percent reduction in RLU was calculated relative to the corresponding week-5 serum sample (shown to be equivalent to preimmunization sera in these assays) for each rabbit. A reduction of 50% is considered significant in this assay. Virus stocks for both assays were generated in phytohemagglutinin-stimulated normal human PBMCs and made cell free by 0.45-μm filtration prior to storage at −80°C.

Phylogenetic method validation.

We first evaluated the ability of likelihood-based phylogenetic methods to accurately predict ancestral sequences using both a simulation and an experimental approach. HIV-1 is widely considered to have a star-like phylogeny, with all external branches radiating from the same central point on the tree (1). However, there is at least some level of substructure in the phylogeny (i.e., some short internal branches at the base of the tree). In terms of tree space, the realized HIV-1 tree (Fig. 1) is between a true star and what can be called a caterpillar tree (i.e., long external branches with short internal branches). To estimate the accuracy of ancestral state reconstruction, we performed 100 simulations with a known ancestor at each of the two extremes of the star-caterpillar continuum. For the caterpillar tree, the MRCA was estimated with 95.4% accuracy. For the star phylogeny, the MRCA was estimated with 98.2% accuracy. Since the true HIV phylogeny falls somewhere between the caterpillar and star structures, we estimate our MRCA reconstruction accuracy to be in the range of 95 to 98%.

We also evaluated viral gene sequences taken from four rhesus macaques infected with a molecularly defined strain of SIV, SIVmac251-BK28 (34), in an effort to assess the reconstruction of the infecting viral sequence. From a phylogenetic tree (Fig. 1b) of 1.1-kb env gene fragments taken between 1 and 3 years postinfection (15), 99.8% of the infecting viral sequence (373 of 376 amino acids), including 98.2% of the 170 variable sites, were accurately predicted (see Fig. S1 at http://ubik.mullins.microbiol.washington.edu/HIV/Doria-Rose2005/).

Derivation of subtype B ancestor.

Using the same procedures, we derived an ancestral nucleotide sequence of the envelope gene of HIV-1 subtype B, the most extensively evaluated HIV-1 subtype to date. We used the sequences listed in Table S2 in the supplemental material and the phylogeny shown in Fig. 1c. This ancestral sequence produced an open reading frame encoding a complete, 884-amino-acid gp160 gene product (AN1-EnvB). The nucleotide sequence is available as Fig. S2 in the supplemental material. The amino acid sequence is shown in Fig. 2, along with potential N-linked glycosylation sites (NXS/T sequons). The amino acid distances between the ancestral sequence and the natural subtype B strains used to estimate it were 12.3% on average (range, 8.0 to 21.0%), while these sequences were 17.3% different from each other (range, 13.3 to 23.2%). At 884 amino acids, the encoded protein is long relative to most sequences (average, 858 amino acids); much of the additional length is in the variable loops. Also unusual is the number of NXS/T sequons: the median for gp120 is 25 (33, 67), while AN1-EnvB has 31 in gp120. Based on V3 sequence features, AN1-EnvB is also predicted to use CCR5 as a coreceptor (30).

In vitro validation of ancestor protein expression.

We next sought to evaluate the functional characteristics of the ancestral env sequence. We reengineered the deduced sequence to reflect the dominant patterns of human codon usage (48) without changing any of the encoded amino acids (codon optimized) and then chemically synthesized the gene and cloned it into a mammalian expression vector, pVR1012 (27). Transfection of this plasmid into COS-7 or 293T cells resulted in the production of high levels of viral protein gp160 and its cleavage products, gp120 and gp41. A truncated gp140 form, which is expected to be secreted but retain the oligomeric properties of gp160, was also generated, expressed, and shown to bind soluble CD4 in an ELISA, as did gp160 (data not shown). Western blot analysis of 293T cells transfected with either the gp160 or gp140 form showed that the expressed proteins were of the predicted sizes and were recognized by anti-HIV antibodies (Fig. 3a). Cell lysates contained both gp140 and gp160, while only the gp140 form appeared in the medium, showing, as expected, that it was secreted. Lysates of cells transfected with gp160 showed bands for gp41, suggesting that Env was cleaved to its mature form intracellularly. The cleavage was not complete, as full-length gp160 was also visible. Little, if any cleavage to gp120 was noted in the gp140 samples. The apparent molecular weight and somewhat diffuse nature of the gp140 and gp160 bands suggest that the AN1-EnvB proteins were highly glycosylated. Western blot and quantitative ELISA showed that levels of the gp160 form of AN1-EnvB gp160 were comparable to levels detected following transfection of the widely employed, humanized Env-gp160 gene of SHIV89.6P (4) and higher than that of nonoptimized HIV-1 89.6 Env (Fig. 3a). Expression of the gp140 form was similar to that of a codon-optimized gp140 from HIV-1 SF162 (3) (data not shown).

Expression of gp160 in transfected cells was confirmed by an immunofluorescence assay (Fig. 3b and c). Cell-cell fusion occurred when GHOST-CCR5 cells (10) expressing the HIV-1 receptors CD4 and CCR5 were transfected with AN1-EnvB gp160 (Fig. 3b). In contrast, no fusion occurred when GHOST-CXCR4 cells (10) expressing CD4 plus CXCR4 were transfected (Fig. 3c), although Env was expressed in these cells. These results indicate that AN1-EnvB used only CCR5, as predicted by its sequence, and that it was folded properly and fully functional for directing membrane fusion. To confirm the functionality of AN1-EnvB, we used pVR1012AN1env (encoding full-length gp160) to complement the env-defective genome Q23Δenv (40). The resulting pseudotyped virus particles were able to infect permissive cMAGI cells (data not shown). We also used the same plasmid to complement the env-defective genome encoded by pNL4-3lucR-E- (28), and the resulting pseudovirus was infectious on TZM-bl cells (data not shown).

Immunogenicity of AN1-EnvB.

To assess the humoral immune response elicited by An1-EnvB, we immunized mice with pVR1012AN1env using a Gene Gun and found binding of antibody to HIV-1 gp120 at titers up to 1:300 (data not shown). We then immunized rabbits with DNA encoding gp160 and gp140 versions of the protein via the Gene Gun (19, 41) followed by boosting with purified recombinant AN1-EnvB gp120 protein produced in CHO cells. As controls, we immunized an additional group of rabbits with DNA encoding a codon-optimized Env gp140 gene from a primary HIV-1 isolate, HIV-1 SF162, and recombinant HIV-1 SF162 gp120 (rgp120) (3), or with the empty vector and adjuvant. All envelope plasmids elicited envelope-specific binding antibody, with endpoint titers in rabbits of up to 1:400,000 (Fig. 4a). Binding antibody titers increased after protein boosting (roughly eightfold when geometric mean titers of all animals at each time point are compared).

Virus neutralization.

We next measured antibody neutralization, first using the cMAGI assay (11, 32) with five heterologous subtype B, primary R5 HIV-1 isolates, SF162, 92US657, 92TH014, 91US056, and 92HT593, and one subtype C primary isolate, 93IN101 (Table 1). As a positive control, we included HIV⁺ sera pooled from several HIV-infected patients from the Seattle, Wash., area. All isolates tested were neutralized by this latter pool, but with differing sensitivities. Sera from the rabbits were analyzed after four or five immunizations with DNA against the same panel of isolates and compared with the preimmunization sera. After four DNA immunizations, all four gp140-SF162-immunized rabbit sera had neutralizing activity with 80% neutralization titers of >8 against the homologous virus. AN1-immunized rabbits had neutralization detected at this time point only using lower cutoff criteria (50 to 75%) (data not shown). After five immunizations, neutralization was assessed against the full virus panel. At this time, sera from only one SF162-immunized animal neutralized homologous virus, and one other rabbit had neutralizing activity against a heterologous virus (92US657). However, sera from this time point had atypically low neutralizing activity (Fig. 4b). In contrast, all eight rabbits immunized with AN1-EnvB genes neutralized isolate 92TH014 and four of eight neutralized 92US657. No other neutralizing activity was observed. Hence, while not very broad, more heterologous neutralizing activity was evident in rabbits that received AN1-EnvB compared to those immunized with SF162 Env. No rabbit sera neutralized 92HT593, 91US056, or the clade C virus 93IN101.

Whereas the above sera had low or no neutralization of HIV-1 SF162 in the cMAGI assay, activity was usually detected by a second neutralization assay that targets the luciferase-expressing 5.25.EGFP.Luc.M7 cells (7), herein referred to as the M7-luc assay. Sera were tested for neutralization activity against HIV-1 SF162 2 weeks after each immunization with DNA or with a gp120 subunit protein boost. Titers were calculated as the dilution that gave 50% reduction of luciferase relative to wells with no antibody. Significant titers of neutralizing antibodies against HIV-1 SF162 were found in all Env-immunized rabbits (Fig. 4b). M7-luc neutralizing activity generally increased over the course of vaccination and reached a mean titer of 800 to 1,000 in the rabbits immunized with DNA expressing gp140 after six immunizations (Fig. 4b). Neutralization titers were boosted roughly twofold with gp120 protein immunization at week 67 and sustained with a second protein boost at 78 weeks. Titers of neutralizing antibody against HIV-1 SF162 were not significantly different between any of the groups at weeks 21, 25, 48, or 80 (Mann-Whitney U test).

Using the M7-luc assay, sera from all rabbits were tested after the final DNA vaccination and after each protein boost against a panel of heterologous B, C, and E primary isolates. In this assay, all sera were diluted 1:10, results were recorded as percent reduction of luciferase compared to week 5 sera, and values greater than 50% were considered significant. Serum from one control rabbit scored at the 51% level, but other samples from controls were negative. Week 5 samples were used as controls because of a scarcity of preimmunization samples; these had no or minimal Env binding activity, and background neutralization titers were very close to those of corresponding preimmunization sera for the same animals (data not shown). Data from samples taken after the last protein boost, reported as %RI, are shown in Table 2. Sera from three of three rabbits that received HIV-1 SF162 immunogens (the fourth died prior to protein boosting) could neutralize Bx08, two were active against Ba-L, and one of those was also active against QH0692. Sera from rabbits immunized with AN1-EnvB gp160 or gp140 also had activity against these three viral isolates. Three of eight rabbit sera neutralized Bx08, and two of those sera also had activity against Ba-L and QH0692. Activity against BaL and Bx08 was also detected at weeks 48 and 70 in several animals (not shown). No rabbit sera neutralized clade B isolate 6101, clade C isolate S080, or clade E isolate CM244.

To confirm that the observed virus-neutralizing activity was due to antibodies and not to nonspecific inhibition by other components of the serum, we purified IgG from sera of six of the rabbits and used these in our assays. Three rabbits had received AN1-EnvB immunogens, and three had received the HIV-1 SF162 immunogens. IgG and the corresponding sera (Table 2) were tested in the M7-luc system. For this subset of samples, we obtained titers for 50% neutralization of each isolate, allowing more precise quantitation of activity in the serum. All week 80 IgG samples had strong neutralizing activity against HIV-1 SF162, with 50% inhibitory concentration (IC₅₀) values from 1.6 μg/ml to 61 μg/ml; likewise, the sera from which they were purified had high titers against this virus, from 1:494 up to 1:4,176. The sera with the highest titers yielded the IgG samples with the lowest IC₅₀ values, as expected. IgG from two of the six also had strong activity, and one showed weak activity, against Bx08. In this set of assays, no significant neutralization of isolate 6101 was found. Weak neutralization of BaL was noted in all serum samples, and sporadic low-level activity against three other isolates was noted in several sera, but not in the corresponding IgG; this may be due to the limits of detection of the assay. IgG samples from week 5 had no detectable activity at the highest concentrations tested. These data demonstrate that the observed neutralizing activity of the rabbit sera is attributable in large part to IgG.

Comparing data across the two assays, we found that five of the eight primary heterologous clade B viruses tested were weakly neutralized by AN1-Env B sera. The clade B viral amino acid sequences used here were, on average, 9% distant from AN1-EnvB and 13% from HIV-1 SF162. The non-B amino acid sequences were on average 24% and 22% divergent from AN1-EnvB and SF162, respectively. Within the clade B group, neutralization of individual viruses by vaccinated rabbit sera did not correlate with total sequence distance between the vaccine immunogen and the tested virus. The number of viruses neutralized by serum from each rabbit was not statistically different between the AN1-EnvB- and SF162-immunized rabbit groups (Mann-Whitney U test). Overall, the neutralizing antibodies elicited both by AN1-EnvB and by HIV-1 SF162 immunogens had modest potencies and breadth.