Despite success in preclinical tests, previous vaccine candidates were ineffective in human trials. This has raised concern about the predictive power of available tissue culture and animal models of HIV infection. The tissue culture infection assays are the only available means to determine whether a candidate vaccine induces Abs capable of neutralizing a sufficiently broad range of HIV strains. Panels of virus strains with varying genetic divergence and resistance to commonly studied Abs to gp120/gp41 have been assembled to assess neutralization breadth and potency [76
]. CCR5-dependent primary strains are substantially more resistant to anti-HIV Abs compared with laboratory-adapted strains [80
]. Abs to the 421–433 epitope neutralize highly divergent HIV strains drawn from different group M subtypes, including difficult-to-neutralize CCR5-dependent strains using the classical peripheral blood mononuclear cell (PBMC) assay (or clinical isolate infection assay) [47
]. This assay utilizes the closest available tissue culture model to the natural infection process, for example, phytohemagglutin-activated PBMCs pooled from humans without HIV infection and HIV isolates grown from the clinical specimens (usually serum) from infected patients that have not been subjected to tissue culture passage in cell lines. Minimizing tissue culture manipulations is advisable in view of changes in viral coat structure upon growth in different types of host cells [81
]. Host cell effects on viral growth rates and susceptibility to neutralization by Abs have also been described [81
]. The PBMCs are composed primarily of lymphocytes with variable levels of monocyte contamination. The natural host cells for HIV-1 are T lymphocytes and cells of the monocytic lineage.
To assess the reliability of the PBMC/clinical isolate assay, we conducted a retrospective analysis of neutralization data gathered using the PBMC/clinical isolate assay and two well-defined Abs to the 421–433 epitope, mIgG clone YZ23 isolated by immunization with E-gp120 and single chain Fv (scFv) clone JL427 isolated from a lupus Ab phage library. The extent of variability was judged from the Ab concentrations needed to reach 50% virus neutralization (IC50 values). The coreceptor CCR5-dependent subtype C strain 97ZA009 was neutralized by IgG YZ23 in 26 out of 26 assays (IC50 range: 0.5–59 μg/ml) and by scFv JL427 in 34 out of 35 assays (IC50 range: 0.003–9 μg/ml) conducted using various host PBMC batches, various virus batches and various purified Ab batches. The variability was reduced using the same infecting virus inoculum obtained from a single large-scale tissue culture passage in PBMCs (). The IC50 values for Ab clones YZ23 and JL427 were spread over a 13-fold and 150-fold concentration, respectively, range using different batches of pooled PBMCs as hosts. The IC50 spread for assays conducted using the same pooled PBMC host cells was smaller. Neutralization assays conducted in parallel using PBMCs isolated from four individual human donors indicated a small IC50 spread for clone YZ23 and a larger IC50 spread for clone scFv JL427 (). In future studies, it is important to determine whether activation of the cells by the mitogen (phytohaemagglutinin) or allogenic PBMCs is a factor governing the potency of Ab neutralization. Provided that the variability is taken into account by including a sufficiently large range of test Ab concentrations, the assay is a reproducible guide to the neutralizing activity of the Abs.
Reliability of antibody neutralization in tissue culture
Endotoxin (lipopolysaccharide) can induce chemokine release from monocytes that may bind chemokine coreceptors and inhibit HIV infection [85
]. Preparations of Abs to the 421–433 epitope-containing endotoxin at concentrations lower than required for HIV neutralization by the chemokine-release mechanism displayed readily detected virus neutralization [37
]. Removal of trace endotoxin amounts in the Ab preparations by ion-exchange methods did not diminish the neutralizing activity [37
]. Moreover, a wealth of immunochemical data and control studies are inconsistent with endotoxin contamination and other trivial causes of HIV neutralization [37
], for example, removal of the neutralizing activity upon immunoadsorption with immobilized E-416–433 but not an irrelevant immunoadsorbent and induction of the neutralizing Abs by immunization with E-gp120.
Reporter cell lines and HIV pseudovirions have been developed for convenient analysis of large numbers of Ab samples, for example, the TZM-bl cell line/pseudovirion assay [87
]. TZM-bl is a genetically engineered HeLa cell line that expresses the HIV receptors and contains the Tat-inducible luciferase gene. The pseudovirions are replication-incompetent particles expressing the HIV envelope proteins. This assay is unsuitable to detect Abs to the CD4BS 421–433 epitope. Homogeneous Abs to this epitope prepared in the authors’ laboratory [37
] did not impede infection of TZM-bl cells by several pseudovirus strains but neutralized the corresponding native HIV strains expressing gp120 with the same sequence (clinical virus isolates) in the PBMC assay in David Montefiori’s laboratory (). The reference anti-CD4BS mAb b12, also displayed discrepant neutralization of one strain (undetectable neutralization of strain 98Du123 in the PBMC/clinical isolate assay and robust neutralization in the TZM-bl assay). Discrepant neutralization by other anti-HIV mAbs in these assays has been noted previously [24
]. For example, the pseudovirion reporter assay does not detect the neutralizing activity of mAb 2G12 validated by in vivo
passive transfer studies [91
]. The level of discrepancy varies depending on the epitope recognized by the Abs. It is possible that excessive expression of the HIV coreceptor CCR5 on TZM-bl cells compared with PBMCs [91
], and nonphysiological pseudovirion interaction with host membrane proteins/lipids permit infection with reduced dependency on the CD4BS 421–433 epitope. The conformational flexibility of gp120 in differing membrane microenvironments is another variable [26
]. Epitope-specific variations in the conformations of gp120 expressed by native HIV versus pseudovirions are conceivable.
Animal model testing is desirable to predict the success of candidate human vaccines. HIV infects chimpanzees transiently. The infection does not progress to AIDS. Immunization of chimpanzees with recombinant gp120 suppressed HIV viremia, but human trials of the gp120 immunogen did not reduce HIV infection risk [6
]. As the HIV and SIV envelope proteins are structurally divergent, direct testing of candidate HIV vaccines in the SIV-infection model is difficult. Hybrid simian–human virus strains (SHIV) containing the HIV envelope proteins grafted into SIV produce viremia in rhesus monkeys. Candidate vaccines that induced cytotoxic T cells protected monkeys from SHIV infection but did not protect humans from HIV infection [7
]. The SHIV/rhesus monkey model was recently suggested to be a useful ‘gatekeeper’ to identify candidate vaccines that induce ‘better immunity’ compared with the failed immunogens [97
]. However, as the precise laboratory tests constituting ‘better immunity’ have remained undefined, it is not possible to predict vaccine success in humans from this animal model.