Back in the 1960s, after the emergence of the devastating Asian H2N2 influenza virus, I started contacting people who stored influenza viruses. I wanted to determine if any of these viruses might be related to the pandemic-causing strain that had killed at least 1.5 million people in 1957. One of the largest collections of influenza viruses – from humans, horses, pigs and birds – was in Helio Pereira’s laboratory at the National Institute of Medical Research in Mill Hill, London.
Pereira was the director of the World Influenza Centre, which was one of the World Health Organization (WHO) laboratories collaborating on influenza research, and he, like me, was a proponent of the animal reservoir hypothesis to explain the origin of human influenza pandemics. He was very interested to know whether we could find any cross-reactivity between the recently identified Asian H2N2 influenza viruses and other influenza viruses from various animals in his collection. Běla Tůmová, an influenza virologist from Czechoslovakia who had made antisera in rats to the H2N2 virus and to some of the animal influenza viruses, joined the study.43 Antisera contain virus-specific antibodies made by the body in response to infection or vaccination; antisera are used to identify unknown influenza viruses.
In the initial study in 1967 we used the rat sera from Czechoslovakia and ferret sera from Mill Hill to determine whether any of the sera to the animal influenza viruses would react with the human influenza virus. We found in three quite different tests that there was a strong reaction between an influenza virus isolated from turkeys in Massachusetts in 1965 and the human influenza virus that had caused the 1957 pandemic. The reaction was so strong that we did not quite believe the results and worried that somehow the human viruses had been contaminated with animal ones or vice versa. Exhaustive studies ruled out that possibility, and we concluded that indeed we had found that the turkey and human influenza viruses had something in common. We were absolutely delighted, for this provided some of the first solid evidence supporting the hypothesis that animal influenza viruses are the source of at least part of the pandemic influenza viruses. Our next step was to determine which part of the turkey and human influenza viruses was shared – the neuraminidase (N) spike, as we thought, or the haemagglutinin (H).44
While these studies were under way in London, my close colleague Graeme Laver in Canberra had successfully separated the two major surface spikes on influenza virus – the haemagglutinin (H) and the neuraminidase (N) – in chemically pure forms. On my return to Canberra I prepared antisera in rabbits to the pure H and pure N components and showed that we could specifically identify each of them. Pereira encouraged me to bring these antisera to London to answer the question of which components were shared by the turkey virus and human pandemic virus.
In a whirlwind two-day trip back to Mill Hill in early 1967 we set up the neuraminidase inhibition test used in the Great Barrier Reef studies and showed that the specific antisera to neuraminidase from the 1957 influenza virus completely inhibited the enzyme activity of the turkey influenza virus. We also found that three other avian influenza viruses in Pereira’s virus collection possessed a neuraminidase very closely related to, or serologically identical with, the neuraminidase of the human 1957 influenza virus. One of these viruses was obtained from turkeys in Wisconsin in 1966, and two were obtained from ducks in Italy in the same year.45 These findings further supported the idea that the human pandemic influenza virus of 1957 had acquired its neuraminidase component from an animal influenza virus.
But was it possible for human influenza viruses to acquire pieces of animal influenza viruses under natural conditions? I knew of the early studies by Frank Macfarlane Burnet and Patricia Lind in Melbourne, who had put two different influenza A viruses into chicken embryos together and found that the viruses reassorted (exchanged) their genome segments, resulting in hybrid viruses.46 If we did a similar experiment using chickens and pigs, we too might generate new influenza viruses. But because the hybrids might be dangerous to pigs or poultry, this research required a highly secure environment. Since we had no high-containment laboratories at St Jude Children’s Research Hospital in Memphis in 1970, I approached the agricultural authorities at the high-containment laboratories on Plum Island, off the northwest end of Long Island, New York.
The laboratories at Plum Island are designed to protect American livestock from ‘exotic’ animal diseases and their causal agents. By studying such exotic agents under high-security conditions, the scientists develop vaccines, antiviral agents and control strategies in case these agents are ever introduced into the country. Jerry Callis, the director of the Plum Island facility, was extremely interested in my proposal. Fowl plague can kill every chicken, turkey or other type of poultry that contracts it (see Chapter 2), and Plum Island had no scientists working on this enormous potential threat to the American poultry industry. Callis invited me to visit the island and present the proposed study to his staff. The staff were also positive, and Charles Campbell agreed to provide space and training so that my colleague Allen Granoff and I could work in his high-security laboratories.
Since Plum Island is approximately 2000 kilometres from Memphis, and these experiments were going to take several weeks to complete, it was necessary to organise travel and accommodation. Travel was straightforward: by air to New York, then by bus to Greenport, Long Island, the nearest small town to Plum Island, and then by private government ferry to the island itself. For the last leg, we needed either to have a security pass or to be travelling with a security-cleared staff member.
Once we had settled in, we realised that the cost of our mainland accommodation was quickly going to become prohibitive. Charles Campbell, my adopted mentor at Plum Island, came to the rescue and asked Callis if we could stay in the safety officer’s accommodation on the island. (One senior scientific officer always stayed on the island every night in case storms prevented the ferry from bringing staff to care for the experimental animals.) Callis agreed, and the arrangement was a tremendous bonus. Although we had to prepare our own meals and clean our rooms, we also had uninterrupted time to discuss science with the safety officer.
Turning up for work each day at the laboratories entailed disrobing and donning clothes to be worn on site only. At the end of the workday we had to leave all work clothing behind, take a thorough shower and put on street clothes. Nothing but people left the laboratories. All the air from the building was filtered to remove any particulate material, including viruses, and all water and waste were autoclaved. Once they were approved as completely sterile, they were discarded into the sea.
In the first experiment, Granoff and I investigated whether two different avian influenza viruses would exchange their haemagglutinin (H) and neuraminidase (N) surface spikes when they were put into a turkey together. Since these experiments were done in the era before genome sequencing became readily available, the only way to identify the H and N protein components on the viruses was with the specific antibodies used in the studies with Pereira.47 The turkeys were infected with the killer fowl plague virus (H7N7) and the turkey influenza virus (H6N2). The latter, which caused mild disease in poultry, shared its neuraminidase (N2) with the virus of the 1957 human influenza pandemic. The turkeys began to die after two days, and we found that one in every four viruses from the respiratory tract had reassorted their H and N proteins to produce H7N2 and H6N7 hybrid viruses. The H7N2 viruses were lethal to turkeys.48
In the second experiment, pigs were co-infected with two influenza viruses, one that could multiply in pigs and one that did not. For the former, we used the classical H1N1 swine influenza virus, the descendant of 1918 Spanish influenza virus that persisted in pigs. For the latter, we used the fowl plague virus. Two days later the pigs had a high fever of 40°C, and in lung samples we found that some of the viruses present had exchanged their H and N surface spikes (Figure 6.1).
In both experiments we found fewer reassorted viruses than parent ones, and the novel hybrid viruses were detected only after we had suppressed the parental viruses with antiserum specific to them. This led us to ask whether selection in nature could result in such new viruses becoming dominant. So, in the next experiments, we put our various infected turkeys in contact with birds vaccinated with the influenza viruses found in nature, with the expectation that the vaccine would suppress the parent viruses. Influenza viruses possessing the H7 protein of the fowl plague virus and the turkey virus N2 protein were detected in the vaccinated contact birds and rapidly killed them.
In the pig version of these experiments, we decided to use the human H3N2 influenza A virus, which had spread to pigs and then been re-isolated from pigs, and the classical H1N1 swine influenza virus. However, this time we made the experiment more like a real-world situation by putting one virus into one pig and the second virus into another pig. After six hours, both pigs were introduced into a group of four contact non-vaccinated pigs. By the seventh day we detected H3N1 hybrid influenza viruses (with the H3 haemagglutinin from the human virus and the N1 neuraminidase from the swine virus) and the other possible hybrid virus (H1N2) in one of the four contact pigs.
Figure 6.1 Reassortment of influenza viruses in pigs. Pigs were given, in the nose, a mixture of fowl plague H7N7 influenza virus (which did not multiply in the pig) and swine influenza H1N1 influenza (which did multiply). Two days later the pigs had a high fever (40°C) and were euthanased. Their lungs contained numerous influenza viruses, including the parental pair, H7N7 and H1N1, and the hybrid viruses H7N1 and H1N7. The hybrid influenza viruses were isolated by growing the mixed viruses from the lungs in the presence of antibodies specific for H7 plus N1, or H1 plus N7.
These experiments showed that when influenza viruses from different species co-infect an animal, new hybrid influenza viruses can emerge. Thus, the N2 on the virus that caused the 1957 human pandemic could have been a hybrid that emerged from animals.
The results from all these experiments were very exciting, and everyone in Campbell’s laboratory at Plum Island realised that we had hit a home run. Influenza viruses in different animals could indeed mix and exchange components under natural conditions, and the resultant hybrids could become dominant.
There was one drama while we were there that was a bit too exciting. During the second series of experiments with pigs and human influenza viruses in 1972, the laboratory technologist holding the animals called in sick two days after the experiment had started. The next day he phoned to say that he was severely ill, with a high fever. We were very concerned. While we had ensured the protection of the animal population and the environment, the protection measures for humans were rather primitive. We wore gloves, masks and gowns, and we had our showering routine, but we lacked the kind of personal protective equipment that would be required for such experiments today, such as a powered air-purifying respirator (PAPR) with a hood and mask, which covers the face entirely.
On the third day the technologist left a message to say he had been diagnosed with mumps, and there was a collective sigh of relief. (Though of course mumps can be a serious infection for an adult male, he recovered with no long-term consequences.) Although the incident did not result in human transmission, it made us aware of how easily influenza viruses could be spread between animals and then exchange components. I was convinced that it would be only a matter of time before we found evidence for this reassortment in the natural world. In fact we did not find it for nearly three decades, until the H5N1 bird flu outbreaks in Hong Kong in 1997 (see Chapters 10 and 11).
In tandem with the reassortment investigations outlined above, we began to look for possible parent or precursor viruses for the H3N2 Hong Kong influenza pandemic of 1968. Since only the H spike of this virus was new, we focused on this spike in our search. Through WHO collaborations we obtained influenza viruses from ducks, pigs and horses from around the world to compare with the H3N2 virus. Two viruses of interest emerged from our comparisons. An influenza virus isolated from horses in Miami in 1963 had a similar H spike to that on H3N2, as did one from ducks in the Ukraine obtained that same year.49
To determine the extent of molecular similarity among the three haemagglutinin proteins, in 1972 Laver isolated the haemagglutinin from all the viruses and ‘mapped’ their peptides. Peptide mapping was a key method of studying proteins before the age of sequence analysis. It produces a map of the peptides (short sequences of amino acids) that make up a protein. Identical proteins have maps that completely overlay each other. If the proteins are completely different, their maps do not coincide at all; if they are slightly different, a few spots do not coincide.
Part of the haemagglutinin of the duck, horse and human influenza viruses was identical (the backbone light chain), while the rest of the molecule had a small number of differences, according to the peptide maps.50 These molecular studies explained the similarities we and others had found among sera targeting isolated H spikes, and they added to an increasing body of evidence from studies on the virus of Asian 1957 H2N2 and the virus of Hong Kong 1968 H3N2 that animal influenza viruses contribute component parts of pandemic influenza viruses.
I learned one of life’s unforgettable lessons after my whirlwind visit to Pereira’s laboratory at Mill Hill to set up the neuraminidase inhibition test to identify which component turkey and human influenza viruses shared. The final test results came out very close to my plane’s departure time, and in the rush to catch the plane, I put all of the data from the experiments in my checked baggage. On arrival in Hong Kong, my luggage was missing, and I became frantic because we had no copy of the data (this was 1967, before the computer age). The airline was very helpful, but it took over an hour to locate my suitcase, languishing at Heathrow because of my late check-in. It was put on a later flight to Hong Kong, and I was very relieved to receive it before boarding my flight to Canberra.
Another life lesson followed shortly thereafter. One of the greatest disappointments for a scientist is to have work that he or she considers earth-shaking rejected for publication by a scientific journal. This is what happened to us with the Plum Island studies. My colleagues at Plum Island and at St Jude were excited about the results, which showed for the first time that influenza viruses ‘in a real-world situation’, when the right conditions were present, could easily reassort their genes and produce a new or novel virus. I was the most excited, for the findings offered a possible explanation for the genesis of both the Asian 1957 and Hong Kong 1968 strains of the influenza virus. Imagine my bitter disappointment when the Journal of Experimental Medicine rejected the article owing to a perceived ‘lack of interest’ to their readers. Granoff immediately suggested sending the article to George Hirst, now the editor of the journal Virology, who had discovered the haemagglutinin of influenza viruses. He would appreciate its significance. Indeed he did, and the paper was accepted with only minor changes for publication. The lesson for young scientists is not to be disappointed by initial rejection. You may just need to try a different, more appropriate journal. And then perhaps try again.