More than 30 years after scientists identified HIV as the cause of AIDS, we still have not managed to devise an effective vaccine against the virus. An array of drugs can usually keep the infection under control for decades, but a vaccine that prevents infection in the first place would be the best weapon—particularly in the developing world, where the costs of drugs and other factors can put effective therapy beyond the reach of many. Without treatment, HIV infection usually runs silently and progresses to severe immunodeficiency (AIDS) and death over the course of several years.

The long delay in developing a vaccine is not for want of trying or even a lack of funding. The problem is that HIV is like no other virus scientists have ever tackled. For any antiviral vaccine to work, it has to arouse the immune system to attack and destroy the virus of concern before it can invade cells and spread through the body. But HIV has evolved many defenses against the human immune system. Devilishly, it kills or impairs critical immune cells that are supposed to coordinate the body's response against it. And it is an unparalleled master of disguise that has so far thwarted the efforts of vaccine makers to teach the body how to quickly recognize and block its many variants from infecting humans.

The three of us and our colleagues have recently managed, after nearly two decades of trying, to create a synthetic protein that should help overcome the difficulties vaccine makers have faced in the past. We have shown that this molecule can elicit a strong response to HIV in animals. To serve as the basis for a vaccine in people, it will need to be modified to become more powerful and able to prevent infection by a much broader range of viral strains. That work will take time. But our laboratory and many others are already pursuing the remaining challenges and are optimistic that we are on the right course at last.

The Vision

The protein we constructed mimics the viral protein called envelope, or Env, more completely than has ever been possible. Env rises from the surface of HIV like a spike and enables the virus to enter immune system cells known as CD4+ T lymphocytes. These T cells normally communicate with other parts of the immune system through various proteins—including two called CD4 and CCR5—that dot their outer surface like signaling towers on a fortress wall. As HIV attempts to enter the immune cells, one of its envelope proteins first latches onto the CD4 protein, which allows it to then bind to CCR5 as well. Next the envelope protein twists and rearranges itself so that the outer membranes of the virus and the immune cell fuse together. As the membranes fuse, the virus releases its genes into the cell, which produces billions of copies of the virus; these viral particles in turn break out of the cell and spread to other cells, where the infective process repeats.

Researchers have long dreamed of preventing HIV infection by blocking the envelope protein's maneuvers. The most logical approach would be to “teach” the body's immune system to produce molecules called antibodies that would specifically recognize and adhere to the envelope protein on HIV. In theory, such antibodies would have two desirable effects. They would form a barrier that would prevent HIV from attaching to CD4 and CCR5 and thereby entering CD4+ T cells, and they would ensure the virus's destruction or clearance by different parts of the immune system. Much the same approach works well for vaccines against other viruses, such as hepatitis B: a protein from the pathogen's surface is produced in the lab through genetic engineering; when injected into a person, such proteins cannot themselves cause disease (because the rest of the virus is absent), but they can induce the immune system to raise antibodies that will home in on and destroy any invading virus that displays the same or similar proteins.

Unfortunately, HIV thwarts the standard approach to developing vaccines because its envelope proteins have a nasty habit of falling apart as soon as they are separated from an intact virus. These pieces include the gp120 subunit (the part of the envelope protein that attaches to CD4) and the gp41 subunit, which anchors the envelope protein in the viral membrane and later facilitates the fusion of the viral and immune cell membranes.

Now you might think that the envelope protein's tendency to fall apart would not be too much of a problem. After all, the virus cannot infect a cell without gp120 attaching itself to the CD4 signaling protein, and the immune system can and does make antibodies against individual gp120 molecules. Indeed, for years researchers tried without success to make a vaccine using gp120 subunits (and some are still trying). It turns out that antibodies made against single gp120 proteins do not trigger a strong immune response against the virus that infects people. Studies of whole Env proteins, in contrast, suggest that antibodies against them are much more effective at targeting HIV for destruction.

Finally, in 1998, one of us (Moore) decided that producing a successful vaccine would probably require abandoning the gp120 route and focusing on making a vaccine based on the full envelope protein. Creating such a vaccine would be hard for many reasons, not the least of which is that each envelope protein is complex: it is a trimer, consisting of three copies of gp120 and gp41 components. Another of us (Sanders) joined the effort a short time later, followed, soon after, by our other co-author (Wilson).

Multiple Challenges

To prevent hiv infections, any vaccine—including one based on our research—will have to meet many challenges. For one thing, it will have to spur the immune system to produce particular kinds of antibodies. The most effective antibodies are those that both recognize an intact virus (in the case of HIV, specifically by homing in on the envelope protein) and that bind to it in such a way that they prevent that virus from entering a cell. Researchers call these crucial defense molecules neutralizing antibodies because they neutralize the infectivity of the virus.

To prevent infection around the world, however, we cannot raise just any old neutralizing antibodies; we need ones that are “broadly active”: able to recognize many different variants of envelope proteins and to stop them from using CD4 and CCR5 to enter immune cells. The ideal neutralizing antibodies would presumably home to parts of the envelope protein that do not differ much among viral strains. Generating broadly active neutralizing antibodies against several different parts of Env might, if possible, be an even better strategy.

Click or tap to enlarge

Credit: FALCONIERI VISUALS (main image); JEN CHRISTIANSEN (vaccine schematic)

Researchers also want the antibodies raised by a vaccine to react to the envelope protein even though it is shrouded in an extraordinarily thick blanket of sugars that essentially mask HIV from the immune system. In the course of untreated HIV infections, the immune system manages to mount a response (including making neutralizing antibodies) that limits viral replication for years, but that response is too slow and weak to eradicate HIV fully. And it can take months or years for the body to figure out how to make neutralizing antibodies that bypass or recognize HIV's camouflaging sugars. In the meantime, the virus destroys more and more immune cells that the body can ill afford to lose.

Trial and Error

Making a trimer that met two of our key criteria—that it would not disintegrate and would trigger neutralizing antibodies against the relevant strains of HIV—took our team multiple attempts (all funded by the National Institutes of Health) and the better part of two decades to accomplish.

We began by isolating Env genes from a particular strain of HIV and using them to synthesize Env proteins. To do this, we eliminated the part that normally anchors the envelope protein to the HIV surface. Our first attempt resulted in a protein that still fell apart. Several other groups of scientists tried to get around the problem by genetically engineering the envelope protein in a way that practically guaranteed the components would stay together. Sure enough, the resulting molecules did not completely fall apart, but their structure was so different from that of the envelope proteins that are found on HIV that they proved incapable of eliciting the necessary antibodies.

It was time to look for clues in other viruses that had some structural similarities to HIV. We realized that the surface proteins on some of them had a kind of chemical strut that linked their equivalent of the gp120 and gp41 proteins with a pair of sulfur atoms. We started looking for places where we could add such sulfur struts to the HIV envelope proteins that we were synthesizing and used what was already known about how the gp120 and gp41 components of the HIV trimer fitted together to make some educated guesses about where to place the sulfur struts to link everything together more strongly. By trial and error, we found the right locations, but the resulting trimer still crumbled—just in a different way than our previous attempts.

We then made a minor tweak to the gp41 component. All proteins are made up of various amino acids, whose electrical charges, among other things, cause the proteins to adopt distinctive shapes. Sanders decided to force the gp41 portion of our artificial trimer to adopt slightly different shapes by making particular amino acid substitutions. Eventually he found one alternative composition (replacing an isoleucine with a proline) that allowed the trimer to stay together. We gave our engineered protein the name “SOSIP” in honor of the two gambits that had made it possible: the first three letters (SOS) refer to the sulfur struts, and the last two (IP) indicate the key tweak we made in the gp41 protein.

And there things stood for many years. Our trimers were stable, but when we put them in a liquid, as would be needed for a vaccine, they just clumped together, making them useless.

Two critical developments finally enabled us to make new progress. First, Andrew Ward of the Scripps Research Institute, then an assistant professor, joined the effort to determine the physical structure of the Env trimer. Ward made highly detailed photographs of our SOSIP trimers with an electron microscope and showed that they were attracting fatty globules, or lipids, that basically made the trimers very sticky, causing them to congeal like chewing gum. And whereas some of our artificial trimers looked like viral envelope proteins, others had adopted very odd shapes indeed. Clearly, we were not consistently making the kinds of spike-mimicking trimers we were after.

Guided by the electron micrographs, we figured out a way to chop off a section at the end of our engineered trimers that was absorbing the meddlesome lipid molecules. We called these truncated trimers SOSIP.664 because of where we now cut them off: each third of the trimer consists of a long chain of amino acids, and we cut them off at the 664th amino acid in the chain. Looking at these slightly shorter trimers through the electron microscope, Ward saw that they all closely resembled the visible part of the spiky structures found on infectious HIV strains.

At this point, SOSIP.664 had the amino acid composition of the envelope protein from one variant of one strain of HIV, but we wanted to construct a trimer that was most likely to elicit production of neutralizing antibodies that had broad activity against many strains.

No one really knows, even now, exactly how to make a trimer that will induce broadly neutralizing antibodies in people. But our best chance of doing so is to make sure, at a minimum, that the trimer can be recognized—that is bound—in lab tests with a collection of broadly neutralizing antibodies that have been gathered from some people infected for many years with different strains of HIV. In other words, for existing broadly neutralizing antibodies to attach to a particular trimer at all, it must appear pretty similar—from a biochemical point of view—to naturally occurring Env proteins. And thus, injecting such a closely matched trimer into uninfected humans might well prompt the immune system to produce similarly powerful antibodies.

Because we could not predict which amino acid composition for the Env protein would give us the properties we wanted, we had no choice but to screen envelope proteins from about 100 different viral strains from patients around the world. We then made SOSIP proteins from all of them to find a variant that mimicked the spike under the electron microscope and could be bound in our lab tests by broadly neutralizing antibodies taken from people.

Eventually we found what we were looking for in samples collected from a six-week-old infant—given the code name BG505—who was born with HIV in Nairobi, Kenya. This particular viral strain was isolated by Julie Overbaugh of the Fred Hutchinson Cancer Research Center in Seattle and her colleagues at the University of Nairobi, and information about its genetic sequence—and thus the amino acid composition of its proteins—was passed to us for screening by the International AIDS Vaccine Initiative (IAVI).

The second development was the invention of a way to make a lot of this particular trimer, which we named BG505 SOSIP.664 (the BG505 trimer, for short), in as pure a form as possible. Among other things, this achievement allowed us to create crystals with the material through which we could shoot x-rays to determine its molecular structure. It also meant that we could make enough of it to test in animals and ultimately in people. Although lab tests of our trimers looked promising, we still needed to confirm the results in animals.

We injected the BG505 trimers into rabbits and monkeys and collected the antibodies against HIV that they made. When we added the antibodies to tissue cultures made up of human cells, we found that they did protect those cells against infection with the BG505 virus but not against other strains. Although the antibodies did not have the breadth of neutralizing activity that will in the end be needed, we had made a good start.

One of the next steps is to repeat these experiments in people. Much of our protein production research to date has been supported by the Bill & Melinda Gates Foundation and IAVI. We are also talking with IAVI and the nih about designing and funding an exploratory clinical trial, which should enroll about 50 volunteers. We will not develop a protective vaccine right off the bat from this first artificial trimer—at least in its current configuration. Although results from lab animals are reasonably predictive of what happens in people, they are not foolproof. Clinical trials in people will teach us about how the human immune system responds to our artificial trimers. That kind of information, along with data from Wilson's lab on how closely the trimers resemble naturally occurring envelope proteins, should help us redesign our proteins to develop a protective vaccine. We will have to tweak what we create, probably more than once. We will also harness recent developments in understanding how the human immune system makes broadly neutralizing antibodies to improve how we deliver current and new trimers to people.

In essence, we have created a working first-generation prototype that we can modify in different ways to determine which configuration will be most likely to produce the most effective antibodies. Our ultimate goal—manufacturing a vaccine that induces broadly neutralizing antibodies against the most common strains of HIV in people—is still far from assured. But the good results that we have achieved so far with our approach in animals and cell tests suggest that the problem is not unsolvable.

And now the research community has the SOSIP tool kit and the methods to help create the best proteins possible for use as a vaccine. Many different groups are currently making their own versions of these spike-mimetic trimers to test their various vaccine designs. The coming years should finally be productive ones for a field that has been battering away at this tough, tough problem for a long, long time.