|Now, when I say 'paradigm-busting', it doesn't necessarily mean that other scientists disagree or that they were/are not aware of the particular issue or question, but rather that these questions were often simply ignored, such that a set of 'conventional wisdoms' that formed the bulk of our knowledge of influenza was developed over time based, in part, on rather shaky foundations (again see examples, also Neumann 2009, below).
One of the most enduring 'paradigm-busters' that was raised in those early days of my education, was the 'n=1' problem, ie that our understanding of how influenza viruses behave in humans was based on one single virus - the 1918 H1N1 pandemic virus - and its descendants, which all contained at least 5 of the 8 genes from the same origin. As a result, it was not possible to determine to what extent observations from this single-sample could be extrapolated to future pandemic viruses with different ancestry, eg to H5N1, which has genes from entirely different origins.
For example, scientists have found that certain mutations seem to affect virulence and/or host range (ie whether a virus can replicate in certain hosts eg humans, which by definition is a required step for a pandemic). One particular mutation E627K in the PB2 protein attracted the most attention because it's been found that this one single change was enough to cause host-switching (Subbarao 1993). When comparing viruses after stable host switching (which in the case of humans means indefinite h2h, whether in pandemic or seasonal flu), the pattern was very consistent - all samples isolated from humans contained K, and all samples isolated from birds contained E (Taubenberger 2005, Finkelstein 2007). Therefore it seemed safe to conclude that E627K is a mammalian or even human adaptation, that is required for host switching, and/or that causes more efficient replication and therefore more severe disease. Indeed there were plenty of studies that supported such a 'rule', and, over time, some scientists even began to propose a possible 'mechanism' to 'explain' such a rule, based on replication efficiency at different temperatures (Massin 2001).
The problem, of course, was that all such studies based on human-adapted viruses before the 2009 pandemic, were by definition plagued by the 'n=1' problem, such that the conclusions may be applicable to this particular lineage, but there is no data to support whether such findings can be extrapolated to other viruses. By extension, it also means that such evidence does not allow us to make meaningful predictions based on the presence or absence of a particular mutation in virus with genes from a different ancestral origin, eg H5N1 or H9N2, and of course the 2009 H1N1, as discussed below.
Indeed, such conservative interpretation seems to be borne out by the overall experience with H5N1. Although initially there were studies based on the 1997 virus that suggested a change from E to K increased virulence in mice, (Hatta 2001, Shinya 2007), subsequent more extensive studies on viruses from 2003 onwards did not show any consistency as to whether a E627K change increased virulence in humans, or in animals such as mice, ferrets, or non-human primates. (Chen 2006, Govorkova 2005, Maines 2005, Salomon 2005, Peiris 2007, Li 2009)
In addition, there's evidence to suggest that rather than single-gene mutations, viruses may need the 'right' combination of mutations (Memoli 2009) or genes (PA, PB1, PB2, and NP) that produce an optimally functioning ribonucleoprotein (RNP) unit (Salomon 2005, Watanabe 2009), in order to remain viable, switch hosts, cause disease, and/or become more virulent etc. The 'rules' for such changes or combinations are, of course, not yet determined!
Despite these limitations and uncertainties, there is no shortage of scientists, even eminent ones, (and of course bloggers!!) who continue to present the "E627K = mammalian adaptation/increased virulence" model as if this is an established universally applicable paradigm or rule. Just to give one example, as recently as June 2009, after the new H1N1 virus has already been sequenced and shown to contain E, a review published in the very prestigious journal Nature (Neumann 2009) described this and other supposedly known molecular determinants of viral pathogenicity (see table below) in an entirely factual manner, without caveats. (The only caveat given was that "the determinants of pathogenicity may differ among animal species.")
Out of that list, at least the first 3 of these 'rules' have now been shown to be not applicable to the 2009 virus. The study regarding PB1-F2 is well explained in this blog 'It's not easy to make the 2009 H1N1 influenza virus a killer' so I won't comment further.
As for the PB2 mutations, several different groups have now published their findings, in the context of the 2009 H1N1 virus. One group studied the effect of the single mutation E627K, and found no difference in virulence or growth rates in mice (Zhu 2010). Another group studied 3 different mutations, E627K, D701N, and E677G, and found none of them made a significant difference in replication, virulence, or transmission, in mice or ferrets! (Herfst 2010) By far the most interesting paper is the one from Taubenberger's lab (Jagger 2010). Because of the study design, they arrived at some really interesting results which deserve more exploration, so I'm going to save that for the next diary (this one is already way too long, sorry!) For now, on the question of whether E627K and/or D701N increased virulence, they found the opposite, that these mutations attenuated (ie weakened) the 2009 H1N1 virus.
That, ladies and genlemen, is 3 strikes out of 3.
To conclude (for now), we now know that:
- E627K (and D701N, PB1-F2 etc) do not enhance virulence in the 2009 H1N1 virus, in mice or ferrets
- (since the virus is clearly pathogenic) the 2009 H1N1 depends on other as-yet-unidentified mutations or sets of mutations to achieve host switching, from (possibly) swine to humans
- we can no longer assume that results from studies of a particular lineage of virus (eg 1918 and descendants) will apply to a new pandemic virus with different origins
In other words, the 'n=1' concern has now been demonstrated to be valid, thus challenging a whole lot of 'conventional wisdoms', things that we used to think we know, about influenza. I've heard more than a few scientists say, when talking about flu, "The more I learn, the less I know." I can't agree more!
The upside of that is, now that we have this new virus circulating in humans, we have a second human-adapted lineage that we can study, ie we now have n=2. 2 is better than 1, obviously, not least because it allows us to re-define the boundaries of our certainties and uncertainties, but, alas, 2 is still 2. I don't know how high n has to go, for us to accumulate enough knowledge to be able to define some vaguely 'universal' rules about influenza (especially for the purpose of making predictions!), but 2 sure is not it!!
I'll close this diary with the following quote, which appears, as a quote (doh!), in a paper authored by Jeff Taubenberger and David Morens The Pathology of Influenza Virus Infections
"We regret very much the fact that an influenza virologist is unable to live say 200 years, so that he himself would be able to see what has developed from his earlier assumptions." J. Mulder and J.F.P. Hers: Influenza (1)