At A Glance
Name: Roland Annan
Position: Assistant Director; computational, analytical and structural sciences department
Head, proteomics and biological mass spectrometry laboratory GlaxoSmithKline, King of Prussia, Pa.
Received his PhD at Northeastern University, studying damaged DNA with mass spectrometry in Paul Vouros’ group.
Postdoctoral fellowship at MIT with protein mass spectrometry pioneer Klaus Biemann, working on derivatizing peptides to get better sequence information, and derivatizing proteins to provide better peptides for sequence coverage. First laboratory to receive a commercial Vestec VT2000 MALDI-TOF mass spectrometer.
Joined SmithKline Beecham, now GlaxoSmithKline, in 1991.
What is your position at GSK, and how are you equipped?
The laboratory here is called proteomics and biological mass spectrometry. We have pretty much one of everything here: we have a Finnigan LCQ, a Micromass Q-TOF, two Sciex triple quads — which we use strictly for phosphorylation analysis work — and two Micromass MALDI-TOF mass spectrometers. We also have a Micromass electrospray TOF, and a Finnigan FTMS. Almost every mass spec has an HPLC hooked up to it. They are all microcapillary, either 75 micron or 180 micron. And then we have FPLCs for doing off-line protein fractionations.
But we are set up to do everything, except for cloning, from start to finish here: We have a small cell culture lab, a protein purification lab, a lab to do 2D gels. However, we actually don’t run many 2D gels these days. That’s largely because the company set up a much bigger group to do that over in the UK [in Stevenage].
What do you focus on in your research?
One of the things we concentrate on is trying to understand signal transduction pathways. For a long time, our laboratory has developed techniques to study protein phosphorylation. Now I think a shift is going on from not just knowing which sites are phosphorylated — that’s actually still relatively difficult to do, although it’s certainly easier than it used to be — but now the real question is, of all the sites that are phosphorylated on a protein, which are the important ones?
How do you go about finding out?
In our lab, we use a technique which we developed over the course of the last eight or nine years. We use a two-dimensional approach to mapping phosphorylation sites. We start with preparative LC/MS, which detects only phosphorylated peptides, and collect the fractions. It’s very much like a radioactivity detector, except there isn’t any radioactivity. The mass spectrometer is set up so that phosphopeptides will lose the phosphate groups, and then the mass spectrometer only detects the phosphate ions. We can tell which peaks are eluting from an HPLC contain phosphate, and therefore that fraction must contain a phosphopeptide.
In the second dimension, we use the same principle, but this time, using a precursor ion scan, we obtain the molecular weights of the phosphopeptides. The mass spectrometer again only detects phosphate ions, but this time it reports out the m/z value, from which we derive the molecular weight of the peptide which lost phosphate. It would be nice if you could just rely on the molecular weight, but more often than not you then have to do some direct sequencing.
What is the challenge for studying phosphoproteins in high-throughput?
There has been a growing interest in analyzing the phosphoproteome, which sounds like an interesting idea, although I think what people don’t realize is that it is likely to be a couple of orders of magnitude more difficult than the proteome. For one thing, phosphorylated peptides are present at a much lower stoichiometry relative to the non-phosphorylated peptides. And then unlike traditional proteomics, where you just analyze whatever you find, if you want to do phosphoproteomics, you have to specifically analyze the phosphopeptides, not just any old peptide. Also, it changes even much faster than protein expression profiles.
Do you see any merit in analyzing the phosphoproteome in its entirety?
Currently I don’t really think that there is much value to be gained by looking at the entire phosphoproteome. I don’t really think that the tools are quite there yet. However, it would seem to me that the tools that we do have could be useful to look at specific pathways rather than the entire phosphoproteome. That’s one of the things that we are keen to do here, what I would call ‘directed phosphoproteomics.’
What about other posttranslational modifications?
One that we have been interested in for a while is ubiquitination, which is probably — I hate to use the word again — ubiquitous. It is not quite as commonplace perhaps as phosphorylation, but certainly we know that ubiquitination serves more roles than just to target proteins for the proteasome, and that a very, very large percentage of proteins probably becomes ubiquitinated at some point in their lifetime. And then of course methylation is an important posttranslational modification. That one is likely to be a little bit trickier to study, because it’s very small, and not labile like a phosphate.
How do you interact with other groups at GSK?
Within the company, we tend to work at the very early stages of programs. There are six major sites, each of which has a different therapeutic area. Layered on top of that is the research component of the company, and below them is the development arm. We are part of the research component. The department that I am in is called computational, analytical, and structural sciences; our group tends to specialize in the analysis of signal transduction pathways and phosphorylation-dependent biology, though we help out in any way we can. So even though in principle we would collaborate with any biology group anywhere in the world, by and large we mostly support the biology here in the Philadelphia area. And that’s oncology, cardiovascular sciences, and bone biology. We also participate in broad-based company-wide research initiatives.
Obviously kinases are an important class of proteins which are being developed by everybody as targets for drug intervention. But many kinases are not well characterized, so their substrates are not known. Also, the way that the kinases are activated is not well understood. Oftentimes we determine phosphorylation sites on a kinase, and then we need to find its substrates, and where their phosphorylation sites are. That helps us set up both high-throughput screens and cell-based assays.
What other proteomics labs are there at GSK?
There are other proteomics groups within worldwide research, in a department called genomic and proteomic sciences. They are probably what you would describe as the more classical proteomics department. They are doing expression proteomics by running 2D gels, and also pathway analysis by yeast two-hybrid and by immunoprecipitation followed by mass spectrometry. Some of that work goes on everywhere, but most of the large scale, high-throughput proteomics projects are done by those groups. They are located in Stevenage, UK, and Research Triangle Park in North Carolina.
Do you collaborate with proteomics groups outside of your company?
We do. We have had a long-running history of collaboration with academic labs. One of the people we have worked with is Ray Deshaies, who is a yeast biologist at CalTech. We have worked with Ray on a number of cell cycle-related projects, where we have mapped and determined functionally important phosphorylation sites, and then Ray’s lab has been able to show that this phosphorylation is critical for regulating some aspect of the cell cycle. [On the corporate side], we don’t have anything like [the relationship between GeneProt and Novartis], because SmithKline and Glaxo were involved in genomic medicine very early on. So I think the company has sufficient internal resources to be able to handle a lot of that itself. We do have a number of collaborations with other proteomic companies.
What is the importance of proteomics within GlaxoSmithKline, and is it still growing?
Obviously the upper management here thinks that it can make an important contribution to all aspects of drug discovery, because the company has invested quite a lot of money in building a proteomics program. I believe that it’s still growing, because I think that the applications for proteomics are still evolving. Proteomics has entered into a period now where we have evaluated most of the existing technologies, and we have decided which are the ones that work and which are the ones that don’t work. People are trying to explore exciting applications with the technology that does work, while at the same time, we recognize there are some shortfalls, so there is the search for a new technology.
Where do you see need for improvement?
I think that one of the greatest needs for proteomics is to be able to do efficient quantitative proteomics. Largely I think that has to do with throughput. There is just not any good way to do this in a throughput that would be comparable with, say, microarrays. The problem is that mass spectrometry, while it can detect all of the peptides at once — so it seems to be a parallel approach — nevertheless each individual sample has to be analyzed in series, unlike, say, a plate reader or a microarray.
I would suspect that protein chips would be a much higher throughput way to do this. On the other hand, protein chips are a very immature technology. And because we are working with proteins and not pieces of DNA, it’s much harder to develop protein chips. We don’t develop them here ourselves, but we have a number of collaborations with companies to look into them.