At A Glance
Name: Alma Burlingame
Position: Professor of chemistry and pharmaceutical chemistry, University of California, San Francisco, since 1978.
Director of Mass Spectrometry Facility, UCSF.
Deputy editor, Molecular & Cellular Proteomics, since 2001.
Background: Research chemist, University of California, Berkeley, 1973-78.
Guggenheim fellow in biomedical mass spectrometry, Karolinska Institute, Sweden, 1970-72.
Worked at NASA Johnson Space Center (involved in Apollo moon-landing mission), 1964-73.
Assistant professor in chemistry, University of California, Berkeley, 1963-68.
PhD in chemistry, Massachusetts Institute of Technology, 1962.
BS in chemistry, University of Rhode Island, 1959.
How did you get into proteomics?
Proteomics existed before it was called proteomics. It became clear in the ‘80s that [people] were having trouble doing internal sequencing using Edman degradation and that mass spectrometry, particularly high quality tandem mass spec, would be able to do this quickly and unambiguously.
We started identifying proteins in about 1987 using tandem mass spec. We did several galectins, which weren’t called galectins at the time, in the late ‘80s and early ‘90s. We did sialyltransferase, and we did some fatty acid binding proteins. By the beginning of the ‘90s, it was clear that a lot of people were using gels as media for separations, and I became aware of Lois Epstein’s work here [at UCSF] — she’s a cancer immunologist working on melanoma, and she had identified about three proteins in 15 years by Edman degradation. We got together and identified about 20 proteins [by tandem mass spectrometry]. So I said “Why don’t we just do them [all]?” and we did; that work came out in [Proceedings of the National Academy of Sciences] in ‘93 and ‘95. I gave talks early on with titles like, ‘Is Tandem Mass Spectrometry Sensitive Enough to Identify Spots on 2D Gels?’ We showed that it was. I think that we were the first ones in the world to show that.
When did you start holding your biannual mass spec meetings?
It started in 1984, and at first we had them every five years. In the mid-’80s, progress was slow. People don’t understand that there was life before MALDI and electrospray, and actually it was pretty effective. [The early meetings] included some protein stuff, [but] more metabolite identification — the kind of stuff that’s become routine in the pharmaceutical industry. There were things like liquid SIMS, tryptic peptide digests, ion optics for magnetic instruments, identification of fossil porphyrins from a geochemical point of view, and a variety of applications to things like marine natural products, some Fourier transform applications, advantages of mass spec in protein chemistry, acetyl carnitines, breast cancer cells, metabolic diseases. [What tied it all together was] mass spectrometry and biology.
What is the purpose of these meetings?
The American Society for Mass Spectrometry had its origins in the American Society for Testing and Materials and provided basically a methods-and-standards protocols kind of adjudication forum. It was pretty heavily attended by technocrats and instrument operators, it was really dominated by the petroleum industry, both in the US as well as internationally. We needed to have a forum to discuss the advances in understanding biological systems that were being brought about by being able to characterize the constituents by better methods of mass spectrometry. The idea was to get together people who were interested in developing these sorts of methods to those sorts of purposes and the people who would be the ones to do the science based on these emerging technologies.
What are you working on now?
We work on a variety of things. Technically, we’ve focused a lot of attention in the last decade on making MALDI sequencing routine. What we’re doing presently is finishing automation of the interpretation of large data sets from experiments similar to what [John] Yates does but with better quality mass spectrometers. We do multi-dimensional chromatographic separations and then we do LC MALDI CID measurements of everything that’s there. We can do this with MALDI off-line, which then allows you to interrogate a wider dynamic range because you don’t have this limitation of having to do the experiment within the constraints of the elution time of the components in an HPLC. It’s really fast, so you can have high throughput. You can couple it to the cleavable ICAT reagent that [Applied Biosystems] has and also other isotopically labeled substrates or metabolic labels. This software that we’re more or less finished with is not just able to do all the LC/MS comparisons at a CID level, but also quantitation if you’re using isotopes in the experiment. We apply that to a variety of problems.
One [thing] we’re very interested in is nucleo-cytoplasmic protein trafficking, in collaboration with Mike Rexach at Stanford. We’re presently going through what happens during changes in the cell cycle in terms of the quantitation of protein trafficking across the nuclear pore complex (NPC) using that methodology.
What have you found so far?
We have a fairly large set of proteins because we’re doing mass spectrometry, whereas he has a set of antibodies to some proteins. He’s done quantitative Westerns on cell cycle changes with his antibodies and we’ve been comparing that with what we find with an ICAT kind of experiment. We find that there’s fairly good agreement for the ones he happens to have. But we found a lot of others where we don’t know the exact role they play. We’re going to have to try to figure that out.
What is the ultimate goal of that exercise?
Antibodies to new proteins just take a long time and are expensive to make. So it would be nice to know that if you do this sort of experiment based on isotopic labeling on a much more global scale with mass spectrometry, that the numbers that are coming out can be trustworthy. At this moment it’s a validation kind of exercise. One has to go through that before we can apply it to more complex systems.
Making MALDI sequencing routine … what still needs to happen?
Not a lot. The last thing in our own hands that really needed some attention was the scoring strategy for reliable automated interpretation of individual CID spectra. This is a problem everyone has to some extent. We went through a data set that had about 3200 CID spectra with a statistical scoring system that we’re developing, and the first pass threw out 962 of the 3200 spectra that didn’t fit in terms of being correctly called or identified. So we spent the last two months going through these 962 spectra to manually figure out why they ended up in that category. Some fraction of them turned out to be miscalled for reasons you would intellectually understand, so we need to modify the algorithm to include features in those spectra that caused them to be thrown out. I think we’re going to end up with about half of them as legitimate throw-outs and the rest are probably artifacts. So the purpose of this is to make reliable calls, so that you can be more sure [when] we interpret spectra automatically that the ones that are called are in fact called correctly and things that have any ambiguity are correctly put in a bin for humans to look at.
When you look at proteomics in general, what are some of the other obstacles that need to be worked on?
The whole front end is undefined in terms of getting from cells or tissues to reproducible protein segregations, whether they be from GST fusion protein pulldowns or TAP tags, or some kind of electrophoretic or some other sort of density gradient or organelle separations. Nobody has a good handle on how you do this right.
[Also,] I think a lot of these issues [like defining the phosphorylation state or glycosylation state of proteins] are being handled on a protein-by-protein basis the way they were over the last 30 or 50 years because we don’t know how to automate them or how to do high-throughput measurements on the complexity we’re dealing with. Obvious strategies that get into segments of this are chemical probes for enzymatic activity or chemical probes for various functional groups, and protein cross-linking [to target interactions] in vivo. I think the mass spec isn’t the problem. It’s the chemistry and the separations. I just think there aren’t enough talented people working on it long enough to really understand and make a lot of headway.