INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Accumulating evidence supports a positive role for neutralizing antibodies and CD8⁺ cytotoxic T lymphocytes (CTL) in the antiviral immune response to human immunodeficiency virus type 1 (HIV-1) and simian immunodeficiency virus (SIV) (33, 44). The ability to generate antibodies that neutralize a broad spectrum of primary HIV-1 isolates has proven to be one of the more difficult challenges for vaccine development (44). Major targets for neutralizing antibodies are the surface gp120 and transmembrane gp41 envelope glycoproteins of the virus (12). Antibodies that bind these viral glycoproteins with adequate affinity and appropriate specificity can prevent HIV-1 from entering host cells (32, 41, 45, 71, 78, 83, 84) and have provided potent protection against infection in passive-antibody experiments in animal models (2, 20, 22, 35, 38, 76). Attempts at inducing cross-reactive neutralizing antibodies through vaccination have, however, had limited success. Most attempts have focused on the envelope glycoproteins of a small number of T-cell line-adapted (TCLA) strains of virus. Although the envelope glycoproteins of those strains generate high titers of neutralizing antibodies, those antibodies mostly target strain-specific epitopes in the third variable cysteine-cysteine loop (V3 loop) of gp120 (10, 45) and have been highly specific for TCLA strains of virus (23, 50). Importantly, the antibodies have failed to neutralize primary isolates (4, 5, 10, 37) and often possess little or no detectable neutralizing activity against heterologous TCLA strains (5, 10).

The B-cell response leading to HIV-1-specific neutralizing antibody production could be fundamentally different for primary isolates compared to TCLA strains. For example, whereas TCLA strains are highly sensitive to neutralization by V3 loop-specific antibodies (26, 68), the V3 loop on primary isolates is occluded by N-linked glycans and tertiary folds on the native gp120 molecule (30, 62, 73-75, 90, 91), making it a poor target for neutralizing antibodies (9, 77, 86). The envelope glycoproteins of TCLA strains and many, but not all, primary isolates also exhibit different coreceptor preferences for virus entry. Generally speaking, primary isolates utilize either CCR5 (R5), CXCR4 (X4), or both coreceptors (R5/X4) whereas all TCLA strains have the X4 phenotype (6, 7, 67). For reasons that are poorly understood, most transmitted strains of HIV-1 have an R5 phenotype (63, 87) and, although no association has been found between coreceptor preference and neutralization sensitivity (31, 43, 81), it seems prudent to target R5 strains when designing candidate HIV-1 vaccines.

Much of what is known about the neutralization epitopes on primary isolates comes from studies of three human monoclonal antibodies (MAbs), immunoglobulin G1b12 (IgG1b12), 2G12, and 2F5. Each of these MAbs can neutralize diverse primary isolates (12) and, when combined, exhibit synergistic neutralizing activity (36). A vaccine that generates antibodies equal to the combination of these specificities is highly desirable and deserving of intense investigation. Unfortunately, the corresponding epitopes have proven to be poorly immunogenic in natural infections and in experimentally immunized animals. Other neutralization epitopes are present on primary isolates that are less conserved. The latter epitopes account for the sporadic, low-level neutralization of heterologous isolates by sera from infected individuals (46, 58, 89). They might also account for the potent neutralization of primary isolates by autologous serum samples that are obtained months or years after the time of virus isolation (29, 48, 56, 58, 89), which would require the cognate epitopes to be immunogenic and well exposed on the virus surface. The presence of neutralization epitopes that are both immunogenic and antigenic on primary isolates affords opportunities for vaccine development that have received little attention.

Part of the reason why the latter epitopes have received little attention for vaccine development relates to their high degree of variability, which may not be suitable for cross-reactive neutralizing antibody induction. Given the paucity of information on the immunogenicity of primary isolate Env, however, it would be premature to conclude that no monovalent Env will generate cross-reactive neutralizing antibodies. Also, should monovalent primary isolate Env be capable of generating antibodies that neutralize the autologous primary isolate, it might ultimately make it feasible to formulate a composite (polyvalent) Env vaccine that will target a broad spectrum of primary isolates.

Few studies have examined the immunogenicity of primary-isolate envelope glycoproteins. VanCott et al. (85) showed that immunization with the gp120 protein from the R5 primary isolate HIV-1CM235 generated antibodies in baboons that neutralized TCLA strains, including the TCLA version of the vaccine strain, but not the autologous primary isolate. Similar results were obtained by Beddows et al. (4), who immunized healthy, HIV-1-seronegative volunteers with the gp120 protein from R5/X4 strain HIV-1W61D. Only Berman et al. (8) have succeeded at generating antibodies that neutralized an R5 primary isolate; this was done by immunizing rabbits with the gp120 protein from HIV-1CM244.

In addition to neutralizing antibodies, increasing evidence supports a critical role for virus-specific CD8⁺ T cells in protective immunity against HIV-1 (11, 15, 55, 61, 64, 69, 70, 92). CD8⁺ T-cell responses to HIV-1 can be both cytotoxic (11, 15, 55, 61, 69, 92) and virus suppressive (80, 88). Recent reports of the effect of CD8⁺ cell depletion on viremia and CD4⁺ cell counts in SIV- or SIV-HIV-infected macaques (27, 40, 72) has given additional in vivo relevance to the importance of the CD8⁺ antiviral response in the control of this viral infection. Even more recently, vaccine-induced, virus-specific CTL strongly correlated with viremia control and the preservation of CD4⁺ lymphocytes in the highly pathogenic SIV-HIV 89.6P-macaque model (3).

The immunogenicity of Gag, Pro, and Env from R5 primary isolate HIV-1_Bx08 was investigated here in rhesus monkeys. Immunogens were delivered in the form of either purified noninfectious virus-like particles (VLP), recombinant DNA and canarypox vectors engineered to express VLP, or a combination of delivery modes. Our results reveal the priming of Gag-specific cellular responses by DNA-based immunization. We also demonstrate that envelope glycoprotein from an R5 primary isolate can be used to generate antibodies in nonhuman primates that are capable of neutralizing the autologous primary isolate.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and viruses. MT-2 is a CD4⁺ human lymphoblastoid cell line that is highly permissive to cytopathic infection with TCLA strains of HIV-1 (24). Stocks of HIV-1 strains IIIB, MN, and SF2 were produced in H9 cells as previously described (47). MT-2 and H9 cells were maintained at 37°C in 5% CO₂ in growth medium consisting of RPMI 1640 medium supplemented with 12% heat-inactivated fetal bovine serum (FBS) and 50 µg of gentamicin/ml. Peripheral blood mononuclear cells (PBMC) were purified from buffy coats of healthy HIV-1-seronegative individuals and stimulated for 1 day with phytohemagglutinin P as described previously (46). Growth medium for PBMC consisted of RPMI 1640 medium supplemented with 20% heat-inactivated FBS, 4% interleukin-2, and 50 µg of gentamicin/ml. Monkey kidney Vero cells used for Bx08 VLP production were grown and passaged biweekly in Dulbecco's modified Eagle's medium (Flow Laboratories, McLean, Va.) supplemented with 10% heat-inactivated FBS, glutamine (2 mM), penicillin (50 IU/ml), and streptomycin (50 µg/ml). Stably transfected cells secreting VLP were maintained in the presence of 0.5 mg of Geneticin (Gibco BRL, Grand Island, N.Y.) per ml.

sCD4 and MAbs. Recombinant soluble human CD4 (sCD4) containing the full-length extracellular domain of human CD4 was obtained from Progenics Pharmaceuticals (Tarrytown, N.Y.). Human MAbs IgG1b12, 2G12, and 2F5 have been shown to neutralize diverse primary isolates (14, 19, 28, 59, 82). IgG1b12 recognizes an epitope in the CD4-binding domain of gp120 that is sensitive to mutations in V2 and C3 (42). 2G12 recognizes an epitope comprising residues within C2-V4 of gp120 that involved sites of N glycosylation (82). 2F5 recognizes a linear epitope in the ectodomain of gp41 having the amino acid sequence ELDKWA (51).

DNA vectors. The nucleotide and amino acid numbering used throughout for HIV was that of Myers et al. (52). All pseudovirion-expressing vectors were constructed from p83-19 (57, 65), which was, in turn, derived from plasmid pMTHIVd25 (66). Plasmid p83-19 was engineered to incorporate several mutations to enhance safety, such as deletion of a 25-bp DNA fragment (nucleotides 753 to 777; LAI sequence) containing viral RNA packaging sequences (1, 34) and deletion of a 1.9-kbp DNA fragment containing most of the pol gene, effectively eliminating the coding sequences for reverse transcriptase and integrase. However, the deletion left intact the gene encoding the viral protease (pro), resulting in the expression of particles with processed Gag antigens. In p83-19, transcription of the HIV-1 coding sequences is regulated by the inducible human metallothionein II_a (MT) promoter and a simian virus 40 polyadenylation sequence. In order to produce pseudovirions with the Env glycoprotein from the Bx08 primary isolate, the HIV-1_LAI env gene was replaced with that of HIV-1_Bx08. To this end, a 2,440-bp fragment containing the entire gp160-encoding gene of HIV-1_Bx08 was amplified by PCR from cells infected with this isolate. The PCR product was then used to replace the corresponding region in p83-19, giving rise to plasmid p133B1.

Production of noninfectious VLP for immunogenicity studies. Vero cells were transfected at passage 141 with plasmid p133B1 by the calcium phosphate method as specified previously (57), and stable transfectants expressing noninfectious HIV-1-like particles were established. VLP were isolated from the supernatants of a stable clone (DD19) after induction of the Vero cells with heavy metals (57). Expression of Gag and Env was verified by Western blot assay using, respectively, an anti-HIV-1 p24 mouse MAb (NEA-9306; Dupont Canada, Inc., Markham, Ontario, Canada) and MAb K3A, which recognizes gp120 envelope proteins from different clades (kindly supplied by Ronald Kennedy). Fully assembled, Env-containing particles were isolated from the supernatants of the stably engineered Vero DD19 cell clone by ultracentrifugation through a glycerol cushion and purified by sucrose gradient fractionation (66). The p24 contents of the various particle species were determined by a p24-specific enzyme immunoassay (Coulter Immunology, Hialeah, Fla.). The cell line secreted approximately 800 to 950 µg of p24 per liter.

Construction of recombinant canary poxvirus vCP1579. Recombinant vCP1579 was engineered to express HIV-1_LAI Gag and Pro, together with HIV-1_Bx08 gp120 fused to a 28-amino-acid fragment of the transmembrane domain of HIV-1_LAI gp41, and a synthetic polypeptide containing Pol and Nef CTL epitopes. This recombinant is similar to vCP1452 (10, 79), except that the gene fragment encoding gp120_LAI was replaced with its cognate sequences from HIV-1_Bx08. Due to the absence of the gp41 ectodomain, the expressed envelope glycoprotein would be expected to consist of membrane-anchored gp120. The construct was generated in several steps. Briefly, the vector-modifying sequences from the pMPC6H6K3E3 insertion vector, encoding E3L and K3L, were inserted into the C6 locus of recombinant vCP1566. Recombinant vCP1566 was, in turn, generated by insertion of an expression cassette encoding a synthetic polypeptide containing Pol and Nef CTL epitopes into the C5 locus of vCP1453. Recombinant vCP1453 was obtained after coinsertion of the genes encoding the HIV-1_Bx08 Env and HIV-1_LAI Gag-Pro products into the ALVAC genome at the C3 locus.

Macaque immunizations. Rhesus macaques were divided into six groups with four animals in each group, as summarized in Table 1. Briefly, animals in groups 1 and 2 were inoculated with either a high dose (3 mg) or a low dose (600 µg) of DNA_Bx08, respectively, together with 50 µg of VLP_Bx08 in QS21 (100 µg) at weeks 0, 4, 24, and 44. Animals in groups 3 and 4 were inoculated with either the high dose or the low dose of DNA_Bx08, respectively, at weeks 0 and 4 and were boosted with ALVAC_Bx08 at weeks 24 and 44 (4 × 10⁸ PFU at week 24 and 4 × 10⁸ PFU at week 44). Animals in group 5 were inoculated with 50 µg of VLP_Bx08 in QS21 adjuvant at weeks 0, 4, 24, and 44. Animals in group 6 received 3 mg of control DNA at weeks 0 and 4 and were boosted with ALVAC_Bx08 at weeks 24 and 44 (same doses as above). All immunogens were administered in a volume of 2 ml intramuscularly. The QS21 adjuvant was prepared and filter sterilized before filling in of single-dose vials and stored at 20°C.

Western blot. Antibodies specific for HIV-1 antigens were assessed with a commercial HIV-1 Western blot kit (Genetic Systems, Sanofi Diagnostics Pasteur, Inc.) that is based on the LAI strain of HIV-1.

Enzyme-linked immunosorbent assay. Bx08-V3 peptide-specific binding antibodies were assessed in Nunc (Roskilde, Denmark) Immunoplates (MaxiSorb F96) using alkaline phosphatase-conjugated goat anti-monkey IgG as described previously (16). Plasma samples were assayed in duplicate at a 1:50 dilution, and values are given as the average A₄₀₅. The Bx08-V3 peptide had the amino acid sequence TRPNNNTRKSIHIGPGRAFYTTGDIIGDIR and was made by SynPep Corporation (Dublin, Calif.). The peptide was judged to be 91.5% pure by high-pressure liquid chromatography analysis. This same peptide was used in neutralization competition assays as described below.

Neutralizing antibody assays. Neutralization of HIV-1_Bx08 and other primary isolates was assessed in human PBMC by using a reduction in p24 Gag antigen synthesis as described previously (46, 58). Briefly, 500 50% tissue culture-infective dose of virus were incubated with various dilutions of serum samples for 1 h at 37°C before the addition of PBMC. Cells were washed three times with 200 µl of growth medium and resuspended in 200 µl of fresh interleukin-2 growth medium 1 day later. Culture supernatants were assessed for p24 content at a time when p24 synthesis in virus control wells (no test sample) was in a linear phase of increase, which is when optimum sensitivity is achieved in this assay (93). Neutralization titers were defined as either the dilution of serum or concentration of sCD4 and MAbs at which p24 synthesis was reduced by 80% relative to that of a negative control. These titers refer to conditions under which serum samples were incubated with virus before the addition of cells. All serum samples were heat inactivated for 1 h at 56°C prior to assay. V3 peptide neutralization competition assays were performed by preincubating serum samples with either phosphate-buffered saline (PBS) or Bx08-V3 peptide (50 µg/ml [final concentration]) for 1 h at 37°C. The serum samples were then assayed at a 1:3 dilution on the virus as described above.

ELISpot assay for IFN- release from antigen-specific PBMC. PBMC were separated from whole blood by Lymphocyte Separation Medium (Cellgro), washed twice, counted, and viably frozen in freezing medium for time course studies. Cells were thawed, washed twice, resuspended at 4 × 10⁶/ml, and stimulated for 48 h with 2 µg of HIV-1_HXB2 Gag peptide pools (National Institutes of Health AIDS Research and Reference Reagent Program) per ml. The peptides were 20-mers overlapping by 10 amino acids and were divided into two pools of peptides with 25 peptides per pool. Pool 1 represented amino acids 1 to 260, while pool 2 represented amino acids 251 to 500. Ovalbumin peptide was used as a negative control, and 50 ng of phorbol myristate acetate (Sigma) per ml plus 1 µg of ionomycin (Sigma) per ml served as a positive control. Ninety-six-well mixed cellulose ester filtration plates (Millipore) were coated with 50 µl of a 5-µg/ml concentration of mouse anti-human IFN- MAb 1-D1K (Biosource International) overnight at 4°C. Plates were then washed four times with PBS and blocked with RPMI (Cellgro) medium supplemented with L-glutamine, penicillin, and 10% heat-inactivated FBS (HyClone) for 1 h at 37°C. Prestimulated cells were then added to the coated plate in 100 µl of R10. Duplicate wells containing either 5 × 10⁵ or 2.5 × 10⁵ stimulated PBMC were plated for each sample and incubated for 16 to 20 h at 37°C. Plates were washed four times with PBS (Cellgro) and then four times with PBS containing 0.05% Tween 20 (Sigma) to remove all cells and debris from the plates. Next, 100 µl of 1-µg/ml biotinylated anti-IFN- MAb 1-D1K (Mabtech) diluted in PBS was added. After 2 h, the plates were washed four times with PBS containing 0.1% Tween 20 and incubated with 100 µl of Vectastain ABC (Vector Laboratories) per well for 1 h. Following four washes with PBS plus 0.1% Tween 20, Stable DAB (Research Genetics) was added to the wells and monitored for color development (approximately 5 min). The Stable DAB reaction was stopped by washing the wells three times with H₂O. Wells were counted visually under a dissecting microscope (Olympus) after the plates had dried.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Biologic and immunologic properties of HIV-1_Bx08. Coreceptor usage was determined by infection of cells that coexpressed CD4 and either CCR5 or CXCR4. HIV-1_Bx08 was able to infect U87-CD4-CCR5 cells but not U87-CD4-CXCR4 cells or MT-2 cells (data not shown). This outcome is consistent with an R5 phenotype for this virus. As a control for the integrity of the cells used for X4 phenotype classification, HIV-1_IIIB was able to infect U87-CD4-CXCR4 and MT-2 cells but not U87-CD4-CCR5 cells.

HIV-1 seroconversion. HIV-1-specific IgG was assessed by Western blot assay at multiple time points throughout the immunization schedule. Figure 1A shows that all of the immunized animals possessed Gag-specific antibodies by week 6 (2 weeks after the second inoculation). Gag-specific seroconversion was strongest in animals that received VLP_Bx08 either alone (group 5) or in combination with the high and low doses of DNA_Bx08 (groups 1 and 2, respectively). Gag-specific seroconversion was variable and less potent at this time in animals that received the high and low doses of DNA_Bx08 (groups 3 and 4, respectively). A similar pattern was noted for Pro-specific antibodies (p31), with the exception of very low levels of anti-Pro antibodies in animals inoculated with the low dose of DNA_Bx08 (group 4). Little or no Env-specific antibody was detected at this time.

View larger version (77K): [in this window] [in a new window]   FIG. 1.   Western blot analysis of HIV-1 antigen-specific antibodies. Serum samples were assayed at a 1:100 dilution. Panels: A, sera obtained at week 6; B, sera obtained at week 46. wks, weeks; HIV, HIV negative; HIV+, weakly HIV positive; HIV++, strongly HIV positive.

Neutralization of HIV-1_Bx08 with sera from immunized macaques. Neutralization of HIV-1_Bx08 was assessed in human PBMC with serum collected preimmunization and after the second, third, and fourth (final) immunizations. These assays were performed without knowledge of the animal group assignments. Figure 2A shows that neutralizing antibodies were detected in three of the four animals in group 1 (high-dose DNA_Bx08 plus VLP_Bx08), four of the four animals in group 2 (low-dose DNA_Bx08 plus VLP_Bx08), three of the four animals in group 3 (high-dose DNA_Bx08 plus ALVAC_Bx08 boosting), and three of the four animals in group 5 (VLP_Bx08). These neutralization titers ranged from 1:2 to 1:48. The highest titer was achieved in an animal (RYl5) immunized four times with VLP_Bx08. Notably, a 1:2 dilution of sera from several animals in groups 1, 2, and 5 achieved >90% neutralization and serum from at least one animal in each of the latter groups achieved 99 to 100% neutralization. The potency of neutralization at a 1:2 serum dilution did not correlate with the 80% neutralization titer.

View larger version (37K): [in this window] [in a new window]   FIG. 2.   Neutralization of HIV-1Bx08 by sera from immunized animals. Bar height represents the percent reduction in p24 relative to the amount of p24 produced in the presence of the corresponding preimmunization serum from each animal as assayed at a 1:2 dilution. Values above the bars are the highest serum dilutions at which p24 production was reduced by 80%. Panels: A, animal groups with detectable neutralizing antibodies; B, animal groups with no detectable neutralizing antibodies.

Absence of cross-reactive neutralizing antibodies. Cross-neutralizing activity was assessed with three TCLA strains (IIIB, MN, and SF2) and four heterologous R5, clade B primary isolates. In only two cases were neutralizing antibodies detected with a TCLA strain as measured in the MT-2 cell assay (data not shown). These cases occurred 2 weeks after final boosting and included sera from animal RLj5 in group 3 and animal REl5 in group 4, both having weak neutralizing activity against HIV-1_MN (titers of 1:29 and 1:50, respectively). Nine serum samples that neutralized HIV-1_Bx08 were further tested for the ability to neutralize heterologous R5 primary isolates in the PBMC assay. The results of the latter assays were all negative (Table 2).

Antibody specificity for the V3 loop. Serum samples obtained 2 weeks after final boosting (week 46) were tested for the presence antibodies that could bind a Bx08-V3 peptide. Sera from animals in groups 1, 2, and 5 and, to a lesser extent, groups 3 and 4 tested positive by enzyme-linked immunosorbent assay (Fig. 3). The detection of V3-specific antibodies did not always predict the ability to neutralize HIV-1_Bx08. For example, at least three serum samples that neutralized HIV-1_Bx08 had no detectable V3-specific antibodies (RZh5, RMg5, and RWh5). In addition, two samples that were positive for V3-specific antibodies (RSg5 and RNl5) failed to neutralize HIV-1_Bx08. The frequent detection of antibodies to the V3 loop of Bx08 is strong evidence of the presence of gp120-specific antibodies that were not detected by the Western blot assay (Fig. 1).

View larger version (35K): [in this window] [in a new window]   FIG. 3.   Bx08-V3 peptide-binding antibodies. Serum samples obtained 2 weeks after the fourth immunization (week 46) were tested at a 1:50 dilution for antibodies reactive with a peptide corresponding to the V3 loop of HIV-1Bx08. Bar height represents the average A450 of duplicate tests. All duplicate values agreed within 5% of the average. The value obtained with serum from an HIV-1-naive macaque is shown by a dashed line. Values greater than twice that of the negative control were considered to be positive. Serum samples that neutralized HIV-1Bx08 are indicated by asterisks above the bars. OD, optical density; grp, group.

View larger version (45K): [in this window] [in a new window]   FIG. 4.   Inability to detect neutralizing antibodies specific for the HIV-1Bx08 V3 loop. Two serum samples that contained antibodies reactive with a Bx08 V3 peptide and that neutralized HIV-1Bx08 were tested for neutralizing activity in the presence and absence of the Bx08 V3 peptide. Samples were preincubated with either peptide (50 µg/ml) or an equal volume of sterile PBS for 1 h at 37°C and assayed at a 1:3 dilution. Both postimmunization serum samples were obtained at week 46 (2 weeks after final boosting). Dark-shaded bars, no V3 peptide; light-shaded bars, with V3 peptide; pre, preimmunization serum; post, postimmunization serum. Each error bar represents the standard deviation of the average of triplicate values.

Analysis of IFN- secretion. PBMC from the vaccinated and control animals were tested for the ability to secrete IFN- upon stimulation with two pools of overlapping peptides representing the Gag sequence of HIV-1_HXB2. Peptide pool 1 spanned amino acids 1 to 260, while peptide pool 2 spanned amino acids 251 to 500. The number of IFN- spots per 10⁶ PBMC for each animal at each time point is shown in Fig. 5, and the average value for each group is shown in Fig. 6. We measured Gag-specific responses 2 weeks after the second immunization, a time when near-peak cellular responses would be expected, and then 3 months later (16 weeks), when memory responses from the first two immunizations would be in effect. The PBMC were then screened 2 weeks after each of the next two immunizations (weeks 26 and 46) to determine boosting effects and finally at week 56 (3 months following the final immunization) for immunologic memory for the HIV-1 antigen. The presence of 20 ELISpots in an assay was considered a positive test.

View larger version (64K): [in this window] [in a new window]   FIG. 5.   Number of HIV-1 Gag-specific IFN--producing cells detected by ELISpot assay with PBMC from individual animals. Solid bars represent cells stimulated with peptide pool 1, covering the amino-terminal half of Gag. Hatched bars represent cells stimulated with peptide pool 2, covering the carboxyl-terminal half of Gag. Values on the y axis are numbers of IFN--producing cells per million cells. ND, not done.

View larger version (28K): [in this window] [in a new window]   FIG. 6.   Average number of IFN--producing cells in each group of animals. The values below the x axis are the fractions of animals that were positive for Bx08-specific neutralizing antibodies at weeks 26 and 46.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Despite a growing desire to generate antibodies that are capable of neutralizing primary isolates, very few primary-isolate Envs have been investigated as potential immunogens for inclusion in an HIV-1 vaccine. Here we show that noninfectious VLP containing the surface envelope glycoprotein of an R5 primary isolate were capable of generating antibodies that neutralized the vaccine strain of virus. The antibodies could be generated by (i) coinoculation with DNA_Bx08 and VLP_Bx08, (ii) priming with DNA_Bx08 and boosting with ALVAC_Bx08, and (iii) inoculation with VLP_Bx08 alone. Most notable was the fact that, in certain cases, neutralizing antibodies were induced by recombinant vectors without the need for protein boostinga milestone that has not been reported previously with respect to primary-isolate neutralization by macaque serum.

Unlike the neutralizing antibodies generated by most of the candidate HIV-1 Env vaccines tested to date, the neutralizing antibodies generated here did not appear to be directed to linear epitopes in the V3 loop of gp120. Two lines of evidence support this conclusion. First, some serum samples with neutralizing activity had little or no antibody reactive with a Bx08 V3 loop peptide. Second, the neutralizing activity of two serum samples with strong reactivity to the Bx08 V3 loop peptide was not reduced in the presence of a high concentration (50 µg/ml) of the V3 loop peptide. These results suggest that the neutralizing antibodies targeted one or more epitopes outside the V3 loop. We note, however, that our experiments with linear peptides do not eliminate the possibility that the epitope(s) consisted of a conformational structure within V3 that was not represented by the linear peptide.

A disadvantage of the neutralizing antibodies seen here was their minimal cross-reactivity. Specifically, the antibodies had little or no neutralizing activity against TCLA strains and heterologous R5 primary isolates. The neutralizing antibody response generated by these candidate vaccines needs to be improved to target a broad spectrum of HIV-1 antigenic variants. We also noted that the Bx08-specific neutralizing antibodies were low in titer (1:2 to 1:48); however, the small magnitude of this response may be less of a concern than the limited strain specificity. For example, results of recent passive-immunization experiments with nonhuman primates indicate that preexisting neutralization titers of 1:10 or less may provide complete or partial protection against intravenous and mucosal challenge with simian-human immunodeficiency virus (35, 38, 76). Given that natural HIV-1 transmission appears to be virus dose dependent (60), where not all exposures result in infection (39), it is conceivable that a 5- or 10-fold reduction in virus infectivity by minimally diluted serum in vitro will correspond to an undiluted potency in vivo that is effective prophylactically. A number of recombinant vaccine vectors are also known to prime B cells for a rapid secondary neutralizing antibody response generated by HIV-1 Env protein boosting (23, 79), and at least one example exists of a dramatic secondary neutralizing antibody response after a virus challenge of macaques that had been primed with recombinant modified vaccinia virus strain Ankara-SIV vaccines (54). It may be expected that DNA vectors expressing VLP will prime for similar secondary neutralizing antibody production in response to HIV-1 infection.

We also detected HIV-1 Gag-specific cellular immune responses in the groups of animals that were inoculated with DNA_Bx08. All groups of animals receiving this immunogen scored positive for IFN--producing cells, and with the exception of a weak positive response to Gag pool 1 in one animal (RWh5, week 56), no IFN--producing cells were detected in animals inoculated with either VLP_Bx08 or ALVAC_Bx08 alone. Thus, as judged by the best overall immune responses (i.e., IFN--producing cells and HIV-1_Bx08 neutralizing antibody), coinoculation with a low dose of DNA_Bx08 plus VLP_Bx08 was the superior immunization modality investigated (Fig. 6). The Gag peptides used for PBMC stimulation were not completely matched to the vaccine strain, making it possible that some strain-specific reactivities went undetected in our assay. Despite this, it is noteworthy that IFN- production could be stimulated in many cases with two overlapping peptide pools covering the amino-terminal and carboxyl-terminal halves of Gag as evidence that at least two specificities of the cellular response were induced by vaccination.

While we consider the combined humoral and cellular immune responses elicited by these vaccine candidates to be encouraging, we also readily acknowledge the limited benefit that can be expected from neutralizing antibodies that are so narrow in specificity. Nonetheless, our results point to several vaccine strategies that may be useful for delivering a more appropriate immunogen for cross-reactive neutralizing antibody induction once that immunogen has been identified. We also did not determine whether the structure of Bx08 Env incorporated into the VLP is optimal for neutralizing antibody induction. For example, it has been argued that the native oligomeric (trimeric) structure of gp120-gp41 heterodimers must be preserved for optimal immunogenicity (13). Although it is possible that the Bx08 VLP expressed by our DNA vector contained native oligomeric envelope glycoproteins, we have not confirmed this. Furthermore, the envelope glycoproteins expressed by ALVAC_Bx08 consisted of membrane-anchored gp120 lacking the gp41 ectodomain, which may not represent the most suitable structure for optimal immunogenicity. Additional studies are required to determine the full potential of VLP as an immunogen for neutralizing antibody induction.