In the workshop’s second session, Dr. Jeffery Taubenberger, Chief of the Viral Pathogenesis and Evolution Section, National Institute for Allergy and Infectious Diseases, National Institutes of Health, and Robert G. Webster, Rose Marie Thomas Chair in Virology, St. Jude Children’s Hospital and Director, World Health Organization Collaborating Laboratory on Animal Influenza, discussed the structure and function of influenza viruses.
Viruses are small, infectious agents consisting of genetic material encapsulated within a protein coat. To reproduce, viruses require host cells to which they attach and into which they inject their genetic material. Viruses overtake and use the host’s cellular machinery to replicate themselves. This eventually causes the host cell to burst and viral particles are released back into the environment.
Viral surface proteins largely define a virus’s identity and behavior. In influenza, surface HA (hemagglutinin) and NA (neuraminidase) proteins have opposite roles. Hemagglutinin enables the virus to bind to the host cell, and neuraminidase enables the virus to release itself from the host cell’s surface to seek a new host cell.
Differences in Virulence and Transmissibility
Influenza strains differ in their ability to cause disease (pathogenicity) and to travel between hosts (transmissibility). The influenza virus, in gen-
eral, when well-adapted to a host, is among the most highly transmissible viruses known. If a highly transmissible strain is also highly virulent (i.e., highly likely to cause serious disease), the mortality rate for those infected can be significant. Most influenza strains are transmissible only between members of the same species. Other strains may pass from one species to another. For a pandemic to occur in humans, a human-transmissible strain is required. It is only when a strain becomes transmissible from an animal species to humans and then gains the ability to travel easily among humans that a human pandemic may result.
Differences in transmissibility are due to a strain’s genetic make-up. The influenza viruses that most concern scientists are strains of influenza A. The genetic material of influenza A viruses is a single strand of RNA with eight gene segments. (See Figure 3-1.) The genes encoding hemagglutinin and neuraminidase are carried on different segments; therefore, the particular combination of HA and NA type is easily changed, both in nature and in the laboratory. If a single host is infected with two different strains of influenza, a new viral strain may emerge, in a process known as antigenic shift, through a process of gene segment reassortment that results in an influenza virus with a new combination of the hemagglutinin and neuraminidase proteins. Variation is also often introduced during viral reproduction. The replication process in viruses is imprecise, and errors are frequently introduced into viral RNA sequences. This imprecision often benefits the virus, since different variations may confer survival advantages in different environments. New genetic signatures can result in altered traits which may improve the strain’s ability to evade the host’s immunological defenses. While, in many cases, the new traits do not adversely affect the human host, they may, in some cases, cause serious harm. Regardless, the influenza virus’s rapid evolution means that the influenza vaccine must be reformulated each year.
Influenza A Viruses and Their Many Hosts
Influenza A viruses infect a wide variety of animals that include—in addition to humans—whales, seals, pigs, horses, domesticated poultry, and wild birds. (See Figure 3-2.) Wild birds are the principal hosts, and throughout the world there is a large reservoir of influenza viruses in hundreds of avian species. In general, influenza-infected birds are asymptomatic. When an infected bird is symptomatic, mild gastrointestinal symptoms are most common.
An influenza strain that crosses a species barrier is cause for greater concern. The pathogenicity of such a strain is often much higher on the other side of the species barrier, as has been the case with H5 and H7 influ-
enza sub-types that have been transmitted from wild birds to domesticated poultry.
Pigs are potential sources of new strains capable of infecting humans, as pigs are subject to infection by avian influenza, swine influenza, and human influenza. If a pig is infected simultaneously with influenza strains adapted to different species, it may thus serve as a mixing vessel wherein the interaction of the assortment of viruses may more likely give rise to a strain that is highly pathogenic for humans.
Human Influenza: Seasonal Outbreaks and Unpredictable Pandemics
During the winter season, influenza strains are transmitted rapidly among human beings and cause major illness and death. It is estimated that between 3,000 and 49,000 people in the U.S. die from seasonal influenza every year.1 The influenza strains that cause seasonal outbreaks continually evolve as they circulate among human populations and can acquire decreased or increased virulence. Pandemic outbreaks, in contrast, originate with an animal-adapted strain. They occur when a human-adapted strain receives a new “HA” gene from an animal-adapted strain or when an entire animal-adapted strain becomes adapted to humans.
Four influenza pandemics have been observed in the past 100 years: “Spanish” flu (H1N1) in 1918, “Hong Kong” flu (H3N2) in 1968, “Asian” flu (H2N2) in 1957, and “swine” flu (H1N1) in 2009. In the intervening years, these four strains have circulated as seasonal flu and “mixed” with other animal-adapted strains to form new variants.
Dr. Taubenberger discussed his research on the Spanish influenza outbreak of 1918, and Dr. Webster discussed past and recent developments in H5N1 research. Drs. Taubenberger and Webster addressed a number of questions, including: What was the scientific rationale for undertaking the research? What security and safety risks were considered prior to beginning the research? How, and by which individuals or entities other than the researchers themselves, were these risks weighed, and what kinds of precautionary or counter-measures were mandated or considered? Was the process for the risk-benefit assessment prior to the research optimal, and what might have been done differently? Why has the reaction to the H5N1 research played out differently from other dual-use research? What is different or distinct about the science, design and conduct of research, policy implications, or governance of the H5N1 research as opposed to earlier work such as the reconstruction of the 1918 Spanish Flu virus? What lessons can we learn regarding the design, conduct, communication, and oversight of future life sciences research of concern? Is there scientific research where the risks of undertaking the research outweigh the benefits of undertaking the research, and who or what institutional or societal processes are or should be in place to make such determinations?
1 Centers for Disease Control and Prevention, “Estimating Seasonal Influenza-Associated Deaths in the United States: CDC Study Confirms Variability of Flu,” June 24, 2011, http://www.cdc.gov/flu/about/disease/us_flu-related_deaths.htm.
The 1918 Spanish Influenza Pandemic
Dr. Taubenberger stated that the Spanish flu, an H1N1 strain, killed 40-50 million people worldwide during 1918-1919, including almost 700,000 people in the United States (1-2 percent of those infected died). At the conclusion of the pandemic, this strain lost much of its virulence but persisted globally, circulating and causing typical seasonal flu. (See Figure 3-3.)
In the mid-1990s, Taubenberger’s research group and others investigated the Spanish flu using genetic and virologic techniques to explore the virus’s origin, pathogenicity, and transmissibility in experimental models. The research was motivated by questions such as: How did the strain evolve and adapt to humans? Could the mutations be used to improve current surveillance strategies? Why was the strain pathogenic? Could data generated as the result of the research be used to develop new therapies and vaccines?
Taubenberger’s team recovered viral RNA fragments from two formalin-fixed samples from autopsy collections in Washington, DC and London and one frozen, unfixed sample from Alaska and used RT-PCR to sequence the fragments. Taubenberger observed that both the public and the scientific community were aware of this research as it proceeded and noted that, in comparison with the reception of the research of Drs. Fouchier and Kawaoka, the research caused little alarm.2 Researchers rebuilt cDNAs, cloned the cDNAs into bacterial plasmids, and, in high containment facilities, reconstructed the virus’s complete genome by reverse genetics.3 Taubenberger’s team also constructed novel viruses that contained one or more of the genes from the original 1918 strain. In 1997, the first genetic sequences were published, and in 2005, the entire genome was published with the details about the methods used to reconstruct the virus.
Taubenberger stated that the research was subject to several forms of regulatory oversight. The research was funded by the Armed Forces Institute of Pathology, the Departments of Defense and Veterans Affairs, the American Registry of Pathology, and the National Institutes of Health,
2 While not discussed explicitly at the workshop, there has been significant discussion about the way that certain H5N1 research results were conveyed and subsequently interpreted by the press. See, for example, Katherine Harmon, “What Will the Next Influenza Pandemic Look Like?,” Scientific American, September 19, 2011, http://www.scientificamerican.com/article.cfm?id=next-influenza-pandemic.; Deborah MacKenzie, “Five Easy Mutations to Make Bird Flu a Lethal Pandemic,” New Scientist 2831 (September 26, 2011): 3.; Editorial, “An Engineered Doomsday,” New York Times, January 7, 2012, http://www.nytimes.com/2012/01/08/opinion/sunday/an-engineered-doomsday.html?_r=0.; Peter M. Sandman, “Science versus Spin: How Ron Fouchier and Other Scientists Miscommunicated about the Bioengineered Bird Flu Controversy,” June 7, 2012, http://www.psandman.com/articles/Fouchier.htm.
3 The genome of the 1918 influenza virus was assembled at Mount Sinai School of Medicine (laboratories of Drs. Basler, García-Sastre, and Palese) and the infectious virus was rescued at the Centers for Disease Control and Prevention (laboratory of Dr. Tumpey).
and was carried out in U.S. government and academic research laboratories (see Box 3-1). In 2005, the NSABB reviewed the research and supported publication of the research results. Since 2005, the reconstructed virus has been regulated by the Centers for Disease Control and Prevention (CDC) as a select agent requiring Biosafety Level 3 enhanced containment.
What knowledge was gained from the reconstruction of the 1918 Spanish influenza virus? Taubenberger discussed how his group identified a number of factors that influenced the virulence and host adaptability of the Spanish influenza virus. Specifically, they discovered the importance of the host’s inflammatory response on the pathogen’s virulence and identified mutations in the virus’s hemagglutinin that facilitated the virus’s binding to upper airway cells in humans.
Taubenberger also noted that this research led to the discovery of a new protein common to all influenza viruses. Phylogenetic analyses have shown that all subsequent human seasonal and pandemic viruses are descendants of the 1918 pandemic virus. Data also suggest that evolutionary
Federally funded research on infectious agents must be carried out in laboratories where appropriate protection and containment measures are in place. The measures employed are based on the pathogen. Guidelines are outlined and regularly updated in Biosafety in Microbial and Biomedical Laboratories, a publication of the CDC and the NIH. These guidelines describe the safeguards that must be in place in a facility, the safety equipment that must be used in the laboratory, and the safety practices that must be used by laboratory personnel.
The current biosafety levels—BSL-1 through BSL-4—are based on variables specific to the pathogen, e.g., its health effects and the existence of vaccines and effective treatments for those infected. The appropriate biosafety level for a specific research program depends on the infectivity and transmissibility of the pathogen, the severity of the disease that it is capable of causing, and the type of work being performed. For pathogens that can cause moderate or severe disease, an additional consideration is whether the pathogen is “indigenous or exotic.”
Influenza research is carried out in facilities classified as BSL-2, BSL-3, or enhanced BSL-3. BSL-2 facilities are generally used for research on currently circulating human strains of influenza and on avian strains with low pathogenicity. BSL-3 facilities are employed for research on human-adapted strains that are not currently circulating and against which human populations may lack immunity and for research on highly pathogenic avian influenza. Enhanced BSL-3 facilities include additional precautions which are customized to the estimated risk; these precautions may include enhanced respiratory protection, HEPA filtration of exhaust air from the laboratory, and personal body showers and additional changes of lab clothing.
pressures exerted on influenza A viruses infecting wild birds are distinct from those exerted on influenza A viruses adapted to non-native hosts, such as humans and other mammals. In wild birds, influenza A viruses exist as a set of transient, unstable genomes with frequent reassortment that produces a great variety of HA and NA subtype combinations. There is evolutionary pressure toward diversity of surface proteins and toward stabilization of the virus’s internal genes. When a virus leaves the large, diverse gene pool of its avian host and encounters a non-avian host, its evolution is more linear, with the genetic composition of the original infecting virus forming the basis for subsequent generations of the virus.
Taubenberger’s team also learned that their original hypothesis—that if they elucidated how the 1918 virus adapted to humans and what the mutations were, they could understand all future pandemics—was incorrect. Instead, they discovered that some of the evolutionary adaptations of the 1918 strain were not present in the 2009 pandemic. The influenza virus, it appears, creates alternate solutions to the same problems.
Dr. Webster presented an overview of the host population for influenza viruses. There are 16 HA subtypes in influenza viruses found in acquatic birds, but only three subtypes—those expressing H1, H2, and H3 hemagglutinins—have successfully established themselves in human populations. H5, H7, and H9 viruses have infected humans but failed to establish permanent presence. Birds are the ultimate source of all influenza A viruses that infect humans. In birds, influenza A viruses are transmitted by the fecal-oral route via water and infect a bird’s intestinal tract. Only strains with low pathogenicity are present in bird populations, and thus wild birds exhibit few symptoms. These strains form the basis of the gene pool from which highly pathogenic strains emerge and spread to other hosts.
Dr. Webster focused his discussion on H5 and H7 viruses and the process by which they are transformed from low to high pathogenicity. In wild birds, H5 and H7 viruses persist in the low pathogenicity form (although there is a question as to whether this is changing for H5). After H5 and H7 viruses are transmitted to domesticated birds, they sometimes become highly pathogenic. Specifically, the hemagglutinin sometimes acquires an insertional mutation of basic amino acids at the cleavage site, which confer greater virulence. Once an H5 virus becomes highly pathogenic in a domestic poultry population, 100 percent of the infected birds die. Webster observed that this knowledge has been used for many years in agricultural surveillance and control programs—the presence of these amino acids signals that the virus must be eradicated.
Control and Preparedness
Webster discussed how experience has shown that culling infected flocks, and compensating bird owners—as is the practice in Japan, South Korea, and European countries—is very successful in containing avian influenza outbreaks. Countries that opted for vaccination, for example China, Egypt, and Indonesia, have been less successful in their containment efforts and have even witnessed a more rapid evolution of the virus.
In 2004, pandemic preparedness programs were undertaken in response to an endemic presence of H5N1 in Eurasia due to the fear that the virus might spread to Australia and the Americas. In actuality, the viral strain that has since spread most rapidly throughout the world is H1N1—a low pathogenicity but highly transmissible form. At least 6000 human deaths resulted from H1N1 infections between April and November 2009.
Research Environment into which the Fouchier and Kawaoka Papers Appeared
Webster proceeded to discuss research questions that are being pursued.
In 2006, the National Institute of Allergy and Infectious Diseases called for researchers to study how influenza viruses move between different animal populations as well as the evolutionary pressures that lead to the emergence of new sub-types, with an emphasis on learning what factors allow the transmission of a subtype to humans.4 In 2009, the WHO recommended that researchers investigate the pathogenicity, infectivity, and transmissibility of influenza viruses.5 Webster noted that various approvals are required before such research may be conducted. These include peer review of the research grant proposal, approval (for some studies) by the NIH Office of Biotechnology Activities, and approvals at the research institution consisting of an Institutional Biosafety Committee and Institutional Animal Care and Use Committee. Webster noted that facilities-related procedures include registration of select agents, influenza-specific enhancements to standard BSL-3 facilities, registration with either the U.S. Department of Agriculture (USDA) or Centers for Disease Control and Prevention (CDC), and repeated inspections of the facilities security and inventory. In addition, material-transfer documentation includes importation permits (USDA and CDC) and export permits (U.S. Department of Commerce). Webster also referred to laboratory security and biosafety measures designed to restrict
4 National Institute for Allergy and Infectious Diseases, “Report on the Blue Ribbon Panel for Influenza Research,” June 2007.
5 World Health Organization, “WHO Public Health Research Agenda for Influenza, Version 1, 2009,” (2010): 9.
access to authorized personnel, protect the health of such personnel, and for tracking the movement of viral inventories.
The H5N1 Papers, The Approval Process, and Public Reaction to Fouchier and Kawaoka’s Papers
Webster noted that Drs. Kawaoka and Fouchier’s research demonstrated that H5N1 strains may potentially mutate to become transmissible between ferrets.6 In the transmissible form produced in the laboratory, researchers identified specific attributes of the hemagglutinin protein and its stability. Furthermore, they recognized the importance of glycosylation and observed that there are multiple routes whereby a strain can become transmissible between ferrets.
In the fall of 2011, the NSABB reviewed the two papers and, in December, recommended that the papers should not be published in full. This decision was based on information that indicated that the resultant strains were not only highly transmissible but also highly pathogenic (the latter in the case of the Fouchier study). However, the characterization of the strains’ virulence was subsequently revised when it was acknowledged that many strains of influenza are highly virulent when inoculated directly into the trachea, the pathway used in these experiments. This additional context was provided at a meeting of international influenza experts convened by the WHO in February 2012.7 On March 29 and 30, 2012, this new information was reviewed at a second meeting of the NSABB, along with national security information. At this meeting, the NSABB reconsidered its earlier recommendation regarding the publication of a redacted version of the manuscripts. As a result, in part, of this new information, the NSABB reversed its original recommendation and recommended that both papers be published in full. However, while the NSABB was unanimous in its opinion that the Kawaoka paper be published in full, it was not unanimous in its decision to recommend publication of the Fouchier paper.8
6 It should be noted that the mutant viruses that became transmissible between ferrets lost virulence in the ferret system.
7 The experts at the WHO meeting called for publication of both papers in full but also called for a continuation of the existing moratorium on such research until there could be more discussion about biosecurity and biosafety issues. See Jon Cohen, “WHO Meeting of Flu Experts Calls for Full Publication of Controversial H5N1 Papers,” ScienceInsider, February 17, 2012, http://news.sciencemag.org/scienceinsider/2012/02/who-meeting-of-flu-experts-calls.html.
8 Although the deliberative activity of the NSABB was not discussed in detail at the workshop, it is important to place the NSABB deliberations in context. Scientific manuscripts often do not provide sufficient information about research projects’ aims or the precautions that were taken, obligating boards such as the NSABB to act without complete information
The question of full or partial publication resonates beyond the scientific community and into the realm of international health policy and relations. Both research groups had utilized viral strains provided by countries in Asia: Fouchier’s from Indonesia and Kawaoka’s from Vietnam. Cooperative agreements between the WHO and public health leaders in these countries made these transfers possible. To stimulate such exchanges, the WHO provides vaccines and antiviral drugs to contributing nations. The redaction of key experimental information might complicate relationships among international parties engaged in, and benefiting from, influenza research.
Benefits and Risks
Webster discussed the benefits versus the risks of undertaking research on H5N1. He believes that the Fouchier and Kawaoka research has produced benefits that include knowledge that H5N1 may acquire mutations that permit transmission between mammals. Webster noted that the risks of conducting such research include high lethality to humans, the intentional or unintentional escape of the virus from the laboratory, and the possibility that the knowledge gained from such research could be used to develop biological weapons. He noted, however, that nature is perfectly capable of creating the same mutations that were generated in the laboratory and that the threat of a pandemic is a real one. He observed that the global community’s task is to manage and not avoid risks.
Webster made several observations about the pandemic potential of H5N1 viruses. He observed that the research community is getting closer to knowing which subset of H5N1 possesses the necessary mutations to cause a pandemic and therefore, potentially, the strains on which efforts at control or eradication might be focused. He suggested that further research might focus on such questions as: Does the virulence of H5N1 in mam-
about the risk involved. Further, the NSABB review process does not dictate that it acquire knowledge of key details about risks and benefits before proceeding to a decision.
A letter sent by a member of the NSABB to the NIH highlights particular concerns about the circumstances of the meeting wherein the NSABB reconvened to reconsider its earlier recommendations. The author notes that no disinterested subject matter experts were asked to speak to the current state of the art in the use of reverse genetics technology or to discuss the implications of such work. Likewise, the author observed that no input was invited from people working on H5N1 surveillance and control as to the benefits for surveillance/control of the publication of the mutational data. According to the letter, a security briefing given at the meeting lacked a basis in knowledge of the threat, both historical and present-day, specific to influenza. Further, the letter states that, while investigative research using interviews with prominent professionals in public health was published in Science and Nature, the work was not incorporated into NSABB’s process. See Michael T. Osterholm, “Letter to Amy P. Patterson, M.D., Associate Director for Science Policy, National Institutes of Health,” April 12, 2012, http://news.sciencemag.org/scienceinsider/NSABB%20letter%20final%2041212_3.pdf.
mals increase or decrease with transmission? Are the transmissible strains of H5N1 more likely to reassort with the pandemic-causing H1N1 strains?
Asked whether important questions can be addressed using less pathogenic strains, Webster replied that some of them can. He observed, however, that many of the key questions can only be answered by using highly pathogenic strains; for example, questions concerning the host inflammatory response and its regulation.
What tools do we currently have for use in the face of a pandemic? The session moderator, Dr. Alice Huang, Senior Faculty Associate in Biology, California Institute of Technology, asked what tools are available for responding to pandemics. Taubenberger and Webster both spoke about the effectiveness of quarantine, vaccination, and antiviral medications. Taubenberger noted that the effectiveness of each strategy depends on the pathogen. With an illness in which infected people shed the virus only after they become symptomatic, controlling an outbreak is easier than with illnesses where those infected begin to shed the virus before they become symptomatic. Those who contracted SARS (severe acute respiratory syndrome) in 2003, for example, exhibited symptoms before shedding the SARS virus,9 and it was possible to recognize and quarantine the affected persons. In the case of influenza, a host sheds the virus before exhibiting any symptoms. Therefore, quarantine as a strategy is not as effective with influenza. Likewise, existing vaccines may not be effective counter-measures against future influenza strains. Influenza strains evolve quickly, and as a result, vaccines must be strain-specific. The formulation of strain-specific vaccines takes months: antiviral medications are, therefore, the best counter-measure. Webster noted, however, that only two families of antiviral medications are available: neuraminidase inhibitors and ion channel blockers. He observed that there is a great need to develop new families of antiviral drugs as well as a universal influenza vaccine.
To the question of whether there is an important distinction to be made between doing research on a naturally occurng pathogen versus a pathogen that has been generated by researchers and may never appear in nature, both Webster and Taubenberger indicated that there is not. Webster believes that all viruses that could be created in a laboratory already exist in nature. Taubenberger agreed, noting that chance determines whether a variant reaches a new host that it is well adapted to.
Taubenberger added that at the beginning of his research on the 1918 strain, it was best to understand a virus that had evolved in nature. Thus,
9 The SARS virus is a member of the coronavirus family not previously observed in humans.
even though there were significant risks, given the knowledge that the strain in question could produce a pandemic, the potential health benefits of reconstructing the virus outweighed those risks. Further, information gained from the research might be particularly useful in informing preparations for a future pandemic. Regarding H5N1, Taubenberger sees value in using reverse genetics to construct influenza strains with specific characteristics in order to study the biology of the pathogen. He believes that valuable information is gained from manipulating the virus’s genetic code and studying the effect on the pathogen’s phenotype.