The Vault Guy

Leonard Rome’s lab discovered an odd, abundant component of cells in the 1980s—and he’s still trying to figure out what it does

By John Travis

Vaults were first visualized in 1986 under an electron microscope. (Image courtesy: N. Kedersha and L. Rome, Journal of Cell Biology (1986) 10.1083

This article was originally published by Science

Leonard Rome switches off the overhead light in the small room, leaving it illuminated only by a computer monitor and the fluorescent screen at the base of a towering electron microscope. Qing Lou, a Ph.D. student who works with the University of California, Los Angeles (UCLA) biologist, points to some ovoid smudges within the circular green glow of the microscope display. With a twist of a dial and a click of a mouse, she brings the shadows into focus and snaps a picture. Dozens, maybe hundreds, of barrel-shaped particles suddenly fill the computer monitor.

“There they are,” Rome says, like a proud father showing off his children.

These are vaults, enigmatic cellular structures that he and his then-postdoc Nancy Kedersha discovered back in 1986, when Rome was a new dad with bushy black hair and a Tom Selleck–style mustache, and Ronald Reagan was still the U.S. president.

Vaults, he and others would show, are the most massive particles made naturally by human cells and among the most abundant. Most of our cells have roughly 10,000 of the structures, with the number rising to perhaps 100,000 in certain immune cells. Many other animals make them, too. Their abundance—and the resources cells must pour into making them—suggests vaults have some essential function. But despite decades of work by Rome and other “vaulters,” their purpose is unknown. “It’s a real puzzle,” says Joana Vidigal, a biologist at the National Institutes of Health (NIH) who recently probed the role of RNA found inside vaults.

Over the decades various hypotheses have been proposed, including that vaults help ferry things around inside cells or clear toxins. And one by one, promising ideas were dismissed or lost momentum as supporting evidence failed to materialize. Initially enthusiastic about Rome and Kedersha’s discovery, NIH lost interest in funding basic research on vaults as the years wore on without answers. “There were periods in my career when I was depressed,” Rome says.

In the mid-1980s, Nancy Kedersha (bottom, right) and others in the lab of Leonard Rome (top, second from right) pondered their discovery of vaults. (Image courtesy: N. Kedersha and L. Rome)

Yet Rome’s fascination with vaults hasn’t faded, even as other researchers—including Kedersha—moved on. And now, with help from other funders and labs, he has turned from basic research on vaults to studies of how they might be exploited in medicine and other fields, as nanoscale vessels for delivering therapies and more.

The ones in the microscope on this day were produced in genetically modified yeast and loaded with an immune signaling molecule called CCL21 that has shown tumor-fighting potential. Vault Pharma, a company co-founded by Rome that works out of an incubator space at UCLA, hopes to start a clinical trial in late-stage cancer patients as soon as this year. It would mark the first time synthetic vaults have been injected into humans, and perhaps the beginning of a new turn in the spotlight for these mysterious organelles.

Rome is now 75, with three grown sons. His hair has turned white and his impressive mustache is long gone. He retired in 2020 but is now officially back, unpaid, at UCLA with one lab bench, a postdoc, and some undergraduates under his wing. (Nobody noticed he left, he jokes, because the COVID-19 pandemic had everyone working at home.)

(Image courtesy: A. Fisher / Science; Data courtesy: H Zou et al., Structure, 26:4 (2018))

From a small office at UCLA’s California NanoSystems Institute (CNSI), which he helped design and, for a period, directed, he spends his days consulting on vault research with other UCLA labs, as well as the small Vault Pharma team. He occasionally fields calls and emails from scientists learning about the structures for the first time. And Rome is still forming new collaborations. In one audacious effort, he and a lab in St. Louis have stuffed viruses into vaults in a bid to solve a major problem in gene therapy.

He’s aided by his infectious enthusiasm. UCLA environmental engineer Shaily Mahendra years ago became a vault convert, loading the synthetic ones with enzymes designed to break down groundwater contaminants. She and Rome now have backing from foundations and federal agencies to see whether the encapsulation helps the enzymes work better and last longer when dispersed across land or in water. “No one says ‘no’ to Lenny,” Mahendra laughs.

Rome debated majoring in art or chemistry as an undergrad when his art teacher gently suggested the latter was a more practical option. He later switched to biochemistry, which at the time involved a lot of cell biology. His first lab, at UCLA, focused in part on a cellular structure: lipid spheres known as clathrin-coated vesicles. They transport enzymes to organelles called lysosomes to help break down other molecules.

When Kedersha joined Rome’s group in 1983, she was fresh out of a doctoral program at Rutgers University, but had previously worked for years as a lab tech. The two were nearly the same age and quickly developed a rapport. Rome assigned her the task of comparing the vesicles entering lysosomes with the ones exiting. When she used a gel to separate different kinds of vesicles based on their electrical charge, she unexpectedly saw an additional band, indicating structures of another kind. Unable to identify them, she tried flooding her cell preparations with a heavy metal stain before looking at them with an electron microscope.

Like islands in an ocean, white ovoid outlines appeared amid the background puddle of stain—vaults. The stain was rolling off the structures, revealing their shape. Follow-up experiments revealed that more traditional stains for the lipid membranes common to many cell components did not bind to vaults’ protein-only shells, one reason they had gone undiscovered for so long.

The lab held a competition to name the unidentified cellular objects, which were shaped like tiny U.S. footballs. Some suggested grenades or raspberries. Romesomes was a popular choice, which Rome quickly countered with Kedershacules. But when the postdoc remarked that she thought the objects’ outline resembled the vaulted ceilings in cathedrals, the name stuck.

A heady time followed, lab members recall. Some of the women donned fake mustaches that mimicked Rome’s as they all toyed with possible explanations for what they had found. Kedersha discussed vaults and other science with Rome on power walks he was doing to lose weight. “I loved working with Lenny,” she says.

Whenever he met another biologist, he’d pull her vault micrographs out of his wallet, where they nestled with his kids’ pictures, and ask what the scientist thought. One suggested a contaminating virus. A hasty experiment disproved that but revealed another surprise: Vaults contained snippets of RNA, albeit ones far too short to be a viral genome.

Confident the lab had something unprecedented, Rome mailed the work to one of the top basic research journals, Cell. Its famously opinionated editor, Benjamin Lewin, rejected the submission without sending the manuscript out for review. The journal would be happy to reconsider, he added, once the lab identified a function for vaults. Crushed, the team pivoted to a more specialized cell biology journal, and the paper was quickly published.

Over the years, Rome’s lab and a few others have built up a more detailed picture of vaults. About 10 million times smaller than a football, a vault is still large for a cell—about three times the mass of the much better known ribosomes that translate RNAs into proteins. Each is made up of 78 copies of the elongated major vault protein (MVP), aligned somewhat like staves in a barrel. Inside are clumps of two other proteins and the short vault RNA (vRNA).

The genes for these vault components are found in diverse eukaryotic organisms—those that pack their DNA in the nucleus and share other cellular features—with notable exceptions that include insects, plants, and fungi. Bacteria also seem to lack them. A 2013 study constructed a family tree of all the organisms known to have vaults and concluded they date back to a hypothetical last common eukaryotic ancestor billions of years ago. Over evolutionary time, some lineages evidently lost them.

Based on their barrel shape, Rome initially wondered whether vaults pick up and released cargo within the cell, perhaps plying routes from the nucleus to other locations. Vaults seemed to gather around gateways known as nuclear pore complexes and might fit their opening. “It’s a perfect match,” he told Science News in 1996.

Kedersha had her own views. She was struck by her observation that macrophages, amoebalike cells in the immune system, had the most vaults of any human cells. She also recalled another lab’s discovery that slime molds—simple organisms that can form a blob and creep along the forest floor—have three copies of the gene for MVP, compared with one in humans, suggesting they need more vaults. She suspects vaults have a role in cell locomotion, perhaps by regulating expression of other proteins that form extensions that help cells get around.

Kedersha and Rome’s paths diverged 5 years after that first vault paper. Kedersha joined a biotech firm called ImmunoGen, honing her talents at photographing and characterizing cancer cells with antibodies and stains. Later she returned to academia to teach and work in Paul Anderson’s lab at Brigham and Women’s Hospital. There, she discovered another previously unknown component of cells called stress granules, cytoplasmic “vortexes” of messenger RNAs (mRNAs) and proteins created when cells face challenges like energy depletion or viral infection. The granules, which seem to sort mRNAs so they can later be translated into proteins or degraded, were the main focus of Kedersha’s research career.

In contrast, Rome went all in on vaults, shedding other lab projects. In 2014, his lab showed vaults are made in an unusual way, by an assembly line of ribosomes. As each new copy of MVP is synthesized, the strand immediately layers onto the ones made before, and the vault shell slowly emerges as if from a biological 3D printer.

The rapid assembly process means there’s virtually no free MVP left in a cell, making it difficult to investigate potential vault functions by targeting and inactivating the proteins. The team also found that MVPs don’t form fixed chemical bonds with their neighbors. Instead, weaker noncovalent interactions bind them into a shell. This lets vaults “breathe,” exposing gaps that might allow relatively small molecules to get inside.

Meanwhile, other labs were racking up provocative and confounding findings of their own. In 1996, European cancer researchers investigating a protein that’s unusually abundant in drug-resistant cancer cells cloned the gene and discovered that it resembled the gene for MVP in rats. Working with Valerie Kickhoefer in Rome’s lab, the group then found that the drug-resistant cancer cells generated many more vaults than nonresistant ones, suggesting the structures might sequester or expel chemotherapies. But to investigators’ frustration, stopping the production of vaults, or MVP, didn’t make the cells more susceptible to drugs.

Underscoring the mystery of vaults, in 2002 a Dutch team disabled the gene for MVP in a line of mice. The rodents lacked vaults yet developed normally, seemed to be healthy, reproduced, and lived as long as regular mice. Subsequent knockouts of the two other vault protein genes also left mice unscathed. And Vidigal’s team at NIH recently disabled the mouse vRNA gene—only to find that those rodents, too, display no major changes, they report in a preprint this week.

“The pathologists found no problems in any of the tissues analyzed,” Vidigal says. “I have never seen a case where a set of genes is exceptionally conserved, highly expressed, produces a huge structure, and when you get rid of them, you see virtually nothing, even at the molecular level. It’s crazy.”

Back in 1987, Rome’s lab easily won an NIH grant to study the newfound objects, with the highest grant score he would ever receive from the agency. A second NIH proposal took two resubmissions before getting approved. A third was finally granted in 1996, after three tries, but when it ended a few years later, the agency rejected Rome’s subsequent vault applications. As NIH developed vault fatigue, Rome got creative, turning to the National Science Foundation, private foundations, companies, and others for funding. He found success by focusing less on what vaults might naturally do and more on what could be done with artificial ones.

In 2001, his lab reported something unexpected: When they added the rat MVP gene to moth cells, they produced virtually normal, but empty, vaults. The three other components were not needed to form the structures. That got him wondering: Could these empty vessels be put to use? At the time Rome, who had become a dean of research at UCLA’s school of medicine, was helping develop the university’s part of a new $100 million statewide nanotechnology effort backed by California’s governor—what would become CNSI. With an apparent easy way to make nanocontainers, Rome was soon talking to anyone looking to deliver enzymes, drugs, or anything else into the body. Vaults are already in human cells after all, so Rome argues they are unlikely to provoke an immune response, unlike vessels made from foreign substances.

UCLA cancer immunologist Steven Dubinett joined the vault bandwagon soon after. He had been pioneering CCL21 therapy for cancers, based on evidence that the protein stimulates immune cell proliferation and other responses. Dubinett’s initial strategy was to inject CCL21 directly into tumors. That elicits a strong antitumor immune response, but it is short lived. Hoping to avoid repeat injections, his lab was looking into more enduring approaches such as engineering a patient’s own cells to make CCL21. Rome suggested what might be a simpler solution: Stuff CCL21 inside vaults, where it might leak out slowly.

There are now a few ways to add molecules inside synthetic vaults, but Kickhoefer developed what has become the favored method. One of the proteins inside vaults, VPARP, has a domain called INT that binds to an interior-facing bit of MVP. When Kickhoefer grafted INT onto other proteins, by adding the DNA for it to their genes, the modified molecules slid into the empty vaults.

More than a decade ago, a biotech company licensed Rome’s work from UCLA, hoping to use vaults to deliver cancer drugs. But it abandoned the effort, and in 2013 UCLA reclaimed the intellectual property and Rome launched Vault Pharma to pursue this strategy himself with CCL21 and other potential drugs. Then a corporate takeover in 2018 dealt a devastating blow. Vault Pharma had partnered with a firm called Protein Sciences to develop a commercial process to make loaded vaults in insect cells. But weeks before production of CCL21 vaults was scheduled to start, a company that had little interest in the project bought Protein Sciences and the vaulters were suddenly cut off.

Vaults (green dots) populate most animal cells by the thousands, including these monkey kidney cells. (Blue is DNA in the cell nucleus.) (Image courtesy: N. Kedersha)

It’s taken years to recover, says Oliver Foellmer, Vault Pharma’s CEO. With help again from Kickhoefer, now semiretired herself, Vault Pharma crafted an arguably simpler, better way to mass produce vaults to clinical trial standards. They add the human MVP gene and the gene for desired cargo, tweaked to include INT, into yeast cells so that loaded vaults are made in one fell swoop. “We can make vaults now by the gram, perhaps kilograms,” Rome says.

Yet Vault Pharma still needs to raise $10 million or so to conduct its CCL21 cancer trial. The study is pivotal to establishing vaults’ safety and to keeping the company going, Foellmer says. “We’re in the ‘valley of death,’” he explains, referring to the treacherous stage when a startup biotech with dwindling funds must demonstrate a major clinical advance to survive. Amid a broad cooldown in biotech investing, venture funds and larger drug and vaccine companies have been reluctant to pick up the trial’s tab.

This virus inside a vault represents a new strategy for gene therapy. (Image courtesy: Wandy Beatty / Washington University in St. Louis)

Undaunted, Rome has other possibilities in mind for vaults. One is helping gene therapy address a major hurdle. The strategy often relies on typically harmless viruses known as adeno-associated viruses (AAVs) to deliver therapeutic genes. But it’s tough to target AAVs to individual tissues. Worse, AAVs often don’t work because many people—40% by one estimate—have already been exposed to the viruses naturally and make antibodies targeting them. “The preformed immunity seems a real showstopper,” says David Curiel, a gene therapy researcher at Washington University in St. Louis. And even patients who don’t have detectable levels of AAV antibodies can have dangerous, even deadly, immune reactions if physicians use high doses of the virus to ensure enough of a gene is delivered.

Several years ago, Curiel and his Ph.D. student Logan Collins were brainstorming ways to “stealth” the delivery of an AAV into a patient’s cells when Collins recalled reading about vaults. “It might have been middle school or high school, maybe on Wikipedia,” he says. “I was so fascinated by their ability as a capsule to hold things.” Attaching targeting molecules such as antibodies to the exterior of vaults might also allow them to be directed to specific tissues.

Curiel had heard of vaults, too, and Collins proposed they try to jam AAVs into one. The idea sat unexplored, however, until Curiel reached out to Rome about 2 years later. Rome, too, had long thought vaults could be a gene therapy vector or carry mRNAs for vaccines—substituting for the lipid nanoparticles in COVID-19 shots, for example. But he and others have had trouble getting synthetic vaults to hold gene-length DNA or RNAs.

Over a Zoom chat, Rome updated Collins and Curiel on vaults and they in turn described a potential way to get a virus to stick inside the structure: a “molecular glue” Curiel’s lab had previously developed. Collins used it to stick INT to the surfaces of AAVs so they latch onto MVP inside vaults. Then he mixed the viruses into a solution of synthetic vaults Rome had shipped from UCLA.

The surprisingly simple plan worked, the team reported in a preprint first posted in November 2023. Under the right conditions, an AAV or two would slip into a vault and stay there. And when those vaults were added to cells, the viruses released their genetic cargo, allowing it to be expressed. As important, when the cells were surrounded with AAV-targeting antibodies, the vaults slipped by them like a stealth aircraft. Curiel thinks the tactic has a real chance to solve the AAV immunity problem.

Other AAV gene therapy researchers find the approach provocative, if preliminary, given the team hasn’t shown the virus-laden vaults work in animals. “It’s a very nice, creative strategy,” says Selvarangan Ponnazhagan of the University of Alabama at Birmingham, who studies AAV therapies for cancers and other diseases.

Stanford University’s Mark Kay agrees. “I’d always wondered what happened to vaults,” he adds. But he and other AAV gene therapy researchers note that dozens of labs and companies are working on other approaches, some further along, to overcome the AAV antibody problem.

Leonard Rome, pictured here readying a grid that will hold a sample of vaults under an electron microscope, has recently taken to YouTube to spread the word about his lab’s befuddling discovery. (Image courtesy: Spencer Lowell)

The classic cell biology textbook, Molecular Biology of the Cell, hasn’t mentioned vaults in any of its seven editions, which span 40 years, says UC San Francisco cell biologist Bruce Alberts, an original and current author (and former editor-in-chief of Science). Alberts doesn’t doubt that vaults could be vital in some way, perhaps serving some very specific function that’s hard to replicate in the lab. “For example, vaults may be part of the massive war with viruses, enabling cells and organisms to survive a type of virus that we don’t even yet know anything about,” he says. But Alberts says he and his textbook co-authors don’t like to flood students with unknowns, so vaults haven’t made the cut.

Kedersha finds that omission “sad.” Even a few paragraphs on the vault mystery in the renowned tome might have inspired a reader to come up with a solution. “Science isn’t a compendium of facts, it’s a process,” she says. But she’s not losing sleep over it. The stress granules she discovered are in textbooks. “That’s a win,” she laughs.

As for Rome, he doesn’t regret devoting a career to chasing what some might see as a white whale. He still has a list of about a half-dozen potential vault functions that intrigue him. And he acknowledges one more possibility: Perhaps they don’t do anything. Instead, they may simply be what he calls “rocks,” caches of amino acids that a cell can mine in times of need. (Each MVP has more than 800 amino acids.)

Hoping to inspire a new generation of vaulters—and shake loose a few more ideas—he launched a series of online videos earlier this year. On the Vault Guy, his YouTube channel, Rome mixes jokes and cartoonish animations with serious discussions of scientific methods, funding, and publishing. With barely more than 100 subscribers to date, the series isn’t exactly going viral. But some viewer just might be the person who finally solves the great mystery.

“The optimist in me is that before I die someone will figure out what vaults do,” Rome says.