Vaccines to Antibodies: Grow Up!

Through a process called affinity maturation, all of the HIV-specific antibodies identified so far have accumulated multiple mutations, some of which are required for them to bind to and neutralize HIV. Researchers are now beginning to grapple with what this means for vaccine design.

By Andreas von Bubnoff

Recently, researchers from the Vaccine Research Center (VRC) at the National Institute of Allergy and Infectious Diseases (NIAID) reported the isolation of three new broadly neutralizing antibodies (bNAbs) against HIV, as well as the structure of the antigen binding portion of one of the bNAbs called VRC01 bound to the HIV gp120 core protein (1,2). These are just the latest in a slew of recently isolated bNAbs (see Research Briefs, IAVI Report, Jan.-Feb. 2010).

VRC01 neutralizes 91% of the HIV strains tested, making it the most broadly neutralizing antibody isolated so far. Researchers found that VRC01 is so efficient at neutralizing HIV because part of the antibody mimics the way the CD4 co-receptor—the main receptor HIV uses to infect CD4+ T cells—binds to gp120. “[HIV] has to keep part of itself absolutely constant in order to recognize CD4, [and] the immune system has sort of counter-attacked that one part of conservation to neutralize the virus,” says Peter Kwong, chief of the structural biology section at the VRC and lead author of the study that described the VRC01 structure.

That study also showed that VRC01 has an unusually high degree of affinity maturation, which indicates how many amino acid residues of an antibody’s variable region differ from the genes that antibody is derived from in the germline as the result of a process called somatic hypermutation (see How Somatic Hypermutation Leads to Affinity Maturation, for details). In VRC01, about 30%, or more than 60 amino acids of the variable region, differ from the germline sequence, much more than the 5% to 10% of affinity maturation observed in most other antibodies, says Kwong. VRC01’s affinity maturation also seems important for binding—a version of VRC01 with all of the affinity-matured amino acids reverted to germline could not bind gp120 well anymore. “If you reverted them you got really, really low affinity,” Kwong says.

VRC01 is the HIV-specific bNAb with one of the highest degrees of affinity maturation known. But, in general, many HIV-specific bNAbs show a high degree of affinity maturation, and recent studies suggest that affinity maturation is also important for Env binding of several HIV-specific bNAbs other than VRC01. Earlier this year, Kwong’s group reported that the bNAbs PG9 and PG16, which were isolated last year by researchers at IAVI and The Scripps Research Institute (TSRI), show about 20% affinity maturation and that versions of these bNAbs that were almost completely reverted to germline show no (PG16) or poor (PG9) breadth and potency of neutralization of a panel of HIV isolates (3). And last year, Dimiter Dimitrov, a senior investigator at the National Cancer Institute (NCI) in Frederick, Maryland, reported that germline-like versions of the bNAbs b12, 2F5, and 2G12, which according to Dimitrov also differ by about 20% from germline in their variable region, don’t show any measurable binding to HIV Env (4).

The observations that HIV-specific bNAbs show a high degree of affinity maturation and that their germline reverted versions don’t seem to bind HIV Env have led to concern that current strategies to develop vaccines that are based on HIV Env as an immunogen might not work, because if Env does not bind to the germline precursors of bNAbs, it won’t induce the affinity maturation process that is necessary to make the bNAbs. “Vaccines based on HIV Envelope may not initiate [an] immune response [that elicits bNAbs],” says Dimitrov. “We simply don’t have germline antibodies in our naive B cell repertoire [that] can bind the highly conserved structures, which are exposed, like the CD4 binding site.”

To address these concerns, researchers are looking for immunogens that can bind to the germline precursors of bNAbs like VRC01 to be able to develop vaccines that can kick-start the affinity maturation process, though not everyone is convinced this will be necessary. Researchers are also taking a closer look at how the B cell immune response and affinity maturation develop in HIV-infected people starting soon after infection.


How Somatic Hypermutation Leads to Affinity Maturation  
In the bone marrow, the body makes millions of different versions of B cells by rearranging antibody genes. These are called naive B cells because they haven’t yet encountered antigen.

Once the B cells come into contact with antigen in lymphoid organs, such as lymph nodes, and that antigen can bind to their B cell receptor, they start to multiply and begin undergoing a process called somatic hypermutation, which causes additional mutations in the antibody genes. The B cells with mutations that result in the most improved binding—or affinity—to antigen multiply again and start another round of somatic hypermutation. These repeated cycles of somatic hypermutation and selection result in increasingly affinity-matured antibodies.

This affinity maturation process enables the body to generate a much larger number of different antibodies than can be generated by shuffling around the body’s germline repertoire of antibody genes. —AvB

Kick-starting affinity maturation

To kick-start the affinity maturation process, Dimitrov argues that a vaccine would have to contain immunogens that are different from HIV proteins, or immunogens that are rare variants of HIV proteins, which can bind the germline precursors of bNAbs. To direct the affinity maturation process toward antibodies that can bind the native HIV Env protein, a vaccine would also have to contain the native Env or gp120, he suggests. “We need [a] variant of the Envelope or another protein which is different from the Envelope to initiate the immune response to reach some intermediate degree of somatic hypermutation,” Dimitrov says. “Then [native] Envelope comes, and this intermediate antibody binds the [native] Envelope [efficiently].” Then, he adds, the intermediate antibody matures further, eventually resulting in bNAbs. If this is not enough to help guide the affinity maturation process, Dimitrov suggests a vaccine might also have to contain one or more additional immunogens that have a structure in between the native gp120 or Env and the version that binds to the germline version of the bNAb the vaccine is intended to induce. He cautions, however, that while this is in principle how it could work, it’s far from proven. “We need a much deeper understanding of how our immune system works,” he says. “Affinity maturation is very complex and guiding immune responses has not been achieved.”

Dimitrov, who says he was the first to publish this immunization strategy last year (5), remembers that initially, his talks on this subject often got little attention. Even though the sequence of mature bNAbs such as b12, 2F5, and 2G12 had been known for many years, he says, “there were no published studies to show how important it is that these bNAbs are so somatically hypermutated. There were no reports that this could be critical.”

But in light of the recent observations that show bNAbs such as VRC01 have a high degree of affinity maturation that is important for binding, that seems to be changing. “It was discovering these very highly affinity-matured antibodies [like VRC01] that had us focus on affinity maturation,” says Lawrence Shapiro, an associate professor of biochemistry and molecular biophysics at Columbia University, who works with Kwong and is thinking about immunization strategies similar to the ones Dimitrov is suggesting. “It’s clear that the focus has moved on to affinity maturation more than it ever has.”

Yet not everyone is certain that finding an immunogen that can bind to germline-reverted versions of bNAbs is really necessary. The germline-reverted precursors of bNAbs might actually bind gp120, albeit at a very low level that might be sufficient to kick off an immune response, says Dennis Burton, a professor at TSRI. “What we don’t know is what the threshold of binding affinity is in order to kick off an antibody response,” Burton says. John Mascola of the VRC agrees. “While it’s true that the germline antibody doesn’t bind well to gp120, it’s still possible and perhaps even likely that [the naive B cell] recognizes HIV in a more physiological setting,” says Mascola. “It may be that a fairly low affinity interaction of HIV with the naive B cell is all that it takes to get the B cell stimulated and get going.” Still, he adds, the idea that an immunogen that shows better binding to germline antibodies might give better initial immune stimulation “is a reasonable hypothesis, and it’s testable.”

To see if the hypothesis is correct, Dimitrov is screening protein libraries to identify proteins that can bind to germline-reverted versions of bNAbs. Sanjay Phogat, a principal scientist at IAVI’s AIDS Vaccine Design and Development Laboratory, says that IAVI will be involved in screens that will look for proteins that can bind versions of the PG9 and PG16 antibodies, which have been largely reverted to their germline versions.

Kwong and colleagues are also trying to engineer an Env-derived immunogen that can bind better to the germline precursor of VRC01. They have already found that a version of gp120, which has been stabilized by adding chemical bonds into the configuration it has when it initially binds to the CD4 receptor, is able to bind the germline precursor of VRC01 better than the natural gp120 (1). This shows that, in principle, engineered changes in gp120 can improve its binding to the germline precursor of gp120.

Currently, Kwong and colleagues are working on solving the structure of the germline-reverted version of the VRC01 antibody. Knowing that structure will allow them to engineer a version of gp120 that can bind the germline-reverted VRC01, says Shapiro, who works with Kwong on the project. “It’s a well-defined problem. Can we design a gp120 variant that’s still a gp120 but is different enough to actually bind to the genomic version of these antibodies and activate the maturation process?” Shapiro asks. “One thing we can do is design mutations in the gp120 surface to change it enough so that it’s recognized by the immature antibody.” Such an engineered protein, he adds, could then be developed into an antigen, for example by masking with sugars the unimportant parts on its surface that could distract the immune system.

Once an experimental vaccine candidate has been developed, the challenge is to find or develop an animal model that is similar enough to the human immune system to test the candidate and see if it can indeed induce the right antibody responses. “There is some question about how exactly to proceed in terms of getting an animal system to work,” Kwong says. According to Dimitrov, several groups are planning to use mice that express human antibody genes. “If we test [this] in mice with the human germline antibodies, maybe we can get a proof of concept,” he says.

Meanwhile, Burton has been developing transgenic mice that only express the genes for the bNAbs b12, 2F5, or 4E10, respectively. He plans to use these mice to test if weak binding of normal gp120 to the germline versions of these bNAbs might be enough to induce the affinity maturation process. If so, then efforts to look for or develop immunogens that can bind the germline reverted versions of bNAbs might be unnecessary.

A recent study from the labs of Burton and Ian Wilson at TSRI suggests that immunogens different from HIV could also drive the affinity maturation process toward an HIV-specific bNAb. The researchers found that an affinity maturation change that enables the 2G12 antibody to bind to clusters of mannose residues on HIV Env does not necessarily have to have been driven by an interaction with HIV. Instead, it could have come from yeast infection because yeast, like HIV, has clusters of mannose residues on its surface (6).

Understanding the process

The isolation of VRC01 from an HIV-infected person shows that in principle, certain people can make this antibody. But researchers are just starting to understand how common such antibodies really are, and exactly how they actually develop over time in HIV-infected people.

“We already know that [the immune system] can make VRC01, although we are not sure that it can be made in everyone,” Kwong says, adding that searches to see how common VRC01-like antibodies are in HIV-infected people are underway at the VRC. “Once you know that many people make VRC01-like antibodies that means that the machinery is there [and] it can be made. [Then] it’s just [about] what are the right triggers of that machinery.”

Researchers are also studying how antibody responses develop in HIV-infected people from the time of transmission to the time they have bNAbs. Barton Haynes, a professor at Duke University Medical Center and director of the Center for HIV/AIDS Vaccine Immunology (CHAVI), is leading such a study, which also involves Dimitrov, as well as Scott Boyd and Andrew Fire of Stanford University. The study, a collaboration between CHAVI and the Bill & Melinda Gates Foundation, involves next generation 454 sequencing of millions of antibody genes of memory B cells taken from HIV-infected people as they develop bNAbs at different time points after infection. The goal is to determine at least part of what Dimitrov has dubbed the “antibodyome”—the sequence of all or most of a person’s antibody genes. Haynes hopes that this will make it possible “to figure out what might have driven [the antibody response] and therefore how you might drive it in a vaccine.”

Pascal Poignard, an adjunct professor at TSRI, also plans to study how bNAbs develop over time by selecting HIV-infected people with bNAbs and then sequencing antibody genes of memory B cells from previous samples at various time points starting early after infection. Mascola says the VRC is also planning a project to study how the bNAb response develops in HIV-infected people, although details still need to be worked out.

Haynes is also involved in studying how B cell immune responses are developing in rhesus macaques after immunizing them over long periods of time with the same or different Envs. One goal, he says, is to “determine if repetitive stimulation over the same length of time that it takes to get broad neutralizing antibodies in humans will induce those types of antibodies in nonhuman primates.”

A daunting challenge?

Although affinity maturation has received more attention from researchers recently, it’s still unclear exactly how much of an obstacle it represents for HIV vaccine design. For example, current data suggest that it seems to take several years for affinity-matured bNAbs to develop in HIV-infected people. If so, the question is whether it will take that long to induce such antibodies with a vaccine. Some researchers say that there are reasons to believe that it won’t.

One reason is that not all of the affinity maturation that is observed in bNAbs like VRC01 may be required for broad and potent neutralization of HIV. “If you came with a vaccine that was exquisite or an immunogen that was very well designed, it’s possible that you could induce the corresponding antibody with much less somatic mutation,” says Burton, adding that he has found that many affinity-matured residues in the b12 antibody can be mutated without affecting binding and neutralization of HIV.

This may also be true for VRC01. When Kwong and colleagues did germline reversions of the 12 affinity-matured amino acids that are found on the interface of VRC01 that directly interacts with gp120, they found that the resulting antibody could still bind gp120 pretty well. “I was surprised to see this,” says Kwong, adding that more research is needed to know exactly how much affinity maturation is required. “We don’t know if it’s just a couple of changes that are required or if extensive affinity maturation [of VRC01] is needed. We haven’t fully answered that question.”

According to Shapiro, it is possible that at least some of the affinity maturation in HIV-specific bNAbs is simply the result of chronic HIV infection, during which constant interaction with HIV causes B cells to continuously change. “Does it require this much affinity maturation to make an antibody that’s effective like this?” he asks. “Or is it simply something that happens because these people have been chronically infected for so many years once they get these antibodies?”

A recent study from the labs of Burton and Wilson suggests that it does not take much affinity maturation to induce important functional changes in the bNAb 2G12 (7). They found that surprisingly few of the over 50 affinity maturation changes are required for 2G12 to assume its unique shape of interlocked or exchanged VH domains, which enables it to bind a cluster of mannose residues on HIV Env. This “suggests that the evolution of a domain exchanged antibody response in vivo may be more readily achieved than considered to date,” the authors concluded. These results, Wilson says, support the view that many of the affinity mutations that are acquired on the pathway to final affinity-matured bNAbs may have little or no residual functional significance in the final end product, as they neither enhance nor destroy antigen binding.

And while currently available data suggest that it appears to take years for these bNAbs to develop in HIV-infected individuals, a closer look might reveal that it doesn’t have to take that long, Phogat says. A recent study in macaques infected with a simian immunodeficiency virus (SIV)/HIV hybrid known as SHIV showed induction of highly potent neutralizing antibodies to an epitope on the native trimeric Env spike after just nine months (8). This suggests, Phogat says, that a SHIV with the PG9/16 epitope, which is also specific for the native trimeric Env spike, should elicit PG-like antibodies in macaques rather quickly as well, and that it shouldn’t take that long to induce PG-like antibodies in humans. “It did not take a very long time in macaques,” Phogat says. “Why should it in humans?”

In addition, Phogat says, it may not take uninfected people, who would be the recipients of a preventive vaccine, as long to develop affinity-matured antibodies as HIV-infected people. The reason is that HIV infection probably inhibits or slows down the process of affinity maturation because B cells need the help of CD4+ T cells, the primary target of HIV, to undergo affinity maturation. “We are talking about a screwed up immune system [in HIV infected people],” says Phogat. “They don’t have proper T [cell] help because those [are the] cells HIV feeds [on].” That’s possible, says Burton, but he cautions that not enough is known yet. It’s also possible that “some feature of [HIV] infection favors the induction of [broadly neutralizing] antibodies in some individuals,” he says. “That’s why it’s important to understand the origins of broadly neutralizing antibodies in natural infection.”

While there is still much to understand, the path forward is pretty clear, Shapiro says. “At the VRC it’s a time of really extraordinary excitement,” he says. “We have what looks to us, at least until we are proven wrong, like clear targets and clear ways to approach them. Right now the path looks pretty well defined.”

1.Science 329, 811, 2010 
2.Science 329, 856, 2010
3.J. Virol. 84, 8098, 2010
4. Biochem. Biophys. Res. Commun. 390, 404, 2009
5. Viruses 1, 802, 2009
6. J. Virol. 2010, doi:10.1128/JVI.01110-10
7. J. Virol. 2010, doi:10.1128/JVI.01111-10
8. J. Virol. 84, 3443, 2010