AT A GLANCE
NAME: Carlito Lebrilla
POSITION: Professor of chemistry, University of California, Davis, since 1989.
BACKGROUND:Post-doc in chemistry, University of California, Irvine, 1987-89.
Post-doc in chemistry, Technical University, Berlin, 1985-87.
PhD in chemistry, University of California, Berkeley, 1985.
BS in chemistry, University of California, Irvine, 1981.
How did you get into proteomics from your background of studying sugars?
We started looking at carbohydrates and oligosaccharides a while back in the early ’90s when I was an assistant professor. We were intrigued with what sugars do, and the difficulty in analyzing them, and so we developed a whole program. And at some point along the way, we realized how important glycomics was to proteomics. So it just became a natural extension of what we were doing. We started looking at oligosaccharide analysis from the point of view of gas-phase chemistry, looking at their fundamental processes and fragmentation. We started actually taking samples from glycoproteins probably about six years ago.
Tell me about how you used FT-MS to study glycoproteomics.
There were two aspects of it. One was looking at N-linked glycosylation and developing a rapid method for doing that. The current method now sort of requires you to digest the whole glycoprotein with trypsin. The problem with that was there was a lot of more prep work to do because you’re producing both peptides and glycopeptides, and you have to figure out which is which. And so you had to send the whole thing through HPLC, and then look through each HPLC fraction for the glycopeptides. The problem was then that the glycopeptides were spread throughout — were mixed with peptides — and so it made it very difficult to actually figure out which glycan was associated with each site.
So our approach was to try to just produce glycopeptides, and we thought the best way to do that was using a non-specific protease. What that did was destroy all the peptides, and since the glycans naturally protected the peptides, it made the peptide-associated glycan protected from the protease. So everything else went down to the constituent amino acids, but a certain amount of the peptides stayed with the glycans. And that was enough to figure out which glycan associated with which peptide. And so we took advantage of the accurate mass of the FT to create a glycan profile and then take that and subtract it from the glycopeptides that were produced. Just from that combination, we were able to determine the peptide associated with the particular glycan. So that was the first aspect.
The second aspect was that in other glycoproteins like mucins, you knew where the glycosylation was because it’s oftentimes grouped in these large, long sequences of serine or threonine, and most of those will be glycosylated. The problem is trying to figure out what the glycans are. So these are O-linked glycans, and O-linked glycans are much more difficult than N-linked glycans, because they don’t have a putative structure that you can work with. N-linked glycans, even though they’re bigger, have certain long sequences that, for example in mammalian cells, they don’t vary much from. But in O-linked glycans, they can take on pretty much whatever structures they want.
So there the problem is, how do you determine the structure of something as complicated as an oligosaccharide, given that you don’t have a lot of information regarding the structure? What we do there is we use the mass spectrometry again to determine the composition. We use tandem MS to determine the connectivity, and then we use a combination of glycosidase based on the sequence to try to target a specific bond. We start from the smallest oligosaccharide and know that structure completely, and then when you go to bigger oligosaccharides, you just see which smaller oligosaccharides they’re an extension of. The thing about O-linked, though, is there [are] a finite number of cores that they can have. So if you can characterize the cores, you’re just looking at the extension, and so you don’t have to do the whole glycosidase reaction for the whole thing, it’s just whatever appendage you have for that core that you can deal with. So it’s a much faster way of characterizing them.
At what point in this project are you now?
One of the hallmarks of glycosidation is [the sugars’] diversity. But a lot of the glycan studies are concentrated on one or two glycans. Ours was, how do you characterize the entire glycan profile of biological samples from a given source? Because we think that the glycan profile holds the keys to a lot of diseases. So what we’re trying to do is look at oligosaccharides as disease markers. We’re starting to do samples from diseased humans at this point.
Which diseases are you looking at?
Cancers, mainly. So we’re starting to look at ovarian cancers for markers there, and breast and prostate. We’re using a lot of mass spec, and we’re using the methods for N-linked and O-linked glycosylation that we’ve developed.
We’ve found ovarian cancer markers on the glycans. We’ve looked at diseased individuals and normal, and it’s clear that we can readily see when a person has cancer or not. And so I think within a year or two we should be able to come up with a diagnostic even for early detection of ovarian cancer, and maybe even breast cancer. It would probably be mass spec-based initially. But once we know more about structures, then I think it can be moved onto HPLC.
The beauty of the glycans is that a lot of people are looking at protein markers, but the proteins are a lot harder to work with. The glycans are actually ironically much easier to work with. So from that point of view, I think that glycan markers will be a lot easier to quantify and to detect than protein markers.
We’re looking at entire glycoproteins, but I can’t tell you how we’re doing that now.
Are you working on any other mass spec-based projects?
In mass spec we’re also looking at detections of biological particles in aerosol. We’re collaborating with Lawrence Livermore and we’re developing methods to look in real time at aerosolized spores, for example anthrax and tuberculosis. The thing there is — again — looking for certain markers, and there also it seems that the small molecule markers are the ones that we’re observing. So we’re sort of trying to fuse metabolomics and proteomics together, and it seems that glycomics is the best place for it, because it’s oftentimes part of proteomics, but it’s also not coded in the genome. So that sort of makes it metabolomics.
What sorts of difficulties have you found particularly with the metabolomics side of things, as opposed to proteomics?
There it’s the structural elucidation. Because the molecules are small, so they’re easy to quantify and they’re easy to separate from something else. But oftentimes you’ll get something where you have no idea at all what it is — so you have no idea of structure — oftentimes there’re some small amino acids or lipids. With oligosaccharides you have some idea, because the precise masses tell you they’re sugars. So for example, if we’ll have a mixture of peptides and glycans for example, the glycans really stand out if you know accurately what the masses are, because those will have very definite mass composition as opposed to the peptides.
Where do you get most of your funding?
We get NIH [funds], we’ve had long NSF funding, and we also get funding from the Department of Defense for the aerosolized thing.
What should proteomics scientists consider when they think about glycomics?
I think oftentimes what happens with glycomics is that it gets neglected. The first thing that people do when they have a glycoprotein is trim away the glycans, and I think they’re actually throwing away the information that they need, because when you look at proteomics, you’re looking at up — and down — regulation of certain proteins. But it turns out that, as I think everyone is finding, to quantitate proteins is a difficult thing to do. But the glycans will totally change in disease states, whereas the peptide will remain essentially the same. So I think in this regard, it’s the glycomics that’s really going to provide, particularly in terms of early detection, a more robust and easily quantifiable tool. So I think you’ll be hearing a lot more about it. There’re a lot more people getting trained in it now.
So you’re finding more interest in glycomics recently?
Oh yeah. I think before, people didn’t really care about glycomics. Now I find myself being invited to a lot of proteomics meetings. Because I think they realize that there’s a roadblock there. If there are 150,000 proteins encoded in 30,000 genes, and they’re finding in serum at most about 1,500 — and maybe only about 500 of those accurately — they’re clearly missing a large chunk of maybe very important proteins.
It’s been estimated that 50 percent of all proteins are glycosylated. The fact that people aren’t looking at them tells you something. I think what people are saying is, ‘let’s solve the proteomic problem first, and then we’ll try to find the post-translational modifications.’ But ironically enough, I think it’s in the post-translational modifications where a lot of the things that people want to find are.