Barcelona, in Essence

Vol. 11, No. 06 - November 2013

Our take on the key talks and takeaways from AIDS Vaccine 2013

By Unmesh Kher

Back in the mid-1980s, when scientists first tried to design AIDS vaccines, the task seemed pretty straightforward. They took a piece of HIV’s envelope—a toadstool-like protein complex that traverses the outer membrane of the virus—injected it into folks and waited to see if the immune system would do the rest. The strategy itself was rational, and far from unprecedented: viral vaccines work mainly by inducing antibodies against such surface molecules, known as antigens. Yet the result, after quite a lot of waiting, was a serious letdown. And the blame lay not so much within the immune system (at least, not completely) as with the HIV antigens used.

HIV, it turns out, is far too variable and cunning to succumb to such old-school strategies. But over the past couple of decades, researchers have learned a lot more about how the virus befuddles the immune response, and how that capability might be disrupted. And if the proceedings at AIDS Vaccine 2013 in Barcelona in early October were any indication, they are beginning to figure out how to mine that knowledge to make potentially better antigens, deliver them more effectively and perhaps—someday, in the not too distant future—make reasonably effective HIV vaccines.

A useful trimer mimic?

The work that made the biggest splash in Barcelona was from a team of researchers led by Ian Wilson and Andrew Ward of The Scripps Research Institute in La Jolla, California; John Moore of Cornell University in New York; and Rogier Sanders of Cornell and the Academic Medical Center in Amsterdam. The team described in a series of talks how they had constructed a promising mimic of the external part of the envelope protein—also known as the trimer—and determined its structure using two distinct techniques. Most importantly, they revealed that the mimic appears to be nearly identical in structure to the external portion of the natural trimer, at least to the extent that this structure is known.

This matters because the envelope is the only target on HIV available to antibodies, and so of great significance to HIV vaccine design. It is, however, a highly mutable protein that has also evolved a number of other tricks to evade the antibody response. To make matters worse, it is terribly finicky and falls apart very quickly, confounding efforts to both probe its functional structure and to use it as a vaccine antigen, which is known as an immunogen. The trimer made by John Moore and colleagues, on the other hand, has been engineered to assemble appropriately and remain stable in solution, permitting its structural analysis.

The trimer mimic, named BG505-SOSIP.664, binds to pretty much all subsets of known broadly neutralizing antibodies (bNAbs), so named because they target the envelope protein in a manner that disables most circulating varieties of HIV in laboratory studies. This has two major implications. First, it is further evidence that the trimer mimic is virtually identical to its natural counterpart, since some bNAbs only bind to appropriately structured trimers. Second, it opens the door to using it as an immunogen that might teach the body how to make bNAbs and so thwart HIV (see box, below).

For now, though, the mimic remains a work in progress for such uses. Still, its creation and structural analysis was largely received as a major achievement both at the conference and after its publication in Science.

The making and possible uses of a trimer

HIV uses a toadstool-like structure on its surface—known variously as the spike or envelope protein—to latch onto and invade T cells. This envelope is, in molecular terms, a heterotrimer. It is made up of three sets of two identical proteins, one known as gp41 that traverses the viral membrane (the toadstool’s stem), and another called gp120 (its bulbous top), which binds to a molecule named CD4 on T cells. Even when on the virus, the complex tends to fall apart, leaving clusters of gp41 stems on the surface of HIV that serve as decoys to the antibody response.

This is one major reason that researchers have not been able to study the structure of the envelope, or to use it as an immunogen. Nor have they been able to crystallize the intact, functional protein, a step needed to assess its structure at a practically relevant resolution. The researchers who made the BG505-SOSIP.664 trimer (seemain story) sought to work around both these problems.

To do so, they took a gene for the envelope from a founder virus—the one thought to have caused infection—isolated from an infant in Kenya. They then engineered the gene to improve the formation and stability of the trimer and lopped off the part of the stem that crosses the membrane to make sure the resulting proteins wouldn’t clump together in water.

Viewed under the electron microscope, it looks from above exactly like the naturally occurring protein. And pretty much every one of the mimics the researchers examined looks this way, attesting to its stability. Closer analysis of its structure using x-ray crystallography and a special kind of electron microscopy confirms this finding.

So do the biochemical studies of the trimer. Not only does the mimic bind broadly neutralizing antibodies (bNAbs) that require an exactly right structure to the trimer, it also fails to bind all those non-neutralizing antibodies that target decoys made by malformed trimers.

All this matters because it suggests that the mimic might be useful as an immunogen to elicit bNAbs. But that will apparently require a little more work. Initial studies in which the mimic was used to elicit immune responses in rabbits resulted in the production of antibodies capable of neutralizing only the virus from which the mimic has been derived. It could not neutralize a random panel of HIV strains that are relatively difficult to block. So the current mimic doesn’t quite fit the bill for an HIV vaccine candidate—at least not yet. But studies continue on this front even with the current mimic.

So how can researchers use this mimic, or another like it, to make a bNAb-inducing immunogen? The research team contemplates a number of strategies to that end. Researchers might, for example, trim or chemically cloak elements of the trimer known to be highly variable. These elicit ineffective antibody responses as well, so their removal might focus the immune system on more effective targets.

They are also planning studies in animals now to use the trimer mimic in combination with another immunogen that has been devised to mimic the specific molecular shape targeted by a potent bNAb. Other strategies along these lines would be to use a series of evolving trimer mimics as boosts following a priming vaccination, or a cocktail of similar mimics derived from other HIV viruses. Those would, however, have to be constructed, and there’s no guarantee other trimers would be as amenable to useful manipulation as the one obtained from the founder virus that gave us BG505-SOSIP.

If any one of these strategies does elicit anything like a neutralizing antibody response in animal and laboratory studies, it would be considered a significant breakthrough. But it bears mentioning that BG505-SOSIP represents just the first step of a long journey. Any immunogen made on the basis of this mimic would still have to run the gauntlet of preclinical proof and clinical trial that are required for the licensing of a vaccine. —UK


Coaxing out the bNAbs

Yet simply making the right immunogen isn’t likely to suffice, since bNAbs take several years to mature and any vaccine worth its salt will have to induce them far faster than that. Penny Moore of the National Institute of Communicable Diseases in South Africa described in what she called “excruciating detail” how a game of cat and mouse between HIV and antibodies in a single HIV-infected person fueled the evolution of bNAbs.

The talk was indeed detailed, but fascinating nonetheless. Moore’s analysis showed how the antibody response in her one chosen subject came in three distinct waves. Each of the first two waves applied pressure on the virus, which mutated to evade the attack. But in doing so, it inadvertently exposed a weakness, which was then targeted by the succeeding wave of antibodies. In the end, three successive waves of antibodies generated a highly potent and broadly neutralizing antibody to HIV.

Moore’s detailed description of the viral and antibody mutations that drove this game to its conclusion add to a growing body of data that might permit the precise design of antigens and vaccination regimens to hasten the maturation of bNAbs. Indeed, Barton Haynes of Duke University presented early data on his development of such regimens.

Interestingly, one additional bNAb-related talk that caused some excitement at the conference was not about the role of these antibodies in vaccine responses, but their uses in therapy and possibly as a tool for cure research (seeGlobal News, this issue).

The Thai trial keeps on giving

Another pair of talks, however, reminded attendees that for all the fuss about bNAbs and their generation, the sole vaccine regimen to have provided any measure of protection from HIV induced not one detectable bNAb. A talk by Merlin Robb of the US Military HIV Research Program summed up what researchers have learned about how protection might have been induced in this trial, RV144, which was completed in 2009 in Thailand and showed a 31% efficacy overall after three and a half years (though protection appears to have been nearly twice that high in the first year following vaccination).

To recap: the analysis of samples collected in that trial revealed that certain classes of antibodies, known as IgGs, that target highly variable regions of the trimer were essential to the reduction in risk among vaccine recipients. Robb also said that more recent evidence suggests similar antibodies to a third variable region of the envelope might have played a role. These antibodies are not “neutralizing,” which is to say they cannot block HIV entry into its target cell.

 Previous research has separately confirmed these findings: mutations HIV had to obtain to escape protection correspond perfectly with the spot on the variable region that appears to be targeted by such antibodies. At the same time, the presence of IgA antibodies against the envelope correlated to increased risk for infection in vaccine recipients.

This makes sense because non-neutralizing antibodies work by directing other components of the immune response to destroy infected cells. IgGs, for example, bind immune cells known as natural killer cells and promote the destruction of infected cells by these “NK cells.” IgAs do not induce this sort of antibody-dependent cellular cytotoxicity (ADCC).  But what they can do, if they cluster around the envelope protein, is keep effective IgGs from doing their work.

Antibodies, sugars, and subtypes

Other studies led by the laboratories of Georgia Tomaras of the Duke Human Vaccine Institute and Galit Alter of the Ragon Institute in Boston, Massachusetts, suggest that a subclass of IgGs called IgG3s appear to have been essential to the protection afforded by the RV144 regimen. A comparative analysis reveals that another trial, VAX003, which failed to find any sign of protection, primarily elicited IgG1, 2, and 4, while only RV144 elicited higher levels of IgG3.

It also turns out that the IgG3s elicited by RV144 are more functional than those elicited by VAX003. That is, they induce ADCC better than the IgG3s elicited in VAX003, and greater functionality in this subclass of IgGs correlated with reduced risk in RV144. Most of the IgG3s induced by VAX003 recruit one or two types of responses from other players of the immune system. The IgG3s from RV144, meanwhile, recruit three or more such responses (with a mean 4.6 “effector functions” induced).

Alison Mahan described a fascinating series of analyses conducted on antibodies collected from the two trials in Alter’s lab, where she is a graduate student. She and her colleagues found that the antibodies elicited by VAX003 tended to be modified by glycosylation—the addition of complex sugar chains—in a manner that resulted in drawing immune responses that promote inflammation. Those induced by RV144, on the other hand, were less inflammatory and more tuned to inducing immune responses that lead to the elimination of virally infected cells.

Mahan and her colleagues also compared samples from the Step trial, which tested a vector based on Ad5, and from a trial that tested an Ad26 vector along the same lines—measuring 67 different immunologic parameters. Intriguingly, responses from the Ad5 vector resembled those obtained in VAX003, while the Ad26 responses clustered with those detected in RV144. But perhaps the most interesting takeaway from this talk was the suggestion that different vaccination strategies can induce different degrees of inflammatory and functional immune responses through subtle changes in the kinds of antibodies induced.

Ad5 has now failed dismally in two large-scale efficacy trials, and will likely never again be touched as a vehicle for the delivery of HIV immunogens. Indeed, results from an ongoing analysis of Ad5 clinical trial data by Peter Gilbert of the Fred Hutchison Cancer Research Center suggests that the higher risk of HIV infection observed in vaccine recipients in the Step trial was real. But all agreement stops there. Even the question of whether that effect was linked to the vaccine candidate itself is being debated, and it remains unclear how the candidate or the regimen might have induced that effect.

Targeting infected cells

Effective HIV vaccines will in all likelihood also have to engage the T-cell response, which is essential to eliminating cells infected by HIV. Some viruses are, after all, likely to evade any antibody elicited by vaccination. Devising such vaccines has proven to be quite a challenge as well, since T-cell targets too are highly variable when obtained from HIV. So researchers have lately tried a few strategies to devise immunogens to focus T-cell responses on those parts of the virus that are not subject to much change.

One such strategy presents the immune system with an immunogen encoding just the conserved sequences of HIV proteins. Thomas Hanke of the Jenner Institute in Oxford presented data from the first clinical evaluation of vaccine candidates carrying such an immunogen, which encode conserved regions of the HIV envelope protein. His data showed both a breadth and a potency of cytotoxic T-cell responses that exceeded those observed previously with vaccines targeting the whole envelope protein.

The field has also witnessed exciting advances in the delivery of immunogens, especially the development of vectors that retain the ability to multiply following delivery—potentially eliciting more sustained and effective immune responses. Louis Picker of the Oregon Health & Science University gave a well-attended talk on the preclinical results his laboratory has obtained with a replicating vector derived from cytomegalovirus (CMV).

Picker’s team has been developing this vector for years, and has moved from success to success. They have found that after they vaccinated monkeys with their CMV vector, all monkeys got infected after repeated challenge with SIV, the simian version of HIV. But of those monkeys who received the vaccine, half appear to have durably suppressed—and perhaps eliminated—the infection, and done so using novel immunological mechanisms. Those mechanisms appear to take effect even when non-replicating versions of the vector are used. Picker said work has begun to develop a cytomegalovirus HIV vaccine vector for evaluation in humans.

Ronald Desrosiers of Harvard University’s New England Primate Research Center has been working on an analogous strategy for vaccination. He and his colleagues have, however, built their SIV vaccine vector from the rhesus rhadinovirus, which also causes a chronic infection in monkeys. Desrosiers reported that three of five monkeys given this vaccine and subsequently infected with SIV suppressed their SIV infection to almost undetectable levels. His analysis also suggests that antibodies might have played a role in the observed protection.

Most of these vectors, and the immunogens being designed to elicit effective neutralizing antibodies, are in the earliest stages of development. Many—perhaps most—will stumble along the arduous path to licensure. But if any single thing was clear in Barcelona, it was that the odds are improving in favor of at least one of these strategies resulting in an effective HIV vaccine.