Research Brief

Studies Showcase Advances in Designing Antibody-inducing Antigens

Antibodies are the reason most, if not all, licensed vaccines provide protection against disease. So it is not surprising that vaccine researchers long ago set their sights on inducing them against HIV. For many reasons, inducing antibodies that could neutralize the vast array of HIV variants in circulation is proving a difficult task. But researchers are now making strides in developing vaccine immunogens designed to induce broadly neutralizing antibodies (bNAbs). A trio of research studies published recently in the journals Science and Cell showcase promising first steps in developing immunogens that are capable of effectively stimulating the immune system and goading it to develop a desirable antibody response.

This progress is due in part to recent advances in stabilizing HIV’s highly mutable and unstable trimeric Envelope (Env) protein that is the target of all antibodies, and in isolating and characterizing naturally occurring bNAb responses in HIV-infected individuals. For the first time, researchers are now developing trimeric proteins that closely mimic the natural structure of the HIV Env glycoprotein and testing them as vaccine immunogens. “This represents more than 10 years of hard work and good virology,” says John Mascola, director of the Vaccine Research Center (VRC) at the National Institute of Allergy and Infectious Diseases (NIAID), who was not involved directly in the new studies. Researchers are also engineering vaccine immunogens based on the conserved viral epitopes targeted by the slew of recently identified bNAbs and their precursors.

Both of these strategies are designed to induce the types of broad and potent antibody responses that develop rarely in natural infection—researchers estimate approximately only 20% of individuals develop such bNAbs—and which occur only after years of infection. Continuous exposure to the constantly mutating virus is what stimulates the process of antibody maturation, eventually giving rise to potent antibodies that are capable of neutralizing a wide swath of HIV variants. Although these antibodies do not help infected individuals control the virus, they can prevent infection in animal models and are now being tested in clinical trials of passive transfer to see if they can do the same in humans.

There are still many obstacles to developing an effective bNAb-based HIV vaccine, but this latest research makes some scientists optimistic. Mascola calls this trio of research papers a “major advance.”

Trying the trimer

After almost two decades of failed attempts to stabilize HIV’s floppy Env protein, John Moore, professor of microbiology and immunology at Weill Cornell Medical College, and colleagues reported successfully stabilizing an HIV gp140 protein designated BG505 SOSIP.664 in 2013 (see Keystone in Rio: Breakthroughs, Predictions, and Surprises, IAVI Report, Winter 2013). This trimeric Env protein adopts a native-like conformation and was based on a clade A virus isolated from a six-week-old Kenyan infant who developed a bNAb response after approximately two years of infection.

The earliest immunogenicity data generated by testing BG505 SOSIP.664 in rabbits was presented early last year (see CROI: Progress on Prevention and Cure, IAVI Report, Vol.18, Issue 1, 2014). Now Rogier Sanders, adjunct assistant research professor of microbiology and immunology at Weill Cornell Medical College and the University of Amsterdam, Moore, and colleagues have published more complete immunogenicity data from five experiments in rabbits and one in macaques testing the BG505 SOSIP.664 trimer, as well as the B41 SOSIP.664 trimer, which is based on a clade B founder virus from an HIV-infected adult (Science 2015, doi:10.1126/science.aac4223). Founder viruses are the transmitted viruses thought to be responsible for establishing an infection.

These data show that the soluble, native-like BG505 trimer protein generated cross-reactive neutralizing antibody responses in rabbits against a panel of viruses classified as Tier-1 viruses—a designation given to those viral strains that are easier to neutralize—and potent neutralizing antibody responses against only those Tier-2 viruses with sequences matching that of BG505. Tier-2 viruses are representative of the most commonly transmitted strains of HIV and are what a vaccine would ultimately need to protect against. In other words, the antibodies induced by BG505 were not able to broadly neutralize Tier-2 viruses. However, in some cases these Tier-2 neutralizing antibodies did target some of the same epitopes on HIV Env that are targeted by bNAbs.

Researchers also compared the neutralizing antibody responses in rabbits with those that developed in the Kenyan infant from whom the BG505 virus was isolated. This comparison shows that the recombinant trimers induce antibody responses similar to those occurring during the primary infection phase of the infant, but the BG505-induced antibodies were not nearly as broadly neutralizing as those detected in the infant after 27 months of HIV infection.

BG505 induced similar immune responses in macaques, however the antibody titers, as measured by ELISA assay, were approximately five-fold lower than in rabbits. This suggests a better adjuvant could be used to boost antibody titers in monkeys, researchers say.

Results with the B41 SOSIP.664 trimer were similar. This clade B protein induced heterologous Tier-1 neutralizing antibody responses in all rabbits studied, and induced autologous Tier-2 neutralizing antibody responses in eight of ten immunized rabbits.

Although BG505 did not induce bNAbs against Tier-2 viruses, the study’s authors suggest that inducing neutralizing antibody responses against an autologous Tier-2 virus is an “excellent starting point for iterative vaccine design.” While almost all Env protein immunogens induce antibodies against Tier-1 viruses, identifying an immunogen that could induce antibodies against an autologous Tier-2 virus was previously a challenge. “Previous Env immunogens did not consistently induce potent Tier-2 neutralization against heterologous Tier-2 viruses, but not even autologous Tier-2 viruses, indicating they are inadequate immunogens when the aim is to induce broadly neutralizing antibodies,” says Sanders, lead author on the paper. 

Mascola agrees. “Before this we haven’t had the ability to express a protein that really has the antigenic structure of the viral trimer. Now we can do that,” he says.

Still, there is a long way to go. “It’s now clear to us that native-like trimers are the best route to neutralization breadth,” says Moore, “so we are going to do all we can to refine their design and learn how to use them better.”

Researchers are exploring several strategies to rationally improve the immunogenicity of the SOSIP.664 trimers, or other soluble, stable trimeric Env proteins. Strategies include removing non-neutralizing antibody epitopes from the proteins to avoid distracting the immune response, immunizing with sequential SOSIP.664 trimers, or immunizing with cocktails of different trimers, among several others. Simply increasing the titer of neutralizing antibodies against easier to neutralize Tier-1 isolates is unlikely to improve the antibody response to Tier-2 viruses, according to the rabbit studies with BG505. These data indicate that the antibodies mediating neutralization of Tier-1 and Tier-2 viruses have different specificities and most likely arise from different B cells.

“Our goals now are to devise ways to broaden the neutralizing antibody response to eventually counter a wide range of heterologous Tier-2 viruses,” says Moore. “Only if we can succeed in doing this will we have a chance of coming up with a practical vaccine that might confer a meaningful degree of protection from infection.”

Engineering a better immunogen

Another approach to inducing bNAbs that is gaining traction is designing vaccine immunogens based on the epitopes of Env that are targeted by these antibodies. Thanks to advances in B-cell isolation techniques, over the past six years researchers have isolated scores of bNAbs from HIV-infected individuals. Many of these antibodies were then fully characterized, and through that process it became clear that there are multiple highly conserved regions of Env that are targeted by antibodies. This is welcome news for vaccine researchers.

One class of bNAbs that is widely studied is those targeting the CD4 binding site on Env—a crucial epitope where the virus binds to CD4 receptors, allowing it to infect CD4+ cells. The antibodies that target the CD4 binding site mimic the way HIV binds to cells. Researchers at the VRC identified the first antibody from this class, known as VRC01. Since then, VRC01-like bNAbs were identified in at least seven different HIV-infected donors.

VRC01, like many of the bNAbs recently identified, has several unique characteristics. One is that this antibody, and others that are similar, are heavily somatically mutated. This means the B cells from which these antibodies are derived have undergone multiple rounds of mutation and selection in the germinal centers in response to chronic exposure to the ever-mutating virus. It is through this process of somatic hypermutation that antibodies mature and develop a higher affinity for HIV. While an average antibody may have a degree of somatic hypermutation in the range of 3%-5%, the VRC01 class of antibodies are 30% mutated. It has been shown that not all of these mutations are required for these antibodies to neutralize HIV so effectively. However, the amount of mutation required is still much higher than what is typically achieved by vaccination. While the germline precursor of VRC01 is unknown, researchers are able to work backwards from this antibody and formulate their best guess of what the germline or immature antibody looks like. This hypothetical germline version does not bind well to native HIV Envs.

Which is why researchers from The Scripps Research Institute (TSRI) in La Jolla, IAVI’s Neutralizing Antibody Center, and the Ragon Institute designed what they call “germline-targeting” vaccine immunogens. These immunogens are capable of binding germline VRC01-class B-cell receptor and initiate immune response, setting off what these researchers hope is the initial step in shepherding the immune system to induce broadly neutralizing VRC01-like antibodies.

Joseph Jardine, a postdoctoral research fellow in Bill Schief’s laboratory at TSRI, and colleagues developed one of these immunogens composed of a stripped down outer domain (eOD) of HIV gp120 that can react with germline VRC01-class antibodies. This eOD was then engineered to form self-assembling 60-subunit nanoparticles that mimic the size and shape of a typical virus. Jardine and colleagues tested this engineered immunogen, referred to as eOD-GT8 60mer, as a priming immunization in a transgenic mouse model developed by David Nemazee of TSRI that allows the mice to develop human antibodies. The first immunogenicity data for the eOD-GT8 60mer were presented earlier this year at the Keystone Symposium on HIV Vaccines (see Opening the Envelope, IAVI Report, Vol.19, Issue 1, 2015), and published recently (Science 2015 doi:10.1126/science/aac5894).

The results are encouraging. Mice immunized with a single injection of eOD-GT8 60mers developed antibodies with characteristics similar to that of the VRC01 class of antibodies, whereas neither a previous version of an engineered “germline-targeting” immunogen or the native-like BG505 SOSIP.D664 trimer did. By 42 days after immunization, many of the antibodies in eOD-GT8 60mer-vaccinated mice had accrued mutations that resulted in more than a 1,000-fold increase in their binding activity to HIV Env, as measured by surface plasmon resonance. And importantly, these antibodies started to bind to the CD4 binding sites of more native-like Envs. This suggests a second immunization with more native-like Env immunogens would be recognized by these antibodies and could further guide their development toward a VRC01-like bNAb response. 

Jardine and colleagues also evaluated different adjuvants and found that selection of adjuvant greatly influenced the responses, resulting in differences in the antibody titers in serum, the mutations selected for by the B cells, and the generation of high-affinity antibodies in this mouse model. “The adjuvant makes a huge difference,” Jardine says, which is something researchers think should be considered when testing vaccine candidates in humans.

In a related study, researchers at The Rockefeller University collaborated with the TSRI researchers to test the same eOD-GT8 60mer immunogen in a different transgenic mouse model. This immunogen also binds the predicted unmutated precursors of the 3BNC60 antibody—a VRC01 class antibody that was identified by researchers at Rockefeller. Like VRC01, reverting 3BNC60 to its presumed germline form results in complete loss of the antibody’s ability to bind to and neutralize HIV.

Immunization with eOD-GT8 60mer resulted in a higher number of B cells expressing antibodies with traits similar to that of 3BNC60 and other CD4 binding site-directed antibodies, and induced somatic hypermutation, providing additional support for this approach (Cell 2015, doi: 10.1016/; pII).

This is a promising first step. Researchers speculate that a series of immunizations with different immunogens that are increasingly similar to the structure of native HIV will be required to induce the type of broad neutralizing activity conferred by VRC01-like antibodies. Still, these results are exciting to those involved. “We’ve initiated this process and we think that’s a big step forward,” says Jardine, who is now plotting studies of this sequential immunization approach, which, if successful, could be used beyond HIV as a new paradigm for vaccine design and development. —Kristen Jill Kresge