In recent years, glycan arrays have been used to identify the cell-surface sugars bound to by pathogens such as the flu virus. In parallel, companies such as CombiMatrix in Mukilteo, Washington, have developed specialized diagnostic instruments to identify pathogens, including tools to differentiate flu strains. Now researchers at the MITRE Corporation in Bedford, Massachusetts, and the University of California, San Diego (UCSD), have developed a method that could blur the line between these tow types of tool.

The device was described in the December issue of IEEE Sensors Journal by MITRE researchers Grace Hwang and Elaine Mullen and UCSD researchers Lin Pang and Y. Shaya Fainman4. It features an array developed at the UCSD with a gold surface that is perforated with nanometre-wide holes. A glycoprotein is attached to the gold surface inside the hole and the pathogen or carbohydrate-binding lectin is added. The instrument detects binding events through surface plasmon resonance (SPR), measuring fluctuations in electron density at the boundary between the metal and a dielectric surface.

Hwang and Mullen use microfluidic delivery channels to place the glycoproteins on the gold surface. When the proteins attach to the gold there is a detectable change in plasmon resonance. Once this reaches equilibrium, the pathogen or lectin is introduced using the delivery channel and any binding that takes place further changes the plasmon resonance. The device is reusable as acids can be used to break the glycan–lectin bonds and clean the array.

“The reason we wanted to use a plasmonic device was because plasmons are very sensitive to perturbations at the metal–dielectric interface,” says Hwang. For studying pathogens such as the flu virus, sensitivity can be an issue. The binding of the virus to different oligosaccharides occurs in the low millimolar range says Ian Wilson, a structural biologist at the Scripps Research Institute in La Jolla, California. As a result, he notes, glycan arrays often need to amplify the fluorescence signal, which requires additional antibodies.

Surface plasmon resonance provides label-free methods to look at carbohydrate interactions. Credit: G. HWANG

Hwang and Mullen's system avoids this problem, as plasmon detection does not require fluorescence to measure binding interactions — potentially opening the instrument up to a wide range of sensitive interactions. At the moment, however, the researchers are still working to improve the device's sensitivity for detecting flu viruses — their calculations suggest that it should be possible to identify up to one million influenza particles per millilitre.

Unlike other glycan arrays, the SPR system doesn't need printing or linkers to attach sugar targets to the gold surface. “The disulphide bonds in the glycoproteins will typically break and then bind to gold spontaneously,” says Mullen. The is helpful because when the sulphide bonds form with the gold surface, the oligosaccharides of the glycoproteins are oriented properly with their bioactive sugars projecting towards the medium.

Nevertheless, using glycoproteins in this way means that Hwang and Mullen have to choose carefully and be confident in the glycosylation patterns of the glycoproteins they use as their target. To help them, Mullen and her colleagues built a database called SugarBindDB (http://sugarbinddb.mitre.org). “We know which glycoproteins to choose by looking at our own database of pathogens and their specific sugar sequences,” says Mullen. “Then we go to the GlycoSuite, a database of oligosaccharides, to determine which glycoprotein it was attached to when it was isolated and what organisms it came from.”

Hwang acknowledges that it is challenging to identify potential glycoproteins that present only the sugars required for selective pathogen sensing. If the glycoprotein displays a mixture of sugars, then it could bind to non-pathogens. It is even more difficult to identify potential pathogen targets displayed on glycoproteins in human tissue, but she and Mullen think this is a challenge not just for their device, but for glycobiologists in general.

“I realized from discussions with other researchers that predictive tools to compute binding affinities between sugars and lectins do not exist today,” Hwang says, noting that this is a gap in glycan research tools that does not exist for nucleic acids and proteins. But she thinks in time, as more biophysical information is gained about glycan structures and properties, glycan arrays will catch up.

N.B.