There once was a nice tidy story about why avian influenza viruses infected birds and human influenza viruses infected humans and pigs were the "mixing vessel" that brought them together so they could reassort their genetic innards. It went something like this. The avian virus can only attach to and thus infect cells that have a specific kind of receptor on their surface. The receptor was a terminal sialic acid with a particular linkage to the underlying cellular glycoprotein (for more on this see our four part primer on glycoproteins beginning here). For birds the linkage was designated α2,3NeuAc and for humans it was α2,6NeuAc. Pigs had both in their respiratory tract so either virus could infect them. Hence the "mixing vessel." Not.
Then avian viruses started to infect humans and we took another look. The story changed. Humans did have α2,3NeuAc receptors but they were deep in the lungs. In order for an H5N1 virus to get that far down it would have to be in aerosol form and the prevailing view was that it was expelled from another flu victim in "droplet" form, i.e., in infective particles too large to get down deep in the lungs where the cells with the right receptors were located. Hence H5N1 was not likely to be easily transmissible for that reason. There were a couple of well publicized papers to that effect and much was made of the explanation in the news. We discussed it here and registered our reservations based on the fact that there were at least two other papers showing that H5N1 was binding to ciliated cells in the respiratory tract and they are only found above the deep lung.
Now further confirmation has come that the old story is, well, just a story. In a new paper in Nature Medicine by Nicholls JM, Chan MC, Chan WY, et al. (7 January 2007; thanks to SW for the .pdf) it appears that both H5N1 and H3N2 bind and replicate quite well to tissues in the upper respiratory tract. The paper also confirmed the previous findings that H5N1 and H1N1 bound to the type 2 pneumocytes deep in the lung and even to the wandering policemen of the immune system there, the pulmonary alveolar macrophages. The earlier paper had purported to show that the same binding didn't happen in the upper tract, and herein lies a cautionary tale. We noted in our earlier post that the researchers had not actually assayed the receptors in the lower and upper tract but had instead used a probe, called a lectin, that is known to bind to specific α2,3NeuAc receptors. The specific leptin used in that work is called MAA2 and indeed the new paper verifies that it binds well to the cells in the lower tract but not in the upper. But there is another form of leptin, MAA1 that does bind to α2,3NeuAc in both upper and lower tract cells. Why the discrepancy?
It turns out that MAA2 binds most efficiently to cells where the "further in" units on the glycoprotein are different. For MAA2 and MAA1 it turns out that it is not just the terminal linkage that determines whether the lectin will bind but the next one in as well. For MAA2 the link is to the 3 position of the next sugar and for MAA2 it is 4 position. So the use of MAA2 as a probe in the earlier work missed the ability of H5N1 to bind in the upper tract. Science is hard.
What's the significance of all this? First, it destroys the tidy receptor difference story as being the explanation for host specificity. The ability to bind to α2,3 or α2,6 may still be important but not quite in the ways we thought and clearly there are other factors involved we don't understand at all. Second, the deep lung story is down the toilet, too. This has implications for whether H5N1 is transmitted via aerosol or droplet or both. The old story implicated droplet. That's why the virus wasn't so transmissible. The new paper puts the aerosol back on the table, at least as far as binding evidence is concerned. Third, because H1N1 and H5N1 both bind to the cells in the deep lung, the severity of H5N1 infection isn't explained by the site of infection alone. Lastly, the lack of ready human to human transmission is not explained by an inability of H5N1 to replicate in the upper respiratory tract.
Once again H5N1 shows us there is much we don't know and also that some of what we thought we knew we didn't.
- Log in to post comments
Mmm, truncated?
Mathematician: Thank you for pointing this out. There was something in the code for the sialic acid designations that did this. I tried about ten different rewritings just now and finally had to eliminate the scientific designations as they were just truncating the end of the post. Probably just as well. Most readers didn't need to know that much.
Back to the drawing board.
Revere, Off topic, but did you see the article in Turkish Daily News today that references Effect Measure?
It's a retrospective of Dogubayazit.
http://www.turkishdailynews.com.tr/article.php?enewsid=63680
The words of Donald Rumsfeld came to me as I read this:
As we know,
There are known knowns.
There are things we know we know.
We also know
There are known unknowns.
That is to say
We know there are some things
We do not know.
But there are also unknown unknowns,
The ones we don't know
We don't know.
Nancy: No, hadn't seen it. Thanks.
David: Yes, I thought of him, too. Of course this falls into yet another category which he repressed: the things we think we know and are wrong about.
David,
Hear! Hear!
revere,
I read this yesterday and had this flash of 20/20 hindsight - we've been saying for a while that the best specimen for testing for H5N1 is a throat swab. Now why should that be unless the virus was doing more than being in transit there?
Another implication - We need to look more at the hosts; we've been very lopsided in our investigation by looking mostly at one side of the equation.
Another comment from one of Taubenberger's lectures, which you probably know already - it isn't just the terminal linkage that is important. The internal configuration of the dissacharides also determines specificity. Thus there are different subtypes of 2,3 and 2,6 receptors, and he speculates whether there are physiological differences in different age groups that affect the distribution of these subtypes in various cells of the respiratory system.
Which, if true, might account for the age distribution of severe H5N1 disease.
Interesting thoughts. Enough to keep lots of scientists busy for a long time, I think.
Hi, Revere
Wasn't this experiment ex vivo? I recall you saying earlier that the presence of mucous in the upper (not lower) airways could be a confounding factor on the binding. Does this experiment account for that as well? Thanks.
Revere, please clarify if MAA1 is meant for one of the MAA2 references for the 3 or 4 position of sugars:
It turns out that MAA2 binds most efficiently to cells where the "further in" units on the glycoprotein are different. For MAA2 and MAA1 it turns out that it is not just the terminal linkage that determines whether the lectin will bind but the next one in as well. For MAA2 the link is to the 3 position of the next sugar and for MAA2 it is 4 position. So the use of MAA2 as a probe in the earlier work missed the ability of H5N1 to bind in the upper tract. Science is hard.
anon_22: Yes, it is not only the next sugar on the chain but the one after that. This is in accord with the glycan chip data from Scripps which we commented on in another post.
epitope: Yes, this is an experiment on tissue ex vivo. Mucus has α2,3 linked sialic acids in it which could be a protective feature. So this experiment doesn't take that into account, but it wasn't meant to. It was to see if the receptors exist in the upper airway, which other data also suggested but some were denying. Thus this is an important paper in helping us understand what factors are involved. IMO we will find that there is much more to infectivity than the receptor story, which will undoubtedly be part of it. It probably involves NA and internal proteins and maybe even the NS protein. Stay tuned, as they say.
LEG: The system seemed to choke on the HTML for the characterizations, which perhaps why this was confusing. The sialic acid links detected by MAA1 and MAA2 are the alpha 2,3 gal links (both of them) but the next link is for a 1,3 galactose link and a 1,4 glucose link, respectively. Not sure if that is important to your understanding. The point is that MAA1 and MAA2 were making a finer distinction than we were previously using and that distinction turned out to be important for the upper and lower airway issue.
Revere. Would there be implications from these findings in choosing vaccine candidates? Thanks
Tom: Don't think so as the receptor sites and the epitope sites are different.
or, if you are not a fan of Rumsfeld, maybe Mark Twain (or Yogi Berra) who said,
"What gets us into trouble is not what we don't know.
It's what we know for sure that just aint so."
Thanks for being at the forefront of unwinding what we know for sure.
Earl: Thanks. Smart guy, that Twain. Or was it Yogi?
Jacking the blogosphere Revere for a moment....Tsunami Warning is now in effect for much of the pacific. Everyone should check the wbsite for applicability to your area http://wcatwc.arh.noaa.gov/eventmap.html
According to Sky TV 10 mins. ago the tsunami is a tiddler.10cm rise in Japan,or thereabouts.Relax.
FYI - newish story on new clinical trial on flu vaccine from NIH. Apologies if already covered -
http://www.techreview.com/Biotech/18048/
Global: Thanks. Did see it. I covered the similar PowderMed trial going on in the UK which started earlier than the NIH one but also a DNA vaccine. We'll see. Things that work in mice often don't work in people. Lots of interesting new vaccine work going on and lots of press releases going on with them.