麻豆传媒

THE press conference was packed out. Hundreds of journalists jostled for position as America鈥檚 most senior health politician and several leading scientists announced a major breakthrough in AIDS research. There would be a vaccine ready for testing in two years, they said. And the vaccine would be ready for worldwide use before a further year was out.

The event took place in 1984. The press conference was held in Washington DC to publicise the discovery that the strange new disease killing gay men in California was caused by a virus, HIV. The journalists demanded to know when this knowledge would translate into a way to stop further deaths. The answer they got was three years.

Nineteen years later, 22 million people have died from AIDS, and there are 42 million infected with HIV 鈥� with the vast majority living in developing countries. And still we have no vaccine.

These days most vaccine researchers fight shy of making specific predictions. But despite numerous setbacks, there is now a mood of cautious optimism. HIV is arguably the most diabolical virus that humans have ever encountered, but it does have tiny chinks in its amour. And recent research pinpointing their location suggests we may be finally ready to take on the virus and win.

Like all viruses, HIV reproduces by hijacking the molecular machinery of the cells it invades. But HIV deviously targets the very cells that help fight off infections 鈥� CD4 cells, named after a receptor molecule on their surface. The virus spreads from cell to cell, turning them into virus factories then killing them. As patients鈥� CD4 cell numbers fall, they become less able to fight other infections. That is why AIDS patients often succumb to humble diseases such as pneumonia and tuberculosis.

With hindsight, perhaps we shouldn鈥檛 be too hard on those who made overly optimistic predictions in the early days of HIV research. Immunologists had had a period of outstanding success since the 1950s, creating vaccines against many viral diseases such as polio, measles and mumps. When the causative agent of AIDS was identified as HIV, people assumed it wouldn鈥檛 take too long to create a vaccine for this virus, too.

But the vaccines that were so effective against other diseases used virus that was either weakened or inactivated, to stimulate an immune response without causing disease. A few HIV researchers experimented with this approach in the early days, but soon abandoned it. The prevailing view was that it was too risky to inject people with a fatal virus like HIV, even in an attenuated state. A weakened virus might mutate and regain its strength. Inactivated-virus vaccines could prove lethal too, if just one virus in the batch were incompletely destroyed.

HIV researchers soon began to explore a more sophisticated approach: rather than injecting whole virus, they would inject just its surface molecules, or parts of them. The surface of HIV is coated with mushroom-shaped protein clusters. Each cluster is made up of six subunits 鈥� the mushroom鈥檚 stalk consists of three glycoproteins of a type called gp41, and the cap is formed by three larger ones called gp120. The mushroom鈥檚 function is to gain entry into host CD4 cells. First, gp120 binds to a CD4 molecule on the cell鈥檚 surface. Then it binds to a second surface molecule, called a chemokine receptor because its natural function is to bind to intercellular messengers called chemokines. Only then can the viral and cell membranes fuse and the viral genes enter the cell.

Scientists quite reasonably assumed that antibodies to gp120 would bind to it and block its attachment to CD4, thus stopping viruses from entering cells. Several firms, including SmithKline Beecham, Chiron and Genentech, began developing vaccines made from gp120, or fragments of it. Tests on monkeys showed the vaccines triggered antibodies that seemed to protect against infection. At the turn of the 1990s there was a mood of optimism among AIDS researchers. It seemed a vaccine was only a few years away.

Then everything started to unravel. The first studies had exposed monkeys to standard lab strains of virus, grown in dishes of immune cells. But when researchers exposed the animals to 鈥渨ild鈥� virus taken from other infected monkeys or humans, their antibodies failed to defeat the virus. It became apparent that lab strains of HIV are unusually sensitive to antibodies. Out in the real world the virus has much stronger defences.

We now know that wild HIV has evolved a formidable array of defences against the antibody response. First, gp120 and gp41 are coated with sugars to an unprecedented extent 鈥� no other protein is so heavily glycosylated. Antibodies cannot bind very well to sugars on viral proteins, so the sugars act like a force field keeping the antibodies at bay from the protein underneath.

A second problem was that scientists had been vaccinating animals with single subunits of gp120. Researchers now know that these have a slightly different shape to those that nestle in the mushroom clusters seen on the live virus. The difference is enough that antibodies generated by single subunits fail to recognise the clusters.

If those aren鈥檛 challenges enough, the gp120 protein also has an extremely high mutation rate in its most exposed areas, which are the parts most likely to trigger antibodies. People with HIV naturally produce antibodies to the virus, but gp120鈥檚 constant shape-shifting in these hypermutable sites 鈥� termed 鈥渧ariable loops鈥� 鈥� presents the immune system with a moving target.

There is one part of gp120 that is much less free to mutate 鈥� the site that binds to CD4. Its shape is crucial in ensuring that the molecules slot together like a lock and key, and mutations would disrupt that fit. But HIV cunningly protects this Achilles鈥� heel. The three gp120 units sit together so that the receptor binding site is shielded by the overhanging variable loops.

The chemokine-receptor binding site is protected in a different way. It is created near to the CD4 binding site only after gp120 has attached to CD4, leaving antibodies very little time and physical space in which to access the site. It may even be that it is completely inaccessible to whole antibodies, although fragments may be able to squeeze in.

These discoveries plunged vaccine researchers into intense pessimism. A biotech firm called VaxGen, of Brisbane, California, stubbornly continued with the single gp120 subunits developed by Genentech, and its first large-scale trial results are due by the end of March. But most researchers gave up on this approach in the early 1990s.

Around this time, a different strategy was starting to take shape. The human immune system is not just about antibodies, there is another branch too 鈥� the T cells. These kill other human cells 鈥� ones that are virus-infected or cancerous 鈥� by latching onto them and unleashing a chemical attack. CD4 cells belong to the T cell clan, but there are other kinds that cannot be infected by HIV, including a type called cytotoxic T lymphocytes (CTLs). As antibody-based vaccines looked increasingly flawed, for a while scientists began to wonder if the answer lay in CTLs.

The shift was fuelled by studies of rare individuals who never caught HIV despite repeated exposure, or who didn鈥檛 develop AIDS though they had been infected with HIV for many years. Researchers had found that CTLs might be keeping the virus at bay in some of these people (see 鈥淏eating the odds鈥�). Could this knowledge be exploited to make a vaccine?

A vaccine that worked through CTLs wouldn鈥檛 prevent infection 鈥� CTLs are only activated once the virus is inside cells 鈥� but it might slow HIV鈥檚 initial rampage, by liquidating infected cells before they could pump out more virus. Perhaps this would give the immune system a head start to defeat the virus, or keep it subdued as a chronic infection.

Since the mid-1990s many biotech firms and academic researchers around the world have been working on CTL vaccines. Most consist of a dose of HIV genes, either alone or piggybacking on harmless viruses or bacteria. The genes are taken up by human cells, converted into HIV proteins and displayed on the cells鈥� surface. CTLs recognise the proteins as foreign and become primed to attack cells infected with HIV, or so the theory goes.

Several such vaccines are now in, or rapidly approaching, human trials. The results should start rolling in over the next few years, but many researchers no longer have the confidence they once had in this approach, partly because a few of the so-called 鈥渆xposed-uninfected鈥� people have eventually succumbed to HIV, despite their CTLs. People began to suspect that a CTL vaccine might not offer complete protection, but merely postpone the onset of AIDS, and might not even do that much in all recipients. In the absence of better products, such a vaccine would still be valuable, but CTLs are no longer seen as the saviour of vaccine research.

All is not lost, however. Despite the disappointing results with antibody vaccines, basic research in this field continued and the past few years have seen several advances that mean the original dream of an antibody vaccine might yet be realised.

The breakthrough came from a handful of cell lines grown in labs around the world. HIV researchers commonly culture antibody-producing B cells from infected people to use as a research tool. The vast majority of these cell lines are unremarkable, but a few turned out to have an unusual and unexpected property. They produced antibodies capable of blocking some 鈥渨ild鈥� strains of HIV. The next question was, would any of them be broadly effective. John Moore, at Cornell University at Ithaca, New York state, recalls: 鈥淰arious people had their own pet antibodies and they were saying 鈥楳y baby鈥檚 prettier than your baby鈥�. It wasn鈥檛 until they were all compared in independent tests that we saw whose was the prettiest and whose were the ugly sisters.鈥�

Five antibodies were found to be outstandingly effective against a wide range of strains. They were dubbed 鈥渘eutralising antibodies鈥�, to distinguish them from the ineffective antibodies that most B cells produce in response to HIV.

Vaccine researchers aren鈥檛 sure why a few rare B cells make broadly neutralising antibodies when the vast majority don鈥檛. The patients who donated the cells probably didn鈥檛 benefit: they would only have made the antibodies after they were infected, and antibodies are not very effective against virus-infected cells. The best guess is that they result from a rare chance event.

Whatever the reason, researchers quickly realised that the broadly neutralising antibodies were a potential gold mine. If they could identify the epitope, the part of gp120 that the antibodies recognised, it might be possible to make a protein fragment that mimicked that site. This approach is known as reverse vaccinology. The engineered proteins could be used as vaccines to prompt B cells into churning out similar neutralising antibodies, which would hopefully prevent infection.

Perhaps the most promising antibody was the one found by Dennis Burton鈥檚 team at Scripps Research Institute at La Jolla in California. The molecular techniques available at the time showed that their antibody, which they labelled b12, recognised an epitope on gp120 that overlapped the protein鈥檚 crucial CD4 binding site. That made sense, as it suggested the antibody was so effective because it blocked the virus from binding to CD4 cells. But many other antibodies recognising broadly similar epitopes were much less effective.

It wasn鈥檛 until 2001 that Burton and his Scripps colleague Ian Wilson published a breakthrough paper in Science that explained b12鈥檚 uniqueness (vol 293, p 1155). The use of X-ray crystallography revealed that the antibody has an unusual finger-like projection of 18 amino acids, 1.5 nanometres in length. The team used computer modelling to work out that the protuberance could poke past the variable loops and reach the CD4 binding site of gp120 鈥� the virus鈥檚 Achilles鈥� heel.

Hermann Katinger, now at the Institute for Applied Microbiology in Vienna, found a second neutralising antibody that attaches near enough to the CD4-binding site so that it gets in the way of the two molecules interacting. Called 2G12, it also has a surprising feature. Burton and Moore separately showed last year that this antibody binds exclusively to three mannose sugar subunits on gp120, and not to protein (Journal of Virology, vol 76, p 7306 and p 7293). Immunologists used to think that unlike the sugars on bacteria, those on viruses would not stimulate antibodies, because they look so similar to host-cell sugars 鈥� they have, after all, been made by host enzymes. The mannose cluster seems to be sufficiently unusual, however, to be recognised as foreign. Quentin Sattentau, an immunologist at Imperial College, London, who is a long-time researcher into neutralising antibodies, says: 鈥淭hat鈥檚 exciting because it means you could think about targeting sugars rather than proteins. We are a bit shocked by it.鈥� Sattentau points out that HIV researchers now need advice from bacteriologists, as they are more familiar with developing antibodies against sugars.

The three other neutralising antibodies bind to the gp41 proteins 鈥� the stalk holding up the mushroom鈥檚 cap. These antibodies, called 2F5, 4E10 and Z13, do not inhibit CD4-gp120 binding, but are thought to block the structural changes that must take place within gp120 and gp41 to allow fusion of the viral and immune cell membranes.

Encouragingly, Burton is making some progress in engineering the protein mimics of these antibodies鈥� epitopes. He has a prototype epitope for b12, and should soon have a paper published that sheds a lot more light on 2G12, he says, although so far all attempts to use mimics of the 2F5 epitope have failed. 鈥淲e don鈥檛 know how easy eliciting these antibodies will be,鈥� he says. 鈥淏ut somebody made these antibodies. I hope we can teach everyone to make them with the right immunogens.鈥�

There are a few clouds on the horizon. Strains of HIV do exist with mutations in the CD4 binding site which don鈥檛 stop the virus from attaching to CD4 but do seem to stop b12鈥檚 finger-like projection from working. So Burton wants to make a vaccine containing several epitope mimics so that recipients will generate a range of neutralising antibodies. It鈥檚 highly unlikely that a single virus will be able to evade all of them.

Recently, there has been a surge of interest in neutralising antibodies. The International AIDS Vaccine Initiative (IAVI) highlighted the field鈥檚 importance at a major HIV conference in Barcelona last year. It launched a new Neutralizing Antibody Consortium to accelerate and coordinate research, with Burton as director. Wayne Koff, IAVI鈥檚 senior vice-president of R&D, says the consortium will strengthen the field. 鈥淏efore, one or two labs were working independently on small-scale operations,鈥� he says. 鈥淚t鈥檚 another thing to come at it on a different scale.鈥�

Many scientists, including Koff, now think that our best bet for an HIV vaccine is to combine elements that elicit neutralising antibodies with ones that trigger CTLs. The aim is that when someone encounters HIV, antibodies mop up most of the free virus, and any that slip through to infect cells are killed by CTLs. Putting the two strategies together won鈥檛 be easy. It looks like the most dangerous virus that humans have ever encountered will require the most complex vaccine ever developed.

Despite the challenges, numerous scientists think the recent advances give grounds for cautious optimism. Koff says the existence of neutralising antibodies in nature proves to his satisfaction that this is a solvable problem. 鈥淎nd when you have a problem which is solvable then you鈥檙e only constrained by your own creativity and resources.鈥�