At A Glance:
Professor of Glycosciences, director of the MRC Glycosciences Laboratory, Imperial College School of Medicine, London
Director of The Glycosciences Laboratory, Northwick Park Hospital, Harrow, UK
MB (Bachelor of Medicine)/BS and MD, University of London, Royal Free Hospital School of Medicine
Published a paper on oligosaccharide microarrays in this month’s Nature Biotechnology
How did you get into studying carbohydrate-protein interactions?
I became interested in autoantibodies triggered by infection in the course of doing a residency in hematology. One particular infection I was interested in is caused by Mycoplasma pneunomiae; it gives rise to antibodies against red blood cells.
In the course of working out the specificities of these antibodies, I found that they recognize a carbohydrate structure, the so-called I antigen, which turned out to be related to the ABO blood group antigens. We later showed that the Mycoplasma actually adheres to a structure which is related to the I-antigen on the respiratory epithelium, and somehow this process triggers the anti-I antibodies.
I also had been working with anti-i autoantibodies, which recognize a fetal carbohydrate antigen on red cells. Thus, the I and i are developmentally regulated antigens. I realized soon that these antibodies were very powerful reagents to study developmentally regulated changes in carbohydrates, and I used them to study the mouse embryo. This was of course in the late 70s, early 80s, and from then on I became interested in the roles of oligosaccharides and predicted that they are bearers of biological information.
This is now being borne out, because increasingly we are aware that specific oligosaccharides are directly involved in a number of important processes: protein targeting, protein folding, certain mechanisms of infection, inflammation, and immunity. We are interested in the way that oligosaccharides confer diversity to proteins and to learn how they modulate the functions of proteins. This in turn makes it important to identify proteins that recognize oligosaccharides.
What is the role of microarrays in studying those protein-carbohydrate interactions?
To establish whether a protein recognizes carbohydrates, and which particular ones, has been one of the challenging areas of cell biology because of the huge heterogeneity of oligosaccharides, their availability in only small amounts, and the fact that you can’t clone them, as they are the products of multiple glycosyltransferases.
There is a great need for microarray technologies for oligosaccharides that would enable systematic and high-throughput analysis of protein-carbohydrate interactions. Two approaches already exist: one is a microarray for polysaccharides, developed by Denong Wang at Columbia University and colleagues, the other one is a microarray for monosaccharides, developed by Benjamin Houseman and Milan Mrksich at the University of Chicago. But there has not been a published procedure for microarraying oligosaccharides, although the NIH sponsored Glycomics Consortium and some companies, for example Glycominds, are beginning to synthesize oligosaccharides chemically for array purposes.
So what is special about the arrays you describe in this month’s Nature Biotechnology?
The important feature of our arrays is that they are sourced from glycoproteins, glycolipids, proteoglycans, polysaccharides, and even from whole cells or organs, as well as chemically synthesized oligosaccharides. I think this is what makes ours a unique approach. We want to use these to explore what is now being called the ’glycome’. The actual numbers of oligosaccharides in the glycome is enormous, and I would say unfathomable. There is a huge variation in their backbone chain length, and also in the way they are linked to one another, and often they have modifications, for example sulfation or phosphorylation. In addition, dramatic changes take place during embryonic development and cell differentiation. It is not practical to isolate and sequence all oligosaccharides in a glycome.
What our approach is able to do is to focus on those oligosaccharides that are bioactive. In our paper, we sourced oligosaccharides from the brain, which we fractionated according to whether they were neutral, sialylated, or sulfated. We arrayed these fractions and then probed them with carbohydrate-binding proteins, in this instance monoclonal antibodies. What we found is that these antibodies could pick out their determinants from these heterogeneous oligosaccharide fractions. What our technique then has the capacity for is what we call ‘deconvolution.’ You pick out a fraction of interest, this is resolved by thin-layer chromatography, and then the bound oligosaccharide can be sequenced by mass spectrometry. The value of this particular approach is being able to work with natural sources – a glycoprotein, a mucin, or a whole organ or cell type or cell line, and being able to pick out the ligands.
How did you make the arrays described in your paper?
The arrays are based on a technology we have been developing for a number of years, so-called neoglycolipid or NGL technology.
The principle of this technology is to generate immobilized oligosaccharide probes by conjugating them to a lipid. The value of the lipid-linked oligosaccharides is they can be immobilized, and when they are immobilized, they are naturally clustered because lipid moieties stick together. Clustering or multivalence is very important in carbohydrate-protein interactions; a one-to-one interaction of an oligosaccharide with a protein that recognizes it is very rarely detectable, there are very few interactions of sufficient high affinity. In order to be able to detect binding by conventional assays, you need clustering and multivalence. Of course in Nature this happens on the cell membrane.
We found that the neoglycolipids are very robust probes when they are displayed on nitrocellulose. They are arrayed onto nitrocellulose and we have so far tested them against monoclonal antibodies, selectins, interferon gamma, the chemokine RANTES, and have shown great selectivity in their ability to pick out their ligands. The detection system that we have used in the manuscript is still a conventional ELISA-type assay.
What are you working on now to improve the system?
At the moment our arrays are mini-arrays, but further miniaturization is under way. In the paper we describe a mini-array where the oligosaccharides are spotted in 2 mm lanes, and we have one that is a microarray with 200 micrometer spots. It still could be further miniaturized by more high-tech methods, for example using a different microarray spotter. What we have to overcome is the solvent: The solvents for lipid-linked oligosaccharides are volatile, and arraying them is not convenient with the modern arrayers. That’s why we have been using a spray arrayer. We are now working toward getting the oligosaccharides into a non-volatile solvent in order to be able to array them in the state-of-the-art way. That is a problem, but not a serious one I think.
What are the main applications for these arrays?
One would be for discovering novel carbohydrate-binding proteins and for identifying and characterizing their binding specificities. I would envisage that with advanced protein expression systems and mass spectrometry, the principle of constructing oligosaccharides from desired sources could form the basis both for identifying proteins in the proteome and mapping the repertoire of the complementary recognition structures in the glycome. Our technology, together with mass spectrometry, may in fact be considered as a prototype technology for glycomics.
The ultimate application of this knowledge will be in therapeutics. I have said for many years that in order to develop oligosaccharide-based therapeutics on a sound basis, it is highly desirable to know the repertoire of carbohydrate-binding proteins in the body and also their cross-reactions. We need to understand their specificities so that oligosaccharide analogues are not likely to have undesirable toxic effects. The most promising disease areas are abnormalities of immunity, disorders of inflammation, and host-microbe interactions.
Do you have plans to commercialize your arrays?
We are, of course, interested in that and discussions with our advisors at Imperial College Innovations and the MRC technology transfer office are in progress. We patented the neoglycolipid technology at the time in 1985, but I think the technology came before its time, and we did allow the patent to elapse. Scientifically, it has been extremely powerful, but it was not commercialized.
Where do you see the carbohydrate array field going?
Of course we are developing it as a scientific tool. I think that once it is made into a high-throughput system, there is great potential for it to be operated in conjunction with advanced protein expression systems and bioinformatics. It will be possible to discover the repertoire of carbohydrate-binding proteins in the proteome – in other words, this is an aspect of functional proteomics. Currently predictions are made as to whether a protein is or is not likely to bind carbohydrates, and some of the predictions are correct, while others may be incorrect. To be absolutely sure, we have to actually test them. The conventional methods are very tedious, and arrays will facilitate that. What this approach will also do is enable you to survey virtually all proteins, known and unknown, and by this means you will hopefully dis-cover novel carbohydrate-binding proteins. This is the value of arrays, isn’t it, to be able to survey without preconceived ideas.