Renowned Weizmann scientist, Prof. Ruth Arnon, recently spoke to the New Yorker Magazine about a universal flu vaccine. Arnon joined the Institute in 1960 and has devoted her career to pioneering groundbreaking medications and vaccines. As Chair of BiondVax’s Scientific Advisory Board—an Israeli biopharmaceutical company developing a universal flu vaccine—and as the Paul Ehrlich Chair in Immunochemistry at the Weizmann, she is globally recognized as a leading expert in her field. Few scientists working today have the decades of rich experience Arnon still brings to the table.
In 2009, global health officials started tracking a new kind of flu. It appeared first in Mexico, in March, and quickly infected thousands. Influenza tends to kill the very young and the very old, but this flu was different. It seemed to be severely affecting otherwise healthy young adults.
American epidemiologists soon learned of cases in California, Texas, and Kansas. By the end of April, the virus had reached a high school in Queens, where a few kids, returning from a trip to Mexico, had infected a third of the student body. The Mexican government closed its schools and banned large gatherings, and the U.S. considered doing the same. “It was a very scary situation,” Richard Besser, who was then the acting director of the Centers for Disease Control and Prevention, told me. Early estimates suggested that the “swine flu,” as the new strain became known, killed as many as fourteen per cent of those it infected—a case fatality rate more than two hundred times higher than typical seasonal flu. The virus soon spread to more than a hundred and fifty countries, and the Obama Administration considered delaying the start of school until after Thanksgiving, when a second wave could be under way. Manufacturers worried about vaccine supplies. Like most flu vaccines, the one for the swine flu was grown in chicken eggs. “Even if you yell at them, they don’t grow faster,” Tom Frieden, who replaced Besser as the director of the C.D.C., said, at a press conference.
In the end, the world got lucky. The early stats were misleading: although swine flu was extremely contagious, it wasn’t especially deadly. Sometimes the reverse is true. Avian flu, which spread across the world during the winter of 2005-06, is not particularly transmissible but is highly lethal, killing more than half of those it infects. Each flu virus has its own epidemiological profile, determined by its genetic makeup, and flu genes shift every year. Howard Markel, a physician and historian of epidemics who, in the early two-thousands, helped invent the concept of “flattening the curve,” compared influenza’s swappable genetic components to “two wheels of fortune.” A double whammy—ease of spread combined with lethality—could make covid-19, or even the 1918 flu, which killed between forty million and a hundred million people, look like a twenty-four-hour bug.
After the swine flu’s relatively harmless nature became apparent, many people asked if the alarm it provoked had been warranted. A Swiss survey found that trust in institutions had decreased. Some scientists and officials accused the World Health Organization of stirring up a “faked” pandemic to justify its budget. But most drew the opposite conclusion from the experience. Trying to prepare for a deadly flu pandemic had left them more alarmed. “There was just a sense of overwhelming relief,” Besser said. “If this had been like 1918, we sure weren’t ready.”
In truth, we’re never fully ready for the flu. We know it’s coming, like the first fall leaf, and yet three times in the past century—in 1918, 1957, and 1968—it has flattened us, killing a million or more each time. Even in ordinary years, the disease infects a billion people around the world, killing hundreds of thousands; one study estimated that it costs the United States economy close to a hundred billion dollars annually. Our primary weapon against the virus, the flu vaccine, is woefully inadequate. Over the last decade and a half in the United States, flu vaccines have prevented illness only forty per cent of the time; in particularly bad years, when vaccines were less fine-tuned to the strains that were circulating, they were only ten-per-cent protective. Today, the coronavirus pandemic is rightfully the object of our most strenuous efforts. And yet, as the infectious-disease specialists David Morens, Jeffrey Taubenberger, and Anthony Fauci wrote, in a 2009 article in The New England Journal of Medicine, that “we are living in a pandemic era that began around 1918,” when the flu used shipping networks to traverse the world. Since the 1918 pandemic, this century-long, multi-wave pandemic has killed roughly the same number of people.
We’ve controlled a vast number of diseases with vaccination—chicken pox, diphtheria, measles, mumps, polio, rabies, rubella, smallpox, tetanus, typhoid, whooping cough, yellow fever—and, to some degree, we’ve added covid-19 to the list. But the pathogens behind those diseases tend to be relatively static compared with the flu, which returns each year in a vexingly different form. For decades, scientists have dreamed of what some call a “universal” flu vaccine—one that could target many strains of the virus. A universal vaccine would save countless lives not just this year but every year; as those numbers add up, it would become one of the greatest medical breakthroughs in history. Until recently, it’s been beyond the reach of molecular biology. But new technologies are extending our abilities, and researchers are learning how to see through the flu’s disguises. Without knowing it, we’re living on the cusp of a remarkable scientific achievement. One of the world’s longest pandemics could soon be coming to an end.
What we call “the flu” is really plural. Every season, several strains circulate. When it’s summer in one hemisphere, flu infections surge in the other. Virologists at the W.H.O. investigate the viruses and share what they learn with pharmaceutical companies; pharmaceutical researchers then often develop quadrivalent vaccines, which target four separate strains simultaneously. It’s the shotgun approach.
It takes more than six months to design, test, and manufacture a season’s worth of flu vaccine. In that time, a lot can change. Out in the world, strains mutate, jostling for dominance; prevalent varieties fade away, and sleepers come to the fore. Arnold Monto, an epidemiologist at the University of Michigan who has advised the Food and Drug Administration on flu-vaccine targeting, told me that choosing strains to target requires “science and a little bit of art.” The selected flu viruses mutate further as a result of vaccine manufacturing. By the time a needle reaches your arm, there’s a good chance that the vaccine might be off target or obsolete.
Each strain of the flu can be seen as plural, too. Morrens, Taubenberger, and Fauci explain that “it is helpful to think of influenza viruses not as distinct entities but as eight-member ‘gene teams.’ ” A flu virus, they write, “must sometimes trade away one or more team members to make way for new gene ‘players’ with unique skills.”
The surface of a virus is covered by a forest of proteins; as the virus’s genes change, the proteins change along with them. From a vaccine perspective, two proteins are of preëminent importance. The first, hemagglutinin (HA), helps the virus break into cells; the second, neuraminidase (NA), helps it break out of them. The cryptic code names given to flu viruses—H1N1, H3N2, and so on—reflect the dozens of numbered variations in which these proteins come. The variations themselves mutate when the virus reproduces, making vaccine targeting even more difficult: flu vaccines focus on the more vulnerable HA proteins, and must be tailored to fit the newest version of the virus.
Some viruses are siloed within a single species. But the flu migrates easily among several species, and this adds to its recombinatory range. “There are hundreds of warm-blooded animals that are routinely infected with flu virus,” Taubenberger, who is the chief of the Viral Pathogenesis and Evolution Section at niaid, told me. “It can move from birds to horses to pigs to humans.” If a bird is infected with two strains of the flu at once, those strains may combine to create a new virus; that virus, in turn, may enter another animal. In 1918, H1N1 infected humans, who passed the disease on to pigs; the swine flu that so alarmed epidemiologists in 2009 emerged when two of the pig strains converged, then returned to humans. The existence of so-called animal reservoirs for the flu makes it more likely that, in any given year, virologists will confront a radically altered opponent. It also means that herd immunity is nearly impossible to achieve. “Viruses like smallpox or measles or polio that are specifically adapted to humans . . . if you vaccinate enough people to generate herd immunity, you can actually eliminate the virus,” Taubenberger said. “But flu can never be eliminated, because it’s in hundreds of species of animals, and it’s constantly moving around. So, we need a better strategy.”
Vaccines contain antigens—complex molecules that prime our immune systems to produce effective antibodies. Sometimes, antigens are synthetic molecules designed to mimic parts of the target virus; in other cases, they are actually real parts of the virus that have been cleaved off. The antigen in the covid-19 vaccines is a version of the spike-shaped protein that sars-CoV-2 uses to enter our cells. The flu’s equivalent, the HA protein, looks like a mushroom. Since the nineteen-forties, we’ve made flu vaccines in more or less the same way: researchers grow the virus in eggs, then split it open using chemicals which deactivate the virus but leave the mushroom proteins intact. Unfortunately, when our immune systems build antibodies in response to those proteins, they tend to target the cap of the mushroom, where its most mutable elements reside. Researchers don’t yet understand why our antibodies aim for the most changeable part of our adversary. But it’s not the best outcome—a more effective antibody would attach to and disable the less mutable stem.
Börries Brandenburg, a scientific director at Janssen Vaccines, which is part of Johnson & Johnson, told me about how the company has approached the cap problem: “We say, O.K., if the immune system keeps getting fooled by the head of the HA, we’re just going to use the French solution and take off the head. We present the immune system with a molecule that is headless.” Creating a capless mushroom—what the Janssen researchers call a “mini-HA”—is harder than it sounds. Proteins, such as HA, are actually combinations of hundreds of smaller molecules called amino acids; these acids emerge linked together in sequence, like pearls on a necklace, from factories within our cells. Eventually, the necklace twists and curls in on itself, forming a 3-D shape. The mushroom, therefore, can’t be sliced and diced; decapitate it, and the pearls go tumbling. A protein designer who severs its loops must find ways to close them off. When we spoke, Brandenburg’s video-chat backdrop was an Escher illustration of staircases going every which way. As the call ended, I commented that the recursive pathways behind him were reminiscent of looping chains of amino acids. He saw them differently: “This illustrates how research goes,” he said.
In 2019, researchers at the Vaccine Research Center, which is part of niaid, began a clinical trial for a universal vaccine that used another approach. On the surface of a nanoparticle called ferritin—a naturally occurring, spherical assemblage of proteins, used by the body to transport iron in the bloodstream—they arranged a number of mushroom stems in a regular geometric pattern. The immune system responds with special alacrity to orderly arrays of foreign particles; it seems to be more alarmed by soldiers marching in formation than by individuals shambling along. Barney Graham, then the deputy director of the Vaccine Research Center, told me that “one of the keys to solving the universal flu problem” might be “finding new ways of displaying the protein to make it more immunogenic.” (He retired in August.) It’s also possible to stud a nanoparticle not with identical stems but with a variety of caps—the “mosaic” approach. In 2019, scientists at the Vaccine Research Center reported that they’d tested a mosaic flu vaccine on mice; each nanoparticle of the vaccine featured the heads of up to eight different flu strains. It successfully teed up the production of antibodies capable of neutralizing a range of flu viruses that had appeared between 1918 and 2009. This year, the same group reported a mosaic vaccine that could also protect people against avian flu variants that are especially dangerous to humans. A version of the vaccine has entered clinical trials.
Other researchers have eschewed decapitation: instead of cutting off the mushroom caps, they replace them. In 2019, niaid distributed fifty million dollars in grant money to launch the Collaborative Influenza Vaccine Innovation Centers, or civics. A team of grant recipients at Mount Sinai, led by the virologist Florian Krammer, used their money to combine a common stem with an “exotic” cap derived from a bird flu. The exotic cap is useful, oddly, because it elicits a smaller antibody response. The immune system responds strongly to things it’s seen before. Presented with a completely unfamiliar head, it mounts a more vigorous response to the more familiar stem.
What unites all of these approaches is the idea that it might be possible to design a perfect antigen—a single entity that inspires universal immunity. Jacob Glanville, a forty-year-old computational “immuno-engineer” who studies antibody targeting, has come up with an alternative path. In 2012, Glanville left Pfizer, where he was a principal scientist, to start an immunology Ph.D. at Stanford and launch an antibody-discovery company called Distributed Bio. Several years earlier, researchers at the biotechnology company Crucell had shown that a few fortunate people already have antibodies that work against many different strains of the flu; Glanville took notice. Somehow, the immune systems of those flu-resistant people had solved most of the universal-immunity problem on their own, without access to designer antigens. Glanville wanted to understand how this was possible, and why it happened so rarely.
Every HA protein is different. Any single antibody that works against multiple versions, therefore, must have found a way to attack a shared weak spot. Ideally, our immune systems, when faced with a range of related foes, would seek out their common Achilles’ heel. And yet they do not seem to be very good at identifying areas on proteins that are “conserved” across many variations. Immunologists have offered a few explanations for this weakness. Some have argued that conserved sites too closely resemble our own cellular structures: it would be risky for the immune system to start attacking them with antibodies.
Glanville thought that the story might be simpler. Using cloud computing, he ran hundreds of millions of simulations of antibodies attaching to HA proteins. He found that only one in a million antibodies successfully docked to a universally conserved site. The problem, it turned out, was that each viral strain contains many more sites that mutate than conserved sites. This was true even for the stems of the HA mushrooms. The stems may be less mutable than the caps, but they, too, differ more than they are alike. This, Glanville thought, did not bode well for attempts to create stem-based universal vaccines.
Still, some people’s immune systems had managed to create near-universal flu antibodies; this suggested that the problem was solvable. Glanville and his team started asking how more immune systems could be persuaded to solve it. They studied every HA protein they had on file from 1918 to 2007, and tried to analyze the differences between them. HA proteins are so complex that assessing their differences is essentially impossible for most human minds. Instead, the team used artificial intelligence to tackle the problem. The A.I. identified the thirty most diverse HA proteins of the past hundred years; afterward, the researchers created an HA cocktail vaccine containing mushroom caps from all of these viruses. Their theory is that, across so many diverse strains, no single mutable part will be consistent. Instead, it’s the conserved features that will stand out. A fugitive in disguise might evade capture, but arrange thirty photos of him in thirty different disguises on a detective’s pinboard, and persistent, identifying features might reveal themselves: one shoulder that’s a little higher than the other, or a nose that’s tilted by half a degree.
During the coronavirus pandemic, Operation Warp Speed has funnelled billions of government dollars into vaccine development. The pursuit of a new flu vaccine is a smaller affair. Glanville grew up in a small town in Guatemala—his father owned a bed-and-breakfast, and his mother worked as an artist—and Centivax, the company spun out from Distributed Bio to focus on the flu vaccine and other efforts, does some of its research there, to save on costs. Glanville’s brother, a construction worker, helped the company build a testing facility for the animal trials its researchers conduct on pigs. In 2019, after administering its vaccine to the pigs, Glanville’s team tested the resulting antibodies against influenza strains from 2009 to 2015. The antibodies neutralized all six seasons of the flu, even though the vaccine had been designed using HA proteins that were only as recent as 2007: its immunity was predictive. Two independent labs, funded by the Gates Foundation, have since replicated their results in pigs and ferrets.
Centivax is now conducting so-called live-challenge trials, in which scientists vaccinate pigs and ferrets and then deliberately expose them to flu viruses. So far, the results are promising, with some groups of protected animals showing no symptoms and some groups of unprotected animals losing more than ten per cent of their body weight. If more studies go according to plan, human clinical trials will follow. Meanwhile, they’re also developing a universal flu vaccine for the pig market.“There’s financial reasons to do it,” Glanville said. “But also, pigs are your major recombination species. It would probably be a requirement for eradication of influenza. Otherwise, it just keeps popping up.”
At one point last year, there were dozens of coronavirus vaccines in development; not all of them have panned out. With vaccines, it’s best to bet on all the horses. Last year, the horse closest to the finish line was Multimeric-001 (M-001), a universal flu vaccine made by the Israeli company BiondVax. That summer, BiondVax’s C.O.O., Elad Mark, gave me a FaceTime tour of the site’s twenty-thousand-square-foot manufacturing facility, in the Jerusalem BioPark. The trials for M-001 were in Phase III, when researchers give a vaccine to thousands of people and perform tests to see if it’s safe and effective. Wearing white protective garb head to foot, Mark walked me through the space, which was crowded with shiny metal machinery for protein production, purification, dilution, and syringe-filling crowded the labs. Most of the action was in the first room, where one white-suited researcher sat at a computer workstation and another collected samples from a tank. “This is Haya,” Mark said, gesturing to the sample collector. “She is our singer. She has an amazing voice, a beautiful voice.”
BiondVax and several other promising vaccine companies have taken the approach of targeting T cells, which are another line of immune defense, alongside antibodies. If a virus gets past a host’s antibodies and infects a cell, the immune system attempts to kill that cell, ripping up the viruses it finds and ingesting their particles, called peptides; it then shows these bits to other parts of the immune system as a kind of heads-up. T cells attend to the warnings; the next time they see those peptides, they attack the cells that contain them. T cells are an appealing object for flu-vaccine research because, unlike antibodies, which target proteins on the surfaces of viruses, T cells identify bits of proteins from the viral interior, which tend to remain constant from one strain of flu to another. They look beneath the mushroom.
Whereas Glanville, at forty, is committed to studying the flu’s past, Ruth Arnon, the longtime chair of BiondVax’s scientific advisory board, has lived it. Now eighty-eight, she began her research in 1957. At first, she studied the manufacture of synthetic peptides. She and her doctoral adviser were curious about whether these artificial bits could elicit antibodies against entire natural proteins, and found that they could. Focussing on T cells, she successfully developed a universal flu vaccine, tested successfully on mice, in the nineteen-eighties, before launching into human research with her colleague Tamar Ben-Yedidia, in the nineties. For two decades, they surveyed the genetic sequences of many strains of flu, and created synthetic peptides that evoked nine conserved regions from three proteins: HA and internal proteins called NP (nucleoprotein) and M1 (matrix 1).
The resultant vaccine, M-001, had been through five Phase II trials with promising results. In one study, the antibodies of elderly participants who’d received the vaccine in 2011 and 2012 were found to neutralize the strains of the flu that struck during the winter of 2014-15. In Phase III, medical professionals administered the vaccine to thousands of participants, age fifty and older, across Eastern Europe in 2018 and 2019. Arnon has become accustomed to sitting on the edge of her seat. Still, last year, as the trial was wrapping up—and as covid was bearing down across Europe—she was in an agony of anticipation. “The tension is dramatic,” she told me. If the results from the trials showed that M-001 offered greater than seventy-per-cent protection against that year’s flu strains, then the F.D.A. and European Medicines Agency could approve it for commercial use within two years—perhaps even sooner, given the risk of a coronavirus-influenza “twindemic.”
Haya’s singing ability aside, the real star of my BiondVax tour was a three-hundred-litre cylindrical fermentation tank. It was about the size of a home fridge, and, inside, genetically engineered bacteria produced the vaccine, a string of peptides. The fermentation process was not much different from making a microbrew; from start to syringe, Mark explained, production took about two days, allowing the facility to produce ten to forty million doses a year. Building it, Mark said, had cost about fifteen million dollars—a risky investment before the vaccine had been approved. He argued that the jump start would avoid a delay to market: “If we waited to get approval, and only then designed, constructed, and validated the facility, it would take us at least three years.”
In 2020, people in South America, Africa, Asia, and Australia showed up at their doctors’ offices with flu-like symptoms. They subjected their nasal cavities to cotton swabs; medical personnel chilled the samples and sent them to National Influenza Centres in Melbourne, São Paulo, Cape Town, and elsewhere. Researchers at those centers genetically sequenced the flu viruses that they found, and sent a selection of samples on to the W.H.O., which performed further testing. From these data, the W.H.O. selected, in February of this year, the flu strains to target with the Northern Hemisphere flu shots. Those shots are available now; perhaps you got one. Meanwhile, the same process unfolded on a staggered schedule in the Global North. In September, the W.H.O. recommended targets for the Southern Hemisphere’s next flu season.
Every six months, it’s the same gamble. The virus turns its wheels of fortune, rearranging its genes; we make our preparations and hope for the best. Arnon, Glanville, and other researchers look to change the game. (Researchers at several companies, including Pfizer and Moderna, are pursuing mRNA flu vaccines, but it’s early days.) The flu, for its part, threatens to mutate in ways for which we’re unprepared.
For now, we remain on the cusp of a breakthrough that could neutralize our foe. A few months after my tour, BiondVax announced the results of its Phase III trial. The researchers found that, in a randomized, double-blind test involving more than twelve thousand people, M-001 did not significantly reduce the incidence or severity of the flu. “For now, the M-001 vaccine candidate remains on the shelf, and we may return to it in the future,” Joshua Phillipson, the company’s director of business development and investor relations, told me. A new C.E.O. has been appointed; in a recent letter to shareholders, he promises to diversify the company’s risk “along several axes” in the future. It has begun a new collaboration with the Max Planck Society, in Germany, focussed on new immune molecules called nanobodies. Arnon, needless to say, was disappointed. On the other hand, she knows from experience how slow the going can be right up until the finish line. In 1996, the F.D.A. approved Copaxone, a drug Arnon developed, for the treatment of multiple sclerosis. It was a huge leap forward in treatment for the disease, and ushered in what felt like an instant transformation for patients all across the world. For Arnon, however, the approval was the culmination of a long and arduous process; developing the drug had taken twenty-nine years. “The first time that you see that you can prevent infection with something that you synthesize by your own hands, there is an excitement that can’t be really explained in words,” she said. Still, “you have to be patient. And you have to live long.”
