For years scientists have been puzzled by the role of neutralizing antibodies in the control of human immunodeficiency virus (HIV) infection. Despite an early and effective neutralizing antibody defense, initial infecting viruses are able to mutate rapidly into resistant viruses, escaping neutralization by antibodies.

Two papers published last week examine how antibodies quickly eliminate neutralization-sensitive HIV early in infection, only to be replaced by populations of resistant virus. One of the research teams discovered that changes in glycan moieties of the viral envelope generate yet another mechanism of viral neutralization escape.

The team led by George Shaw, from the University of Alabama at Birmingham, believes that they have found a new way by which HIV escapes recognition by neutralizing antibodies. The paper, published in last week’s Nature [PubMed], describes a model in which the virus is able to resist neutralization by evolving a “glycan shield” mechanism of neutralization escape. According to the model, mutations in resistant viruses had primarily occurred at N-linked glycosylation residues of the env gene rather than at the well-studied neutralization epitopes. Shaw and his team were able to experimentally support their model by constructing neutralization-resistant site-directed mutant envelopes from early neutralization-sensitive viruses. The mutated viruses were tested against autologous plasma and neutralizing monoclonal antibodies.

A second team lead by Douglas Richman, from the University of California, San Diego, in collaboration with ViroLogic Inc., in South San Francisco, reported similar results. The findings were published in the Proceedings of the National Academy of Sciences [PubMed]. Using a recombinant virus assay developed by ViroLogic, the group studied serial plasma specimens from patients with primary HIV infection. They found that a significant neutralizing antibody response to autologous viruses quickly developed. However, plasma viruses mutated at a very high rate, probably due to the selective pressure exerted by neutralizing antibodies. As a result, the virus acquired the ability to rapidly change into forms that are not sensitive to neutralization by autologous antibodies.

Neutralizing antibody responses to heterologous viruses or to laboratory strains were negligible or did not develop at all during the first year of HIV infection. This lack of cross-neutralization antibody response against heterologous primary isolates in the early stages of infection is of great concern, they indicated. That's because vaccine research is focused on immunogens that can get a broadly reactive antibody response. They add, however, that the prophylactic eliciting of a narrowly focused but potent neutralizing antibody response, able to tackle very high levels of viral replication, may be effective and protect against a modest inoculum.

We had the opportunity to talk to Dr. Shaw about his research, and he kindly gave us some additional insights on these recent experiments:

[AIDScience] Were you aware that the ViroLogic group was working concurrently on the same [neutralization escape] problem?

We found out about each other’s work a year ago at a meeting where we were both presenting our work, and since that time we have been aware of it. We are both studying exactly the same topic, that is, the development of neutralizing antibodies to HIV-1 in humans. We have used slightly different methodologies and have made different observations that have been mutually reinforcing.

Why do you think it has taken so long to study this aspect of neutralization escape?

I think our work — as well as the work by the Richman’s group — applied different experimental tools and strategies to look at an old question. The whole field of neutralizing antibodies against HIV — as well as other viruses — is a very well established discipline. In the past, people have looked at the issue of neutralizing antibodies against HIV using the tools that were at hand which were, typically, phytohaemagglutinin (PHA)-stimulated human lymphocytes and virus obtained from peripheral blood mononuclear cells as the source of virus.

Early in 1995, we published a paper in Nature [PubMed] where we showed the dynamics of HIV turnover using plasma virus. We noticed that plasma virus has such a short half-life that the study of its genetic composition could provide a real-time assessment of the biological selection pressure acting on the virus, whether it was drug or cytotoxic T lymphocytes (CTL) effects, or as in the present case, neutralizing antibodies. In the 1995 paper we looked at the effects of drugs on the selection of mutant viruses. It was a very rapid process, and we could infer the effects that drugs were having by changes in the virus quasispecies. In 1997, in a Nature Medicine paper [PubMed] we asked very similar questions with regard to CTL escape. We found the same process again. When we looked at acute infection, as we have done in the present case, but at CTL epitopes, we found that plasma virus turned over very quickly and achieved escape by epitope-specific CTL.

Several years ago I recognized that if neutralizing antibodies are there and are exerting a biologically significant role, we should be able to infer it. For that, we needed a more sensitive test for the neutralizing antibodies themselves. Additionally, we should apply that test to look at changes in the viral quasispecies. Our test is not completely novel because people have used recombinant-based entry tests similar to the one we describe to answer many questions on HIV.

I was going to mention that the techniques that you used have been available for some time.

Correct. But the cell line that we use is a new HeLa-based cell line, JC53BL-13, which are genetically modified to express CD4, CCR5 and CXCR4. It is very sensitive to infection by both R5 and X4 viruses. We also engineered those cells to express beta-galactosidase and luciferase under the control of the HIV-1 LTR as the indicator. The important properties of this cell line — I would not say that they are entirely unique — allowed us to use them as a sensitive indicator of virus entry using naturally occurring viral envelopes. This differentiates it from many other cell lines that may not work as well.

We obtained viruses directly from plasma, extracted the viral RNA, and amplified the envelope gene. After sequencing and determining that they were biologically functional, the envelopes were used to construct pseudotype viruses. These pseudotype viruses were then used to look for neutralizing antibody effects, which we describe in detail in our paper.

How do you think your findings can be quickly translated to HIV vaccine research?

I believe they will be quickly translated to HIV vaccine research for several reasons. The competing paper by the ViroLogic group uses an assay initially developed to measure antiretroviral drug resistance in a single entry cycle. This assay is different than ours but used to answer the same questions. Both their assay and ours can be applied to the vaccine effort because they can both detect neutralizing antibodies in a very sensitive fashion.

People who are interested in asking whether or not candidate vaccines are inducing neutralizing antibodies now have a different assay that provides a slightly different aspect of neutralizing antibodies in vaccinated subjects. In both of our papers we have shown that the neutralizing antibodies that we are detecting in vitro have a biological correlate in vivo. Now we know what a 10-fold difference in neutralization susceptibility will do to the viral quasispecies in vivo. We have developed a test which measures an activity in vitro that, potentially, is a surrogate for neutralizing antibody efficacy in vivo.

When you amplified the virus from the early plasma samples, how did you know that those envelopes were representative of the quasispecies present in the samples?

That is a really good question. In our paper we typically amplified 10 or 15 different envelopes from each plasma sample. We then looked at the sequences of those envelopes and found that they were representative. In some cases we also tested biologically for susceptibility to neutralization. If we had representative viruses, we predicted that all early envelopes would behave in one manner, and all the late envelopes would behave in a different manner, and that is in fact what we found.

In this aspect our studies differ from those described by the Richman’s group. While we cloned individual envelopes into expression vectors and made pseudotype viruses that were genetically pure, the ViroLogic group took plasma samples and essentially shotgun-amplified everything that was in plasma and introduced them into HEK293 cells. These are a human epithelial kidney cell line with a vector that lacks env. Through these cells they are able to recover a mixed viral stock potentially more representative of plasma virus at any one point in time.

There are pros and cons of using both assays. In our case, we know exactly what the envelope is in our virus stock and we can do structure/function analyses. This is how we determined the glycan shield mechanism. We had a biological phenotype in terms of virus neutralization susceptibility and we knew, unequivocally, what env sequences corresponded to that phenotype, which lead us to the glycan shield model.

ViroLogic’s methodology is not only different in this aspect, but their conclusions are different too. We spent a lot of time talking about the evolving glycan shield, while they just basically commented that they do not know why their viruses differ phenotypically although they have many genetic changes.

Do you think vaccine developers will start worrying more about glycosylation rather than just protein sequence/structure?

In our study we showed an additional method by which HIV-1 evades neutralization, among three or four others that also play important roles such as epitope variation, oligomeric exclusion, and the immunologically “silent face” which is covered by glycans and where antibodies are not elicited.

^*Senior Editor, AIDScience