What can 1081 influenza viruses tell you?

By Jonas Waldenström

Today we published a major article in a well-respected journal. The reason why I write major is not to brag (although I am very pleased). No, the reason for that epithet is that the paper is based on such a huge long-term effort. In fact, in this paper, ten years of fieldwork, laboratory work, and statistical analyses are boiled down into nine glossy pages!

As frequent readers of this blog probably know, mallards and flu is our main study system. Through repeated captures, samplings and recaptures of ducks at a migratory stopover site we have built very large datasets that we now can analyze for long-term patterns in virus-host interactions. The title of the current paper is: “Long-term variation in influenza A virus prevalence and subtype diversity in migratory mallards in northern Europe”

Influenza A virus prevalence was in part determined by peaks of mallard migration. Photo by Serget Yeliseev under a CC BY-NC-ND 2.0 license.

Influenza A virus prevalence was in part determined by peaks of mallard migration. Photo by Sergey Yeliseev under a CC BY-NC-ND 2.0 license.

What we did was to screen all 22,229 samples collected in the period 2002-2010 for the presence of influenza A virus RNA. Positive samples were then inoculated in eggs in order to obtain virus isolates. After this process, we had a virus bank consisting of 1081 viruses of 74 different subtypes, ranging from H1N1 to H12N3. As you can see from the figures above, influenza virus research is time-consuming and costly, and the travel from sample to RRT-PCR-positive to characterized virus could be described as a negative logarithmic function. It is all about big numbers! You need a lot of samples to get the statistical power to say something about virus ecology and epidemiology at the level of subtypes. You also need to be stubborn as a mule.

There are three major results that I would like to share with you.

First, we were able to fit a model of how influenza A virus varied with season in the sampled mallard population. The resulting figure very neatly shows how the virus starts low in spring, becomes more or less absent during the breeding season, and how it suddenly increases in frequency in August when the first wave of migrating mallards arrive at Ottenby. The August peak is followed by a second peak in October-November, likely consisting of mallards with a Finnish or Russian origin. Actually, the plot looks like a camel!

Influenza A virus prevalence showed two distinct peaks in autumn, one in August and one in October-November.

Influenza A virus prevalence showed two distinct peaks in autumn, one in August and one in October-November.

However, plotting prevalence rates over time has been done before. The strength with our analysis is that it includes and accounts for the variation in prevalence induced by year effects. Mallards are migratory birds, but their timing of migration is rather flexible. In years characterized by mild autumns they arrive late at our study site, and in years with harsh autumns they are early. The final model accounted for approximately half of the variance in prevalence, which is pretty good all considered.

Second, I would like to stress the incredible diversity of subtypes! The two surface proteins hemagglutinin (16 variants) and neuraminidase (9 variants) sit on two different RNA-segments in the genome and can theoretically be combined in 144 different ways, or subtypes as we call them. We found 74 different HA/NA subtypes. In addition, some subtypes are likely not functional, or would have to include a hemagglutinin (like H14 or H15) that is restricted to areas outside Europe. This plethora of genotypes is a world record from a single site. Or to put it in perspective: more than half of the possible subtypes have been found in mallards trapped in our little duck pond on the southern point of the island Öland, in the SW part of the Baltic Sea, in Northern Europe. A speck in the ocean, but a global diversity of viruses.

Further, the 1081 viruses were not evenly distributed on subtypes. Rather, some subtypes were very common, such as the H4N6, the H1N1, or the H2N3 subtypes. Others were rare, including the famous combinations H5N1 and H7N9, both which were only found once, and not in the pathogenic forms known from elsewhere. Interestingly, the high frequency of certain combination, and a low frequency of other combinations despite the HA and NA being common in other virus constellations suggests that some subtypes have low fitness. Consider for instance H4N3 that was found only 5 times, while the H4 hemagglutinin was found in 291 viruses, and the N3 neuraminidase in 116 viruses.

A cute mallard couple. Photo by Chuq Von Rospach under a CC BY-NC-ND 2.0 license

A cute mallard couple. Photo by Chuq Von Rospach under a CC BY-NC-ND 2.0 license

Third, and perhaps most interestingly, we found a heterosubtypic effect at the virus population level. By grouping viruses in classes depending on their HA relatedness we could see that the different virus classes peaked at different times within an autumn. The virus type that was common in early autumn was rare in late autumn and vice versa. Understanding how individual and herd immunity processes affect influenza A virus dynamics in nature is highly warranted, as that would aid our capacity to predict how the virus population could change over time. Viruses in wild birds remain an important pool from which genotypes could be seeded in domestic animals, and even humans.

Finally, I would like to say how incredibly fortunate I am to have had the opportunity to work in such a hard-working and persistent research group. The work we presented today has been collected by a small army of duck trappers, a score of laboratory staff, a handful PhD-students, a couple of postdocs and a quartet of PIs from Kalmar, Uppsala and Rotterdam. And the most important of all was Dr Neus Latorre-Margalef, who carried this publication from start to finish! Well done!

Link to the article:

Latorre-Margalef, N., Tolf, C., Grosbois, V., Avril, A., Bengtsson, D., Wille, M., Osterhaus, A.D.M.E., Fouchier, R.A.M., Olsen, B. & Waldenström, J. 2014. Long-term variation in influenza A virus prevalence and subtype diversity in migratory Mallards in Northern Europe. Proceedings B, online early.


If you enjoyed this post, or other posts on this blog, why not follow the blog via email, Feedly or get updates via Twitter by following @DrSnygg?

How bats in Peru change our view of flu (and it rhymes!)

By Jonas Waldenström

I am the real Bat Man and here to bite y'all

I am the real Bat Man and here to bite y’all

One of the major news in the virology community last year was the publication in PNAS describing a completely new influenza A virus. In line with the taxonomy traditionally used for influenza viruses it got the name H17N10, illustrating that it possessed novel hemagglutinin (H17) and neuraminidase variants (N10). However, it wasn’t the numbers that was the ground breaking news, it was the fact that the virus was detected in a Central American bat, and not in a bird. A tropical bat is very far from the ‘normal’ diversity of influenza A viruses seen in wetland birds and waterfowl. Although bats and ducks both have wings, in evolutionary terms they separate a very, very long time ago in the age of dinosaurs. In fact, there are more differences than similarities between bats and gulls in ecology, physiology and aspects of cellular biology. Hence, the bat flu was a remarkable observation. A real shaker. In one sweep, the whole flu field needed to come with terms that not all viruses are bird viruses.

The initial findings also hinted that the first bat influenza virus was unlikely to be alone. An influenza-iceberg, of sorts, made up of fluffy, winged mammals. This week, a first follow-up was published in PLOS Pathogens. A crew of (mainly American) scientists analyzed samples from bats sampled in the Amazonian parts of Peru in 2010, collected as part of CDC’s tropical pathogen surveys. In total, 114 individuals of 18 bat species were taken out from the freezers and different sample types were screened with a molecular method designed to broadly pick-up the RNA of any influenza A virus. They got one hit from a fecal sample in a single bat! A lucky shot at the Tivoli, given the low sample size. Prompted by this, the authors brought in the big machinery and sequenced the totality of the genetic material in the samples from this poor, long-dead bat and used bioinformatic tools to resolve the genome of the virus that had infected its intestines. When bit by bit was added it became clear that it was indeed a completely new influenza A virus, very different from avian viruses, and similar, but still distinctly different from the earlier H17N10 bat virus. And the name? H18N11 of course!

Please take a close look at the figure below. It shows the phylogenetic relationships of each of the influenza A virus’ eight RNA segments – in black are all ‘non-bat viruses’ and in red the two new bat viruses H17N10 and H18N11. For all the segments coding for ‘internal’ proteins, i.e. those involved in the polymerase machinery or the structural properties of the virus, you see that the two bat viruses are always found in a neat little red outgroup. This signals a long evolution away from other known influenza A viruses. It is a little prematurely to say exactly how long, but the branch lengths indicate that this happened a long time ago.

Phylogenetic trees for the 8 different IAV segements, see http://www.plospathogens.org/article/info%3Adoi%2F10.1371%2Fjournal.ppat.1003657

Phylogenetic trees for the 8 different IAV segements, see http://www.plospathogens.org/article/info%3Adoi%2F10.1371%2Fjournal.ppat.1003657

Now look at the hemagglutinin and the neuraminidase trees (HA and NA, respectively). The same pattern is repeated for the NA, but not the HA. In fact, the two novel hemagglutinins are nested within avian hemagglutinins. How can we interpret this? At first this doesn’t make any sense, but one has to remember that influenza viruses don’t evolve in the same way you or me, trees, shrimps or ferns do. Influenza viruses can reassort, meaning that if two viruses of different origin infect the same cell the different RNA segments can be put in new combinations in the resulting virions. Imagine two decks of cards being shuffled, one red and one black, and that each virion randomly consists of a draw of card from the combined shuffled deck, sometimes red, sometimes black, and sometimes mixed.  This is a rapid way in which new variants can arise, and a reason behind the genesis of pandemic flu in humans.

Returning to the bats, it seems that bat and avian viruses have met in a not too distant evolutionary past, and that a HA variant have sailed into the bat influenza gene pool. It will be interesting to see how the picture changes when more bat viruses are sequenced. Has there been one reassortment event, followed by drift and a subsequent separation into H17 and H18? Or, has there been many? Are there, perhaps, avian H17 and H18 to be found in South American birds? What about bats in North America, Europe, Africa and Asia?

One thing we can be sure of is that there are more viruses waiting to be detected and described. One sign of this comes from the current paper. The authors used the sequenced genomes to construct recombinant HA and NA molecules (using fancy virologist tricks) and used these to build assays (ELISAs) where bat sera could be screened for antibodies against the new HA and NA variants. Where the molecular screening yielded one positive bat, the serology approach found 55 of 110 bats showing signs of having been infected with flu earlier in life. This clearly indicates that influenza viruses are widespread in Peruvian bats, and likely in other parts of the world too. Moreover, they found cases of bats with antibodies to one of the recombinant HA or NA, but not to the other, suggesting that are more combinations of HA/NA to be found.

Finally, and perhaps the most interestingly of all results was that the hemagglutinin of bat influenza viruses does not to behave in the same way as avian hemagglutinins. When a virus is to infect a cell it needs the hemagglutinin protein to serve as a key, docking with a sialic acid receptor – the lock – on the cell. If the key and the lock don’t fit infection will not occur. For instance, a major division between human flu and avian flu is the preferred conformation of a galactose residue on the sialic acid receptors. This little difference makes it hard for avian viruses to infect humans, and vice versa. But with bat viruses it seems sialic acid receptors are not used at all! Instead bat HA uses an unknown receptor for cell entry. Holy Moses!

More to follow shortly, I suppose. Major obstacle at present is the lack of a culturing method for bat influenza viruses. Neither cell lines nor eggs have worked so far. Without the means to grow the virus it is very tricky to study it. But there are many clever virologists out there, so it is likely not too far away.

But I still prefer feathers to fur, and will stick with ducks.

Links to articles:

Tong S, Zhu X, Li Y, Shi M, Zhang J, et al. (2013) New World Bats Harbor Diverse Influenza A Viruses. PLoS Pathog 9(10): e1003657. doi:10.1371/journal.ppat.1003657

Tong S, Li Y, Rivailler P, Conrardy C, Castillo DA, et al. (2012) A distinct lineage of influenza A virus from bats. Proc Natl Acad Sci USA 109: 4269–4274.