At A Glance
Name: Leonard Foster
Position: Post-doc with Matthias Mann, Protein Interaction Laboratory, University of Southern Denmark, Odense, Denmark, since Aug. 2001.
Background: PhD in biochemistry, University of Toronto, 2001.
BS in biochemistry, Simon Fraser University, Vancouver, Canada, 1996.
How did you get involved with proteomics?
It was sort of through some complicated nepotism. I did my PhD in Toronto in a cell biology lab and my wife was doing her PhD at the same time in another molecular biology lab, and her supervisor happened to start up a mass spec company in Toronto — the precursor to MDS Proteomics. My current supervisor, Matthias Mann, was also involved in a startup here in Denmark, and then MDS bought both of those startups and made it MDS Proteomics. That’s when the link to the lab here was made.
What’s it like being a Canadian in Denmark?
We’re enjoying it immensely — for a native English speaker there’s no easier non-English [speaking] country to come to, because everyone speaks English and all the science is done in English, and it’s very easy to get along. We’ve [also] had the opportunity to do a lot of traveling in Europe.
Tell me about some of the things you’re working on now.
Our lab has started to focus on different quantitative methods in proteomics, so we’ve gotten into a lot of stable isotope labeling, and I’ve been involved with a study on lipid rafts where we use a drug that disrupts rafts, [and] combine [it] with some quantitative proteomics to identify a more definitive list of which proteins are actually in rafts. This wasn’t possible before, due to imperfect biochemistry preparations. More re-cently, we’ve been looking at focal adhesion — the complexes that mediate cell adhesion to sub-cellular matrixes — and using quantitative proteomics there, we identified several novel proteins that weren’t expected to be in those preparations, and we’ve done a lot of cell biology to follow up on those. We’re now moving into larger-scale quantitative studies on different domains.
Tell me more about the stable-isotope labeling by amino acids in cell culture (SILAC) method.
Most of what we’re doing involves stable isotope labeling with specific amino acids. Typically we will grow one set of cells in a labeled amino acid — so usually we choose arginine with six C13 atoms in it, and we grow the cells for about six population doublings to fully incorporate the label. Then we have, growing in parallel, a population that is grown in normal isotopic abundance amino acids. Then those two populations are stimulated in some way or exposed to some antagonist to make some differential condition, and the cells are harvested. In some cases, like for the lipid rafts, we’d do some biochemical fractionation to increase the purity of our preparation in some way to some extent, but we don’t do really involved purifications because we’re relying on the stable isotopes to tell us which proteins are important. After that, it’s fairly standard mass spectrometry analysis — we digest it and then we’ve developed a lot of software that allows us to process the data much more easily.
There’s nothing really commercially available to handle quantitative proteomics very well. So we’ve developed software that makes a more user-friendly interface between the data and the user and allows us to quantitate hundreds or thousands of proteins a day fairly easily. The only thing available that really works is to do things manually — so to go into the mass spectrum, and just read off the intensities of the isotopic peaks. And then if you’ve got a preparation that has 1,000 peptides in it, then you have to do that 1,000 times, and it’s very time consuming. So this just speeds that process up a lot.
Do you have plans to commercialize the software?
It’s still in development, but I would say no. If we’re leaning in any direction it would be to make it open source, but even that is a long ways away.
Why does your lab use SILAC for quantitative labeling rather than ICAT?
We had some ideas on better ways to do quantitative proteomics than ICAT. So some of the disadvantages of ICAT are, you need to start with enormous amounts of sample, and you also need fairly large quantities of the labeling reagent, and the cost for doing those kinds of experiments can be quite high. In some cases, the yield can be quite low, because ICAT labels cysteines specifically, and only somewhere around 20 to 25 percent of tryptic peptides have cysteine in them at all, so you’re already limiting — although there are some advantages to doing it that way also. We thought we had an idea where we could label more peptides and therefore get better statistics for each of the proteins we’re looking at, plus one of the major advantages of labeling biologically — growing the cells in the isotope — is that the starting amounts that you require are much lower, so the costs are lower, and it’s generally easier. [It also] has less inherent errors in it than chemical labeling, because after you introduce your differential condition by stimulating the cells or whatever, we combine our samples immediately, and all the downstream steps — biochemical fractionation or even down to the digestion — all the preparation for mass spec are all done on a single sample so there’s no error between the samples introduced, whereas in chemical labeling you have to keep the samples separate for quite farther into the workflow before they’re combined for the final analysis in the mass spec.
Tell me about your work with cell adhesion.
Some of the biological background is that there are some fairly enormous protein complexes that mediate adhesion of the internal cytoskeleton of a cell to the subcellular matrix. Through the membrane, that interaction is usually through the integrin proteins, but inside the cell there’s really an enormous complex of proteins that are involved and there’s never been a proteomics-level study on these sorts of complexes. We thought with our quantitative approach we might be able to shed some new light on this. So what we did was to again grow one population of cells in a heavy isotope-labeled amino acid and one of the populations was lifted from the growth substrate and incubated in solution for awhile. This allows the cell adhesion complexes to break up. The theory is that the protein-protein interactions will disassemble. Then if you isolate one of the proteins that should be in the complex in the cells that are still attached, these complexes will be together and you’ll bring the entire complex down. In the cells that are no longer attached, those complexes will be broken up, so you’ll only bring down the protein that you are targeting. So it’s an immunoprecipitation approach. From the labeled set of cells we would get the complex, and from the unlabeled cells, we would get no complex. And then we can easily measure for all the proteins we identify in that immune complex which ones that have a ratio different than one, and those are the ones that are defined as being in that adhesion complex.
It gave some very surprising results. We didn’t get many of the proteins we expected that other people have found in the literature to be involved in these complexes. But we did find a very large number of unexpected proteins of different types and then we did a lot of follow-up biology with that. We’re in the middle of trying to get that published right now.
What sorts of surprising proteins did you find?
Mostly they are RNA-binding proteins. We don’t have any idea [why they’re there]. We still have no idea. We’re quite sure it’s not [an artifact] because of the quantitative data that supports it, and they are not one of the typical proteins that you find sticking to everything. So we’re not doing so much proteomics on that particular project right now — we’re trying to do a lot of follow-up biology and establish why.
What would you like to work on when you have your own lab?
My training in my PhD was in cell biology and biochemistry and I’m still quite interested in that. My goal is to have a proteomics-focused lab, but my biolo-gical systems of interest will be sub-cellular organelles and how those change under physiological or pathological conditions. That will all involve quantitative proteomics. I’ve really bought into the idea of quantitative proteomics and how much more information it can give you in a single experiment than just identification-based proteomics.
I think one of the frustrations of proteomics in general now is that protein identification isn’t really enough anymore. A well-equipped lab can easily identify 500 or 1,000 proteins in a complex sample of whatever type, but it turns out — and this varies from case to case — that a lot of the proteins that are identified have no business being there. Without a lot of experiments it’s hard to tell that. Whereas with a quantitative experiment, you can tell that kind of thing much more easily, or you can overlook and get past those things. For instance, if you were to stimulate those kinds of cells with a growth factor and then immunoprecipitate the receptor, you’re going to get all kinds of things coming down in the immune complex. But only those things that change between the stimulated and the unstimulated case are the ones you’re interested in, and you can get at those very easily with quantitative proteomics. Then you don’t have to either spend a lot of time doing validation biology types of experiments — like Western blots or follow-up function studies on every protein you identify — or publish that list and have others do that validation for you, [which] just wastes a lot of time.