At A Glance
Name: Peter Roepstorff
Position: Professor of Protein Chemistry, University of Southern Denmark, Odense
Prior Experience: Developed first commercial peptide sequencer, pioneered the use of plasma desorption and fast atom bombardment ionization in the early ’80s
How did you get involved in proteomics?
It goes all the way back to 1966. I constructed in ’66 or ’67 the first fully automated peptide synthesizer to be commercialized. But our peptides didn’t come out active, or with [regular] activity, so I had to figure out what went wrong. I had read about some very early mass spec investigations, so I applied for a mass spectrometer, and by a miracle I also got it. Since then I’ve been a protein mass spectrometrist. We quickly found out what was wrong with the synthesizer; it was a minor technical problem that resulted in partial blockage of the peptides in each synthesis cycle. Then my question was: ‘If we couldn’t see such a simple thing with normal protein chemistry techniques, how did we know the proteins were not modified?’ That [led me] into protein modifications in the years coming. Initially that was with quite primitive mass spectrometry techniques — electron impact mass spectrometry. But I was always looking out for new techniques which could help us look at intact proteins and underivatized peptides. That came in the very beginning of the 80s — I was among the pioneers of plasma desorption and fast atom bombardment. Most of the techniques used in proteomics today were developed with these two techniques in the ’80s. When MALDI and electrospray came along by the end of the ’80s, we [adopted] them immediately by building home-built instruments. Then, we were really in business because we also had sensitivity plus extended mass range. So I’ve been in the field for 35 years now.
How come you studied in France for part of your graduate studies?
Close to the end of the graduate studies I went to France for a year. The main reason was really because my girlfriend had broken up with me so I wanted to do something really dramatic, so I went to France! There I studied Raman spectroscopy of small molecules, and also obtained a French university degree. Then I went to finish my studies at the Technical University of Denmark.
What was the first mass spec you worked with?
It was an instrument produced by Perkin Elmer called the Perkin Elmer 270 that was a rather small electron impact mass spectrometer. It required volatile derivates. That was the case throughout the 70s. We needed derivatives of peptides in order to get them into the gas phase before we could analyze them. As soon as other techniques came along, such as chemical ionization around ’70, and field ionization desorption around ’74-’74, I also got them included in my laboratory immediately. But there were still so many limitations for what we could do that I also had to dream about what should be possible. I was trying new techniques all the time.
What’s your definition of proteomics?
I have a rather restricted view. First of all, it requires that we can analyze complex protein samples. It requires that we quantitate, or at least make relative quantitative measurements between two situations. It requires that we have a very high dynamic range, that is, the dynamic range between the lowest expressed and the highest expressed proteins in a cell. [The dynamic range of] proteins expressed in a cell can be as large as 107. Then we require at least a certain throughput — I’m not a high throughput fan, but we need to be able to have a certain throughput per day, and not work on a sample for a week or a month. We need at least what people would call medium throughput! These are the four requirements: complex protein samples, possibility of quantitation, very large dynamic range, and a reasonable throughput. If not, it’s protein chemistry.
What would you say are the major challenges of proteomics?
There are many challenges. The main and most important challenge is to define a proteome of a given organism, especially a multicellular organism like a human. The human proteome is probably infinite because we have so many different types of cells and so many different situations that [they can be in]. It’s very hard to define a human proteome because it’s a dynamic, constantly changing thing. It changes with your health status, your age, the phases of the sun and the moon, what you eat, where you go — there are so many things. The major challenge is to reach a point where we can get an overall picture of what goes on in the living cell, which is especially complicated in a multicellular organism. We’ve not yet reached that point. I consider proteomics a tool — a very efficient tool — but you need to ask good questions. If you don’t ask good questions you can’t get any answers. [In that sense] I’m a little in opposition to the real high-throughput people who believe that if they generate enough data they can get information just by mining the data. I don’t think it works that way. [You must] you start with a good hypothesis and a good question.
What is your approach to proteomics?
For us we have the tools in parallel with the biological question — addressed either by us or by collaborators. We’re also mainly focused on methodology development. We are always trying to figure out what the questions [will be] two years from now, and then trying to develop the methodologies that will be able to answer the questions when they come. It’s difficult to always be on the front line but [so far] it has worked quite nicely. Our main focus now is on posttranslational modifications, which everyone speaks about in proteomics but very few [tackle head on]. We are working on a concept we call modification-specific proteomics, where we can go in and look for specific types of modifications in the proteome, and see how they change as a function of cell type, cell cycle, or whatever. We [could] call it modificomics but my students say that’s not a serious word! So we say modification-specific proteomics.
All kinds of posttranslational modifications?
We are focusing at present on phosphorylation, glycosylation, and protein processing. We are also starting to look at isolating proteins with different lipids. We are trying to develop methods for selectively looking at each of these types of modifications. The one that we can’t do specifically is protein processing because these proteins don’t have any explicit characteristics; they just get shorter. That’s a tough problem, but we have our ways to deal with it now although they require a lot of work.
Are there particular tools that are in development that would assist you?
I would like to have a number of specific labels to use. ICAT is very limiting because you can only look at the cysteine-containing peptides and you lose the modification information. So I would like to develop a number of tools that could cover the proteins much better. Then we will have the ability to also [measure] the degree of modification, which we have no real way [of doing] at present except for when we work with 2D gels. At present, 2D gels are the only [tools] giving us differences in this respect.
In what area do you think proteomics will have the greatest impact?
The real problem now is that people have far too great expectations for proteomics. It’s a little bit like it was with genomics: people believed that once we had the genome we could understand the living organism. Now that people realize they can’t do it on the genome level, they hope to do it on the protein level. It’s not that easy. But I’m absolutely sure we’ll find a number of new drug targets, and we’ll find rather good diagnostic markers. Of course, once we have identified the diagnostic markers, they won’t be applied using conventional proteomics in the clinic. Instead, you will design specific chips, just like the pregnancy test, to identify and look at these markers. But proteomics will give us a better understanding of what to look for.
There’s a field just emerging which might be quite interesting called environmental proteomics. You monitor organisms in different environments and, based on the protein production in these organisms, you can get a good idea of the level and type of pollution. Also [proteomics will find applications] in the understanding of microorganisms. We will find a wealth of different functional proteins in microorganisms suited for carrying out all types of biotechnology processes. That’s also an emerging field. Only 1 percent or less of microorganisms are known today or characterized in any way so I think a lot of things will go on there. I think neither my students nor their children will be out of work!
But I think we’ll see a dip in proteomics now because many people are disappointed. Eventually, we will find a steady state where the level of interest will be reasonable. I discussed this with Brian Chait of Rockefeller University, and Brian made an analogy with a huge tidal wave approaching the coast. When it reaches the coast, it breaks, and there will be chaos. Only the good surfers will survive, and I think that’s a little how it’s going now. We will see that a lot of the startup companies won’t be able to live up to the expectations, because it takes a hell of a lot of time to develop drugs. Once you have identified an interesting protein, you must then go through all the validation to understand what it does. That may be a job of years. Proteomics is just the beginning.
What I can say is, it has a been a hell of a lot of fun to be in this field, but it was much more fascinating when we were pioneers than now when we are in the middle of a hot field. When we were pioneers nobody believed us, so we could say what we wanted! Now people take us seriously, so we have to be careful what we say. That has changed a lot.