Friday, March 24, 2006

The papers in Science and Nature

The newswires have been buzzing about the papers just outin Science and Nature concerning the location of H5N1 attachment in the human respiratory system. Most stories jump to the punchline: what it might mean for human susceptibility and transmissibility. That part, however, is interpretation and speculation. Let's look to see what these papers actually say.

The central issue is the question of host range (or host specificity), that is, what makes one influenza virus primarily a bird virus and another one a human virus? This also bears on the ease of bird to human transmission and possibly human to human transmission, although the latter is where the speculation rather than the scientific results enter. First a bit of background. If you want even more, we did a four part series not so long ago that lays the science out in even more detail here, here, here and here. I'll crib a little from them to save time (after all, I'm just plagiarizing myself).

The influenza (or any) virus needs to get inside a host cell in order to make new copies of itself. Reproducing is essentially its only task in life. We know that viruses and other pathogens don't usually infect all animals (they have a specific host range) and within an animal, usually infect only specific tissues. So cells from different animals and different tissues must somehow look different to the virus. How does a virus "recognize" the right cell?

The first view the virus gets of the cell is a surface covered by a dense canopy of sugars linked to cell surface proteins. This outer fur-like sugar surface is called the glycocalyx and plays an important biological role, including cell-cell recognition and communication, interacting with and binding of cells to the material that glues cells together (the extracellular matrix), altering or modulating the response of immune cells and proteins, and, most important for our purpose, protecting against or determining sensitivity to pathogens like the flu virus. The influenza virus has learned to recognize one of these projecting sugars and uses it to grab onto the cell and initiate the process of getting inside it.

The particular sugar we are interested in is called sialic acid. It often rests at the tips of a sugar chain in turn attached to proteins that are part of the cell surface (see the earlier posts for pictures and a lot more explanation). How the sialic acid is attached to the other sugars in the chain is the key to what the papers are about. For our purposes there are two ways this attachment can be done, designated either an α-2, 3 or an α-2, 6 linkage. These denote two different ways to attach the sialic acid tip to another sugar, galactose, which is next in line in the sugar chain. Other sugars and configurations further in may also be important but the papers at issue consider only the Sialic Acid α-2, 3 or α-2, 6 Galactose linkages (often written NeuAc α-2, 3 Gal or NeuAc α-2, 6 Gal because another name for sialic acid is neuraminic acid; the two papers use the designations SAα-2, 3Gal and SAα-2, 6Gal). This all may seem somewhat esoteric, but it turns out that avian viruses like to bind to sialic acid linked α-2, 3 while human viruses like the ones linked α-2, 6.

The story used to be fairly simple. Birds had cells with SAα-2, 3Gal visible and available for binding to the virus in their intestines and influenza in birds is primarily an intestinal infection. Human beings have SAα-2, 6Gal in their respiratory tract and the viral subtypes that looked for that linkage were the ones that infected humans. Alas, like everything else related to influenza, the story is more complicated. For one thing it was discovered several years ago that humans also have SAα-2, 3Gal on some of their cells. But which cells? That's where these papers come in.

In 2004 Matrosovich and colleagues used a tissue culture system derived from the human respiratory tract to try to figure out which cells had α-2, 3 and which ones had α-2, 6 (Proc. Nat. Acad. Sci. 101:4620 - 4624. 2004). Their conclusion was that the avian virus infected ciliated cells of the respiratory tract while the human viruses infected the non-ciliated ones. The ciliated cells are higher up and in the upper, mid and a bit of the lower respiratory tract, decreasing in density as you go deeper. There are non-ciliated cells throughout, too, but deep in the lungs, especially in the areas where the gas exchange is taking place (the alveoli) the cells are non-ciliated. The number of α-2, 6 cells in humans seemed to be higher, but there were significant numbers of α-2, 3 throughout as well. This was a bit of a puzzle. Why didn't the avian virus infect humans more easily?After all, there were sufficient numbers of α-2, 3 cells in humans, too. Moreover, in human influenza infections, ciliated cells throughout the upper respiratory tract are infected. The story was unclear.

Matrosovich didn't use an intact human respiratory tract but a tissue culture model of one, that is, one that had the same kinds of cells as the respiratory tract but not organized into actual respiratory tract tissue. The two new papers were designed to answer the question of exactly which cells in the intact human respiratory tract the H5N1 virus preferentially bound.

The Dutch group didn't bother with determining α-2, 3 or α-2, 6 characteristics but instead incubated inactivated virus with a label on it with archived formaldehyde preserved tissue sections and then looked to see what part of the lung and which cells bound the virus. The results were that H5N1 mainly bound to a type of cell in the deepest part of the lung where gas exchange takes place (see the earlier posts for an explanation of lung structure). The cell is called a type II pneumocyte and it secretes a protein that helps the lung stay expanded by decreasing its surface tension. Secretory cells make abundant protein, which may be an advantage to the virus because it hijacks the cell's protein making machinery to make copies of itself. The virus also bound to wandering immune system cells called pulmonary alveolar macrophages, which play an important part in fighting pathogens and other garbage in the delicate tissues of the deep lung where gas exchange is the main order of business.

The Japanese group (Nature) did look for α-2, 3 cells and found them in the same place that the Dutch group saw viral binding: deep in the lungs in the type II pneumocytes. Additionally they showed that avian viruses bound to those human cells, as would be expected because they had α-2, 3 sialic acid linkages. Cells higher in the human respiratory tract (nose, throat, bronchial tubes that lead down into the lung) had abundant α-2, 6 linkages. Human viruses tended to bind to and infect cells higher in the respiratory tract, although some type II pneumocytes also had α-2, 6 linkages.

These findings show virus infecting non-ciliated cells just as in Matrosovich's study, but don't mention ciliated ones. So the story isn't complete. Nor does it explain a number of things, for example, if there are α-2, 3 and α-2, 6 studded cells throughout the human respiratory tract, why do the avian viruses seem only to prefer the ones down deep and the α-2, 6 viruses the ones further up? There are quantitative differences in the density of these cells but the details aren't known at the moment and other factors besides the sialic acid linkage might be involved. The finding that H5N1 (an avian α-2, 3 virus) destroys the deep part of the lung has been found clinically and is an explanation for the remarkably virulent nature of this disease. Orange over at the excellent bird flu site The Coming Influenza Pandemic? has resurrected several earlier stories identifying type II pneumocytes as the site of the primary lesion in fatal cases of bird flu, so this isn't a completely new finding.

Finally we come to the question of what it means. My answer may be disappointing. At this point we don't know. The investigators speculate (in the news stories more than the papers themselves) that the reason bird flu is not as "catchable" as ordinary flu is that its residence deep in the lungs makes its transmission more difficult. There is no mucus in the gas exchange units so coughing and sneezing is less likely to create a virus-containing aerosol. Or so it would seem. In truth, however, we don't know the main routes of transmission. Gas from the alveoli (the deep air sacs) is certainly expelled on exhalation, likely contains virus, and once outside the humidified environs of the respiratory tract would rapidly dessicate (dry up) and could form droplet micronuclei. The assumption that virus deep in the lungs is less transmissible might be correct but it has not been shown. Other factors might be involved. Neither paper tested the transmissibility question, which remains pure speculation (although not implausible). The Japanese paper also points out that if the virus were to develop the ability to dock with α-2, 6 cells, either in addition to or instead of α-2, 3, we could have a nasty actor on our hands. One isolate from Hong Kong in 2003 seems to have this ambidextrous character, although most H5N1s do not.

That's my initial read of these two fascinating papers. They are a step forward but leave much to be discovered. Contrary to the more optimistic interpretations, it is too early to conclude that H5N1 is not likely to be easily transmissible from person to person soley on account of its location deep in the lungs. The papers do not show this nor do they conclude this.

This is exciting science, but also urgent science. Much, much more to learn.