Moving beyond metagenomics to find the next pandemic virus

I was asked to write a commentary for the Proceedings of the National Academy of Sciences to accompany an article entitled SARS-like WIV1-CoV poised for human emergence. I’d like to explain why I wrote it and why I spent the last five paragraphs railing against regulating gain-of-function experiments.

Towards the end of 2014 the US government announced a pause of gain-of-function research involving research on influenza virus, SARS virus, and MERS virus that “may be reasonably anticipated to confer attributes to influenza, MERS, or SARS viruses such that the virus would have enhanced pathogenicity and/or transmissibility in mammals via the respiratory route.”

From the start I have opposed the gain-of-function pause. It’s a bad idea fostered by individuals who continue to believe, among other things, that influenza H5N1 virus adapted to transmit by aerosol among ferrets can also infect humans by the same route. Instead of stopping important research, a debate on the merits and risks of gain-of-function experiments should have been conducted while experiments were allowed to proceed.

Towards the end of last year a paper was published a paper on the potential of SARS-virus-like bat coronaviruses to cause human disease. The paper reawakened the debate on the risks and benefits of engineering viruses. Opponents of gain-of-function research began to make incorrect statements about this work. Richard Ebright said that ‘The only impact of this work is the creation, in a lab, of a new, non-natural risk”. Simon Wain-Hobson wrote that a novel virus was created that “grows remarkably well” in human cells; “if the virus escaped, nobody could predict the trajectory”. I have written extensively about why these are other similar statements ignore the value of the work. In my opinion these critics either did not read the paper, or if they did, did not understand it.

Several months later I was asked to write the commentary on a second paper examining the potential of SARS like viruses in bats to cause human disease. I agreed to write it because the science is excellent, the conclusions are important, and it would provide me with another venue for criticizing the gain-of-function pause.

In the PNAS paper, Menachery et al. describe a platform comprising metagenomics data, synthetic virology, transgenic mouse models, and monoclonal antibody therapy to assess the ability of SARS-CoV–like viruses to infect human cells and cause disease in mouse models. The results indicate that a bat SARS-like virus, WIV1-CoV, can infect human cells but is attenuated in mice. Additional changes in the WIV1-CoV genome are likely required to increase the pathogenesis of the virus for mice. The same experimental approaches could be used to examine the potential to infect humans of other animal viruses identified by metagenomics surveys. Unfortunately my commentary is behind a paywall, so for those who cannot read it, I’d like to quote from my final paragraphs on the gain-of-function issue:

The current government pause on these gain-of-function experiments was brought about in part by several vocal critics who feel that the risks of this work outweigh potential benefits. On multiple occasions these individuals have indicated that some of the SARS-CoV work discussed in the Menachery et al. article is of no merit. … These findings provide clear experimental paths for developing monoclonal antibodies and vaccines that could be used should another CoV begin to infect humans. The critics of gain-of-function experiments frequently cite apocalyptic scenarios involving the release of altered viruses and subsequent catastrophic effects on humans. Such statements represent personal opinions that are simply meant to scare the public and push us toward unneeded regulation. Virologists have been manipulating viruses for years—this author was the first to produce, 35 y ago, an infectious DNA clone of an animal virus—and no altered virus has gone on to cause an epidemic in humans. Although there have been recent lapses in high-containment biological facilities, none have resulted in harm, and work has gone on for years in many other facilities without incident. I understand that none of these arguments tell us what will happen in the future, but these are the data that we have to calculate risk, and it appears to be very low. As shown by Menacherry et al. in PNAS, the benefits are considerable.

A major goal of life science research is to improve human health, and prohibiting experiments because they may have some risk is contrary to this goal. Being overly cautious is not without its own risks, as we may not develop the advances needed to not only identify future pandemic viruses and develop methods to prevent and control disease, but to develop a basic understand- ing of pathogenesis that guides prevention. These are just some of the beneficial outcomes that we can predict. There are many examples of how science has progressed in areas that were never anticipated, the so-called serendipity of science. Examples abound, including the discovery of restriction enzymes that helped fuel the biotechnology revolution, and the development of the powerful CRISPR/Cas9 gene-editing technology from its obscure origins as a bacterial defense system.

Banning certain types of potentially risky experiments is short sighted and impedes the potential of science to improve human health. Rather than banning experiments, such as those described by Menachery et al., measures should be put in place to allow their safe conduct. In this way science’s full benefits for society can be realized, unfettered by artificial boundaries.

TWiV 321: aTRIP and a pause

On episode #321 of the science show This Week in Virology, Paul Duprex joins the TWiV team to discuss the current moratorium on viral research to alter transmission, range and resistance, infectivity and immunity, and pathogenesis.

You can find TWiV #321 at www.microbe.tv/twiv.

What we are not afraid to say about Ebola virus

sneezeIn a recent New York Times OpEd entitled What We’re Afraid to Say About Ebola, Michael Osterholm wonders whether Ebola virus could go airborne:

You can now get Ebola only through direct contact with bodily fluids. If certain mutations occurred, it would mean that just breathing would put one at risk of contracting Ebola. Infections could spread quickly to every part of the globe, as the H1N1 influenza virus did in 2009, after its birth in Mexico.

Is there any truth to what Osterholm is saying?

Let’s start with his discussion of Ebola virus mutation:

But viruses like Ebola are notoriously sloppy in replicating, meaning the virus entering one person may be genetically different from the virus entering the next. The current Ebola virus’s hyper-evolution is unprecedented; there has been more human-to-human transmission in the past four months than most likely occurred in the last 500 to 1,000 years.

When viruses enter a cell, they make copies of their genetic information to assemble new virus particles. Viruses such as Ebola virus, which have genetic information in the form of RNA (not DNA as in other organisms), are notoriously bad at copying their genome. The viral enzyme that copies the RNA makes many errors, perhaps as many as one or two each time the viral genome is reproduced. There is no question that RNA viruses are the masters of mutation. This fact is in part why we need a new influenza virus vaccine every few years.

The more hosts infected by a virus, the more mutations will arise. Not all of these mutations will find their way into infectious virus particles because they cause lethal defects. But Osterholm’s statement that the evolution of Ebola virus is ‘unprecedented’ is simply not correct. It is only what we know. The virus was only discovered to infect humans in 1976, but it surely infected humans long before that. Furthermore, the virus has been replicating, probably for millions of years, in an animal reservoir, possibly bats. There has been ample opportunity for the virus to undergo mutation.

More problematic is Osterholm’s assumption that mutation of Ebola virus will give rise to viruses that can transmit via the airborne route:

If certain mutations occurred, it would mean that just breathing would put one at risk of contracting Ebola. Infections could spread quickly to every part of the globe, as the H1N1 influenza virus did in 2009, after its birth in Mexico.

The key phrase here is ‘certain mutations’. We simply don’t know how many mutations, in which viral genes, would be necessary to enable airborne transmission of Ebola virus, or if such mutations would even be compatible with the ability of the virus to propagate. What allows a virus to be transmitted through the air has until recently been unknown. We can’t simply compare viruses that do transmit via aerosols (e.g. influenza virus) with viruses that do not (e.g. HIV-1) because they are too different to allow meaningful conclusions.

One approach to this conundrum would be to take a virus that does not transmit among mammals by aerosols – such as avian influenza H5N1 virus – and endow it with that property. This experiment was done by Fouchier and Kawaoka several years ago, and revealed that multiple amino acid changes are required to allow airborne transmission of H5N1 virus among ferrets. These experiments were met with a storm of protest from individuals – among them Michael Osterholm – who thought they were too dangerous. Do you want us to think about airborne transmission, and do experiments to understand it – or not?

The other important message from the Fouchier-Kawaoka ferret experiments is that the H5N1 virus that could transmit through the air had lost its ability to kill. The message is clear: gain of function (airborne transmission) is accompanied by loss of function (virulence).

When it comes to viruses, it is always difficult to predict what they can or cannot do. It is instructive, however, to see what viruses have done in the past, and use that information to guide our thinking. Therefore we can ask: has any human virus ever changed its mode of transmission?

The answer is no. We have been studying viruses for over 100 years, and we’ve never seen a human virus change the way it is transmitted.

HIV-1 has infected millions of humans since the early 1900s. It is still transmitted among humans by introduction of the virus into the body by sex, contaminated needles, or during childbirth.

Hepatitis C virus has infected millions of humans since its discovery in the 1980s. It is still transmitted among humans by introduction of the virus into the body by contaminated needles, blood, and during birth.

There is no reason to believe that Ebola virus is any different from any of the viruses that infect humans and have not changed the way that they are spread.

I am fully aware that we can never rule out what a virus might or might not do. But the likelihood that Ebola virus will go airborne is so remote that we should not use it to frighten people. We need to focus on stopping the epidemic, which in itself is a huge job.

TWiV 302: The sky is falling

On episode #302 of the science show This Week in Virology, the TWiVers discuss the growing Ebola virus outbreak in West Africa, and an epidemic of respiratory disease in the US caused by enterovirus D68.

You can find TWiV #302 at www.microbe.tv/twiv.

Reconstruction of 1918-like avian influenza virus stirs concern over gain of function experiments

ferretThe gain of function experiments in which avian influenza H5N1 virus was provided the ability to transmit by aerosol among ferrets were met with substantial outrage from both the press and even some scientists; scenarios of lethal viruses escaping from the laboratory and killing millions proliferated (see examples here and here). The recent publication of new influenza virus gain of function studies from the laboratories of Kawaoka and Perez have unleashed another barrage of criticism. What exactly was done and what does it mean?

According to critics, virologists should not be entrusted to carry out gain of function studies with influenza virus; they are dangerous and of no scientific value. The headline of a Guardian article is “Scientists condemn ‘crazy, dangerous’ creation of deadly airborne flu virus” (the headline is at best misleading because the influenza virus that was reconstructed by Kawaoka and colleagues is not deadly when transmitted by aerosol). The main opponents of the work appear to be Lord May*, former President of the Royal Society; Harvard epidemiologist Mark Lipsitch; and virologist Simon Wain Hobson. They all have nasty things to say about the work and the people doing it. To his credit, author of the Guardian article Ian Sample (who likely did not write the headline) does present both sides of the study, and attempts to explain what was done. He even quotes Kawaoka on the value of the work. But much is left unsaid, and without a detailed analysis of the study, its importance is not readily apparent.

The work by Kawaoka and colleagues attempts to answer the question of whether an influenza virus similar to that which killed 50 million people in 1918 could emerge today. First they identified in the avian influenza virus sequence database individual RNA segments that encode proteins that are very similar to the 1918 viral proteins.

Next, an infectious influenza virus was produced with 8 RNA segments that encode proteins highly related to those of the 1918 virus. Each RNA segment originates from a different avian influenza virus, and differs by 8 (PB2), 6 (PB1), 20 (PB1-F2), 9 (PA), 7 (NP), 33 (HA), 31 (NA), 1 (M1), 5 (M2), 4 (NS1), and 0 (NS2) amino acids from the 1918 virus.

The 1918-like avian influenza virus was less pathogenic in mice and ferrets compared with the 1918 virus, and more pathogenic than a duck influenza virus isolated in 1976. Virulence in ferrets increased when the HA or PB2 genes of the 1918-like avian influenza virus were substituted with those from the 1918 virus.

Aerosol transmission among ferrets was determined for the 1918-like avian influenza virus, and reassortants containing 1918 viral genes (these experiments are done by housing infected and uninfected ferrets in neighboring cages). The 1918 influenza virus was transmitted to 2 of 3 ferrets. Neither the 1918-like avian influenza virus, nor the 1976 duck influenza virus transmitted among ferrets. Aerosol transmission among ferrets was observed after infection with two different reassortant viruses of the 1918-avian like influenza virus: one which possesses the 1918 virus PB2, HA, and NA RNAs (1918 PB2:HA:NA/Avian), and one which possesses the 1918 virus PA, PB1, PB2, NP, and HA genes (1918(3P+NP):HA/Avian).

It is known from previous work that amino acid changes in the viral HA and PB2 proteins are important in allowing avian influenza viruses to infect humans. Changes in the viral HA glycoprotein (HA190D/225D) shift receptor specificity from avian to human sialic acids, while a change at amino acid 627 of the PB2 protein to a lysine (627K) allows avian influenza viruses to efficiently replicate in mammalian cells, and at the lower temperatures of the human upper respiratory tract.

These changes were introduced into the genome of the 1918-like avian influenza virus. One of three contact ferrets was infected with 1918-like avian PB2-627K:HA-89ED/190D/225D virus (a mixture of glutamic acid and aspartic acid at amino acid 89 was introduced during propagation of the virus in cell culture). Virus recovered from this animal had three additional mutations: its genotype is 1918-like avian PB2-627K/684D:HA-89ED/113SN/190D/225D/265DV:PA-253M (there are mixtures of amino acids at HA89, 113, and 265). This virus was more virulent in ferrets and transmitted by aerosol more efficiently than the 1918-like avian influenza virus. The virus recovered from contact ferrets contained yet another amino acid change, a T-to-I mutation at position 232 of NP. Therefore ten amino acid changes are associated with allowing the 1918-like avian influenza virus to transmit by aerosol among ferrets. Aerosol transmission of these viruses is not associated with lethal disease in ferrets.

Previous studies have shown that changes in the HA needed for binding to human sialic acid receptors reduced the stability of the HA protein. Adaptation of these viruses to aerosol transmission among ferrets required amino acid changes in the HA that restore its stability. Similar results were obtained in this study of the 1918-like avian influenza virus, namely, that changes that allow binding to human receptors (HA-190D/225D) destabilize the HA protein, and changes associated with aerosol transmission (HA-89D and HA- 89D/113N) restore stability.

The ability of current influenza virus vaccines and antivirals to block replication with ferret transmissible versions of the 1918-like avian influenza virus was determined. Sera from humans immunized with the 2009 pandemic H1N1 strain poorly neutralized the virus, indicating that this vaccine would likely not be protective if a similar virus were to emerge. However replication of the ferret transmissible 1918-like avian influenza virus is inhibited by the antiviral drug oseltamivir.

Examination of influenza virus sequence databases reveals that avian viruses encoding PB2, PB1, NP, M, and NS genes of closest similarity to those of the 1918-like avian virus have circulated largely in North America and Europe. The PB2-627K change is present in 168 of 4,293 avian PB2 genes (4%), and the HA-190D change is in 9 of 266 avian H1 HA sequences (3%), and one also had HA-225D.

Most of the viral sequences used in this work were obtained quite recently, indicating that influenza viruses encoding 1918-like proteins continue to circulate 95 years after the pandemic.  Now that we have discussed the work, we can summarize why it is important:

  • An infectious 1918-like avian virus can be assembled from RNA segments from circulating viruses that is of intermediate virulence in ferrets. Ten amino acid changes are sufficient to allow this virus to transmit by aerosol among ferrets.
  • Confirmation that transmissibility of influenza virus among ferrets depends on a stable HA glycoprotein. This result was a surprising outcome of the initial studies on aerosol transmission of H5N1 avian influenza viruses among ferrets, and provided mechanistic information about what is important for transmission. Experiments can now be designed to determine if HA stability is also important for influenza virus transmission in humans.
  • We understand little about why some viruses transmit well by aerosol while others do not. Transmission should be a selectable trait – the virus with a random mutation can reach another host by aerosol, where it replicates and can transmit further. Why types of mutation allow better transmission? Why don’t avian influenza viruses become transmissible among humans more frequently? Are there fitness tradeoffs to becoming transmissible? These and similar questions about transmission can be answered with sutdies of the types discussed here. The list is not confined to influenza virus: aerosol transmission of measles virus, rhinovirus, adenovirus, and many others, is poorly understood. This work shows what can be done and will surely inspire similar work with other viruses.

Do these experiments constitute an unacceptable risk to humans? Whether or not the 1918-like avian influenza virus, or its transmissible derivatives, would replicate, transmit, and cause disease in humans is unknown. While ferrets are a good model for influenza virus pathogenesis, they cannot be used to predict what will occur in humans. Nevertheless it is prudent to work with these avian influenza viruses under appropriate containment, and that is how this work was done. The risk is worth taking, not only because understanding transmission is fundamentally important, but also because of unanticipated results which often substantially advance the field.

The Guardian quotes Lipsitch as saying that “Scientists should not take such risks without strong evidence that the work could save lives, which this paper does not provide”. The value of science cannot only be judged in terms of helping human health, no matter what the risk. If we only did work to improve human health, we would not have most of the advances in science that we have today. One example is the biotechnology industry, and the recombinant DNA revolution, which emerged from the crucial discovery of restriction enzymes in bacteria – work that was not propelled by an interest in saving lives.

The results obtained from the study of the reconstruction of a 1918-like avian influenza virus are important experiments whose value is clear. They are not without risk, but the risk can be mitigated. It serves no useful purpose to rail against influenza virus gain of function experiments, especially without discussing the work and its significance. I urge detractors of this type of work to carefully review the experiments and what they mean in the larger context of influenza virus pathogenesis. I understand that the papers are complex and might not be easily understood by those without scientific training, and that is why I have tried to explain these experiments as they are published (examples here and here).

In the next post, I’ll explain the gain of function experiments recently published by the Perez laboratory.

*May’s objection is that the scientists carrying out the work are ‘grossly ambitious people’. All scientists are ambitious, but that is not what drives Kawaoka and Perez to do this work. I suggest that Lord May read the papers and base his criticism on the science.

TWiV 287: A potentially pandemic podcast

On episode #287 of the science show This Week in Virology, Matt Frieman updates the TWiV team on MERS-coronavirus, and joins in a discussion of whether we should further regulate research on potentially pandemic pathogens.

You can find TWiV #287 at www.microbe.tv/twiv.

Virologists plan influenza H7N9 gain of function experiments

A group of virologists lead by Yoshihiro Kawaoka and Ron Fouchier have sent a letter to Nature and Science outlining the experiments they propose to carry out with influenza H7N9 virus.

Avian influenza H7N9 virus has caused over 130 human infections in China with 43 fatalities. The source of the virus is not known but is suspected to be wet market poultry. No human to human transmission have been detected, and the outbreak seems to be under control. According to the authors of the letter, the virus could re-emerge this winter, and therefore additional work is needed to assess the risk of human infection.

The research that the virologists propose involve gain-of-function experiments which provide the H7N9 virus with new properties. The isolation of avian influenza H5N1 viruses that can transmit by aerosol among ferrets is an example of a gain-of-function experiment.

The proposed gain-of-function experiments fall into five general categories:

  • Determine whether viruses with altered virulence, host range, or transmissibility have changes in antigenicity, or the ability of the virus to react with antibodies. The results of these studies would suggest whether, for example, acquisition of human to human transmissibility would have an impact on protection conferred by a vaccine produced with the current H7N9 virus strain.
  • Determine if the H7N9 virus could be adapted to mammals and whether it could produce reassortants with other influenza viruses. The results of this work would provide information on how likely it is that the H7N9 virus would become better adapted to infect humans.
  • Isolate mutants of H7N9 virus that are resistant to antiviral drugs. The purpose of these experiments is to identify how drug resistance arises (the mutations can then be monitored in clinical isolates), determine the stability of drug resistant mutants, and whether they confer other properties to the virus.
  • Determine the genetic changes that accompany selection of H7N9 viruses that can transmit by aerosol among mammals such as guinea pigs and ferrets. As I have written before, the point of these experiments, in my view, is not to simply identify specific changes that lead to aerosol transmission. Such work provides information on the mechanisms by which viruses can become adapted to aerosol transmission, still an elusive goal.
  • Identify changes in H7N9 virus that allow it to become more pathogenic. The results of these experiments provide information on the mechanism of increased pathogenicity and whether it is accompanied by other changes in properties of the virus.

I believe that the proposed gain-of-function experiments are all worth doing. I do not share the concerns of others about the potential dangers associated with gain-of-function experiments: for example the possibility that a virus selected for higher virulence could escape the laboratory and cause a lethal pandemic. Gain-of-function is almost always accompanied by a loss-of-function. For example, the H5N1 viruses that gained the ability to transmit by aerosol among ferrets lost their virulence by this route of infection. When these experiments are done under the proper containment, the likelihood that accidents will happen is extremely small.

All the proposed experiments that would use US funds will have to be reviewed and approved by the Department of Health and Human Services:

The HHS review will consider the acceptability of these experiments in light of potential scientific and public-health benefits as well as biosafety and biosecurity risks, and will identify any additional risk-mitigation measures needed.

While I understand that the authors wish to promote a dialogue on laboratory safety and dual-use research, I question the ultimate value of the communication. Because the letter has been published in two scientific journals, I assume that the target audience of the letter is the scientific community. However, the letter will clearly have coverage in the popular press and I am certain that it will be misunderstood by the general public. I can see the headlines now: “Scientists inform the public that they will continue to make deadly flu viruses”. The controversy about the H5N1 influenza virus transmission studies in ferrets all began with a discussion of the results before the scientific papers had been published. I wonder if the publication of these letters will spark another controversy about gain-of-function research.

In my view, science is best served by the traditional process known to be highly productive: a grant is written to secure funding for proposes experiments, the grant proposal is subject to scientific review by peers, and based on the review the work may or may not be supported. The experiments are done and the results are published. I do not understand why it is necessary to trigger outrage and debate by announcing the intent to do certain types of experiments.

I am curious to know what the many readers of virology blog – scientists and non-scientists – feel about the publication of this letter. Please use the comment field below to express your views on this topic.

Inefficient influenza H7N9 virus aerosol transmission among ferrets

ferretThere have been 131 confirmed human infections with avian influenza H7N9 virus in China, but so far there is little evidence for human to human transmission. Three out of four patients report exposure to animals, ‘mostly chickens‘, suggesting that most of the infections are zoonoses. Whether or not the virus will evolve to transmit among humans is anyone’s guess. Meanwhile it has been found that one of the H7N9 virus isolates from Shanghai can transmit by aerosol among ferrets, albeit inefficiently.

Ferrets were inoculated intranasally with influenza A/Shanghai/02/2013 virus or A/California/07/2009, the 2009 pandemic H1N1 virus. One to two days later the ferrets developed fever, sneezing, coughing, and nasal discharge; both viruses induced similar clinical signs. Virus was shed in nasal secretions for 7 days. Six infected ferrets were then divided among three separate cages, and each group was housed with a naive ferret, and a second uninfected animal was placed in an adjacent cage. Airflow was controlled so that air flowed from the cage of infected animals towards the cage of naive animals. Transmission of infection was measured by observing clinical signs, and measuring virus shedding in nasal secretions and hemagglutination-inhibition antibodies in serum.

Of the three ferrets housed in the same cage with H7N9 virius-infected animals, all three had signs of infection (sneeze, cough, nasal discharge), shed virus in nasal secretions, and developed anti-HA antibodies. All three ferrets in neighboring cages developed signs of infection, but only one shed virus in nasal secretions, and two of three seroconverted. From these data the authors conclude that H7N9 virus is ‘efficiently transmitted between ferrets by direct contact, but less efficiently by airborne exposure’. In contrast, transmission of H1N1 virus to naive ferrets by contact or aerosol was efficient (3/3 animals in both cases).

The authors also found that pigs could be infected intranasally with A/Shanghai/02/2013 virus: the animals shed virus in nasal secretions and developed clinical symptoms. However the infected pigs transmitted infection inefficiently to other pigs by contact or aerosol, or to ferrets by aerosol.

The  authors’ equivocal conclusion that “Under appropriate conditions human to human transmission of the H7N9 virus may be possible” could have been reached even before these experiments were done. Their results provide no information on whether the virus can undergo human to human transmission because animal models are not definitive predictors of what might occur in humans. I disagree with the authors’ statement on page 5, “Efficient transmission of influenza viruses in ferrets is considered as a predictor of human to human transmissibility’. While many influenza virus strains that transmit among humans by aerosol also do so in ferrets, this does not mean that human transmission of a novel virus can be predicted by animal experiments.

Infection of ferrets with A/Shanghai/02/2013 or or A/California/07/2009 virus results in mild disease with no mortality. In contrast, 32 humans infected with H7N9 virus have died, and many humans have died after H1N1 infection. These findings further emphasize the differences in influenza virus pathogenesis in ferrets and humans.

TWiV 233: We’re surrounded

On episode #233 of the science show This Week in Virology, Vincent, Rich, Alan and Kathy review aerosol transmission studies of influenza H1N1 x H5N1 reassortants, H7N9 infections in China, and the MERS coronavirus.

You can find TWiV #233 at www.microbe.tv/twiv.

Further defense of the Chinese H1N1 – H5N1 study

Robert Herriman of The Global Dispatch interviewed me this week on the H1N1 – H5N1 reassortant study that has been in the headlines:

There was much written concerning the research published earlier this month in Science, where researchers from China’s Harbin Veterinary Research Institute reported creating an  avian H5N1 (highly pathogenic) and pandemic 2009 H1N1 (easily transmissible) hybrid, that according to them, achieved airborne spread between guinea pigs.

Read the rest of the article at The Global Dispatch.