|Home | About | Journals | Submit | Contact Us | Français|
We previously reported that soluble, stable YU2 gp140 trimeric human immunodeficiency virus type 1 (HIV-1) envelope glycoprotein immunogens could elicit improved breadth of neutralization against HIV-1 isolates compared to monomeric YU2 gp120 proteins. In this guinea pig immunization study, we sought to extend these data and determine if adjuvant could quantitatively or qualitatively alter the neutralizing response elicited by trimeric or monomeric immunogens. Consistent with our earlier studies, the YU2 gp140 immunogens elicited higher-titer neutralizing antibodies against homologous and heterologous isolates than those elicited by monomeric YU2 gp120. Additionally, the GlaxoSmithKline family of adjuvants AS01B, AS02A, and AS03 induced higher levels of neutralizing antibodies compared to emulsification of the same immunogens in Ribi adjuvant. Further analysis of vaccine sera indicated that homologous virus neutralization was not mediated by antibodies to the V3 loop, although V3 loop-directed neutralization could be detected for some heterologous isolates. In most gp120-inoculated animals, the homologous YU2 neutralization activity was inhibited by a peptide derived from the YU2 V1 loop, whereas the neutralizing activity elicited by YU2 gp140 trimers was much less sensitive to V1 peptide inhibition. Consistent with a less V1-focused antibody response, sera from the gp140-immunized animals more efficiently neutralized heterologous HIV-1 isolates, as determined by two distinct neutralization formats. Thus, there appear to be qualitative differences in the neutralizing antibody response elicited by YU2 gp140 compared to YU2 monomeric gp120. Further mapping analysis of more conserved regions of gp120/gp41 may be required to determine the neutralizing specificity elicited by the trimeric immunogens.
The human immunodeficiency virus type 1 (HIV-1) exterior envelope glycoprotein, gp120, mediates entry by binding to the viral primary receptor CD4 (8, 29, 38) and the coreceptors CCR5 (1, 7, 9, 11, 12, 18) or CXCR4 (50, 51, 54). The transmembrane glycoprotein, gp41, contains the oligomerization domain (5, 58) and mediates virus-to-cell membrane fusion. These glycoproteins are derived from gp160 precursor proteins that, following glycosylation, folding, and trimerization in the endoplasmic reticulum-Golgi, are cleaved into the noncovalently associated gp120-gp41 heterodimeric, trimeric spikes (2, 14, 15, 35, 45, 48, 57). Due to their exposed location on the surface of the virus (or infected cells), the gp120 and gp41 proteins are the sole viral targets for neutralizing antibodies. Since effective neutralizing antibodies are likely to be a critical component of a successful HIV vaccine, a great deal of effort has focused on how to more efficiently elicit antibodies of breadth and potency capable of neutralizing a broad array of primary isolates. The first clinical trial utilizing monomeric gp120 as an immunogen failed to demonstrate any level of protection (19); hence, the focus has shifted to design of molecules that more closely resemble the trimeric spike found on the virus (3, 4, 13, 16, 20, 27, 52, 59-61).
We previously reported that gp140 (−/GCN4) trimeric immunogens could elicit improved, although limited, breadth of neutralization against HIV-1 isolates compared to monomeric gp120 immunogens (22, 61). In this study, we sought to confirm and extend these observations in another animal model and to examine if adjuvant could further enhance the neutralizing response. Thus, we compared YU2-based gp120 and gp140 immunogens emulsified in the commercially available Ribi adjuvant or in one of several newer adjuvants that have undergone extensive optimization with more modern technologies to improve their efficacy.
Adjuvants function in at least two distinct ways. In a relatively nonspecific manner, adjuvants increase the in vivo half-life of the immunogen by a “depot effect” that increases the persistence of the immunogen at the site of inoculation. Many oil-in-water or water-in-oil adjuvants accomplish depot, or deposition, through the formulation of an immunogen-containing emulsion that slowly releases the protein immunogen to interact with the host immune system. Adsorption of the protein to alum precipitates provides another means to accomplish protein deposition, and currently alum still remains the most widely used adjuvant for clinical applications.
Besides the “depot effect,” many adjuvants contain other components that activate innate inflammatory and adaptive responses, including humoral responses, by targeting known or not-yet-defined “danger signal”-sensing receptors to improve immunogenicity. For example, monophosphoryl lipid A (MPL), the active component of lipopolysaccharide that interacts with Toll-like receptor 4 (17, 39), is a component of Ribi adjuvant and two of the other adjuvants tested here. In this study, we analyzed antibody responses to the trimeric immunogens compared to monomeric gp120. We also compared Ribi adjuvant to three adjuvants developed by GlaxoSmithKline Biologicals (GSK, Rixensart, Belgium), called AS01B, AS02A, and AS03, to assess if the GSK adjuvants could elicit enhanced immune responses to the immunogens. These adjuvants have undergone extensive optimization to increase both humoral and cell-mediated immunity (32, 53). Ribi adjuvant contains the Toll-like receptor 4 agonist in a metabolizable oil, as well as natural and synthesized microbial components. The adjuvants from GSK are well defined and have been used in clinical trials. GSK AS01B is comprised of liposomes, QS21 and MPL™, GSK AS02A is an oil-in-water adjuvant containing QS21 and MPL™, and GSK AS03 is composed of oil in water.
For each of the adjuvants tested in this study, the YU2 gp140(−/GCN4) immunogens elicit higher-titer neutralizing antibodies than monomeric YU2 gp120. Mapping of the antibody responses indicates that neither immunogen elicits homologous V3-directed neutralizing responses, although significant levels of V3 loop antibodies are produced. The elicited V3 loop antibodies appeared to neutralize some heterologous isolates but not all. The homologous neutralization elicited by YU2 gp120 is completely inhibited by a single peptide located in the C-terminal region of the YU2 V1 loop. However, this is not true for the limited heterologous neutralization activity elicited by YU2 gp120. The homologous neutralizing activity generated by YU2 gp140(−/GCN4) trimers is much less sensitive to V1 peptide inhibition. This report identifies a novel focus of the homologous neutralizing activity elicited by monomeric gp120 and outlines a nascent strategy to dissect the complex binding and neutralizing specificities elicited by Env-based immunogens. Although we can better map the neutralizing specificities elicited by the monomeric immunogens, the trimeric immunogens elicit qualitatively different and more diverse specificities. Further mapping analysis of more conserved regions of gp120 or gp41 will be required to determine neutralizing specificity elicited by the trimeric immunogens.
The envelope glycoproteins were expressed by transfecting the 293F cell line that has been adapted to serum-free medium (Invitrogen, Carlsbad, CA). In brief, the 293F cells were seeded in T225 flasks at a density of 1 × 105 cells/cm2 and transfected with the expression plasmid YU2gp120/pcDNA3.1(−) for monomeric gp120 or YU2gp140(−/GCN4)/pcDNA3.1(−) DNA for trimeric gp140(−/GCN4) proteins. Each of the encoded protein sequences contained in-frame sequences encoding a six-His-tag at the C terminus as described previously (22). The cell culture supernatants were collected, centrifuged at 3,500 × g to remove cell debris, exchanged into phosphate buffer (20 mM phosphate, 0.5 M NaCl) with 10 mM imidazole, pH 7.4 by dialysis, and applied to a 5-ml His-trap nickel affinity column (Amersham, Piscataway, NJ). For monomeric gp120, the nickel affinity column was washed with phosphate buffer containing 10 mM imidazole, pH 7.4, and phosphate buffer containing 40 mM imidazole, pH 7.4, sequentially, and the proteins were eluted by phosphate buffer containing 300 mM imidazole. For trimeric gp140(−/GCN4) proteins, to minimize copurification of monomeric forms of gp140 in the culture supernatant, we increased the wash stringency prior to elution. The nickel affinity column was washed with phosphate buffer containing 30 mM imidazole, pH 7.4, followed by phosphate buffer containing 60 mM imidazole, pH 7.4, and the protein oligomers were eluted with phosphate buffer containing 300 mM imidazole. The protein eluates were dialyzed against phosphate-buffered saline (PBS), pH 7.4, and concentrated with Amicon Ultra 30,000 MWCO centrifugal filter devices (Millipore, Bedford, MA). The purified proteins were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, blue native gel, and gel filtration analysis, and the purity was verified to approach 95% homogeneity. By gel filtration, the gp140(−/GCN4) proteins purified by this nickel affinity procedure contained no detectable monomer and approximately 70% trimer; the remaining proteins were higher-order oligomers (not shown).
Hartlan guinea pigs (females, ~7 weeks of age) were inoculated intramuscularly with 20 μg of either monomeric YU2 gp120 or trimeric YU2 gp140(−/GCN4) proteins emulsified in either 0.5 ml of Ribi (Corixa, Hamilton, MT) or each of the GSK adjuvants. The protein-adjuvant emulsion was always prepared within 1 h of inoculation into the animals. Boosting inoculations occurred 6, 10, 14, and 21 weeks after the initial inoculation. To isolate serum, the blood was incubated at room temperature (RT) for 2 h to clot and centrifuged for 10 min at 2000 × g at RT to separate the liquid phase from the clotted components. The serum was collected and incubated at 55°C for 1 h to heat-inactivate complement and stored at −20°C until subjected to analysis.
The titers of the animal serum binding antibodies were determined by using an enzyme-linked immunosorbent assay (ELISA) method. In brief, protein or synthesized peptide was adsorbed onto each well of a Maxisorp high binding plate (Nunc) in PBS, pH 7.4, overnight at 4°C. For protein ELISA, 100 ng of YU2 or ΔV1/2 gp120 protein expressed from stable Drosophila S2 cells and purified to near homogeneity by antibody affinity chromatography was adsorbed to each well. The anti-C1/C5 region antibody C11 was used to confirm that similar levels of each protein were applied to the wells. For peptide ELISA, 40 ng/well of overlapping V1 to V2 loop peptides or 100 ng/well of V3 peptides was coated onto the ELISA plate surface. All wells were blocked by blocking buffer containing PBS, 2% dry milk, and 5% heat-inactivated fetal bovine serum. Serum was serially diluted in blocking buffer and incubated at RT for 1 h. The wells were washed five times with PBS containing 0.2% Tween-20, and a secondary anti-rabbit immunoglobulin G (IgG)-horseradish peroxidase antibody (Jackson Labs) was added to all wells in PBS-0.2% Tween 20 at a 1:10,000 dilution and incubated for 1 h at RT. After five washes with PBS-0.2% Tween 20, 100 μl of the colorometric TMB (3,3′,5,5′-tetramethylbenzidine) peroxidase enzyme immunoassay substrate (Bio-Rad) was added to each well, and the reaction was stopped by adding 100 μl of 1 M H2SO4 to each well. The optical density was read on a microplate reader (Molecular Devices) at 450 nm, and the end-point titers of the serum antibodies were defined as the last reciprocal serum dilution at which the optical density signal was greater than twofold over the signal detected with the preimmune serum.
HIV-1 primary isolates, SF162, 89.6, Bx08, and the T-cell line-adapted HIV IIIB, were obtained from the National Institutes of Health AIDS Research and Reference Reagent Program. Viral stocks were prepared and titrated in phytohemagglutinin and interleukin-2-stimulated human peripheral blood mononuclear cells (PBMC). The replication-competent molecular clone BR07 was provided by Dana Gabuzda of the Dana-Farber Cancer Institute. This virus is a chimeric infectious molecular clone of NL4-3 that contains the nearly full-length env genes from HIV-1 strain BR07 (41). After initial plasmid transfection of 293T cells, HIV-1 BR07 was expanded in PBMC as described above.
As previously described (37), this assay used PBMC depleted of CD8 T cells as targets for HIV-1 infection. After CD8 depletion by magnetic beads, the cells were activated with phytohemagglutinin and interleukin-2. In most cases, the viruses tested were uncloned virus stocks derived from PBMC cultures, as described above. However, virus YU2 was a recombinant Env-pseudovirus encoding a green fluorescent protein (GFP) reporter gene (37). Virus neutralization was measured using a flow cytometric single-round infection assay that detects HIV-1-infected T cells by intracellular staining for HIV-1 p24 Gag antigen or expression of GFP (used for the YU2 pseudotyped virus). For replication-competent viruses, an HIV-1 protease inhibitor (indinavir) was used to prevent secondary rounds of virus replication. The percent neutralization was derived by calculating the reduction in the number of p24-Ag- or GFP-positive cells in the test wells with immune sera, compared to the number of HIV-1-positive cells in wells containing preimmune sera from the corresponding animal. In some cases, serum samples were screened for neutralizing activity at a single dilution. All assays included at least two positive control reagents: an immune globulin from HIV-seropositve donors and an HIV-positive serum pool. Prebleed values at the equivalent serum dilution were subtracted before calculating the percentage of postimmune serum HIV-specific neutralization to adjust for nonspecific serum effects on viral entry. To calculate the dilution of serum that neutralized 50% of infectious virus (IC50), the serum was serially diluted and the dose-response curve was fit with a nonlinear function (four parameter logistic equation) using GraphPad Prism software (San Diego, CA).
The assay was performed as previously described (33, 37) with minor modifications as noted here. Briefly, a HeLa cell line that expresses CD4, CXCR4, and CCR5 was used as target cells for HIV-1 infection. These cells, initially called JC53-bl and now termed TZM-bl cells, were obtained from the National Institutes of Health AIDS Reference and Reagent Repository, as contributed by John Kappes and Xiaoyun Wu (10, 46, 55). The cells contain Tat-responsive reporter genes for firefly luciferase and Escherichia coli β-galactosidase under regulatory control of an HIV-1 long terminal repeat. In our assays, the level of HIV-1 infection was quantified by measuring relative light units (RLU) of luminescence that are directly proportional to the amount of virus input. The assay was performed in a 96-well microtiter plate using the same format as the PBMC assay described above, except that 10,000 TZM-bl cells were used in place of PBMC. Approximately 48 h after virus infection, the cells were lysed, and RLU were measured using black solid OptiPlates-96F plates (PerkinElmer, Boston, MA) and a Veritas Luminometer (model 1420-061; PerkinElmer) that inject luciferase assay substrate (Promega) into each well. Pseudoviruses were prepared by cotransfecting 293T cells with an Env expression plasmid containing a full-length gp160 env gene and an env-deficient HIV-1 backbone vector (pSG3ΔEnv). Pseudotyped virus-containing culture supernatants were harvested 2 days after transfection, filtered through 0.45-μm-pore-size filter, and stored at −80°C. For neutralization assays, each pseudotyped virus stock was diluted to a level that produced approximately 100,000 to 500,000 RLU. The cloning and construction of the full-length gp160 Env expression plasmids used to make the pseudoviruses are described in detail in a report by Li and colleagues (33). The Env pseudovirus based on the BR07 viral strain has been previously described (37, 41). Additionally, we isolated two functional Env clones from the DNA of PBMC infected with the primary isolate BaL. The Env clones BaL.01 and BaL.26 differed at 16 amino acid positions in gp160 (data not shown).
Peptide inhibition neutralization assays were done in the same assay format as the PBMC fluorescence-activated cell sorting (FACS)-based neutralization assay or the pseudotype assay, except that the control or test peptide was added to serum 30 min prior to the addition of virus. The concentration of peptide reported (usually 15 or 30 μg/ml) was that present when peptide, serum, and virus were incubated together as previously described (22). Control assays demonstrated that the peptide itself did not significantly affect virus entry. “No peptide” (defined as PBS of equivalent volume) was used as a negative control, as were several irrelevant peptides as described below. The effect of the peptide on virus neutralization was reported as the percent inhibition of neutralization. This was calculated as follows: [(percent neutralization observed with no peptide − percent neutralization with test peptide)/percent neutralization with no peptide] × 100. Several peptides were used in the competition studies. The YU2 V3 peptide was a 23-mer peptide (TRPNNNTRKSINIGPGRALYTTG) and synthesized by SynPep (Dublin, Calif.) (22). As controls, a scrambled V3 peptide (IGPGRATRPNNNFYTTGTRKSIH) (22) and a corresponding HIV IIIB V3 peptide (24 mer), purchased from Sigma-Aldrich, were used. The potent anti-V3 monoclonal antibody 447-D was used as a control to confirm the ability of the V3 peptides and not the scrambled peptide to inhibit antibody-mediated neutralization as previously described (22) (see Fig S3 in the supplemental material). As additional controls, a mixture of 22 peptides (15 mers overlapping by 9 amino acids spanning the Ebola [Zaire] viral glycoprotein sequence) was used in some assays to confirm the specificity of V3 peptide inhibition. For the V1 and V2 region, 15-mer peptides overlapping by 5 residues were synthesized by New England Peptide, Inc. (Gardner, MA). These included YU2a01 (TDLRNATNTTSSSWE), YU2a02 (SSSWETMEKGEIKNC), YU2a03 (EIKNCSFNITTSIRD), YU2a04 (TSIRDKVQKEYALFY), YU2a05 (YALFYNLDVVPIDNA), YU2a06 (PIDNASYRLISCNTS), and YU2a07 (SCNTSVITQACPKVS) (see Fig. Fig.55).
Guinea pigs, four animals per group (animal 47 in group B died during the course of the study), were inoculated with either monomeric YU2 gp120 or trimeric YU2 gp140(−/GCN4) protein immunogens emulsified in Ribi or the GSK adjuvants, AS01B, AS02A, or AS03. Animals were inoculated a total of five times, with 4- to 7-week intervals between inoculations. Test bleeds were collected 10 days after each inoculation, and the isolated sera were subjected to ELISA and HIV-1 neutralization assays.
After the fourth inoculation, sera were collected and tested for binding activity to YU2gp120 proteins by ELISA. All groups generated high titers of anti-gp120 IgG titers, with the end-point titers ranging from 3.2 × 105 to 7.5 × 10 6 (Table (Table1),1), which is consistent with the results from our previous studies in mice or rabbits (22, 61). Although most animals generated relatively high antibody titers, the animals from group A (monomeric gp120 in Ribi adjuvant) and both groups of animals inoculated with either monomer or trimers in the AS03 adjuvant had slightly lower end-point binding titers (Table (Table1).1). Both the AS01B and AS02A adjuvants stimulated approximately fivefold higher titers of binding antibodies than Ribi adjuvant in most animals inoculated with either monomeric or trimeric glycoproteins (Table (Table11).
Because much of the neutralizing response elicited by YU2 gp120 was directed against the V1 region (see below), we sought to determine if these elicited binding antibodies were predominantly elicited against the V1/V2 variable loops. We first compared the antibody binding titer against gp120 and a gp120 with the V1/V2 loop deleted (gp120ΔV1V2). We observed that for the 15 gp120-inoculated animals, 7 exhibited a fivefold decrease in recognition of gp120ΔV1V2 compared to recognition of wild-type gp120 (Table (Table1,1, animals 41, 42, 44, 45, 52, 54, and 55 of groups A, B, and C). In the group D animals (gp120 in the AS03 adjuvant), which had lower titers to gp120, there was not an observable binding decrease to gp120ΔV1/V2 proteins. In the animals inoculated with YU2 gp140(−/GCN4) proteins, 5 out of 16 animals displayed a fivefold decrease in binding titer to the gp120ΔV1/V2 proteins compared to wild-type gp120 (Table (Table1).1). These data suggest that although there are significant levels of binding antibodies directed toward the V1/V2 region, it is not the single dominant antibody response elicited by either gp120 or gp140 glycoproteins (Table (Table11).
As shown below, much of the neutralizing activity elicited by monomeric YU2 gp120 can be inhibited by a single YU2 V1 peptide but much less so by the YU2 V3 peptide; we sought to determine how much of the overall binding antibody repertoire was directed against these variable loop regions. By ELISA, we tested for the serum binding activity to the pool of V1/V2 peptides, to the single 15-mer V1 a02 peptide, and to the V3 peptide (Table (Table11 and data not shown). Most of the animal sera showed high binding end-point titers for the V1 a02 peptide, ranging from 3.13 × 105 to 1.5 × 10 6; the exceptions were that all the group A animals and animals 59 and 61 had lower binding titers to this peptide (Table (Table1).1). All the animal sera demonstrated similar binding patterns to the V1/V2 peptide pool, suggesting that most of the binding antibodies are directed toward the V1 a02 peptide. The binding titers of most of the sera to the V3 peptide were reasonably high (1.25 × 10 4 to 1.5 × 10 6) but were less than the levels obtained for binding to the V1 peptides (Table (Table1).1). Taken together, these data suggested that both YU2 gp120 and YU2 gp140(−/GCN4) proteins elicited significant levels of V1- and V3-loop-directed binding antibodies.
After the fourth injection, sera were collected and assayed for neutralization activity in a titration format against the homologous YU2 virus and the relatively sensitive virus SF162 using a FACS-based PBMC neutralization assay. We selected sera from animals from groups D and H that had been inoculated with monomeric YU2 gp120 or trimeric YU2 gp140(−/GCN4) proteins in the same AS03 adjuvant. Not all groups were analyzed to conserve limited sera for subsequent analyses. Sera were tested at an initial dilution of 1:10 and serial fivefold dilutions. For YU2, at the initial 1:10 serum dilution, the trimeric YU2 gp140(−/GCN4) animal sera had a mean (± standard error of mean) neutralization activity of 89% ± 8.0%, while monomeric YU2 gp120 animal sera displayed an average neutralization activity of 56% ± 20% (Fig. (Fig.1A).1A). The IC50 values for the reciprocal dilutions of the sera elicited by the gp140 glycoproteins were at least fivefold higher, ranging from 37 to 186, than the values for monomeric YU2 gp120-elicited sera, which ranged from 6 to 32. Similarly, for SF162, the sera from YU2 gp140(−/GCN4) protein-injected animals (group H) displayed approximately fivefold higher IC50 values than the YU2 gp120 protein-injected animals (group D) (Fig. (Fig.1B1B).
To assess the neutralization breadth of the sera elicited in selected adjuvants, we tested immune sera obtained after both the third and fourth inoculations of protein-adjuvant. Sera were assayed at a 1:5 dilution against the homologous YU2 virus and several heterologous viruses, as shown in Fig. Fig.2.2. Within the gp120-adjuvant groups, the percent neutralization of the YU2 and SF162 isolates generally increased from the third to the fourth inoculation (Fig. (Fig.2).2). After the third inoculation, sporadic neutralization of IIIB and 89.6 was detected, but this diminished following the fourth inoculation. Similarly, there were only two serum samples that possessed greater than 50% neutralization of the Bx08 and BR07 isolates. In general, sera isolated from the animals inoculated with gp140(−/GCN4) in the GSK adjuvants displayed more frequent and more potent neutralization of the YU2, IIIB, and 89.6 isolates than was achieved by the gp120-inoculated animals (Fig. (Fig.2).2). From the sera isolated after the third inoculation compared to after the fourth, increased neutralization of the YU2 and SF162 isolates was observed, whereas IIIB neutralization diminished slightly and 89.6 neutralization diminished dramatically. Sporadic, low-level neutralization of the BR07 and Bx08 isolates was observed in the gp140-GSK adjuvant groups, with the YU2 gp140(−/GCN4)-AS02A group exhibiting the most neutralization breadth, albeit of relatively modest potency (Fig. (Fig.22).
We sought to examine the breadth of neutralization following a fifth inoculation of the gp120 and gp140(−/GCN4) proteins in the recently standardized assay designed to permit the cross comparison of neutralization elicited by diverse immunogens (33, 36). Due to the analysis described in the present report, there were no sera remaining from the third and fourth protein inoculations; therefore, we compared the sera following the second inoculation to that following the fifth. Initially, we tested the ability of the sera to neutralize a panel of clade B Env-pseudotyped viruses at a single dilution of 1:5. As seen in Fig. Fig.3A,3A, the sera from the gp140(−/GCN4) protein inoculated in the GSK adjuvants efficiently neutralized YU2, BaL, JR-CSF, and JR-FL deleted of the 301 glycan (JR-FLΔ301 ). In general, and as before, the sera derived from YU2 gp140(−/GCN4) emulsified in the GSK adjuvants displayed more potent neutralizing activity than did the sera elicited by gp120 in the GSK adjuvants. The difference in neutralization between the gp140- and gp120-elicited sera was statistically significant as determined by a nonparametric Mann-Whitney test. For most viruses, the P value derived by this analysis ranges from 0.0001 to 0.056 (see Fig S1 in the supplemental material).
By comparing the serum neutralization potency from samples after inoculations 2 and 5, we observed a general increase in homologous neutralization and, when achieved, heterologous neutralization generally increased (Fig. (Fig.3A).3A). Interestingly, in contrast to this trend, the YU2 gp140(−/GCN4)-elicited sera after inoculation 2 (and after inoculation 3 in the previous analysis) neutralized the 89.6 isolate; however, most of this activity declined following inoculations 4 and 5. After inoculation 2, all sera elicited by the YU2 gp140(−/GCN4) in AS02A adjuvant (group G) neutralized the 89.6 virus at values greater than 80% (Fig. (Fig.3A).3A). We assessed the potency of these sera by serial dilution and 89.6 neutralization to derive IC50 values in excess of 1:640 for the individual serum (see Fig. Fig.6A).6A). In general, the sera in group G (YU2 gp140 in AS02A) displayed the greatest potency, as most individual animal serum samples from this group could neutralize 7 of the 10 isolates tested at the level of 50% or greater (Fig. (Fig.3A).3A). As before, we performed a titration of the neutralizing activity against the homologous YU2 virus from sera elicited in the matched AS03 adjuvant; the gp140(−/GCN4)-inoculated animals displayed significantly higher neutralization titers with average IC50 values 15-fold higher than the gp120-inoculated animals (1:12 for the monomer compared to 1:186 for the trimer) (Fig. (Fig.3B).3B). Consistent with the homologous neutralization titers, when the heterologous virus JR-CSF was tested in the titration assay, a moderate threefold higher IC50 value was observed for the gp140-GCN4-inoculated animals compared to that for the gp120-inoculated animals (data not shown). In summary, in both the PBMC-based assay and the standardized Env-pseudotyped virus neutralization assay, we confirmed that trimeric gp140(−/GCN4) proteins elicit more potent and modestly broader neutralization than the monomeric gp120 proteins.
By inspection of the neutralization data presented in Fig. Fig.22 and and3A,3A, it is apparent that especially with the gp140(−/GCN4) immunogens, the GSK family of adjuvants is superior to the commercially available Ribi adjuvant. After inoculation 4 (Fig. (Fig.2),2), the sera from three out of four animals inoculated with YU2 gp120 in Ribi adjuvant could not neutralize either the homologous isolate YU2 or the lab-adapted isolate IIIB or the sensitive primary isolate SF162 (Fig. (Fig.2).2). From this group, only animal 42 had relatively potent neutralizing activity against the YU2 and SF162 isolates. The sera from most animals in all groups injected with monomeric gp120 in the GSK series of adjuvants had more potent neutralization activity against YU2 and SF162 but less neutralization activity against IIIB. The 2 animals in group B (AS01B adjuvant) were the exception and did neutralize IIIB in the 60 to 70% range at the 1:5 dilution values tested (Fig. (Fig.22).
When the mean neutralizing values elicited by the gp140 proteins in each of the four adjuvants were statistically compared by a Mann-Whitney nonparametric analysis, it was apparent that the GSK adjuvants all elicited higher neutralizing activity at the single dilution point analyzed (Fig. (Fig.4).4). For animals inoculated with the trimeric gp140(−/GCN4) proteins, the percentage of homologous YU2 neutralization ranged from 91 to 99% in the AS01B, AS02A, and AS03 adjuvants and ranged from 0 to 68% in Ribi adjuvant group (P < 0.03) (Fig. (Fig.22 and and4A).4A). For the lab-adapted isolate IIIB and the sensitive primary isolate SF162, the percent neutralization values ranged from 50 to 93% in the AS01B, AS02A, and AS03 adjuvants, while it was below 50% in the Ribi adjuvant (group E) (P < 0.03) (Fig. (Fig.22 and 4B and C). In summary, the GSK family of adjuvants improves the potency of the neutralizing antibody response with either the monomeric or trimeric YU2 envelope glycoprotein immunogens compared to the commercially available Ribi adjuvant.
To define the neutralization specificity elicited by monomeric gp120, we performed peptide inhibition assays. Initially we focused on the V3 region and utilized an assay that we had established previously (22). We used the anti-V3 loop monoclonal antibodies 447-D and 2442 to validate the assay, as previously described (22). Briefly, animal sera were preincubated with synthesized peptides derived from the YU2 V3 region. In sera, these peptides could potentially form a complex with the V3 loop-directed antibodies and thus block the neutralization activity against V3-specific neutralization determinants. The percent inhibition of neutralization was obtained by comparing the percent neutralization in the presence or absence of peptide to identify V3-directed neutralization present in a given serum sample.
For the V3 analysis, serum samples remaining in greatest abundance after the previous analysis were selected. Ten serum samples from the YU2 gp120-inoculated animals were analyzed in this experiment. As shown in Table Table2,2, the presence of V3 peptide in the assay did not greatly inhibit the homologous YU2 neutralization activity for most of the sera (30% or less). The exception was the serum from animal 54 in the GSK AS02A adjuvant group, which showed a 47% activity decrease following incubation with the V3 peptide. However, many of the gp120-elicited sera that possessed relatively weak neutralization also displayed some minor fluctuations in neutralization activity with and without the V3 peptide. Because of the increased variability in the assay, and by inspection of the robust V1 peptide inhibition data below, we interpret the results to indicate that the homologous neutralization activity elicited by YU2 gp120 was not predominantly against the V3 loop.
When we performed similar V3 peptide inhibition assays with the more potent YU2 gp140(−/GCN4)-elicited sera, the data clearly demonstrated that little homologous neutralizing activity was V3 directed (Table (Table2).2). The percent inhibition of neutralization mediated by the V3 peptide was less than 10% for 11 of 12 animals, and the sera from animals 61 and 63 displayed V3-mediated inhibition levels of 24% and 17% (Table (Table2).2). These data were consistent with our previous rabbit immunogenicity, study which demonstrated that the homologous neutralization activity elicited by trimeric YU2 gp140(−/GCN4) protein was not predominantly directed toward V3 loop.
We then attempted to examine if the specific neutralization activity was focused on the V1 or V2 loops. To develop a peptide inhibition assay for this gp120 region, we first selected two serum samples (animals 44 and 46 from group B) and preincubated each of the sera with a pool of overlapping 15-mer peptides derived from the YU2 V1 and V2 loops (Fig. (Fig.5A)5A) (see Materials and Methods). Somewhat surprisingly, we found that the V1/V2 peptide pool inhibited most of the neutralization activity of these two serum samples. For animal 46, 100% of the neutralization activity was inhibited by the V1/V2 peptide pool (Fig. (Fig.5B5B and Table Table2),2), and for animal 44, 61% of neutralization was inhibited (Table (Table2).2). These results suggested that monomeric YU2 gp120 elicits neutralizing antibodies predominantly focused on V1 and/or V2 loops. To better define the specificity, we divided the V1 and V2 peptides into three test groups, the first containing two peptides, designated V1V2 pool 1. The second group contained a single peptide V1 a02 that we designated pool 2 (we had found that a similar peptide derived from HXBc2 inhibited HXBc2 homologous neutralization; J. Mascola and G. Nabel, unpublished observations). A pool of the four most-C-terminal V1/V2 peptides was designated V1V2 pool 3 (Table (Table22 and Fig. Fig.5A).5A). We then tested the inhibition effects of these three peptide sets on 10 selected animal serum samples. We found that the single peptide V1 a02, derived from the C terminus of V1 loop, inhibited homologous neutralization of 7 of 10 serum samples at the 100% level, 2 of 10 serum samples at the 70% level, and 1 serum sample at the 40% level (Table (Table22 and Fig. Fig.5B).5B). These results strongly suggested that the neutralization activity against the homologous isolate YU2 elicited by monomeric gp120 glycoprotein was focused to a single epitope within the V1 loop. Consistent with this observation, most of the other peptides within the V1 and V2 loop did not show significant inhibition of neutralization. The one exception was that the serum from animal 48 was inhibited 35% by the V1V2 pool 1 and 47% by the V1V2 pool 3 (Table (Table2).2). However, this particular serum sample displayed diverse effects, and the data were difficult to interpret. By definition, the inhibition of neutralization should not exceed the 100% level, which was already achieved by the V1 a02 peptide itself.
To examine if the YU2 gp140(−/GCN4) elicited neutralization activity toward V1 and V2 loops, we preincubated the sera with the three V1/V2 peptide groups. As shown in Table Table2,2, 11 of 12 animal serum samples had less than 20% activity inhibition by both V1V2 pool 1 and pool 3. Compared to peptide pools 1 and 3, there was more inhibition of neutralization inhibited by the peptide a02. Seven of 12 samples had less than 50% inhibition of neutralization by V1 peptide a02, while 5 of 12 samples displayed 50 to 83% inhibition. These data suggested that trimeric YU2 gp140(−/GCN4) elicited neutralization activity against the autologous virus YU2 that was partially, but not predominantly, directed toward the V1 and V2 loops, in contrast with neutralizing specificity elicited by monomeric YU2 gp120. In a similarly designed assay, we also analyzed selected sera elicited by the gp140(−/GCN4) glycoproteins for inhibition by preincubation with a 2F5-binding peptide. No reduction in homologous neutralization activity by the 2F5-binding peptide could be observed, although this peptide could efficiently inhibit 2F5-mediated neutralization in the same experiment (data not shown).
We observed that although homologous YU2 neutralization and heterologous neutralization of several isolates increased with repeated inoculations, neutralization of 89.6 declined after inoculations 2 through 5 (see Fig. Fig.22 and and3A).3A). Less dramatically, neutralization of the IIIB isolate also declined slightly after inoculations 3 to 4 (Fig. (Fig.2).2). To determine where the initial 89.6 neutralizing activity was directed, we performed adsorptions of the sera using the YU2 V1/V2 peptide pools and with the V3 peptide prior to performing neutralization assays. As shown in Fig. Fig.6A,6A, when we performed the adsorption/neutralizations over a range of serum dilutions from 1:10 to 1:640, most of the neutralization could be removed by the V3 peptide (in this case, a 40% reduction in neutralization is roughly a threefold difference in viral entry), but the V1/V2 peptides did not affect neutralization. From these data, IC50 values for 89.6 neutralization could be determined and were greater than 1:640 for animals 68 to 72. Given that we could now detect some V3-directed heterologous neutralization, we analyzed selected and relatively potent sera against other selected isolates. In contrast to 89.6, heterologous neutralization of JR-CSF elicited by either the trimer or the monomer in ASO2A adjuvant following the fifth inoculation was not V3 directed (Fig. (Fig.6B).6B). We also mapped selected sera against BaL.01 following the fifth inoculation and found that similar to 89.6, neutralization was mostly adsorbed by the YU2 V3 peptide (approximately 10-fold reduction of viral neutralization) (see Fig S3 in the supplemental material), whereas most JR-FLΔ301 neutralization following the fifth inoculation was not V3 directed (data not shown).
Previously we had observed that emulsified in Ribi adjuvant, the YU2 gp140(−/GCN4) molecules elicited neutralizing antibodies more efficiently than YU2 gp120 monomers. In that study, the maximal breadth was observed after repeated protein boosting without any observed increase in ELISA binding antibody titers after the inoculation (22). These data suggested that perhaps repeated boosts either improved the quality or increased the neutralizing titer. In this study we then sought to confirm that the gp140 molecules better elicited neutralizing antibodies in comparison to the monomers with larger numbers of animals and in a different species. We also asked if the adjuvant further improved either the quantitative or qualitative neutralizing responses to the trimeric immunogens. The YU2 gp120 monomeric proteins were included in parallel as a comparative control. We focused on a family of proprietary GSK adjuvants, one of which (AS02A) has shown potent induction of cellular and humoral responses in clinical settings (32, 53).
Here we have shown that the binding titers were slightly but not substantially increased by emulsification of the YU2 Env proteins in the GSK adjuvants, but the neutralization potency was enhanced. Importantly, the YU2 gp140(−/GCN4) trimers consistently, albeit in some cases modestly, outperform the YU2 gp120 molecules in terms of both neutralization potency and breadth, as determined by two independent neutralization assays, and the values are shown in parentheses and italics. These data are in agreement with those of our previous studies (22, 59) and consistent with the data reported in a similar investigation comparing ADA monomers to trimers by Kim et al. (27). Mapping of the homologous YU2 neutralization demonstrated that there was a distinct difference in the specificity of neutralization activity elicited by the YU2 gp120 monomers compared to the gp140 trimers, indicating that modestly improved neutralization elicited by the gp140 trimers is qualitative and not simply quantitative in nature. Another major finding is that the GSK adjuvants outperform Ribi adjuvant in terms of neutralization elicited from either the monomeric or trimeric molecules. Taken together, these two observations suggest that further modifications of the trimer, in the more potent GSK adjuvants, may continue to elicit better antibodies with increased neutralization breadth. There was a trend for a slight increase in the neutralization breadth of the sera elicited by the YU2 gp140(−/GCN4) trimers in the AS02A adjuvant in both the PBMC-based assay and the pseudotype neutralization assay against IIIB, SF162, and JR-CSF that was marginally statistically significant (P = 0.057; data not shown). Further study of selected adjuvant-protein combinations is certainly warranted.
For some isolates, the neutralizing antibodies elicited by monomeric gp120 were directed toward the accessible and immunogenic V3 loop (25, 26, 40, 49, 62). The V2 loop has also been previously reported as a neutralization determinant (24, 44). Interestingly, for the monomeric YU2 gp120 immunogen, homologous neutralization mapped to a single region in the C terminus of the V1 loop. However, based upon our peptide neutralization inhibition data, the heterologous neutralization of SF162 and IIIB viruses appeared not to be mediated through V1-directed antibodies (data not shown). By comparison of homologous V1 residues (using the IIIB molecular clone HXBc2) coupled with the knowledge that IIIB and SF162 heterologous neutralization was not V1 directed, it appears that the YU2 homologous neutralization is likely directed at the N-terminal region of the V1 a02 peptide. More direct proof of this deduction will require further V1 peptide fine mapping. Strain-restricted, V1-directed antibodies have been reported in Abgenix XenoMouse animals inoculated with SF162 gp120 (23).
The trimer protein elicited homologous neutralizing activity that is not yet defined: only V3, V1, and 2F5 have been ruled out (by negative data) from the inhibition of neutralization assay with the appropriate homologous peptides. Heterologous neutralization may be likely achieved by some other means for both the monomer and the trimer. In fact, for a relatively V3-sensitive molecular clone such as BaL.01, substantial levels of neutralization elicited by both the monomer and the trimer were inhibited by the V3-derived peptide. For less-V3-sensitive strains such as JR-CSF, heterologous neutralization was not V3 mediated (Fig. (Fig.66 and Fig. S3B in the supplemental material). By process of elimination and by the relatively limited neutralization breadth, it is possible that relatively nonpotent CD4 binding site antibodies typified by F105 or b6 (42, 47) or CD4-inducible responses typified by 17b (54) have been elicited and may mediate neutralization of IIIB. However, these responses may vary over the course of immunization as the 89.6 isolate, which was not neutralized at all time points, is weakly neutralized (if at all) by most CD4 binding site antibodies except IgGb12. In fact, 89.6 neutralization, when observable, was mostly V3 directed. The waning of 89.6 neutralization (and to a lesser extent the non-V3 neutralization of IIIB) is interesting but potentially disturbing in terms of boosting selected responses at the cost of others. After two inoculations, only the trimer-elicited sera could neutralize this isolate; however, the neutralization waned precipitously between the inoculations 3 and 4 and was also deficient in the sera after inoculation 5. This is in contrast to the more potent and consistent homologous neutralization of YU2 or when comparing neutralization after inoculations 2 to 5 of the heterologous BaL.01, BaL.26, or JR-CSF isolates. For selected trimer-elicited sera, we mapped the majority of the 89.6 neutralization to the V3 loop. As shown in Fig. S3B, BaL.01 neutralization also maps to V3 but does not wane; in fact, it increases with repeated inoculations. Taken together, the data suggest that initially the elicited V3 loop antibodies are directed against the V3 tip, a region that is conserved between 89.6 and BaL.01. Following inoculations 3 to 4, for reasons that are not clear the specificity shifts toward tip-flanking residues of V3, which are not conserved between 89.6 and BaL.01, and the ability to neutralize 89.6 via V3 antibodies is lost. The more modest decreases in IIIB neutralization must occur by a different mechanism as this virus appeared not to be neutralized by YU2-elicited V3 antibodies. This study highlights that even in sera of modest breadth, there are multiple neutralization specificities present. The development of tools to map these diverse specificities is of a high priority, and a systemic approach to decipher the complex specificities in broadly neutralizing sera by the design of novel adsorption assays and viruses of defined sensitivity/resistance merits immediate and further investigation. Since there is evidence and modeling that V1 and V3 are proximal in the HIV functional spike (6, 28, 31, 34), it is possible that some combination of V1-directed plus V3-directed neutralizing responses might be elicited and neutralize a greater array of isolates. However, epitopes that bridge across Env protomers and that are already variable themselves are likely to be extremely strain restricted. The recently described trimer-dependent V2- and V3-specific and strain-restricted 2909 neutralizing antibody may be of this type (21).
In conclusion, the relatively robust homologous neutralization elicited by the YU2 gp140 trimers (IC50 of 1:186 dilution) suggests that even for a relatively resistant isolate such as YU2, there are “chinks in the viral armor.” However, such accessible regions may be more strain specific, determined by variable loop interactions (and maintenance of function) and selection molded by concurrent circulating antibodies (56). This aspect may be worth further investigation of human sera exhibiting breadth of neutralization (such analyses are ongoing). These data emphasize the more general nature of the rare, broadly neutralizing antibodies (IgGb12, 2G12, 2F5, and 4E10) in guiding immunogen design, since they recognize more “general” chinks in the armor that are displayed by a relatively wide array of HIV-1 isolates. An obvious follow-up to this study is to mask (or delete) the V1 region (and perhaps V3) in both the context of the monomer and especially the trimer to see if the elicitation of antibodies can be shifted toward the targets implicated in conferring greater neutralization breadth. The general strategy of altering glycans to impact on a more focused antibody response has been suggested in two recent studies (43, 44). These and other protein modifications will be the focus of future studies directed toward further improvements of the gp140 soluble, stable envelope glycoprotein trimeric immunogens.
We thank Brenda Hartman and Toni Miller for help with the figures and thank Gary Nabel for helpful discussions. The GSK adjuvants were obtained via a Materials Cooperative Research and Development Agreement (MCRADA) between the National Institute of Allergy and Infectious Diseases (NIAID) and GlaxoSmithKline.
The analysis described in this study was supported in part by the Intramural Research Program of the Vaccine Research Center, NIAID, National Institutes of Health, Bethesda, MD.
†Supplemental material for this article may be found at http://jvi.asm.org/.