Login
Before COVID, an NIH lab chief proposed making a SARS-like bat virus more infectious
Vincent Munster and Michael Letko (photos from NIH and WSU)
By Lewis Kamb | U.S. Right to Know | October 15, 2026
On a Sunday morning in October 2019, the chief of the virus ecology unit at the National Institutes of Health’s Rocky Mountain Laboratories emailed a postdoctoral scientist in his lab with an idea.
Vincent Munster and his postdoc, Michael Letko, had spent that year building a system to rapidly test whether dozens of SARS-related bat coronaviruses could infect human cells.
That week, Munster read a new preprint from prominent virologists Vineet Menachery and Ralph Baric’s lab groups showing that the protease trypsin could unlock entry for certain coronaviruses that otherwise would not infect human cells.
Letko had forwarded him the paper two days earlier with the note, “Seems like more groups will be converging on this idea soon.”
“You should be able to make S with a more broader cleavage signature (like furin-like) to enhance infectivity,” Munster wrote back on Oct. 6, 2019. “You can include a motif like this R-X-[K/R]-R↓ Might be cool to get some synthesized.”
In virology parlance, “S” is the coronavirus spike protein, which allows the virus to latch onto the cells it infects, and R-X-[K/R]-R is a furin cleavage site motif, which can make viruses better able to infect cells and spread.
Munster was suggesting that his postdoc engineer a furin site into a coronavirus spike protein to make it more infectious.
Eleven weeks later, a coronavirus carrying a similar furin cleavage site emerged in Wuhan, China, spawning the COVID-19 pandemic. It remains the only known sarbecovirus, the SARS-related group of coronaviruses that Munster and Letko were then screening, to have one.
The email exchange has never been made public before. It appears in a 270-page batch of records recently released by the National Institutes of Health (NIH) to U.S. Right to Know in response to a request made under the Freedom of Information Act.
The request sought specific correspondence from Munster’s lab in Hamilton, Montana, from March 2019 through Jan. 22, 2020, the date when his group published one of the first papers showing that the new coronavirus used a human cell receptor called ACE2 as a doorway into cells.
Two new threads run through the records: Weeks before the pandemic began, a senior NIH virologist proposed in writing to build into a SARS-like spike protein the feature that would become the central argument for COVID-19’s laboratory origin: a furin cleavage site.
And, through the same year, his lab was asking virus hunters for unpublished bat coronavirus spike sequences, and getting them, including from researchers in the U.S.-funded consortium called PREDICT, whose New York-based lead partner, EcoHealth Alliance, had a long-standing collaboration with the Wuhan Institute of Virology (WIV) to collect and study SARS-related viruses.
Each goes to a different key point in the origins debate. The furin cleavage site is the single genetic feature most cited in the arguments that SARS-CoV-2 did not arise entirely in nature.
Unpublished sequences are the other: because the pandemic virus matches nothing in the published scientific literature, most laboratory-origin scenarios depend on some lab having held a virus the rest of the world had never seen.
Nothing in the records shows that Letko followed through on Munster’s suggestion to insert a furin cleavage site in a SARS spike protein, however. In fact, the furin cleavage site found in SARS-CoV-2 differs from the motif, or short sequence of amino acids, that Munster suggested.
Nor do the documents indicate that Rocky Mountain Labs possessed SARS-CoV-2 or a close relative, or that either Munster or Letko directly communicated with researchers in Wuhan while working on their project.
Their research also used synthetic gene fragments, not live viruses, employing particles that cannot replicate or spread. Such work generally poses lower biosafety risks.
The two then-unpublished bat virus spike sequences obtained from PREDICT, one from Malaysia and one from Uganda and Rwanda, are also too distantly related to SARS-CoV-2 to have given rise to it.
Letko, who is now an associate professor at Washington State University, declined to comment for this story, saying in an email “(y)ou will write what you write, regardless of the truth or what I have to say.”
Munster did not respond to requests for comment. He was indicted earlier this year for bringing mpox specimens into the U.S. without proper documentation and giving false statements to law enforcement officials as part of unrelated NIH research. He has pleaded not guilty.
The NIH press office also declined to answer questions about why the agency had not previously told the public that its scientists had solicited and obtained unpublished SARS-like coronavirus spike sequences, or proposed inserting a furin cleavage site into a SARS-like spike, before the pandemic.
Converging on the idea
Munster’s suggestion to add a furin cleavage site is the second known instance of scientists proposing to insert this distinguishing feature into a SARS or SARS-like spike within roughly two years before COVID-19 emerged.
In 2018, EcoHealth Alliance, Baric’s lab and researchers at the WIV had proposed the same kind of modification to the Defense Advanced Research Projects Agency (DARPA), in work that would have used live, replicating viruses. Their application, known as the DEFUSE proposal, was denied.
The concept is notable because the furin cleavage site that helped the virus cause the pandemic has become central to the debate over the pandemic’s origins.
Some proponents of a laboratory origin view DEFUSE as a blueprint for creating SARS-CoV-2, while prominent supporters of a natural origin have argued that deliberately engineering such a feature would have been unlikely and unnecessarily cumbersome. Munster’s email gives the lab-origin supporters another example of such a furin site proposal.
The revelation about Munster’s lab seeking unpublished sequences speaks to a different issue in the COVID-19 origins debate.
Because SARS-CoV-2 descends from no virus previously described in the scientific literature, any scenario in which it was engineered requires assuming that some laboratory held SARS-related sequences that had never been made public.
In a 2021 review of SARS-CoV-2 in Cell by 21 scientists called that implausible. “There is no rational experimental reason why a new genetic system would be developed using an unknown and unpublished virus,” they wrote.
But the records show a U.S. government laboratory doing just that two years earlier.
Letko’s platform was a new genetic system, and he sought unpublished sequences precisely because they were unpublished. Sharing them, he wrote in April 2019, “will surely increase the impact of our story by providing a proof of principle with unpublished sequences.”
Researchers in the U.S.-funded virus hunting program supplied two within eight weeks. One of the researchers, Simon Anthony, was also an author of the Cell review.
Alina Chan, a molecular biologist who has advocated for a thorough investigation of laboratory origin scenarios, recently reviewed the Munster-Letko email records at the request of U.S. Right To Know.
“What this shows is that different groups of virologists were converging on the idea of inserting cleavage sites in the spikes of potentially dangerous coronaviruses,” she said.
“Assuming a lab origin of COVID-19, if a lab in Wuhan hadn’t caused the pandemic, it’s plausible that a lab in the U.S. would’ve eventually followed the same path.”
Soliciting sequences
Munster’s emailed proposal didn’t come out of the blue. He and Letko were in the middle of a yearlong research project that the newly released records document from its start.
Seven months earlier, in March 2019, Munster and Letko set out to answer a question: of the hundreds of SARS-related coronaviruses in research databases worldwide, which ones could infect people?
Letko built a workaround that he and Munster described as a “rapid and scalable platform,” according to the records.
Rather than build each virus individually, they took only the receptor-binding domain — the fragment of the spike that grips onto the host cell’s receptor — and spliced it into a SARS-CoV spike backbone.
They then loaded these chimeric spikes onto harmless reporter particles and used them to infect cells expressing hACE2, the receptor SARS-CoV uses to enter the human airway.
A glow from the bioluminescent enzyme luciferase indicated successful cell entry. With Letko’s method, what typically had taken months to test now took about three weeks.
The limiting ingredient to test their method was the spike sequences. Letko had exhausted the SARS-like sequences already published. So, he and Munster started looking for others that hadn’t been, the records show.
On March 8, 2019, Munster emailed EcoHealth Alliance President Peter Daszak, Vice President Kevin Olival and Columbia University virologist Simon Anthony, attaching a confidential abstract with a chart and figures illustrating their project.
“While we have now tested all of the variation(sic) currently published for lineage B beta-coronaviruses for their ability to enter primate cells, we are seeking additional, unpublished lineage B beta coronavirus sequences to test in our platform,” Munster wrote.
He asked them not to circulate the attachment, because the work was “still unpublished and quite competitive.”
Of the three recipients, only Anthony, then one of the core leaders on PREDICT, appears in the released correspondence. He replied later that evening: “Hey! Looks super cool.”
Two months of negotiation followed, the emails show. On April 5, Letko thanked Anthony “for agreeing to share betacoronavirus lineage B spike sequences with us,” adding that unpublished material “will surely increase the impact of our story.”
On May 2, Anthony reported back from a PREDICT meeting in Vancouver, British Columbia. His team had detected “around 5 or 6 novel 2b viruses,” he wrote, but had “only managed to get spikes for 2 of them so far.”
Because neither had been published, Letko’s paper “would be the first report of these viruses in the literature,” so long as he credited PREDICT and its partners as co-authors, Anthony said.
Letko agreed. Two days later, Anthony sent the sequences. He identified them as PREDICT_CoV-51, isolate PDF-0663, from Malaysia (which Anthony described as “SUPER divergent. Could be really interesting,”) and PREDICT_CoV-43, from Uganda and Rwanda.
“Absolutely stellar!” Munster replied. “Will keep you posted on the results!”
Anthony was generally correct. A small fragment of the Malaysian virus had been deposited in GenBank, the NIH’s public sequence database, in 2016, and referenced the following year in a survey paper Anthony co-authored with Daszak and others. But neither its full spike sequence nor its complete genome was public.
Nothing at all was public about the African virus. In neither case could anyone outside PREDICT obtain the spike sequence, the part that determines whether a virus can enter human cells and the only part Letko’s platform could use.
A step further
Five months later, after reading the new preprint about trypsin, Munster proposed taking Letko’s project a step further.
For a coronavirus to infect a host cell, binding the receptor is only the first step. The spike must then be cut open by one of the host’s own enzymes before the virus can fuse with the cell and get inside. These enzymes that cut protein apart are called proteases, and each one targets different spots.
The protease a spike is built for largely determines what tissues it can infect. SARS-CoV depends on proteases that are not available everywhere in the body. Furin, however, is expressed in the cells of most tissues.
The Menachery-Baric preprint that sparked Munster’s idea had shown that adding trypsin from the outside could unlock entry into human cells for coronaviruses that otherwise couldn’t infect them.
Trypsin is a far less selective protease, cutting a protein after common amino acids rather than requiring a particular short sequence, or motif. What Munster proposed to Letko — engineering a furin site into the spike itself — would let the virus cut itself open.
Instead of waiting for an outside enzyme to do the job, a furin site could clear the hurdle in one step to “enhance infectivity,” as Munster wrote.
The idea had been proposed before.
EcoHealth Alliance submitted the DEFUSE proposal to DARPA, under a program called PREventing EMerging Pathogenic Threats, or PREEMPT. Its key partners included Baric’s laboratory at the University of North Carolina (UNC) and Zhengli Shi, the leading coronavirologist at the Wuhan lab.
The proposal aimed to introduce human-specific proteolytic cleavage sites, including furin sites, into SARS-related bat coronavirus spikes to evaluate and enhance their ability to infect human cells. DARPA declined to fund it, faulting the proposal over the risks of “gain-of-function” research.
Munster was not named on the proposal submitted to DARPA. But according to official congressional correspondence, he had been part of the DEFUSE team two months earlier.
In April 2024, Sen. Rand Paul (R-Ky.), then ranking member of the Senate Homeland Security and Governmental Affairs Committee, wrote to the U.S. Department of Health and Human Services (HHS) and DARPA saying he had obtained documents showing that “Vincent Munster of NIAID’s [National Institute of Allergy and Infectious Diseases] Rocky Mountain Laboratory was listed as a partner in the DEFUSE proposal presented at the PREEMPT Proposers Day,” a January 2018 DARPA event at which Daszak pitched the project to representatives of at least 15 federal agencies.
Rocky Mountain Laboratories, Paul wrote, “was at one point listed as a partner” alongside the WIV, UNC, Columbia University’s Lipkin Lab and Duke-NUS, a disclosure Paul said appeared to contradict former National Institute of Allergy and Infectious Diseases director Anthony Fauci’s testimony to Congress that NIH knew nothing of DEFUSE.
By the time the proposal reached DARPA that March, Munster and his lab were off it; the final list of partners included EcoHealth, the WIV, UNC, Duke-NUS, the USGS National Wildlife Health Center and Palo Alto Research Center.
Paul demanded from HHS records of “all communications between Vincent Munster and any individual from EcoHealth Alliance, WIV, University of North Carolina, Columbia University, or Duke-NUS.” Paul has not released his underlying documents, and neither HHS nor DARPA has publicly answered his requests.
The newly released NIH records show that about a year after DEFUSE was submitted, Munster asked EcoHealth for unpublished sarbecovirus sequences; that he received them from a virologist at Columbia’s Center for Infection and Immunity, the center Ian Lipkin directs; and that in October 2019 he proposed building a spike with a furin-like cleavage site to enhance infectivity.
Steven E. Massey, a professor of bioinformatics at the University of Puerto Rico and a researcher with DRASTIC, the independent group that first published the DEFUSE proposal, called Munster’s suggestion “significant,” saying it “shows the [furin site] idea was circulating outside the EHA-WIV-UNC group,” a reference to EcoHealth Alliance, the WIV and UNC.
Munster’s email does not mention DEFUSE. His idea appears instead as a reaction to the Menachery-Baric preprint, framed as a potential next experiment for his and Letko’s work.
And the motif he proposed is not the one found in the pandemic virus that emerged less than three months later. SARS-CoV-2’s site has alanine where Munster’s pattern called for lysine or arginine.
A suboptimal site
That difference is key. The furin site in SARS-CoV-2 substantially improves the virus’s ability to enter human airway cells and is widely regarded as a major contributor to its transmissibility. Its absence from every other sarbecovirus found in nature is also one of the observations lab-origin proponents cite most often.
The 2021 review in Cell also directly addressed whether the site could have been engineered. Its authors, including Kristian Andersen, Robert Garry and Simon, argued that the site’s own sequence weighs against a lab design.
SARS-CoV-2’s motif, RRAR, “does not match its canonical form (R-X-R/K-R),” they wrote, and is “suboptimal” compared with those of other human coronaviruses.
“There is no logical reason why an engineered virus would utilize such a suboptimal furin cleavage site, which would entail such an unusual and needlessly complex feat of genetic engineering,” the paper stated.
The only prior work inserting a furin site at that position in a SARS-CoV spike, they noted, had used an optimal sequence in a pseudotype system.
Massey disputes that framing. He said excessively efficient cleavage can destabilize the spike and reduce the virus’s ability to enter cells, so a site that looks suboptimal on paper may not be suboptimal for the virus.
The newly released records show Munster reasoning along the Cell authors’ lines nearly two years before their review appeared. The motif he proposed matches the canonical form they describe, not the one the pandemic virus carries, and his lab ran a pseudotype system of the kind they cite.
Read one way, the email corroborates their logic: that a scientist contemplating a furin-like site reached for the textbook motif, and SARS-CoV-2 does not have it.
Read another way, it establishes something the review did not address: by October 2019 the idea of adding such a furin cleavage site to a SARS-related spike to enhance infectivity was circulating not only in an unfunded DARPA proposal but also inside an NIH laboratory.
What was published
When Chinese scientists ultimately posted the genetic sequence of the novel coronavirus spreading in Wuhan on Jan. 10, 2020, Letko and Munster had everything they needed to test it in their system.
The platform did not require a live virus, only the sequence of the small piece of the spike that latches onto a human cell. Within hours of the sequence going up, Neeltje van Doremalen, another scientist in Munster’s lab, pasted it into an email to Letko and Munster.
Letko expected results by Jan. 26 or 27. They came back ahead of schedule. On Jan. 21, Munster decided where to send the final manuscript showing their research. “Nature it is,” he wrote to Letko. “Lets(sic) do 2 authors and speed this up.”
The paper went out that afternoon, to Nature and to the preprint server bioRxiv. Only 12 days had passed since the sequence appeared online.
The lab made the new method’s speed its selling point. Their method “reduces the time from the sequence being published to having real in vitro data to under 12 days,” the cover letter told Nature’s editors, “allowing the scientific community to gain critical information on virus tropism in almost real-time with the outbreak.”
The work drew notice at the top of NIAID.
“First data from my lab confirming ACE2 as the receptor for 2019-NCoV,” Munster emailed an NIAID coronavirus response list that afternoon.
Greg Folkers, then chief of staff to then-NIAID Director Anthony Fauci, replied: “When this hits bioRxiv might be worth a tweet.”
The paper that was submitted gives no sign that Letko took up Munster’s suggestion to engineer a cleavage site into a spike. But it does show that the scientists used the protease trypsin rather than the furin Munster had proposed.
The paper’s second figure is titled, “Trypsin enhances lineage B entry in various cell lines.” It reports that bat coronavirus spikes that could not get into human, primate or pig cells on their own did get in once the particles were bathed in the enzyme.
The trypsin experiments appear nowhere in the project before Munster and Letko’s October exchange about the Menachery-Baric preprint. They are also absent from the abstract that Munster previously circulated in the spring. The published paper cites the Menachery-Baric preprint that sparked their exchange conversation.
Neither of the sequences that Anthony sent in May 2019 appears in the paper, and neither he nor PREDICT are acknowledged. The only trace of PREDICT in the 18-page manuscript is a footnote citation.
The panel of viruses the paper did report testing consisted of 29 receptor-binding domains obtained from a public database administered by the NIH.
One of the two viruses resurfaced a year later, when Columbia’s Center for Infection and Immunity deposited its complete genome in GenBank on July 6, 2020, five and a half months after the preprint that was to have carried it.
In February 2021, the bat coronavirus from Uganda and Rwanda also appeared in the journal Virus Evolution, in a study of how SARS-related viruses acquired the ability to use the human ACE2 receptor.
Letko was second author, Munster was listed near the end, Anthony was the senior author and the Ugandan and Rwandan field scientists and PREDICT leadership were included as co-authors, as Anthony had said in 2019 would be required. Run through Letko’s platform, the virus could not enter human cells through ACE2 at all.
The Malaysian virus, which Anthony had called the more interesting of the two, has never appeared in a published study. Nothing in the records says why. Anthony, now a professor at the University of California, Davis, did not respond to requests for comment.
On May 24, 2021, two years after he emailed the virus’s spike sequence to the Montana lab, Columbia deposited its complete genome in GenBank on behalf of the PREDICT consortium, crediting Malaysia’s Sabah Wildlife Department and EcoHealth Alliance.
Lewis Kamb is an investigative reporter who specializes in using freedom of information laws and public records to uncover wrongdoing and hold the powerful accountable.
