Tapping the Sanguine Humor

Broadly neutralizing antibodies stole the show at Keystone

By Regina McEnery

A vaccine that elicits broadly neutralizing antibodies (bNAbs) to HIV has long seemed at least as elusive as the Holy Grail. But as more and more such antibodies are isolated from HIV-infected volunteers—revealing rare vulnerabilities on HIV—researchers are increasingly hopeful of coaxing that coveted humoral response.

Some 400 attendees at the Keystone Symposia, which paired the annual HIV Vaccines meeting with the Viral Immunity and Host Gene Influence meeting, got an update on the quest to make such vaccines. A number of talks at the conference, held March 21-26, illuminated how structural and computational biology are being applied to reverse-engineer immunogens that might induce bNAbs. Others described how next-generation DNA sequencing technologies are unraveling the genetic origins and evolution of those antibodies.

Researchers are also making significant progress toward solving the unliganded structure of the Envelope trimer, the lack of which has long hampered immunogen design (see A Slew of Science in Seattle, this issue). Yet the insights gleaned from such breakthroughs, and from other lines of inquiry into the ontogeny of the B cell response to HIV, have also exposed just how hard it will be to elicit appropriate antibodies. Indeed, Gary Nabel, director of the Vaccine Research Center (VRC) at the US National Institute of Allergy and Infectious Diseases (NIAID), was moved to quote Sir Winston Churchill while describing progress in that endeavor: “This is not the end, it is not even the beginning of the end. But it is perhaps the end of the beginning.”

If so, the start hasn’t been half bad. The field, as Nabel noted, now has access to more than 100 bNAbs against HIV, many of which have not yet been described in the literature. Structural, genetic and biochemical analyses of these exquisitely precise molecules are providing valuable clues to the structure-assisted design of vaccines.

The sole target for those antibodies, the HIV envelope trimer, is a wonder of evolutionary refinement. The few functional trimers scattered over HIV’s surface, for example, encode a number of immunodominant decoys for antibodies, most notably in the variable loop domains. These provoke a vigorous but generally useless antibody barrage that drowns out potentially neutralizing responses. Beyond that, the spike is covered with a thick fur of complex carbohydrates—often called the “glycan shield”—that largely resemble the complex sugars found on human cells, and are thus often ignored by the immune system. The bulky sugar chains also restrict antibody access to underlying peptide epitopes that might otherwise elicit neutralizing responses.

Nabel and his colleagues at the VRC have sought to pierce these defenses with their immunogens. They have covered up regions on the envelope that are highly immunogenic but essentially serve as decoys, and have removed glycans that hinder access to conserved epitopes. They have rearranged glycans on the surface of the protein, and altered some known immunodominant regions by modifying the variable loops, so as to bias the response in favor of neutralizing antibodies. Yet the responses induced thus far in mice, rabbits and nonhuman primates (NHPs) suggest they have a ways to go.

Not, however, for want of creativity. One immunogen developed at the VRC contains a re-engineered monomeric gp120 stripped of its inner domain, an excision intended to improve antibody responses to the CD4 binding site. Unfortunately, the excision destabilized the gp120 protein. So the VRC team secured the required structure by introducing disulfide bonds via directed mutagenesis. They then studded the surface of a Chikungunya virus-like particle with an array of these stabilized proteins—the immune system recognizes and responds better to large protein particles—and tested their immunogen in rabbits (see A Bangkok SurpriseIAVI Report, Sep.-Oct. 2011).

Nabel and his colleagues at the VRC have found that the scaffolded subunit immunogen, whether administered alone or in a prime-boost regimen, induces antibodies in rabbits that target the CD4 binding site. “The bad news,” said Nabel, “is that the antibodies are hitting mostly tier 1 viruses.” These are the most sensitive of HIV variants, and are not representative of those that circulate naturally.

One reason for the limited neutralization, said Nabel, could be that the addition of glycans to the resurfaced core of gp120—done to block immunodominant regions and refocus the response on the CD4 binding site—was not extensive enough. A number of antibodies that are not broadly neutralizing recognize regions of gp120 similar to those bound by b12 and b13. The target areas of partially masked immunogens might therefore have to be further constrained to elicit bNAbs like b12 and b13, which approach the CD4 binding site from very specific angles.

While much immunogen design has thus far focused on that highly conserved site, the discovery last year of a new family of bNAbs—some of them extremely potent—that mostly appear to be glycan-dependent has given researchers a new target for immunogen design (1). The antibody PGT128 is among the most potent in this family of so-called PGT antibodies, said Dennis Burton, a professor of immunology and microbial science and the director of IAVI’s Neutralizing Antibody Center at The Scripps Research Institute. It is capable of neutralizing 70% of a panel of 162 HIV pseudoviruses that represent all HIV subtypes currently in circulation. Burton noted that PGTs 135-137, if not as broadly neutralizing as PGT128, are still notably potent. Glycans at positions 332 and 392 on the trimer appear to be critical to their binding, said Burton. But a crystal structure of PGT135 bound to the gp120 core shows that it also makes contact with three other glycans.

“For the CD4 binding site, it’s clear you have to have quite a precise angle of approach,” said Burton. “This does not seem to apply to the glycan shield.” An antibody that approaches the CD4 binding site reaches out and “kisses” its target, he said. Glycan-dependent antibodies, on the other hand, only lightly embrace their sugar targets so as to make firmer contact with the peptide.

A base approach to neutralization

Some bNAbs—such as 4E10 and Z13—disable HIV by latching onto the very bottom of the HIV spike, a region of the trimer known to scientists as the membrane-proximal external region (MPER). Jinghe Huang, a researcher in the laboratory of Mark Connors, chief of the HIV-specific immunity section at NIAID, detailed the isolation of a new bNAb that binds the MPER stalk in a rather unusual way.

This new bNAb, 10E8, appears to target an epitope that overlaps the one bound by 4E10 and Z13. 10E8 also appears to bind two additional amino acids within MPER, both of which are highly conserved residues. MPER antibodies are known to bind phospholipid, are autoreactive, and bind poorly to envelope prior to CD4 docking. However, Huang noted that the 10E8 antibody doesn’t bind phosopholipid and does not appear to be autoreactive. It binds envelope on the surface of infected cells, and that binding is more resistant to washing of virions compared to other MPER antibodies, meaning that its epitope is accessible on the unliganded HIV Envelope trimer.

The neutralization potency and breadth of 10E8 was tested against a panel of 181 HIV pseudoviruses in parallel with five other bNAbs: 4E10, 2F5, VRC01, PG9 and PG16. At low concentrations, 10E8 could neutralize 72% of the tested viruses, similar to VRC01 (75%) but considerably more potent than 4E10 (37%) and 2F5 (16%). When infused into rhesus macaques, 10E8 provided 100% protection from mucosal challenge with simian human immunodeficiency virus (SHIV). “Discovery of 10E8 certainly puts gp41 back on the table in terms of vaccine targets,” said Huang.

A shocking strategy

HIV immunogens are hard enough to make, but even the most promising ones typically elicit sera that neutralize only the most sensitive viruses in established assays. Richard Wyatt, director of viral immunology in IAVI’s Neutralizing Antibody Center, explored whether a judiciously selected immunogen in the DNA prime—delivered via electroporation—and a boost using an adjuvanted, soluble trimer could improve outcomes. His team assessed three prime-boost regimens in rhesus macaques. Most of the primes consisted of a DNA plasmid encoding cell-surface anchored, cytoplasmic-tail deleted gp160JR-FL, the externally exposed part of a viral Envelope complex from a strain of HIV that is particularly difficult to neutralize. However, two of the 12 animals received DNA encoding the soluble gp140 foldon as a prime.

Wyatt’s team electroporated the DNA plasmid to boost its uptake, administering it at 0, 1 and 2 months. Some of the priming constructs were codon-optimized to improve protein expression. The boost was a soluble gp140JR-FL trimer with a C-terminal tag, or foldon, to improve stability. It was formulated with an adjuvant of carboyhydrate polymers derived from yeast and administered at months 5 and 7.

Wyatt and his team found that DNA priming after the third injection achieved reasonable levels of binding antibody titers. However, “the big question is not [whether we get] binding antibodies but, do we get neutralization?” said Wyatt.

To find out, he and his team conducted a standard neutralization assay using TZM-bl, a HeLa cell derivative engineered to be highly susceptible to HIV. Serum collected after the boost neutralized the usual tier 1 viruses, and DNA priming didn’t appear to contribute much to the scale of antibody neutralization. The DNA prime and a protein boost were required for tier 2 neutralization, and only three tier 2 viruses were neutralized. But Wyatt said studies using an A3R5 cell line—derived from CEM human lymphoblastoid cells—that is 10 times more sensitive than the TZM-bl assay, found substantial neutralization of the more resistant tier 2 viruses. In those assays, the DNA prime alone or with a protein boost elicited tier 2 neutralization. For most sera, all eight clade B and C tier 2 viruses were neutralized, but with decreasing potency on the more resistant isolates. (The A3R5 assay has been developed by David Montefiori, director of the Laboratory for AIDS Vaccine Research and Development at Duke University.)

Wyatt said his team now plans to repeat the experiment in a larger group of macaques to better quantify the results. “We also want to make the DNA prime better by increasing immunogenicity of the DNA candidate, and we want to generate trimers that preferentially present the neutralizing determinants, and not so much the non-neutralizing determinants.”


RV144: Take 2, Sort of...   

The central unanswered question raised by the RV144 trial is how on earth the prime-boost combination it tested actually worked. After all, when evaluated separately, neither showed even a hint of possible efficacy. Together, however, they provided 31.2% protection against HIV.

Researchers have been attacking this troubling conundrum from a variety of angles, not least by immunizing nonhuman primates with simian immunodeficiency virus (SIV) vaccine regimens that mimic those used in the RV144 trial. At Keystone, Poonam Pegu, a postdoc in the laboratory of Genoveffa Franchini, chief of the animal models and retroviral vaccine section at the US National Institutes of Health’s National Cancer Institute, presented results of a pilot study of 21 Indian rhesus macaques, half of which were vaccinated with an ALVAC-SIV/ SIVgp120 prime-boost combination that followed the same dosing schedule as RV144. The animals were subsequently challenged rectally with a low-dose of the highly pathogenic SIVmac251.

The animals were challenged with SIV once a week for five weeks. After five weeks, all of the controls had become infected, but in the vaccination arm of the study, about a third of the animals remained protected—a rate almost identical to that achieved in the RV144 trial. The viral load levels and CD4+ T-cell counts were no different in SIV+ animals, whether they had received the vaccine or not. The vaccinated animals protected from infection also had equivalent T-cell responses. But the antibodies they generated against gp120 bound with higher avidity than those of the unprotected animals.

Rectal challenges were stopped after five weeks because the unvaccinated animals all became infected by that time point. “We could have continued the challenge, but statistically we wouldn’t have gained any power,” said Franchini. She added that the primary purpose of this pilot study was to see whether the specific dilution of the challenge virus was able to transmit just a few SIV variants. A larger study that will include 75 animals is planned for August to try and identify a correlate of protection. —RM

Orchestrating the right response

Structure-based vaccine design is often described as a sort of molecular sculpting. But Nancy Haigwood, director of the Oregon National Primate Research Center at the Oregon Health and Science University, thinks it should instead resemble a symphony. “Viral vaccines that have worked in the past,” she explained, “have been symphonic in nature because they generate all types of immune responses.” Current approaches to HIV vaccine design are, in her view, more analogous to the work of musical soloists or, at best, chamber groups. “What we really need is a symphony: rich, iterative and sustained.”

In pursuit of that orchestral complexity, Haigwood’s lab is assessing the antibody responses elicited by immunogens that contain multiple Envelope variants from a viral quasispecies population—the genetic HIV variants that exist within a single HIV-infected individual. Her strategy derives from the observation that B cells are programmed during affinity maturation to develop bNAbs through exposure to a series of Envs encoded by variants of viral quasispecies. Two years ago, Haigwood and colleagues used clones of native, longitudinal sequential Env variants that arose over years of infection to elicit bNAbs. The study tested three different immunization strategies in rabbits and found that a sequential vaccine approach using variants from a viral quasispecies population best replicated features of the humoral immune response in the subject from which these Envs were cloned (2).

Haigwood and colleagues have since taken this approach one step further. They have exposed rabbits to a collection of Env variants representing the viral quasispecies of a single HIV-infected elite neutralizer who developed bNAbs relatively quickly—within two years of infection—with the breadth of the response further increasing over three years.

The researchers cloned fifty full-length functional Envelopes by single genome amplification at nine longitudinal timepoints, selecting 18 of them as vaccine candidates. The immunogens, which consisted of gp160 DNA and gp140 trimeric protein, were co-administered to rabbits using four different immunization strategies. Neutralizing antibodies were detected at six weeks, after only two immunizations, and increased after additional immunizations. Rabbits immunized with sequences that were representatives of Envelopes present at the time of first evidence of bNAb development achieved the highest neutralization potency of the four groups. This suggests that it might be possible to educate an immune system that is naive to HIV by methodically presenting it with immunogens derived from HIV quasispecies known to be present preceding the development of bNAbs.

The roots of the right response

Scientists have found that approximately 25% of HIV-infected individuals develop antibody responses that can neutralize a diverse array of primary viruses. But only a small percentage of this select group develops neutralizing responses that can be called both very broad and potent (3; 4). Yet, while such antibodies are more common than previously believed, much remains unknown about how they evolve in response to HIV infection.

It isn’t clear, for example, when exactly bNAbs first turn up in HIV-infected individuals. This question isn’t easily answered because most serum samples from which bNAbs are isolated came from individuals who had been living with the virus for years before they offered their blood up to science. But some rare cohorts of folks followed from the earliest stages of infection do exist, and two research groups separately parsed sera from such volunteers to trace the evolution of bNAbs.

A research effort led by the University of Amsterdam had previously established that approximately 33% of HIV-infected enrollees in the Amsterdam Cohort had cross-reactive neutralizing activity (CrNA) in sera approximately 35 months following sero-conversion. It has also found that evidence of neutralizing activity does not correlate with slower disease progression (5).

In their latest study, the Dutch team combed the sera of six HIV-infected men who have sex with men (MSM) previously enrolled in the Amsterdam Cohort Studies on HIV and AIDS, which began in 1984. The clade B HIV-infected men in this sub-analysis, who all joined the cohort in the late 1980s, had the most potent CrNA at 35 months after seroconversion. One of them was an elite neutralizer, which is to say his serum neutralized five pseudoviruses out of a panel of six pseudoviruses that have been statistically screened to predict neutralization breadth (6). For their study, researchers obtained serum samples at three monthly intervals in the first year after seroconversion and at multiple intervals thereafter, and tested the sera for neutralizing activity.

Because it is too difficult and expensive for researchers to identify specific neutralizing antibodies in individuals, researchers used CrNA as a measure of bNAb production, said Hanneke Schuitemaker, who led the study before joining the Dutch-based Crucell last fall as its senior vice president for viral vaccine discovery and early development. Schuitemaker presented recently published data at Keystone that showed five of the men had developed CrNA 20-35 months following seroconversion. CrNA activity peaked in the men at 35 months, revealing the long time-frame in which such antibodies mature. The elite neutralizer, by contrast, showed signs of CrNA just 10 months after seroconversion (7).

Researchers analyzed peripheral blood mononuclear cells (PBMCs) from the same six individuals with potent CrNA as well as PBMCs from three HIV-infected men who lacked CrNA to examine autologous clonal virus variants that appear over the course of infection. They did so because some quality of the transmitted founder virus that establishes infection is thought to influence an individual’s ability to mount a cross-reactive neutralizing response. In all of the men, CrNA coincided with neutralizing activity against autologous strains that were isolated less than 12 months after seroconversion, while viruses from later points had already escaped autologous neutralizing activity.

The researchers are, of course, interested in learning more about the elite neutralizer. Schuitemaker said epitope mapping continues, but that new evidence contradicts the indications of preliminary findings that pointed to an epitope that includes the N322 position on the spike, associated with a glycan epitope. In collaboration with Lynn Morris, chief specialist scientist and head of the AIDS unit at the National Institute for Communicable Diseases in Johannesburg, Schuitemaker’s team found that the serum neutralized a mutant virus that lacks N332. Further, mutations at the N332 position did not contribute to viral escape.

The rapid development of glycan-specific antibodies against position 332 of Env gp120 has previously been shown in monkeys (8). “If we could have confirmed this epitope specificity in our elite neutralizer, it would have provided proof of concept for rapid development of this antibody specificity in humans,” said Schuitemaker.

Help from the Relatives  

Attendees at Keystone also got an update on the first clinical study of a chimpanzee adenoviral vector—ChAdV-63—bearing a universal HIV-1 immunogen. The hope is that such vectors will provide an alternative to human Ad vectors, which as a class have had a bit of a rough ride as potential vehicles for HIV immunogens.

Most famously, analyses of the Phase IIb STEP trial indicate that the MRKAd5 viral vector vaccine candidate actually led to an increased risk of HIV acquisition in a subgroup of volunteers who were uncircumcised and had pre-existing adenovirus serotype 5 (Ad5) antibody immunity. Further, such pre-existing immunity dampens HIV-specific cellular immune responses to MRKAd5. As a consequence, a few research groups have been designing HIV vaccines built on vectors derived from human adenoviruses that are less common worldwide, such as Ad26 and Ad35 (see Adenovirus Vectors: Promise and Possible PitfallsIAVI Report, Jan.-Feb. 2012).

Others are looking at chimp Ad vectors as alternatives. A recent study named ChAdV-63 as being one of several such vectors that have an “immunological potency equivalent” to human Ad5 (see 9). Tomáš Hanke, professor of immunology at the University of Oxford, presented immunogenicity results at Keystone of a pair of prime-boost vaccine regimens evaluating combinations of a universal immunogen, delivered in a modified vaccinia virus Ankara (MVA) vector, a DNA plasmid and a non-replicating ChAdV-63 vector. The HIVconsv universal immunogen is a chimeric protein consisting of the 14 most conserved regions from clades A-D, with no more than a 6% variation between clades.

The ongoing trial enrolled 32 volunteers from the UK. In the safety arm, two volunteers received one shot of the low-dose ChAdV63.HIVconsv vaccine candidate. In the second arm, eight individuals were given the ChAdV63.HIVconsv candidate followed by the MVA.HIVconsv candidate at week 8. In the third arm, eight volunteers received the DNA.HIVconsv at weeks 0, 4 and 8, followed by ChAdV63.HIVconsv at week 12 and MVA.HIVconsv at week 20. And in the fourth arm, eight volunteers received the DNA.HIVconsv vaccine at weeks 0, 4 and 8 followed by MVA.HIVconsv at week 12 and ChAdV63.HIVconsv at week 16. In each of the last three arms, two additional volunteers received a placebo vaccine.

Preliminary data show that the heterologous prime-boost regimens were well-tolerated and highly immunogenic. Vaccine-induced T-cell frequencies among volunteers who received the ChAdV-63/MVA combination reached a median of 5150 SFU/10(6) PMBCs—with a range of 1,475 to 16,495—as measured by IFN-y Elispot. That response was specific for epitopes that are dominant when induced by the conserved vaccine but subdominant during natural infection, when they are overcome by responses to variable regions. The responses were composed of both CD4+ and CD8+ subtypes. DNA prime pushed these frequencies to a median of 7023 SFU/10(6) PBMCs. By comparison, vaccine-induced T-cell frequencies in the STEP trial ranged from 136-686 SFU/106. —RM

Out of Africa

A cohort of HIV-infected heterosexual men and women from Kenya, Rwanda, Zambia, South Africa and Uganda is providing its own clues on bNAb development. It derives from IAVI’s Protocol C cohort, a longitudinal study of more than 600 donors that has been evaluating immunological and virological markers from the earliest stages of infection.

Elise Landais, a research associate at the IAVI Neutralizing Antibody Center at The Scripps Research Institute, presented neutralization data at Keystone on 328 of the 611 enrollees in Protocol C. At year two, only two of the individuals in this subgroup had developed broadly neutralizing activity, as defined above. By years three and four, the percentage of enrollees who possessed such activity had increased to 9% and 19%, respectively.

It appears, Landais said, that the development of cross-clade neutralizing activity is a gradual process. By 24-30 months, over 20% of enrollees were already able to neutralize three to four of the pseudoviruses—a capability that she considered indicative of moderate broadly neutralizing activity. “We know from previous studies that you won’t find bNAbs early,” said Landais, “so the point of this longitudinal study was to find out exactly when they develop. It was interesting to us that there seemed to be this intermediate step of moderate breadth. Maybe there is some kind of learning process for the antibody, that it evolves later on to neutralize other strains.”

Once achieved, neutralization breadth appeared to be relatively stable over time, said Landais, though the potency of the neutralization was relatively low. “We are now wondering if breadth develops first and potency later,” said Landais. “This is the type of question we want to answer.”

The study also found that serum antibodies in six of 21 top neutralizers were sensitive to a mutation at the N332 glycan at the base of the V2 loop, one of the sites commonly targeted by broadly neutralizing antibodies. Landais and her colleagues now plan to sort antibodies from three or four of the most interesting samples collected from this cohort to see if they are similar to PGT123 or PGT128—a bNAb that engages glycans attached to N332 or N301. Some individuals also appear to possess PG9-like neutralizing activity in their serum; PG9 targets an epitope exactly where the V1 and V2 loops connect. “We are going to try and isolate antibodies of known specificities and retrace the history of their development, which is essential in order to make a vaccine that mimics the virus,” she said.

1. Nature 477, 466, 2011
2. J. Virol. 85, 5262, 2011
3. J. Virol. 83, 188, 2009
4. J. Virol. 83, 7337, 2009
5. J. Infect. Dis. 201, 1045, 2010
6. J. Virol. 83, 7337, 2009
7. J. Virol. 86, 2045, 2012
8. Proc. Natl. Acad. Sci. 108, 20125, 2011
9. Sci. Transl. Med. 4, 115ra2, 2012