At A Glance
Name: Thierry Rabilloud
Age: 41
Position: Staff researcher, Laboratoire de Bio Energétique Cellulaire et Pathologique (DBMS / BECP), CEA, Grenoble, France
Prior Experience: Starting in the early 1980s, began developing protocols for separating proteins via 2D gel electrophoresis
How did you get started in proteomics?
As a matter of fact I started with my PhD. I recently went through my old lab books, and I did my first 2D gel in ’81, because at the place where I was studying, the Ecole Normale Supérieure in Paris, [my advisor] was strongly building in that field. That was not proteomics at all, merely 2D gel electrophoresis. The important point was that something important could be drawn from large-scale protein measurements. At the time, it was something of a new technique. You could see plenty of spots, but you had no idea who was who, except for the major ones. My advisor [Philippe Tarroux] was really persuasive and I started to believe that he was right. That’s how I started.
Who else was involved in large-scale protein measurements at the time?
To put it nicely, in the proteomics field there are five evangelists, and a little bit more than 12 apostles. The evangelists can be defined as those who believed in the field in the ’70s, mainly [Sam] Hanash, [Joachim] Klose, [Jim] Garrels, [Leigh] Anderson, and [Julio] Celis. The apostles were those who believed in the field in the ’80s, and I belong to [that second group]. In the ’80s there was the major revolution of the molecular biology of the gene. Many people flew away from the protein field to the DNA field — with a lot of success. There were quite a few people [who did that]. When you think of the so-called proteomics field in the ’80s, there were no more than 100 people. That was the whole community in the world of those who believed that protein expression levels [measured] on a large scale could be biologically relevant. All of us had to tackle a number of technical problems. The proteomics field per se evolved when the 2D electrophoresis field could match the protein chemistry field. They could join together. That was the very end of the 80s, with the work of [Ruedi] Aebersold, [Joël] Vandekerckhove, and all those people doing micro Edman sequencing techniques in the gas phase. That was the real turning point of proteomics. The mass spec revolution was only a productivity-throughput revolution. But from a conceptual point of view, that’s not a revolution. Mass spec has just been able to put names on proteins, things we could do before with Edman sequencing. From a conceptual point of view, proteomics existed in the very late ’80s.
What’s your particular line of expertise with respect to 2D gels?
As a matter of fact, I worked in two major fields in 2D which are related to protein solubility — how to improve protein solubility in 2D gels. That pertains [in particular] to membrane and nuclear proteins. [The other area I worked in was] protein detection after 2D gels: Silver staining, interfacing with mass spec, and fluorescence staining. That’s where my own contributions to the 2D field are.
What kinds of biological problems are you studying?
For quite a while, I was involved in cellular differentiation studies. I did that in a variety of models, including white blood cells, B-lymphocytes, and red blood cells. Quite recently, I moved into mitochondrion proteomics, especially [how] it is related to oxidative stress response in cells.
Using differential proteomics?
Exactly. We would like the scope of the proteins we can look at to be perfect. We would like to be able to look at everything, but we know that’s not the case, especially with 2D gels. But provided we can find differences we are happy, because the differences point out things that are interesting. That’s the basic situation of what we can call differential proteomics. There is plenty of work related to that, [particularly] with disease markers.
How does your lab relate to the CNRS?
I am very strange. I am paid by the CNRS, but I work in a lab that belongs to the CEA, the atomic energy commission, where I do the biology I want to do. It’s a question of money. [The CEA] has more money for doing experiments.
What kind of mass spec facilities do you have there?
As a matter of fact, there is a mass spec [at the CEA], but I don’t work with those people. I work with other collaborators in Strasbourg, with a guy named Alain van Dorsselaer — a real mass spec guy. He started his PhD on petrol and coal fractions, something very far from biology, but he moved to biology. In the mass spec field, you have two kinds of people: the kind of people who were first protein chemists, and then went to mass spec — the typical example is Ruedi Aebersold — and then you have people who started in mass spec and went to biology, such as Peter Roepstorff, Al Burlingame, Peter James, Alain van Dorsselaer, and John Yates.
In general, is there technology you would like to see that would help you in your work?
I would like to see available a perfect quantitative separation technique. Because, as a matter of fact, what we all are doing is the following: We are first separating either proteins or peptides, or first proteins, then peptides, and then analyzing [them] by mass spec. And mass spec is no longer the limiting factor. The computing power after the mass spec is limiting, according to what the mass spectrometrists say, because it limits the throughput and accuracy of what we can do. The instrument per se is very good, but the computer that comes with that is never powerful enough, nor are the databases big enough. We have a problem of how to interpret the data.
But up front, we have the problem of how to be able to face the real complexity of the proteins in the sample. For the moment there is no technique that is able to fix that. That’s the real problem. My feeling is that the computer problem is not a real problem. With Moore’s law, that problem will be solved, provided we can put enough money into it. When you see the computer programs that some people [have developed for] supercomputers and supercomputer networks, I’m confident that that issue can be solved.
But you have something more subtle, and that’s how to recognize what’s in Nature’s code. I will give an example: In genes, you have introns and exons, and nature knows who is who. But we have not found a real code that enables us to [distinguish introns and exons] with decent precision on a complete genome. On one gene, we can more or less [determine that], but genome -wide, we are no longer able to say, ’Here is an intron, here is an exon.’ We have the information, because it’s somewhere in the sequence or in the structure. It’s somewhere, but only nature can find it. Maybe it’s just a problem of computing power, but maybe it’s also an issue [of not having found] the algorithm that enables us to find what we want to find. Maybe the same holds true in proteomics. Maybe we don’t know exactly how we should process our data. So, on the computing side there are two different problems. First the power problem, and second, do we know exactly how to mine our data?
But completely up front, if you want to be as comprehensive as the genetic techniques are with genome scanning, we are far from that in the proteomics field. We are miles away. That’s the second real bottleneck.
That’s why I’m working in 2D gels. As a matter of fact, I believe 2D gels will never be perfect, but I believe we still can improve them to a point where they will deliver something interesting. Given the improvements that have been made in the last 10 years, I am still confident that we can make major improvements in the field.
What would those be? Imaging techniques?
No. Separation chemistry techniques. Separation science [exists], but it’s not that well developed, and that holds true for electrophoresis of proteins and for chromatography, especially peptide chromatography. People have done a lot of scaling down to be able to cope with the very tiny quantities of sample that are available in proteomics. [But] if you are able to separate 10,000 peptides, then we should be able to separate a million peptides if we really want to be comprehensive on the cell scale. We are lacking two orders of magnitude, and the same holds true for 2D gels. We are lacking one to two orders of magnitude in the number of objects we can handle.
Is there just a fundamental limit to the resolution of 2D gels?
Probably there is. There are chemistry limits that we will probably face, one day or another, but we are far from them right now. That’s the point of chemical diversity. With DNA, you can handle it, because it’s a messenger molecule and you can [apply one method to] sequence it [and another method to] clone it. Proteins are the real stuff. It can be very diverse from a chemical point of view. That means there obviously won’t be a unique technique, and there may not be even be a unique [suite] of related techniques that will be able to handle everything. We will have to mix different separation techniques to face the complexity of Nature on the protein side. It’s very difficult because proteins “live” in various environments in the cell. They live outside [the cell], they live in serum, and there are proteins in the nucleus that are different from those in the cytoplasm —not to mention those inside the membrane. It’s a little bit as if you wanted to see fish without any diving equipment, or any goggles in the water. You might see something fuzzy, but you don’t see the detail. You see something move, but you don’t see it [in total] because it lives in a different environment from the one you are used to.