A new study from a glycobiology laboratory at MIT is creating a buzz in the flu community (see the MIT Press Release here). A great deal of molecular biology and virology studies what happens when the virus gets into a cell to use the cell’s own machinery to make copies of itself. Glycobiology is a relatively new area, concentrating on the straight and branched chains of sugar units that make up a great deal of the “stuff” one finds outside of a cell. What do these sugars have to do with influenza?
In earlier posts (see here and here and links therein) we showed how the influenza virus recognizes the cells it is going to infect by the cell’s outward appearance. That appearance, akin to a suit of clothes, is made up largely of a covering of sugars attached to proteins that stick on or through the cell membrane. There are a number of different sugars and each can have chemical modifications and unlike the amino acid units that make up a protein they can branch. The result is a tremendous diversity of clothes and accessories a cell can “wear” to change its appearance.
Studying these sugar chains (also called glycans) is a formidable technological challenge. We’ve known for a long time glycans are important in influenza because they determine whether and where the virus can grab onto the host cell in a way that opens the door to get inside. The copying machinery the virus needs to make copies of itself — essentially its only activity — is inside the cell. The way in is through a door marked by a specific glycan on the cell surface. About 25 years ago it was determined that the virus grabbed on to sugar chains with a specific sugar at its tip. That sugar is called sialic acid (pedantic digression: strictly speaking it is one particular sialic acid, the N-acetyl derivative of the core 9 carbon sugar neuraminic acid, often abbreviated Neu5Ac; the 5 designates carbon 5 of the 9 carbons, the place where the acetyl group attaches). From the virus’s perspective, the sialic acid (aka Neu5Ac) also has to be connected to a specific sugar unit as the next unit in on the sugar chain. That sugar is called galactose (it is one of the two units that make up milk sugar, lactose). The sialic acid can hook on to the galactose either of two ways, designated alpha 2, 3 or alpha 2, 6 (the alpha tells us that the linkage is below the plane of the sugar ring form while the 2, 3 or 6 tells us that the connection is between the number 2 carbon of sialic acid to either the number 3 or number 6 carbon of galactose).
The story for the last 25 years has been that influenza viruses of birds like to latch on to alpha 2, 3 glycans (specifically the sugar chains with a sialic acid at the tip and the 2, 3 linkage to the galactose next in line) while influenza viruses circulating in mammals, including humans, prefer alpha 2, 6 linked glycans of the same sort. The new work doesn’t change that story but it modifies and refines it. We’ve known for quite a while there must be more to it because there are examples of influenza viruses that prefer host cells that have alpha 2, 6 glycans on their surface but they don’t transmit well in humans. The suspicion has been that features of the glycan further upstream also affect virus behavior and the MIT work confirms this and spells out a feature the researchers believe is one of the differences. Their paper is not an easy read for a non-specialist, but here is some of what is in it (without most of the messy details).
The first (of four) analyses looked to see where in the human bronchial tract (the conducting system that brings gases two and from the outside) there were cells with alpha 2, 6 linkages. Here’s a detail, but an interesting one. Sugars hook onto proteins in either of two ways, called N-linked and O-linked (we’re now at the other end of the glycan [sugar] chain where it hooks onto the protein; the sialic acid is at the opposite [free] end, the tip, waving around at the outside). The N-link hook up is always to the same amino acids, asparagine and either serine or threonine for O-links. It’s more constrained than that. The only places that sugars can hook up are when an asparagine and a serine/threonine are at either end of a sequence of three amino acids, the middle one being variable (this amino acid sequence of asn-X-ser/thr is called a sequon). Thus you can always find potential sites where sugars might hook up by looking at the amino acid sequence of a protein (these are called potential glycosylation sites). Not all sequons are glycosylated, however, and they can be glycosylated with all sorts of sugar chains.
Interestingly the MIT team found alpha 2, 6 marked cells distributed throughout the human respiratory tract but O-linked glycans were very localized onto goblet (mucus producing) cells. By concentrating on the N-linked glycans they determined there was a significant diversity of sugar chains, i.e., the chains varied significantly in length and make-up. But they did have an interesting commonality: they tended to have long branches containing multiple lactosamine repeats. Lactosamine is just like lactose (a galactose and glucose hooked together) but instead is made up of units of galactosamine and glucosamine, where the simple sugars are slightly modified by an amine group attached to each). Determining all this was not easy. The investigators used mass spectrometry to glean this information, often a tricky and difficult technique.
After getting this far they looked at what happens when you fix one end of the chain when the sialic acid fits into a niche in the virus’s hemagglutinin (HA) molecule, the protein on the viral surface that is the docking component corresponding to the glycan “door” on the host cell. By comparing the differences in shape when the fixed sialic acid is connected either via an alpha 2, 3 or alpha 2, 6 linkage to the galactose next in line they determined that the alpha 2, 3 link permitted further contacts with viral HA contact points only in a space roughly cone shaped and essentially only involving the sialic acid and the neighboring galactose units. That was also true if an alpha 2, 6 link was followed by a fairly short sugar chain before hooking to the protein root, but if the chain was longer (at least four sugars) and had lactosamine repeats, the space of further contact with the HA was much broader. In the alpha 2, 6 linkage it was possible for the more distant sugars to bend back toward the area where the sialic acid and galactose were making contact and themselves participate in contacts with the viral HA. This produced an “umbrella like” topology, i.e., a docking shape for the viral HA that was different than either alpha 2, 3 (avian) glycans or alpha 2, 6 short chain glycans. Thus not all alpha 2, 6 are the same. Some, the ones with short sugar chains, are more like alpha 2, 3 in their docking shapes than the longer chain alpha 2, 6 glycans.
The MIT team then verified that the binding preferences of known influenza viruses were consistent with this picture by examining data from arrays of glycans of known structure. The short chain alpha 2, 6 and alpha 2, 3 bound the bird viruses while the longer chain alpha 2, 6 the human viruses. This strongly suggests that it is the shape of the docking zone on the host cell that must be matched by the viral HA, not necessarily the alpha 2, 3 or alpha 2, 6 linkage.
This is fascinating work but we are still at the beginning. It has implications for how research is carried out in the future, for example, how glycan microarrays are characterized and populated. But the headline that we now “know” the secret to what determines human infection is perhaps an exaggeration. What we now know is that the linkage itself may not be that important and we are further directed to some very fruitful areas to look to see what is important.