INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The surface antigens of human immunodeficiency virus type 1 (HIV-1) have been the primary targets in attempts to develop an AIDS vaccine in the last decade (24, 54). The efficacy of envelope-based vaccines has been demonstrated largely in the chimpanzee model against tissue culture-adapted viruses such as HIV-1 IIIB (10, 11, 15, 26, 43, 57) and SF2 (14, 22). Infection by these virus isolates is generally self-limiting and does not lead to AIDS-like diseases. Furthermore, the scarcity of these animals and the expenses involved limit the value of this model for the evaluation of multiple vaccine approaches under different conditions.

To circumvent these limitations, a number of investigators have used macaque models with simian immunodeficiency viruses (SIV) that vary in the rapidity with which they induce CD4⁺ cell depletion or death (33, 52). Using a pathogenic cloned virus, SIVmne E11S, we demonstrated complete protection against the homologous virus with gp160 vaccines in a combination immunization regimen that consisted of recombinant vaccinia virus for priming and subunit protein for boosting (30, 32). However, the protective efficacy of this immunization approach remains controversial, since a number of investigators using similar regimens have reported conflicting results. Abimiku et al. (1) showed that macaques immunized with recombinant canarypox vaccines and boosted with subunit HIV-1 proteins were partially protected against infection by HIV-2, a divergent albeit nonpathogenic virus. Hirsch et al. (28) showed that immunization with a modified vaccinia virus-based trivalent SIV vaccine followed with inactivated SIV failed to protect against infection by a more pathogenic challenge virus, SIVsmE660, but was able to reduce the virus load, resulting in prolonged disease-free survival in infected macaques. On the other hand, Giavedoni et al. (25) and Daniel et al. (19) showed that combination immunization regimens with recombinant vaccinia virus priming and subunit antigen boosting resulted only in reduction of the viral load in a minority of animals challenged with a highly pathogenic virus, SIVmac251, with no apparent benefit in disease outcome. Direct comparison of these studies, however, is hampered by the divergent nature of the challenge viruses and by the different vectors, immunogens, and immunization regimens used.

To study the protective efficacy of the combination immunization strategy more systematically, we first investigated the limits of the protective immunity elicited by the envelope antigens alone in the SIVmne model, where complete protection was achieved against a pathogenic cloned virus, E11S. In this communication, we demonstrate that the protective immunity elicited by recombinant gp160 vaccines was restricted primarily to the homologous cloned virus. Only partial protection against the uncloned virus SIVmne was achieved. Analyses of SIV-specific antibodies failed to reveal any correlation between protection and serum neutralization. However, analysis of "breakthrough" viruses in infected animals indicates that protection against SIV infection may be attributable in part to immune responses directed to the V1 hypervariable region within the envelope antigen.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Immunogens and immunization regimen. Recombinant vaccinia virus vac-gp160 (v-SE5) contains the coding sequence of the full-length gp160 of SIVmne molecular clone 8 in a New York City Board of Health strain (v-NY) of vaccinia virus (30, 32). Molecular clone 8 is a derivative of the SIVmne single-cell clone E11S and has an identical env sequence to the parental virus, E11S (17, 48a). V-SE5 was plaque purified and propagated on African green monkey kidney (BSC-40) cells (32). Nineteen cynomolgus macaques (Macaca fascicularis) were each inoculated with 10⁸ PFU of the recombinant virus by skin scarification at two or three sites along opposite sides of the midline of the back. Booster immunizations were done by intramuscular injections at 2 to 3, 10 to 18, and 12 to 26 months with gp160 produced either in recombinant baculovirus-infected insect cells (experiment 1) (31) or in BSC-40 cells infected with recombinant vaccinia virus (experiments 2 and 3). Similar to their HIV-1 counterparts (13, 38), SIVmne gp160 produced in these systems are glycosylated, have an apparent molecular mass of 160 kDa by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and bind to recombinant human CD4 (data not shown). Size exclusion chromatography has shown that the majority of the gp160 produced in the vaccinia virus expression system are oligomeric, mostly trimeric or tetrameric (data not shown). Each booster dose contained 250 µg of total protein (corresponding to approximately 125 µg of gp160) formulated in Freund's incomplete adjuvant.

Challenge virus and conditions. SIVmne was isolated from a pigtailed macaque (M. nemestrina) with lymphoma and was propagated on HuT 78 cells (6). E11S is a single-cell clone of SIVmne-infected HuT 78 cells that produced large amounts of envelope glycoproteins (8). Challenge was performed 4 weeks after the last immunization by an intravenous injection. Challenge with SIVmne clone E11S was performed with 10 to 100 animal infectious doses (AID) of virus grown on either HuT 78 cells or macaque peripheral blood mononuclear cells (PBMC). Challenge with uncloned SIVmne was done with 10 to 100 AID (experiment 1) or 2-20 AID (experiments 2 and 3) of virus grown on HuT 78 cells. Some of the animals protected from the E11S challenge were held for 1 to 2 years to confirm their virus-negative status before they were given further booster doses with gp160 and rechallenged 4 weeks later with uncloned SIVmne. Blood samples were collected on the day of challenge, at 2, 4, 6, and 8 weeks after challenge, and monthly thereafter. Plasma and serum samples were collected and stored until used at 70 and 20°C, respectively. Lymph node biopsy specimens were obtained at the indicated times after challenge and were frozen at 70°C for DNA analysis or fixed for histological examinations. The animals were housed in the Washington Regional Primate Research Center and were under the care of licensed veterinarians. Euthanasia was performed on the basis of the following criteria: (i) AIDS; (ii) termination of the experiment; or (iii) unrelated cause. Euthanasia is considered to be AIDS related if the animal exhibits CD4⁺ cell depletion in the peripheral blood and two or more of the following conditions: wasting, unsupportable diarrhea, opportunistic infections, proliferative diseases (e.g., lymphoma), and abnormal hematological profile (e.g., anemia, leukopenia, or thrombocytopenia).

Virus isolation. PBMC were isolated over Histopaque-1077 (Sigma Chemical Co., St. Louis, Mo.) as described previously (7, 9). Briefly, 4 × 10⁶ PBMC were cocultivated with 5 × 10⁶ AA-2CL5 cells, and cultures were maintained for 8 to 9 weeks. Virus was detected by reverse transcriptase assays performed as described previously (9). A positive value means positive results in reverse transcriptase assays, and a negative value means no reverse transcriptase activity detected after 8 to 9 weeks of cocultivation.

Serum neutralization assays. Neutralizing antibodies against uncloned SIVmne and SIVmne clone E11S were measured in CEM-X174 cells by methods similar to those described by Montefiori et al. (46). The uncloned SIVmne used for neutralization studies was grown on HuT 78 cells and was identical to the challenge stock but was prepared at different times. The E11S clone used for neutralization assays was grown on macaque PBMC after initial isolation and propagation in HuT 78 cells (6). It was derived from the same stock as the challenge virus but grown on PBMC from a different M. fascicularis animal. Neutralization of a related but heterologous virus, SIVmac251, was measured in HuT 78 cells as described by Langlois et al. (40). Twofold serum dilutions (heat inactivated at 56°C for 30 min) were made in 96-well plates. The neutralization titer is expressed as the reciprocal serum dilution that inhibits 50% of SIVmne-induced cytopathic effect in CEM-X174 cells or 90% of the syncytium formation by SIVmac251-infected Hut-78 cells.

ELISA. SIV-specific antibodies were measured by an enzyme-linked immunosorbent assay (ELISA) as described previously (29), except that gradient-purified and disrupted whole SIVmne clone E11S virion was used as the antigen in ELISA. End-point titers were determined as the reciprocal of the highest serum dilution that resulted in an optical density reading threefold greater than that obtained with negative control sera.

Nested PCR analysis. PBMC were isolated from EDTA-treated blood by Hypaque-Ficoll gradient centrifugation, and nucleic acid was extracted by standard techniques. A 1-µg sample of total nucleic acid was used as a template for a two-step amplification by PCR with a nested set of oligonucleotide primers specific for the envelope regions. The conditions for the first and second rounds of amplification were as described previously (30). The final amplified fragment was approximately 642 bp. Amplified products were resolved by agarose gel electrophoresis and visualized by ethidium bromide staining. Results for a subset of samples were also confirmed by PCR with primers from env and long terminal repeat (LTR)-gag regions.

Semiquantitative PCR analysis of the proviral DNA load. The amount of proviral DNA in PBMC was measured by PCR with radiolabeled primer incorporation for quantification (50) and was expressed as the number of copies of proviral genome detected per million PBMC. Briefly, 1 µg of DNA from each sample was amplified in a PCR mixture that contained 0.2 µM each primer, 200 µM each deoxynucleoside triphosphate, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, and 1.0 U of Taq polymerase (Perkin-Elmer Cetus, Branchburg, N.J.) in a volume of 50 µl. The reaction was subjected to 30 cycles of denaturation for 1 min at 94°C, annealing for 2 min at 60°C, and elongation for 3 min at 70°C. The oligonucleotide primers used were derived from the nucleotide sequence of SIVmne (GenBank accession no. M32741 [27]). They consist of a primer pair specific for the env region, env10 (7191 to 7211 [sense]) and env12 (7541 to 7561 [antisense]), and a primer pair specific for the LTR-gag region, S1 (228 to 251 [sense]) and S8 (536 to 559 [antisense]). The amplified products for the env and the LTR sequences are 370 and 330 bp, respectively. One oligonucleotide of each complementary pair was 5'-end labeled with [³²P]ATP by the use of polynucleotide kinase (New England Biolabs, Inc., Beverly, Mass.). The ³²P-labeled PCR products obtained by amplification were analyzed by electrophoresis on 8% nondenaturing polyacrylamide gels and quantified by autoradiography by phosphorimager analysis (50). The amount of SIV DNA was determined by using a standard curve generated with known quantities of a plasmid clone of E11S.

RT-QC-PCR determination of plasma viral RNA. Plasma viral RNA was prepared as described by Watson et al. (56). The viral RNA samples were serially diluted in a 96-well PCR microplate into a reaction buffer containing a fixed copy number of a competitor RNA containing an internal deletion. The template and the competitor were subjected to reverse transcription (RT) followed by quantitative competitive PCR (QC-PCR). The primers used are from the SIVmne gag sequence: 5' primer (5G) from nucleotides 675 to 698 (AAAGCCTGTTGGAGAACAAAGAAG) and 3' primer (3Diii) from nucleotides 993 to 1011 (AATTTTACCCAGGCATTTA). The internal RNA control contains a deletion of 82 bp, which enables products amplified from the viral (336-bp) and the control (254-bp) templates to be distinguished. The conditions for the RT reaction and QC-PCR were as described by Watson et al. (56).

Sequencing of the env hypervariable regions of uncloned SIVmne. Infected HuT 78 cells from which the uncloned SIVmne challenge stock was derived were used as the source in a determination of viral genomic sequence and complexity. The region in env encompassing the hypervariable regions V1 to V5 was amplified by PCR and subcloned into M13 vectors for sequence analysis. The method was based on that described by Overbaugh et al. (49). Briefly, genomic DNA was isolated from infected cells and the V1 to V5 region in the SIV env gene was amplified by two rounds of PCR with nested sets of primers: for the first round, Env1 (6346 to 6367, 5'-ATAGGTACCCTCTTTGAGACC TCAATAAA-3' [sense]) and Env8 (7575 to 7594, 5'-ATAGAATTCCCAATTGGAGTG ATCTCTAC-3' [antisense]), and for the second round, Env7 (6364 to 6382, 5'-GACGGTACCTAAAGCCTTGTGTAAAATTA-3' [sense]) and Env4 (7524 to 7544, 5'-GAATTCAGTTCTGCCACCTCTGCACT-3' [antisense]). The inside primers were designed to contain restriction sites at the 5' ends for subsequent cloning into M13 vectors for sequence analysis. To minimize possible bias introduced by PCR amplification, we performed six independent amplifications for each DNA sample and generated three or four subclones for each amplified DNA sample for sequence analysis.

Analysis of the proviral DNA sequence in PBMC or lymph node cells from animals infected with uncloned SIVmne. Proviral DNA sequences in infected macaques were analyzed by one or more of the following methods. The first was nucleotide sequencing. The same method used for the analysis of uncloned SIVmne virus stock was used to analyze proviral sequences present in the PBMC or lymph node cells from infected animals. The second was PCR amplification with radiolabeled primers. Results from nucleotide sequence analysis indicate that the majority of the sequence variations in the env gene of the uncloned virus are present in the V1 region. A total of 85% of the amplified V1 sequence in the challenge stock is identical to that of E11S (E11S type), while the remaining 15% shares a consensus sequence in V1 that is different from E11S (variant types). Based on this information, we designed two oligonucleotide probes (nucleotides 6471 to 6499), one specific for the E11S-like sequence (E11Sp, 5'-TTTATTGCCTCTGCTTTTGTTGGTATTGC-3' [antisense]) and the other for the variant-type sequences (Variantp, 5'-TCTATTTTCTTTGTTGTTGGTTTTGGTGT-3' [antisense]). The E11Sp probe hybridizes with the proviral cDNA of E11S and uncloned SIVmne, while the Variantp probe hybridizes only with the latter. With primers specific for the E11S or the variant type sequences, we amplified V1 sequences in the PBMC or the lymph node cells of infected macaques. A radiolabeled primer specific for the V1 region (Env71, 6097 to 6120, 5'-TTATCGCCATCTTG TTTCTAAGTG 3' [sense]) was used in combination with the antisense primers specific for the E11S or variant sequences. The amplified product, which was 397 bp, was resolved by electrophoresis on 8% nondenaturing polyacrylamide gels, and the relative abundance of the E11S-type and the variant-type sequences was determined by autoradiography using phosphorimager analysis as described above. For each PCR amplification, DNA from E11S-infected and uncloned SIVmne-infected cells were used as controls. The third method was the heteroduplex mobility assay. Heteroduplex formation conditions were used as described by Delwart et al. (20). Briefly, DNA from infected cells (1 µg per reaction) was amplified by nested PCR. The primers for the first round were Env1 and Env 8, and those for the second round were Env7 and Env14 (6740 to 6757, 5'-CTAATAGCATCCCAATAA-3' [antisense]). Of the 50-µl first round reaction mixture, 2 µl was used for the second-round amplification. The final product is 394 bp and spans the V1 to V2 region. For heteroduplex formation, we combined 4.5 µl of the PCR products amplified from the proviral DNA of E11S-infected cells and an equal volume of PCR product from the test DNA sample with 1 µl of 10× annealing buffer (1 M NaCl, 100 mM Tris-HCl [pH 7.8], 20 mM EDTA). DNA in the mixture was denatured at 94°C for 2 min and annealed by cooling on ice as described by Delwart et al. (20). Reannealed products were resolved by electrophoresis on a 5% polyacrylamide gel and stained with ethidium bromide.

Lymphocyte subset analysis. Cell surface immunofluorescence was quantified with a FACScan flow cytometer and Lysis II software (Becton Dickinson Immunocytometry Systems, San Jose, Calif.). Lymphocyte subsets (CD4, CD8, CD2, and CD20) of whole heparinized blood samples were evaluated by conventional methods.

Statistical analysis. Differences in proportions were tested by Pearson's chi-square test or Fisher's exact test. Differences between medians of continuous variables were tested by the Mann-Whitney U test.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Immunization regimen and outcome of challenge. The protective efficacy of gp160 vaccines was evaluated in a total of 19 macaques in three separate experiments (Table 1). All animals in the experimental group received one or two inoculations of recombinant vaccinia virus expressing the gp160 gene of SIVmne clone 8 (GenBank accession number M32741 [27]) followed by two or three booster immunizations with subunit gp160 prepared either from recombinant baculovirus-infected insect cells (Table 1, experiment 1) or from recombinant vaccinia virus-infected mammalian cells (Table 1, experiments 2 and 3). In experiments 1 and 3, the animals were challenged first with homologous virus E11S clone grown either on the HuT 78 cell line or on macaque PBMC. The protected animals were then rechallenged with uncloned virus SIVmne. In experiment 2, the animals were challenged in parallel with the cloned or uncloned virus. All challenges were performed by intravenous injection as described. Infection was monitored by nested PCR, virus isolation by coculture, measurement of the anamnestic response, and measurement of seroconversion to nonvaccine antigens. The outcome of the challenge is summarized in Table 2.

View larger version (26K): [in this window] [in a new window]   FIG. 1.   PBMC proviral load in macaques after challenge with SIVmne E11S clone (top) or uncloned SIVmne (bottom). Proviral load was determined by PCR analysis with radiolabeled primer incorporation, as described in Materials and Methods. Values are expressed as copies of proviral genome per 106 PBMC. Proviral DNA was measured by using an external E11S DNA standard of known quantities.

SIV-specific responses in immunized animals. SIV-specific antibody responses were monitored throughout the study by ELISA with disrupted whole virions, by immunoblotting (data not shown), and by serum neutralization assays against the challenge virus and a heterologous SIV strain before and after the challenge inoculation. As observed previously, all animals developed low levels of SIV-specific antibody response, which increased 10- to 30-fold after the first subunit protein immunization (reference 30 and data not shown). By the time of challenge, all the immunized animals developed SIV-specific antibodies, although the titers varied (Fig. 2). The sera of these animals were also able to neutralize a heterologous virus, SIVmac251, passaged in HuT 78 cells. However, some of the immunized animals had only low levels of neutralizing antibodies against the challenge virus SIVmne E11S. As shown in Fig. 2, there was no correlation between the challenge outcome in these animals and their SIV-specific antibody titer at the time of challenge as measured by the methods described.

View larger version (28K): [in this window] [in a new window]   FIG. 2.   SIV-specific antibody responses in immunized macaques. Sera collected on the day of challenge were analyzed for neutralizing activities against the homologous challenge virus (E11S or uncloned SIVmne) and a heterologous virus SIVmac251. Neutralizing titers are expressed as the reciprocal serum dilutions that resulted in >90% reduction of syncytium formation by SIVmac251 on HuT 78 cells (center) or >50% cytopathicity of E11S or uncloned SIVmne infection in CEMx174 cells (left). Serum reactivity with disrupted SIVmne E11S virion proteins was analyzed by ELISA, and the results are expressed as end-point titers (right). (a) Animals challenged with SIVmne E11S. (b) Animals challenged with uncloned SIVmne. Solid symbols denote protected animals; open symbols denote the persistently infected ones; crossed symbols denote the transiently viremic ones. The upper and lower dotted lines represent, respectively, the titers of positive and negative control serum samples in each assay. Nab, neutralizing antibody.

View larger version (25K): [in this window] [in a new window]   View larger version (29K): [in this window] [in a new window]   View larger version (28K): [in this window] [in a new window]   FIG. 3.   SIV-specific antibody responses in immunized and control macaques after challenge. Dilutions of macaque sera collected at the indicated times were incubated with disrupted, gradient-purified SIVmne virion proteins immobilized on microtiter plates. End-point titers were defined as the reciprocal of the highest  dilution that gave an optical absorbance value at least threefold higher than the average values obtained with SIV-negative macaque sera. (A) Experiment 1. (B) Experiment 2. (C) Experiment 3. For each panel, the top row gives results for animals challenged with SIVmne E11S and the bottom row gives results for animals challenged with uncloned SIVmne.

Analysis of viral sequences recovered from infected animals. To gain further insight into the basis for the success or failure of the gp160 immunizations, we examined proviral sequences recovered from infected animals after the uncloned virus challenge and compared them with sequences present in the inoculum. We first analyzed proviral sequences in HuT 78 cells infected with the same uncloned SIVmne virus stock used for the challenge studies. We used nested sets of primers in a two-step PCR to amplify env sequences encompassing the entire region containing V1 to V5. Amplified fragments were subcloned in M13 vectors for nucleotide sequence analysis. Three or four subclones from each of six independent amplification reactions were used to minimize any potential bias introduced by the amplification and cloning procedures. Of 20 clones analyzed, 12 had V1 to V5 sequences identical to that of SIVmne biological clone E11S or its derivative molecular clone 8 (8). Five clones had one or more nucleotide substitutions, resulting in two nonsynonymous amino acid changes in one clone and a single amino acid change in two others. All were unique within the V1 and V2 regions. Of the remaining three clones, all had multiple amino acid changes clustered in the V1 region. Although there were one or two amino acid changes unique to each clone outside this region, all three clones had the same sequence (TPKPTTTKKIE) distinct from that of E11S or molecular clone 8 (AI-PTKAEAIK, corresponding to amino acid residues 134 to 143) (Table 4). Using ³²P-labeled oligonucleotide primers specific for these distinct sequences, we amplified the V1 env fragments directly from the proviral genomes of the uncloned viruses. Although the E11S-specific primer amplified both E11S and uncloned viral sequences, the variant-specific primer amplified only the latter (Fig. 4A). Quantification of the amount of ³²P label hybridized indicated that 85% of the fragments amplified from the uncloned virus had E11S-like sequences while the remaining 15% were of the variant type, in agreement with results of the sequence analysis of individual clones. Results from heteroduplex mobility analyses also confirmed the relative complexity of these viral stocks (Fig. 4B).

View larger version (24K): [in this window] [in a new window]   View larger version (47K): [in this window] [in a new window]   FIG. 4.   Composition of V1 sequences in the uncloned SIVmne challenge virus and tissues from infected animals. (A) PCR analysis with radiolabeled primers. Radiolabeled primers specific for the E11S (E) or the variant-type (V) V1 sequence were used to amplify a 397-bp fragment from the proviral DNA from the following sources: cells infected with E11S (lanes 11 and 12) or with uncloned SIVmne (lanes 13 and 14); lymph node (LN) of an immunized animal, macaque J90304 (lanes 9 and 10), or control animals, macaques 91064, 91070, and 92168 (lanes 3 to 8), at the indicated weeks after challenge with uncloned SIVmne. All DNA samples used for analysis contained at least 50 proviral copies per reaction. (B) Heteroduplex mobility analysis. Proviral sequences present in cells infected with E11S or uncloned SIVmne and the PBMC of infected animals were analyzed after PCR amplification of the V1 to V2 region with nested sets of primers as described in Materials and Methods. The results for PBMC obtained 2 weeks after challenge are shown. PCR-amplified fragments from each sample were annealed for the formation of heteroduplexes, which were resolved by gel electrophoresis as described in the text.

Clinical outcome of challenge. Infected animals were monitored for up to 5 years after challenge to determine the clinical outcome of infection and the effects of immunization. The animals were monitored periodically for lymphocyte subsets, hematology, blood chemistry, body weight, opportunistic infections, and proliferative diseases. Figure 5 summarizes their peripheral blood CD4⁺ cell levels and survival after challenge. Euthanasia was performed either because the infection progressed to cause AIDS or because the experiment was terminated. A few animals died of reasons unrelated to AIDS. Histopathological examinations were performed upon necropsy.

View larger version (39K): [in this window] [in a new window]   FIG. 5.   Peripheral blood CD4+ T-lymphocyte numbers in immunized (A and C) and control (B and D) macaques following challenge with SIVmne E11S (A and B) or uncloned SIVmne (C and D). Animals sacrificed because of AIDS are labeled A; animals euthanized at the end of the experimentation period are labeled E; and animals that died of causes unrelated to AIDS are labeled U. The last datum point for each animal represents the time of death or termination of the experiment (euthanized, alive or rechallenged).

View larger version (21K): [in this window] [in a new window]   FIG. 6.   Analysis of an immunized animal that was transiently virus positive early after challenge with SIVmne E11S. (Top) Viral load in plasma and CD4+ T-lymphocyte counts in peripheral blood. (Bottom) Proviral load in the PBMC and SIV-specific antibody titers in serum. Detection of proviral DNA in lymph nodes (LN DNA) and isolation of virus from peripheral blood (VIRUS) at different times after challenge are as indicated at the bottom.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We confirmed that immunization with recombinant SIVmne gp160 vaccines in a "prime-and-boost" regimen is highly effective against an intravenous challenge by the pathogenic homologous virus clone E11S. We also achieved complete protection against the uncloned SIVmne, which has greater genetic complexity and pathogenicity than the E11S clone, in about one-third of the immunized animals. However, protection against the uncloned virus was only partial in some animals and not apparent in others, indicating the potential limitation of similar vaccine strategies based solely on envelope antigens of a single genotype.

Our results thus provide further insight into the apparently discordant observations concerning the efficacy of the prime-and-boost approach with envelope-based vaccines reported in a number of primate lentivirus models (16, 19, 25, 33, 34, 47, 48). The protective efficacy of a given vaccine approach depends not only on the quantity and quality of the immune responses generated by the vaccine, but also on the biological properties and the genetic complexity of the challenge virus. Failure to achieve protective immunity with similar prime-and-boost immunization protocols against SIVmac251 may be due in part to the highly virulent nature of the challenge virus, which often induces AIDS and death in monkeys within a year (5, 19). Recent data indicate that monkeys infected with SIVmac251 have a high plasma viral load, usually with peak values of >10⁸ and steady-state values of 10⁶ to 10⁷ (5, 16, 21). These values are 10- to 100-fold greater than those observed in most of the SIVmne-infected animals (reference 55 and data not shown) or HIV-1-infected typical progressors (44, 45). The use of a highly virulent challenge virus such as SIVmac251 may therefore underestimate the efficacy of candidate vaccines that are capable of protecting against less virulent viruses. On the other hand, the SIVmne challenge virus appears to have only limited genetic complexity. Immunity generated by molecularly cloned vaccines may be more effective in controlling infection by a few closely related viruses than that by a "swarm," especially if it contains highly replicative and virulent viruses. However, it remains an open question whether results from any of these models will be predictive of the relative efficacy of comparable vaccine strategies against HIV-1 infection in humans. Such a question can be answered only with efficacy data from clinical trials.

Despite considerable efforts, correlates of protection in the SIV model remain elusive. Protective immunity elicited by whole killed SIV vaccines has been attributed to immune responses against cellular antigens (2, 3, 41, 53). Results from our studies indicate that cross-reactive immune responses to cellular antigens did not play an important role in protection by the recombinant gp160 vaccines, because protection was observed against viruses grown on macaque PBMC as well as those grown on human T-cell lines. Among the SIV-specific responses examined, we were unable to identify any correlate of protection. Neither total antibody response (as measured by ELISA) nor neutralizing antibodies (including those against the homologous challenge viruses, cloned or uncloned) correlate with protection. In a subset of immunized animals examined for SIV-specific CTL responses, none showed any detectable level of cytolytic activity prior to challenge (37, 37a). It is possible that protection is mediated either through a combination of these mechanisms or through factors yet to be determined.

Another approach to study the mechanisms of protection is to analyze viruses recovered from immunized animals after challenge infection. Although our data do not distinguish breakthrough infection from preferential amplification early after infection, they indicate that such viruses were more likely to have variant sequences in the V1 hypervariable region (i.e., different from the V1 region of the vaccine strain). If further confirmed, this observation will provide indirect evidence that immune responses against determinants within the V1 region may play an important role in protection, by either preventing or limiting infection by viruses with the homologous sequences. This notion is supported by findings that the V1 region of the SIV envelope proteins contains targets for both neutralizing antibodies and cytotoxic T lymphocytes (18, 23, 33, 36, 51). Recent studies in the SIVmne model have shown that mutations in the V1 region allow the virus to escape neutralizing-antibody recognition and that such mutations are selected over the course of persistent infection in macaques (18, 49, 51). The "variant-type" V1 sequences we found in immunized but infected animals have the same canonical O-linked and N-linked glycosylation sites as those observed in neutralization escape mutants and in variants generated during in vivo infection. These findings indicate an important role for neutralizing antibodies in the selection or preferential outgrowth of variant viruses in animals challenged with the uncloned virus. However, it remains to be shown if and how such responses contribute to protection. For instance, while we failed to observe a correlation between protection against the uncloned SIVmne and neutralization of the homologous virus, this does not preclude the possibility that such a correlation exists for antibodies that neutralize the variant viruses specifically. Further work is needed to determine whether the failure of the E11S-based vaccines to protect against viruses with variant V1 sequences is due to differences in these putative V1 determinants per se or to differences in their biological properties (such as infectivity and pathogenicity) that are independent of V1 (i.e., variant sequence serving only as a "signature" for such differences), or both. The finding that vaccine-induced immunity may bias the type of virus transmitted after exposure also underscores the importance of identifying the relevant viruses to protect against in natural transmission as well as developing vaccines that will elicit broadly protective immunity. In this context, it is relevant to point out that by using the same combination immunization strategy, we have been able to protect against mucosal infection by uncloned SIVmne with gp160 vaccines and against intravenous infection by the same virus with vaccines consisting of both envelope and core antigens (unpublished data).

Finally, a key issue in vaccine development is how vaccine efficacy is defined. It has been recognized that "sterilizing immunity" (i.e., immunity that prevents the initial infection) may not be an easily attainable or a desirable goal (4, 39, 52). Since the correlation between disease-free survival and low viral load in plasma in HIV-1-infected individuals was demonstrated, it has been proposed that reduction of viral load may be a more realistic goal for HIV vaccine trials. This notion, while supported by some studies with nonhuman primates (5, 12, 28, 35, 42, 47), is not supported by others (19, 25). Results from our study failed to show any statistically significant difference in the clinical course in control and immunized animals after infection, perhaps due to the limited number of animals and duration of experimentation. However, we did observe that two of the five immunized macaques that showed transient viremia in the peripheral blood early after challenge infection eventually developed a high viral load, CD4⁺ cell depletion, and AIDS within the same time frame as the controls. Findings such as these should be considered in the design of HIV-1 vaccine trials in which transient viremia and reduction of virus load are to be used as surrogate markers for vaccine efficacy.