# News this Week

Science  13 Sep 2013:
Vol. 341, Issue 6151, pp. 1156
1. # Around the World

1 - St. Petersburg, Russia
New Curbs on Greenhouse Gases
2 - Bilbao, Spain
Neutron Facility's Fate Up in the Air
3 - New Delhi
Parliament Comes Down Hard on HPV Trial
4 - Canberra
New Government Promises Research Grant Overhaul

## St. Petersburg, Russia

### New Curbs on Greenhouse Gases

The G20 group of industrialized nations has agreed to limit emissions of hydrofluorocarbons (HFCs), which are potent greenhouse gases. Production of HFCs, used in refrigerators and air conditioners, has increased in recent years as a substitute for chlorofluorocarbons, an ozone-destroying chemical targeted for eventual elimination under the 1987 Montreal Protocol. The new agreement represents a major concession by India and China, who are big HFC manufacturers and have previously opposed new limits on these organic compounds. The United States and China also signed a bilateral agreement on HFCs.

The new emission limits will rely on "the expertise and institutions of the Montreal Protocol" but will not amend its text—an important clause aimed at preserving the 1987 treaty. "Keeping HFC emissions from increasing to problematic levels is a key component of any sensible overall climate change mitigation strategy," says NASA climate scientist Drew Shindell. The Washington, D.C.–based nonprofit World Resources Institute said the agreements mark "progress … in coming together to curb climate change."

## Bilbao, Spain

### Neutron Facility's Fate Up in the Air

Plans to build a particle accelerator and neutron source in Spain's Basque Country came to a sudden halt when the governing board of the European Spallation Source (ESS)-Bilbao removed the project's scientific director, Javier Bermejo, and did not renew the contract of ESS-Bilbao's executive director, retiring physicist Joan Bordas. The moves highlight a debate over whether ESS-Bilbao should be a freestanding facility primarily serving Spain, or an R&D test bed for ESS, a $2.4 billion, high-power neutron facility expected to be operational in Lund, Sweden, by 2025. ESS-Bilbao was launched in 2009, after Spain lost a bid to host ESS. In exchange for a 10% ownership of ESS, the Spanish Ministry of Science and Innovation committed$240 million to developing the Bilbao facility to provide components and serve as a technology test bed for the Lund accelerator. But Bermejo had plans for a more ambitious facility that would build its own light-ion linear accelerator and neutron source for local scientists.

What direction the Bilbao facility will take is up in the air. Spain's competitiveness ministry, which now oversees science and innovation, has opened talks with Sweden on redirecting ESS-Bilbao to the original agreement. http://scim.ag/Spainspall

## New Delhi

### Parliament Comes Down Hard on HPV Trial

In a 30 August report, an Indian parliamentary panel excoriated the Seattle-based nonprofit Program for Appropriate Technology in Health (PATH) and the Indian Council of Medical Research (ICMR) for alleged ethical violations in a trial of a vaccine to protect against cervical cancer caused by the human papillomavirus (HPV). In 2009, the groups launched a $3.6 million HPV trial in 24,777 adolescent Indian girls. Several months into the trial, the government pulled the plug after news outlets reported seven deaths, but state investigations found that the deaths were unlikely to be related to the vaccine. The parliamentary report blasts PATH and ICMR for alleged ethical lapses un covered in 2011 by a panel appointed by the Indian health ministry and also accuses PATH of being a tool of foreign drug companies such as Merck and GlaxoSmithKline, which donated almost$6 million worth of vaccines to the trial.

The fallout from the report could be severe. "After this episode, it will be harder to do clinical trials in India," predicts geneticist Krishnaswamy VijayRaghavan, secretary of the Department of Biotechnology. http://scim.ag/IndiaHPV

## Canberra

### New Government Promises Research Grant Overhaul

A victory for Australia's conservative coalition on 7 September could mean an overhaul of the way certain research grants are distributed. In a press release issued just 2 days before the national election, the incoming government—led by Prime Minister-elect Tony Abbott—promised to scrap Australian Research Council (ARC) grants in research areas it deems unnecessary. In the release, Member of Parliament Jamie Briggs singled out awards in philosophy, religious history, and the intersection of art and climate change, calling them "ridiculous research grants that leave taxpayers scratching their heads."

The coalition, Briggs said, won't cut ARC's total budget, and the new government has promised ample funding for dementia and tropical medicine research.

Micromanagement of grant selection is not going to go down easy: Scientists are unified in their distaste for government intervention, contends Will J. Grant, an expert on science and policy at Australian National University in Canberra. "It's not for politicians to make a decision" on what types of research are important, says Grant, who notes that the Australian government already sets research priorities for ARC. Once the review process begins, he adds, "the government should stop meddling."

2. # Random Sample

## Lasker Awards Highlight Basic Neuroscience, Cochlear Implants

This year's Lasker Awards, announced Monday, recognized two pioneers in the study of neural signaling and three researchers responsible for modern cochlear implants. Richard Scheller of the biotech company Genentech and Thomas Südhof of Stanford University in Palo Alto, California, received the Albert Lasker Basic Medical Research Award for their investigation into the cellular mechanisms for releasing neurotransmitters—the brain's chemical messengers. The two worked independently in the late 1980s to identify individual proteins that mediate the process, then developed genetically altered mice lacking these proteins.

The clinical research award went to Graeme Clark, Ingeborg Hochmair, and Blake Wilson for their work to restore hearing to the deaf. Hochmair and Clark were the first to insert multiple electrodes into the human cochlea to stimulate nerves that respond to different frequencies of sound. Wilson later refined cochlear implants with a strategy known as "continuous interleaved sampling," which enabled them to process speech clearly. Some see the prestigious award, which includes a $250,000 prize for each category, as a hint at future Nobel contenders. http://scim.ag/Laskers2013 4. # Alarm Over Autism Test 1. Emily Underwood A hypothesis that maternal antibodies can impair fetal brains stirred plenty of doubts—and that was before the researchers set out to turn it into a test for predicting autism. Jackie Murphy didn't worry that her son Fintan was a late talker, at least at first. Her other two children had been slow to say their first words, so it was only when the former California nurse noticed that her 20-month-old wasn't responding to his name, or even reacting to loud noises, that she became concerned. "One day, I dropped a toy xylophone behind him and he didn't even flinch," she says. "That's when I knew something was wrong." Fintan didn't have a hearing problem—he had autism, his mom finally learned after more than 6 months of searching for a diagnosis. A few months later, Murphy enrolled Fintan in the Autism Phenome Project at the MIND Institute at the University of California (UC), Davis, a long-term assessment of children, as many as 1800, aimed at teasing out subtypes of the complex disorder. Murphy also became a research subject, donating a blood sample. One of the project's researchers, Melissa Bauman, soon informed Murphy that her blood had tested positive for antibodies that react to fetal brain proteins. Bauman asked her to donate more blood for studies exploring the provocative idea that some of Murphy's antibodies had slipped through the placenta and into Fintan's developing brain, affecting its maturation. At that point, Murphy says, she and her husband made a big decision: Fearing that the immune proteins in her blood would harm another baby, they decided that she would not again get pregnant. Many more women could face a similarly difficult choice. In July, immunologist Judy Van de Water and her team at UC Davis, which includes Bauman and Daniel Braunschweig, bolstered the hypothesis that maternal antibodies cause some autism with two studies, including one showing autismlike symptoms in monkeys injected with such antibodies. And women may soon be able to check whether they have the suspect antibodies: California company Pediatric Bioscience announced that it is moving forward with a new diagnostic test, based on patented antibody screening techniques licensed from Van de Water and UC Davis. According to a press release issued by the company at the time, a positive result on the test, estimated to cost roughly$800 and be ready in about 18 months, could tell a woman that she has a "99 percent likelihood" of having a child with autism if she became pregnant. It could also tell parents if the child they just had is likely to develop autism, especially if there are already signs of a developmental delay. Such a scenario, says company President Jan D'Alvise, would let parents enroll a child in early interventional therapies, even before symptoms develop. Already, "people are e-mailing me, saying, I would like to get the test, where can I get the test," Van de Water says.

Pediatric Bioscience's announcement, however, has alarmed many autism researchers. "This is very, very premature—this research has come out of one group, and basically one study. I'm amazed that they're going ahead at this point and trying to commercialize a test," says autism researcher George Anderson of Yale University. He and others say that Van de Water's data are too preliminary, and her statistics too weak, to support such clinical uses. They are skeptical of the mechanism she has proposed for how maternal antibodies could damage the fetal brain. And before any antibody test for autism is launched, they say, her results need to be extensively replicated. "This whole thing could be a house of cards," warns Thomas Wassink, a geneticist at the University of Iowa in Iowa City who studies autism.

Perhaps the most controversial aspect of the proposed test is that some already pregnant women might decide to abort their babies on the strength of the results. Van de Water emphasizes that Pediatric Bioscience plans to limit the "intended use" of the maternal antibody test to women who are not currently pregnant. The company is planning to develop and conduct the test exclusively, she says, thus allowing for control over its use and proper counseling for families. Van de Water is also seeking validation of her group's initial studies, with three large prospective trials under way, but she nonetheless feels compelled to push ahead with Pediatric Bioscience on the test. "The community in autism is a very frustrated group of people. They feel like we've done a lot of research but none of that is impacting our lives right now. … If you wait until I have thousands of samples that's kind of late," she says.

## Controversial origins

This isn't the first time purported connections between the immune system and autism have ignited controversy. In 1998, the British physician Andrew Wakefield claimed that eight of 12 children with neurodevelopmental delays had experienced a sudden onset of autism symptoms after receiving the measles, mumps, and rubella (MMR) vaccine. In a press conference held in coordination with the publication of a study in The Lancet, he hypothesized that the triple dose of vaccines could trigger an immune reaction that damages the brain. The study was later retracted, Wakefield was stripped of his medical license for falsifying the medical histories of the children, and more than a dozen epidemiological studies have failed to find any connection between childhood vaccines and autism. But the link has yet to be dissolved in the public imagination, says Emanuel DiCicco-Bloom, a child neurologist and neuroscientist at Rutgers University in Piscataway, New Jersey.

That history puts an extra burden of responsibility on researchers considering immune hypotheses for autism. It has also made exploring such hypotheses more difficult, notes Karoly Mirnics of Vanderbilt University in Nashville, a neuroscientist who studies immune changes in autism, schizophrenia, and other mental disorders. "Ten years back, you could not get funding" for this type of research, he says. "You could not publish it. You were sort of considered to be on the fringes of science."

One place where researchers interested in immune connections to autism have found support is the MIND Institute, which was founded in 1998 by influential California families who had a "big interest" in the MMR hypothesis, Van de Water says. Vaccine research, however, is no longer a focus at the institute, according to Van de Water, and her work does not directly involve vaccines.

An emerging body of evidence unrelated to childhood vaccines suggests that a mother's immune system may play a role in offspring developing autism, says Paul Patterson, a biologist at the California Institute of Technology in Pasadena who studies interactions between the nervous and immune systems. For example, epidemiological studies have shown that several types of infection during pregnancy can increase the risk that a child will develop schizophrenia or autism, he says. Families with autoimmune disorders such as rheumatoid arthritis and celiac disease are also more likely to have children with the disorder. Patterson's own lab has conducted animal studies showing that inflammatory molecules produced by a pregnant mother in response to an infection such as the flu can affect fetal brain development. "It's a very reasonable hypothesis to be out there testing because of the epidemiologic connection between learning disabilities and mothers with a variety of autoimmune diseases," says Daniel Geschwind, director of the Center for Autism Research and Treatment at UC Los Angeles. "It looks like there's some epidemiologic signal there—the question is, is that a causal relationship?"

Van de Water's interest in maternal antibodies was sparked by a 2003 study by researchers at the University of Oxford, which focused on one 38-year-old mother. The woman had two relatively normal children—a girl with high-functioning autism spectrum disorder (ASD) and a typically developing boy. However, her third child, a 6-year-old boy, had appeared fairly normal until about 18 months and then quickly regressed to severe autism, losing all language. When the researchers extracted serum from the mother, they found that some of her antibodies bound to neurons taken from a developing mouse brain, and they suggested that the antibodies could have triggered her son's condition.

Other researchers have pursued the maternal antibody hypothesis over the past decade, but Van de Water's team has taken it the farthest, Patterson says. In 2008, the group expanded on the Oxford study by extracting serum from 61 mothers of children with autism. About 11% of the mothers carried antibodies that reacted strongly to a group of unidentified brain proteins in human fetal brain tissue, and the level of reactivity corresponded with the severity of the child's autism symptoms. None of the 102 control mothers showed the same pattern of reactions.

Van de Water's latest work, published in the journal Translational Psychiatry, expands the sample still further, to 246 mothers of children with autism. She and her colleagues found that 23% carried antibodies that react to fetal brain tissue, compared with only 1% of 149 mothers who have typically developing children. This time the team pinpointed seven specific proteins that the antibodies target, including ones linked to the proliferation of neurons, neural migration during development, and neural branching (see diagram). "Each works at some stage in the development of a neuron," Van de Water says.

It's not clear why a woman would generate such reactive antibodies. "We may never know," Van de Water says. Some women may be genetically predisposed to produce the antibodies in response to an immune insult such as an infection, or even a vaccine, she suggests. Although that's pure speculation for now, Murphy, who has become a close friend of Van de Water, is already persuaded that a vaccine she received contributed to Fintan's autism. "He's the only one I got vaccinated with the flu shot during pregnancy," she says.

In her other new study, Van de Water teamed up with Bauman and neuroscientist David Amaral, director of the MIND Institute, to link the autism-associated antibodies to symptoms. The team reported that when pregnant rhesus monkeys were injected with these human maternal antibodies, their infants developed behavioral and brain development problems. The researchers say the symptoms resemble some of those seen in people with autism. For example, the young monkeys whose fetal brains had been exposed to the antibodies directly approached unknown animals far more often than their peers did, Bauman says—a sign that there was something socially "off" about them.

## Strong reactions

Van de Water's latest work is "an important step that lots of people have been waiting to see," Patterson says. If she's right, "we're at least doubling" the number of autism cases we can explain, adds Andrew Zimmerman, a pediatric neurologist at the Kennedy Krieger Institute in Baltimore, Maryland. Now, only 15% to 20% of autism cases can be traced to a specific cause, he says—primarily genetic mutations.

But other reactions range from skeptical to sharply negative. Zimmerman himself says that absent replication of Van de Water's clinical results in much larger groups of women, "it isn't clear yet what the [autism] risk to a particular pregnancy is if a mother has the antibodies." He and others also question the maternal antibody hypothesis itself. For example, it assumes that the so-called blood brain barrier, which normally protects the brain from molecules such as antibodies, is permeable enough in fetuses for antibodies to enter and for them to do damage—a fact that has yet to be rigorously established, DiCicco-Bloom says.

Betty Diamond of the Feinstein Institute for Medical Research in Manhasset, New York, says the idea that maternal antibodies directly cause autism in some cases is "plausible." In her own lab, Diamond has demonstrated that pregnant mice injected with antibodies made by women with lupus that target DNA-associated proteins and certain cellular receptors on the surface of neurons bear offspring with abnormal brain structures and cognitive problems. And just last month, Diamond published a study of blood from 2700 mothers, which found that roughly one in 10 mothers of children with autism carried what her team calls "anti-brain" antibodies—a ratio roughly four times higher than controls. Preliminary studies suggest that the antibodies targeted proteins different from those that Van de Water has identified, however, and Diamond herself is not ready to claim that these antibodies actually cause autism.

Indeed, Van de Water's dramatic claim that maternal antibodies could play a role in a quarter of all autism cases disturbs many researchers. Because the serum of the tested mothers displayed varying patterns of reactivity to the seven fetal brain proteins—some binding to just one and others to up to five—the researchers added up many different "specific combinations" of reactivity to get their figure of 23%. No specific pattern of reactivity was observed in more than 7% of the mothers of autistic children that they tested, notes Steven Goodman, a biostatistician at Stanford University in Palo Alto. He described the approach as a form of data dredging—sifting data for the largest effects—which at best exaggerates predictiveness, and at worst finds patterns that aren't real.

Ultimately, DiCicco-Bloom says, Van de Water's studies will need to be replicated in thousands of women from different backgrounds to establish whether the maternal antibodies have any predictive value. That should be fairly easy to do, now that the fetal brain proteins targeted by the antibodies have been identified, notes epidemiologist Ian Lipkin, of Columbia University, principal investigator of the Autism Birth Cohort, a large Norwegian study of more than 100,000 children.

Van de Water agrees that her new work needs to be replicated by independent groups. She adds that she is involved in three large prospective studies now in progress at UC Davis, all funded by the National Institute of Environmental Health Sciences. "I didn't want people to get the impression that this was final," she says of her most recent work. "This was really a discovery paper, and not a clinical validation."

## Testy questions

Then why, critics ask, is Pediatric Bioscience already trumpeting a test? Van de Water argues that developing the test in parallel with the research means it will be available sooner for doctors to order. "We'll also know if it's not going to work more quickly," she says. She adds that she is facing competition from other groups, such as Diamond's lab.

Anderson is not persuaded. "They're overstating what they have, and then they're proceeding too quickly," he says. "You don't need to commercialize something to make it available as a research tool."

The researcher's corporate ties add to the unease. "She has a patent, and you could perceive a conflict of interest there," DiCicco-Bloom says. Van de Water emphasizes that her relationship with Pediatric Bioscience has been thoroughly vetted by UC Davis, which allows her to split a 30% share of whatever royalties the company pays to the university for use of the patent with other investigators on the team. As chief scientific adviser to Pediatric Bioscience, Van de Water cannot perform the clinical validation studies herself because that would violate the UC Davis policies guarding against conflicts of interest; she is also prohibited by the university from accepting free or discounted stock in the company, she adds.

Financial considerations aside, the key to any medical diagnostic test is its positive predictive value, a number that indicates how often a test, if positive, is right, Anderson says. In a disorder such as autism, which has a prevalence in the general population of about one in 88 births, even a few false positives very quickly render a test of limited value, he says. Based on his calculations using Van de Water's data and the prevalence of autism, "I get a positive predictive value of 16.5%, which is pretty bad." At that rate, only one out of six positive tests would actually be correct, he explains.

Goodman says that he came up with the same low positive predictive value after crunching the data provided in Van de Water's paper. "That assumes that you accept these numbers at face value, and I don't," he says. Based on the design of the study and his own analysis of the group's published data, he describes Pediatric Bioscience's claim that a positive result on a test for these antibodies will mean that a woman has a "99 percent" likelihood of having a child with autism as "completely false."

Van de Water and D'Alvise respond that the company's press release was "a bit unclear as written" and that likelihood "was not meant to convey likelihood in the statistical sense, but rather the 99% accuracy with which the study demonstrated specificity of the biomarkers for ASD." They also say that test is not meant to screen the general population but would be for women at higher risk of having autistic offspring, such as those who are older or who already have a child with developmental issues.

The U.S. Food and Drug Administration will not be required to review the accuracy or clinical validity of the Maternal Antibody Related Autism (MAR) test, as the company calls it, before it goes to market. Instead, D'Alvise says, Pediatric Bioscience will set up a blood testing lab under the Clinical Laboratory Improvement Amendments, a set of alternative federal standards that regulates diagnostic tests that only one laboratory has the expertise to run. "Typically this is the way new diagnostic tests get offered before they become mainstream," D'Alvise says.

Van de Water believes her test has predictive power, based on the relatively small population she's tested so far. But she acknowledges that it might not hold up in larger, more diverse populations. "I hope it is a valuable area to pursue—we think it is—but down the road who knows what will happen."

DiCicco-Bloom still isn't convinced that the maternal antibody test will bring much benefit and suggests that it could do more harm than good. "Pretty soon, everybody who worries about autism is going to be getting [tested], some for good reason, some because they're highly educated, motivated people," he says. "The moment that this test is given to physicians, it's not going to be controlled. No medical test ever is."

5. Immunology

# Bound for Glory

1. Jon Cohen

The discovery of antibodies that foil almost every HIV variant has transformed the AIDS vaccine search.

In 2006, the International AIDS Vaccine Initiative (IAVI) launched a worldwide hunt to find the rarest of immune system warriors against HIV—antibodies that could thwart almost every known strain of the virus. IAVI, a New York City–based nonprofit, aptly gave the project a Mission Impossible–sounding moniker: Protocol G. If it succeeded, Protocol G promised to invigorate the failure-plagued HIV vaccine field and just possibly help design a product that could end the AIDS epidemic.

HIV at that point had thoroughly humbled vaccine researchers, time and again dodging any immune attack they could devise. More than 2 decades of effort had yet to bring a vaccine to market. But researchers had already glimpsed a weapon they hoped might turn the battle: so-called broadly neutralizing antibodies (bNAbs). Only a handful had been discovered, and even the best of them worked against only a few dozen of the hundreds of HIV strains that a vaccine would have to outwit to make a dent in the global epidemic.

The Protocol G team made an all-out effort to find more. Researchers analyzed 1800 blood samples taken from HIV-infected people in Zambia, South Africa, India, Thailand, Nigeria, Côte d'Ivoire, Rwanda, Uganda, Kenya, Australia, the United Kingdom, and the United States. After 2 years of combing through the sera, an effort led by IAVI team member Dennis Burton, an immunologist at the Scripps Research Institute in San Diego, California, homed in on one woman. By individually probing 30,000 of the woman's B cells, the immune actors that produce antibodies, they isolated two remarkable antibodies from her serum.

The two Protocol G antibodies jumped out because they "neutralized," or rendered incapable of establishing an infection, more than 70% of 162 divergent HIV strains, even at very low doses, Burton and co-workers explained in the 9 October 2009 issue of Science. Suddenly, creating a universal HIV vaccine seemed like more than just a wild dream. "It dawned on us that our little world was never going to be the same," Burton says.

Their little world had indeed turned upside down, and it had big world consequences. In short order, bNAbs became one of the hottest niches not just in HIV research but in immunology itself. Since 2009, researchers have identified more than 50 new HIV bNAbs, several of which look far better than those two from Protocol G.

Dozens of labs are now trying to unravel the intricate pas de deux between HIV and the immune system that leads it to produce bNAbs, a process they hope to mimic one day with an immunogen—a harmless version of the virus used in a vaccine. And the activity in the HIV field has triggered something of a gold rush to find bNAbs for other diseases, including influenza (see sidebar, p. 1171), hepatitis C, dengue, and West Nile viruses. But researchers have no illusions about the challenge of coaxing the immune system to produce a type of antibody that it naturally generates only rarely and in tiny quantities. "If somebody came up with a real immunogen that would reproducibly elicit a broadly protective response, that would be fabulous," says David Baltimore, a Nobel Prize–winning virologist whose own group at the California Institute of Technology in Pasadena is attempting to genetically engineer stem cells to make bNAbs. "But I see a very difficult road."

## Inauspicious start

Shortly after investigators proved in 1984 that HIV causes AIDS, the key obstacle to making a vaccine emerged. HIV reproduces at a breakneck speed, sloppily copying its genetic code each time. Many of the resultant mutants thrive, and the variants display different protein antigens. Antibodies, which normally target a single antigen with exquisite specificity, simply can't keep up with the ever-changing virus. As a result, neither the natural immune response nor the response induced by a conventional vaccine is effective.

Contrast this with viruses that vaccinemakers conquered. Measles virus has so little antigenic variation that an antibody response against one virus works against every relative. Even polio, an RNA virus like HIV that mutates at the same rate, can be derailed with a combination of antigens from three distinct strains of the virus. Unlike HIV, these viruses pay a steep fitness price when their antigens change, making it difficult for them to dodge antibody attacks.

In 1990, researchers from Repligen, a Massachusetts biotech, electrified the AIDS vaccine field with a report in Science of the first bNAb against HIV, far more powerful that any antibody seen to date. More astonishing still, they described the exact viral component, or epitope, that stimulated the production of the antibody in animal experiments. A mere six amino acids at the tip of HIV's surface protein, gp120, taught the immune system how to stop diverse strains of the virus, they reported.

It sounded too good to be true. It was.

As it turned out, this bNAb was a laboratory artifact that worked only against strains of the virus grown in culture dishes, and it had no impact on "primary" isolates taken directly from infected people. That 1993 discovery (Science, 12 November, p. 980) scuttled hopes about the Repligen product. But the next year, Burton and colleagues at Scripps showed that by aspirating bone marrow from an HIV-infected man and then fishing out antibody genes, they had isolated a bNAb that worked against primary isolates.

To establish an infection, HIV first slips gp120 into a CD4 receptor on the surface of white blood cells. Some antibodies can bind to gp120 and neutralize it such that the virus can't enter the CD4 cell. But as HIV mutates, gp120 changes shape, evading antibodies. Burton's bNAb attached to a "conserved" part of gp120 that doesn't tolerate mutations because it is required by many variants to bind CD4 receptors. This bNAb crippled 17 of 24 primary isolates of the virus at what then seemed liked fantastically low doses (Science, 11 November 1994, p. 1024).

At roughly the same time in Austria, another group discovered bNAbs that worked on a different part of HIV. A team led by Hermann Katinger at the Institute of Applied Microbiology in Vienna identified two bNAbs that acted on gp41, a protein that spans the viral membrane and attaches to gp120, and were capable of neutralizing many HIV isolates. This finding suggested that the virus has more than one Achilles' heel.

Over the next decade, structural biologists like Ian Wilson at Scripps and Peter Kwong at the U.S. National Institute of Allergy and Infectious Diseases (NIAID) in Bethesda, Maryland, created high-resolution depictions of how these bNAbs bound to both gp120 and gp41. These exquisitely detailed images enabled the investigators to determine the precise mechanisms by which these antibodies tripped up the virus—and why other antibodies had such limited potency and breadth. But few other HIV bNAbs surfaced, and no one found antigens that could trigger their production in animal models or humans. At the same time, antibodies fell out of fashion with many vaccine researchers, who put more stock in stimulating killer cells, a kind of T lymphocyte. While antibodies prevent infection, killer cells identify and destroy cells that the virus manages to enter. Some groups contended that a T cell vaccine alone could stop the virus; others put their stock in a combination T cell/antibody vaccine. Still others rejected the notion of rationally designing a vaccine, pointing to a long history of vaccinologists who succeeded simply by testing various formulations in animals and people, paying little heed to immune mechanisms.

## Working backward

Then, two developments reinvigorated the bNAb field.

First, high-profile, T cell–based vaccines against HIV failed miserably (Science, 10 September 2004, p. 1545 and 5 October 2007, p. 28), tilting more attention toward antibodies. On the second front, thanks to Protocol G and similar initiatives, the number of bNAbs available for study skyrocketed. Burton and others say that they owe their success to better lures to fish bNAbs from a sea of other antibodies. They also developed improved techniques to clone large numbers of antibodies from single B cells, first amplifying antibody genes and then putting them into viruslike vectors that spit out the products in bulk. For the first time, they could separate these faint immunologic signals from the noise. "Single cell antibody cloning changed the field," says immunologist Michel Nussenzweig, whose lab at Rockefeller University in New York City helped pioneer the technique for HIV bNAbs (see Review, p. 1199). "Otherwise you didn't even know what was in the human repertoires." An extraordinary level of multi-institution collaboration has given the endeavor a major boost, too, through IAVI's Neutralizing Antibody Consortium and two Centers for HIV/AIDS Vaccine Immunology and Immunogen Discovery (CHAVI-ID) created by NIAID.

With these advances, researchers can now find bNAbs, albeit at exceedingly low levels, in up to 25% of HIV-infected people. The bNAbs offer scant help to the already infected, as they come too late, when the virus has already infiltrated many blood cells and tissues, and in quantities too small to make a difference. But a vaccine that elicited bNAbs at the right moment, in sufficient quantities, might foil an infection. So researchers are now asking how the immune system makes these antibodies and what gives them their unusual powers.

Like other antibodies, HIV bNAbs emerge as the B cells first exposed to the virus multiply to produce clones that steadily develop more "affinity" for the invader. Called "affinity maturation," this process of mutation and selection ultimately favors the fittest immune response. But unlike other antibodies, bNAbs appear to evolve slowly over years, accumulating an unusually high number of mutations—three times as many as run-of-the-mill antibodies—which gives them their potency and breadth. The process appears to involve stimulation from different epitopes at different points in time. "There's an arms race between HIV and the immune system," Nussenzweig says. "As antibodies chase the virus around, it keeps changing, and the antibodies have to make more mutations."

Most mutations that give bNAbs their powers take place at the tips of the Y-shaped antibody molecules, which have loops that lasso viral epitopes. Among other things, bNAbs have evolved to become stickier than other antibodies, Nussenzweig and colleagues explained in the 30 September 2010 issue of Nature. That appears to be a response to one of HIV's unusual properties. The gp120 molecules studding HIV's surface form clusters of three called trimers. Each HIV has only about 10 trimers on its surface; on influenza viruses, in contrast, the surface proteins form about 450 trimeric spikes. "HIV is almost bald," Nussenzweig says. This dearth of trimers hampers antibodies: They bind most tightly to a virus when each arm of the Y can separately grab a different trimer. But the extra-sticky bNAbs can compensate by latching on to lipids that make up the HIV viral membrane or even to sugars that coat gp120.

Other peculiar features help HIV bNAbs stymie the virus. As Burton has shown, the loops at the tips of antibody arms are often extra long, increasing their ability to lasso an epitope. Nussenzweig discovered a second, related quirk. Antibody arms typically have a region that remains constant to provide a sturdy framework for the highly mutable loops. But Nussenzweig's group reported in the 28 March issue of Cell that HIV bNAbs shun convention, accumulating many mutations in the framework region that increase breadth and potency. For unknown reasons, these malleable framework regions don't compromise the stability an antibody needs to function. "It goes totally against the norm," Nussenzweig says. "Destabilizing something and making it better is counterintuitive."

Creating a vaccine that will elicit these features is a daunting assignment, say some researchers. "We're going to have to make this vaccine with nature's pathways and guidance," says Barton Haynes, an immunologist at Duke University in Durham, North Carolina, who heads a CHAVI-ID. "And it's a vaccine the immune system doesn't favor making."

Nussenzweig is optimistic. "People don't invent vaccines de novo," he says. "In every case, you take a lesson from nature and try to copy it." Granted, people with HIV make potent bNAbs in only small amounts, but the discovery that dozens of different ones exist, each of which came about independently, provides hard proof that the human immune system knows how to make them. "If it can happen once," Nussenzweig says, "it can happen again."

## The making of a bNAb

In 2006, a man in Malawi came to a clinic for sexually transmitted diseases that had a research project under way to try to find people within weeks of becoming infected with HIV. He met the criteria and joined the "acute infection" study, repeatedly donating blood, which enabled researchers to document changes over time in the gp120 of his virus, his antibody response, and ultimately the emergence of a bNAb. Now, this window into antibody evolution is inspiring a strategy to design a bNAb-based vaccine.

A large collaboration led by Duke's Haynes and John Mascola from NIAID's Vaccine Research Center studied the man's sequential blood samples. As the researchers explained in the 25 April issue of Nature, they traced the evolution of the antibody from the initial B cell response all the way to the bNAb, which emerged after more than a year. The original, or germline, antibody did not have much breadth, but it did bind avidly to the original "founder" virus that infected the man. A vaccine, they reason, should use gp120 from the founder virus to kickstart a B cell lineage that can ultimately produce a bNAb. Booster shots could then attempt to guide the mutation of the antibody genes in the cell to expand their breadth by using the same sequence of gp120 variants that naturally created the bNAb in the Malawian man.

In essence, the strategy amounts to carefully selecting a puppy from a litter and giving it the intensive training needed to one day win the best of show. "It's ingenious that you'd prime with one epitope and boost with another epitope to chaperone the B cell response until the point that it's truly making a broadly neutralizing antibody," says immunologist Anthony Fauci, NIAID's director.

Burton, working with Scripps immunologist William Schief, also tracked the emergence of a bNAb, but they did so retrospectively. They started with a powerful bNAb and used genetic modeling to predict its ancestor's characteristics. Guided by what they learned, they hope to design a single immunogen that both tickles the right B cells into action and steers them to move from the germline antibody toward the mature bNAb. Unlike the Haynes group, Burton and Schief 's team does not want to hold the immune system's hand and guide the germline B cell to make the bNAb. "Taking that too literally is likely to be asking for too much micromanagement," Schief says.

As the group explained in the 10 May issue of Science, a screen of massive gp120 libraries led to one that strongly bound both their putative germline antibody and the mature bNAb that evolved from it. The investigators then groomed this gp120 to contain little more than the epitope that both antibodies target. They now hope to show that they can elicit a bNAb response by "priming" the immune system with this artificial gp120, then following with a booster that contains trimer spikes in the most natural configuration possible.

Both the Burton/Schief team and Haynes's group have monkey studies under way to test their ideas. But even if they learn how to generate bNAbs, Fauci notes that a fundamental, proof-of-principle experiment has yet to happen: No one has shown that giving bNAbs to a person can prevent an infection. To that end, the Vaccine Research Center at NIAID hopes to soon launch human studies that passively give monoclonal bNAbs—which theoretically could last in the body a few months—to newborns of HIV-infected mothers and, separately, to adults at high risk of becoming infected. "Whether they're going to prevent acquisition of HIV is an open question," Fauci cautions.

Even if the bNAb strategy passes all these hurdles, years of testing will still be needed to prove that the vaccine protects people and is safe. But Fauci and many others at the forefront of the HIV vaccine search say bNAbs have brought a level of hope and common purpose that this fractious, battle-weary field has rarely, if ever, enjoyed.

6. # A Once-in-a-Lifetime Flu Shot?

1. Jon Cohen

Novel antibodies suggest a surprising formula for a universal flu vaccine.

No more yearly flu shots. An end to the threat of deadly pandemics. Farfetched as those possibilities may sound, they could become reality if a hot new area of immunology pans out.

Flu vaccines trigger production of antibodies that attach to hemagglutinin, a protein on the surface of the virus that helps it infect cells. But hemagglutinin mutates so rapidly that antibodies to one human variant have limited power against another, requiring vaccinemakers to reformulate their shots each year. And when a novel animal flu jumps from birds or pigs into humans, existing immunity offers little defense and a pandemic can arise. Stopping it would require a new vaccine, which inevitably can't be developed quickly enough.

Recently, researchers have found a possible solution: "broadly neutralizing antibodies" (bNAbs) in humans to hemagglutinin, able to bind most, if not all, variants. Like bNAbs discovered for HIV (see main article, p. 1168), they have sparked provocative ideas about how to make a single vaccine that could thwart all strains of the virus.

In 2008, a Dutch biotech showed that such antibodies existed and that they worked by an unorthodox mechanism. In both test-tube and mouse experiments, the bNAb neutralized H5N1 and H1N1. (The "H" refers to hemagglutinin.) These highly divergent, dangerous influenza viruses caused, respectively, "bird flu" and the 1918 pandemic in humans. Unlike traditional antibodies to influenza, which bind to the highly mutable head of hemagglutinin, this one bound to more immutable amino acids in the highly conserved stem of the molecule.

Hopes ran high that a headless hemagglutinin could be the basis of a vaccine capable of triggering bNAbs that protected against all 17 different H subtypes. But creating a stem-only vaccine has presented formidable technical challenges. "Removing the head is extremely difficult if not impossible," says Antonio Lanzavecchia, an immunologist at the Institute for Research in Biomedicine in Bellinzona, Switzerland, whose team reported in the 12 August 2011 issue of Science the discovery of a stem-binding antibody that neutralized every H subtype. "You can't just cut it out." Basically, headless hemagglutinins fall apart.

Virologist Peter Palese's lab at the Icahn School of Medicine at Mount Sinai in New York City described a creative way around the problem in the June Journal of Virology. Even though people fail to make high levels of antibodies to the stem, they do make some and thus develop what's known as immunologic memory. So they reasoned that a vaccine might work if it had an engineered, chimeric hemagglutinin made from the stalk of a virus that humans had seen, triggering a robust memory response, with an "irrelevant" head that had circulated only in other species. They mimicked this chimeric vaccine in a mouse model, and it protected animals from distantly related flu viruses.

A research team at the National Institute of Allergy and Infectious Diseases has had impressive results in mice and ferrets with a different tack. The bNAbs to influenza are slow to develop in part because hemagglutinin naturally crowds the viral surface, hiding the stem regions of the protein from the immune system. Led by virologist Gary Nabel, the group described in the 4 July issue of Nature how it created an artificial, self-assembling nanoparticle called ferritin, an iron-storage protein, that expresses hemagglutinins at an unnatural angle, exposing their stalks. This new presentation of the protein leads to a potent bNAb response. Nabel, now chief scientific officer at vaccinemaker Sanofi, has high hopes that the days of seasonal flu shots are numbered. "We have the ability to steer the immune system back to where we want it to be," he says.