Wuhan: the origin we still cannot explain six years on

Science has reconstructed SARS-CoV-2 with extraordinary precision, yet six years on it has still not managed to reconstruct its first jump into a human being. Zoonosis remains a perfectly plausible explanation, even though the animal that started the chain never turned up and the information needed to rule out a research-related accident is still incomplete. Between bat coronaviruses, reverse genetics, international funding and the opacity of institutions that should have preserved every record, one question remains a good deal more uncomfortable than some were willing to admit in 2020: what actually happened in Wuhan.

More than six years have passed since those first cases of pneumonia appeared in Wuhan and went on to cause a pandemic, millions of deaths and a temporary transformation of the way we live. Yet one question is still not satisfactorily settled: where exactly did SARS-CoV-2 come from?

During the first months of 2020 the natural-origin hypothesis took hold very quickly, most likely through the jump of an animal coronavirus into humans. It was a perfectly reasonable possibility and it remains the explanation with the widest support across a substantial part of the scientific literature. What I always found harder to accept was the speed with which some ruled out a laboratory-related accident almost entirely.

I read molecular biology many years ago. I am a biochemist, not a microbiologist, and I have no intention of presenting myself as one, but I know molecular genetics well enough to understand a recombination, a restriction enzyme, an infectious clone, or what it means to rebuild a virus from complementary DNA. That is precisely why one argument I heard repeatedly in 2020 never quite convinced me: that if the virus had been engineered, the genome would carry the scars of that work. The trouble is that by then this had not necessarily been true for years.

The question is not whether Wuhan had a BSL-4 laboratory. It is which viruses were being handled, where, and under what conditions.

The Wuhan Institute of Virology had a BSL-4 laboratory, the highest level of biological containment. A facility of that kind is designed to handle especially dangerous pathogens through differential pressure systems, air filtration, decontamination, restricted access and, in certain cases, pressurised suits with an independent air supply. On paper, everything is designed so that a biological agent never leaves the building.

But there are still people inside the system, and people make mistakes. More importantly, the fact that the institute had a BSL-4 does not mean that all of its coronavirus research was carried out at that level of containment. Part of the work was done in lower-grade facilities. The interesting question, then, is not how a virus could escape a BSL-4 fortress, but which virus was being handled, where, and under what conditions.

Investigating how a virus might become more dangerous carries the risk of creating the very thing you are trying to anticipate.

This is where the notorious phrase gain of function appears. It has become a political term, although in biology it simply means that an organism acquires a new function or enhances one it already had. In a virus that can mean a greater capacity to replicate, better use of a cell receptor, a wider host range or increased transmissibility.

Why would anyone want to do that? Because knowing which changes an animal coronavirus needs in order to infect humans efficiently can help us watch for those same changes in nature, develop antivirals or prepare vaccines. The scientific logic is real. The problem is that in order to learn experimentally how a virus can be made more dangerous, we may end up creating precisely what we were trying to anticipate.

This debate did not begin with the pandemic. In 2014, under the Obama administration, the United States temporarily suspended federal funding for certain influenza, SARS and MERS research where there was a reasonable risk of increasing pathogenicity or transmissibility. It was not a blanket ban on all gain-of-function work, as has often been claimed, but it shows that the biosafety problem was already a serious concern years before Wuhan.

American public money ended up funding coronavirus research in Wuhan. That much is documented.

The National Institute of Allergy and Infectious Diseases, then led by Anthony Fauci, funded EcoHealth Alliance to study the emergence risk of bat coronaviruses. EcoHealth, led by Peter Daszak, subcontracted part of that work to the Wuhan Institute of Virology. There is therefore a documented chain of American public funding that ended up paying for coronavirus research in Wuhan.

From there to claiming that Fauci knowingly funded the creation of covid is an enormous leap, and there is no evidence sufficient to make it. It is a separate matter that part of the work received American money and that the Inspector General of the Department of Health later concluded that oversight of certain grants and subcontracts had been deficient.

That research also involved chimeric coronaviruses. The phrase sounds a great deal worse than it is. It essentially means taking part of one virus and inserting it into another to find out what it does. You can, for instance, take the backbone of one coronavirus and replace its spike protein with that of another, then observe whether the new spike allows the virus to use human receptors or changes its behaviour. Even then this was neither science fiction nor clandestine work. Ralph Baric, a researcher at the University of North Carolina, had been publishing this kind of research for years and collaborating with Chinese scientists.

The absence of genetic scars does not allow engineering to be ruled out.

Here, to my mind, lies one of the most interesting elements of the whole story. Coronaviruses have enormous genomes for a virus, around thirty thousand nucleotides. At the beginning of this century Boyd Yount, Ralph Baric and other researchers developed reverse genetics systems capable of splitting the genome into fragments of complementary DNA, modifying them and then reassembling them to recover an infectious virus.

In 2002 they published a system called No See’m cloning. The point of the method was precisely that the sites used to join the fragments could disappear once the reconstruction was complete. A year later the same group published the reconstruction of SARS-CoV from a full infectious cDNA. We are talking about 2002 and 2003.

So when, in 2020, the absence of certain genetic scars was used as an argument to rule out engineering, the case was being oversimplified. Molecular engineering had had procedures capable of making certain modifications without leaving the old cloning signatures for almost two decades. That an intervention can be carried out without leaving those marks does not prove that SARS-CoV-2 was engineered. It simply means that the absence of scars does not allow us to exclude it.

Reverse genetics makes something conceptually extraordinary possible: starting not from the virus but from its sequence. From there we can build DNA, obtain RNA and end up recovering an infectious virus. We have travelled the road backwards. It is a formidable tool, although it greatly complicates any later forensic analysis, because a modification made in a laboratory can reproduce exactly a change that might also have arisen through natural mutation or recombination.

Neither the laboratory hypothesis nor the zoonotic one has yet found the decisive piece.

The engineering hypothesis has one enormous problem: we know of no publicly documented virus held at the institute before 2019 that can be identified as the immediate ancestor of SARS-CoV-2. RaTG13, the most famous coronavirus in its collection, shares around 96 per cent of the genome, but across thirty thousand nucleotides that means more than a thousand differences. That is an enormous gap, and the missing piece has still not appeared.

What is striking is that the zoonotic explanation has not closed the circle either. More than six years on we still have no specific animal of which we can say that it passed SARS-CoV-2 to the first human being. The World Health Organization holds that the weight of available evidence currently favours a zoonosis, while acknowledging at the same time that it has been unable to assess the possibility of a laboratory accident because information is still missing. That reads more like a political answer than a scientific one.

There is also a third possibility that is frequently forgotten. The virus could have been entirely natural and still have reached the population through a research activity. A researcher collects bat samples, transports them, processes them and is accidentally infected during one of those stages. In that scenario there would be no genetic engineering at all and yet the start of the epidemic would still be connected to scientific research. The question natural or artificial is too narrow. What matters is reconstructing the chain of events that carried the virus from its reservoir to the first infected human.

Opacity does not prove an origin, but it does make it impossible to close the argument.

In time another rather unedifying matter emerged. Emails from David Morens, a scientific adviser at the institute, showed the use of private accounts for certain communications and conversations about how to avoid freedom of information requests. This does not prove that Fauci knew the virus had come out of Wuhan, but it is hard to argue that such behaviour helped build confidence. When documents are missing, suspicion moves straight into the empty space.

Something similar applies to China. If in those first weeks the authorities suspected that the virus might be connected to a scientific activity, they had powerful reasons not to admit it. Such an admission would have meant accepting the political cost of a biosafety failure and opening an international discussion about financial liability on a scale that is difficult to imagine. There is an even more disturbing thesis: that Beijing had no interest in halting international travel while China alone bore the economic cost of the epidemic. The economic logic exists, but here it is wise not to run ahead of the evidence. We know that China withheld information, suppressed some early warnings, and that the virus was circulating internationally before Wuhan was sealed off. What we do not know is whether allowing that spread was a deliberate decision or simply the consequence of having tried for too long to conceal a problem that was slipping out of their hands.

My subjective probability of a research-related accident is higher today than it was in 2020.

After this accumulation of facts, doubts and behaviour that is hard to explain, the question that really matters arrives: what do I think happened? I cannot prove that SARS-CoV-2 came out of the Wuhan Institute of Virology, and neither can I prove that it was genetically engineered. Anyone stating either of those as scientifically established fact is going beyond the available evidence.

If I am asked today what probability I assign to a research-related accident compared with what I would have assigned in January 2020, my answer has changed considerably. Not because of a single genetic signal, but through accumulation. Wuhan housed one of the world’s leading centres for bat coronavirus research. Its scientists had spent years collecting and characterising them, used reverse genetics, built chimeric viruses, worked under different levels of containment, and part of their research was funded through a system that later turned out to be inadequately supervised. Add to that the impossibility of accessing certain records, and the existence, for many years, of techniques capable of making some modifications without leaving the old molecular scars.

Professional habit makes me think about these things in fairly Bayesian terms. We began in 2020 with a prior probability, and new data have appeared since. My subjective probability of a research-related accident is now clearly higher than the one I would have assigned at the start of the pandemic, even though none of those data points proves it on its own.

Viruses keep no memory of what we do to them. Laboratories should.

We may never know exactly what happened. It is also possible that the virus jumped from an animal to a human being and that the presence in Wuhan of a large coronavirus research centre was an extraordinary coincidence. But even if that were proved tomorrow, an uncomfortable conclusion would remain: over recent decades molecular biology has developed tools capable of manipulating viruses with extraordinary precision, while our systems of control, traceability and supervision have advanced a good deal more slowly.

We can reconstruct genomes, swap proteins, introduce mutations and recover infectious viruses from a sequence. Future biosafety cannot be limited to pressurised suits, airtight doors and filters. We need complete records, sample traceability, deposited sequences, independent audits and transparency in funding.

Perhaps we spent too long looking inside the virus for a scar that would solve the mystery. After six years my impression is that, if the answer exists, it was probably in the records, the samples and the laboratory notebooks. Viruses keep no memory of what we do to them. Laboratories should.