AT A GLANCE
Name: Denis Hochstrasser
Position: Chairman of Scientific Advisory Board, GeneProt, and Professor, Clinical Chemistry, University of Geneva
Prior Experience: Developed 2D gel image analysis software. Helped found GeneBio and Swiss Institute for Bioinformatics
In the first of a two-part interview, Denis Hochstrasser discusses how he became involved in proteomics and the founding of GeneProt. Next week, Hochstrasser will describe the company’s proteomics platform and goals.
QHow did you become involved in proteomics?
AIn 1978, I was doing a clerkship in endocrinology and the professor asked me to connect a videocamera to an Apple IIE to analyze 2D gel images. I had no idea what a 2D gel was so I went to the library and I looked at some papers, such as the [Patrick] O’Farrell paper. I said, ‘That’s interesting, it could one day become a molecular scanner.’ It was one of those things you put in the back of your mind, you forget about it, and one day it comes back up.
I finished my MD and did my internship residency at UNC, Chapel Hill, came back [to Geneva], finished my board of internal medicine, and [worked] as an internist on the ward of medicine. When the chairman of the department of medicine asked me what I wanted to do for an academic career, I told him I wanted to work on protein separation science and bioinformatics. He said, ‘everyone is working on DNA. It’s time to sequence genomes.’ I said, ‘It’s like the story of the turtle and the hare, if we start early on the proteins one day we’ll be in a very good position.” I knew in ‘84 that people would need to look at proteins anyway.
QHow would you describe proteomics?
AProteomics has four legs. The first leg is protein separation techniques. The second leg is mass spectrometry — to identify proteins and partially characterize them on a fast mode. The third leg is bioinformatics. But the data will not be useful and you cannot interpret the data if you do not have the fourth leg, which is genome and protein databases, because you need to relate the mass spectra to some kind of information.
QHow did GeneProt come about?
AIn June 1999, I was invited by the Novartis Research Foundation to present what proteomics was, and the president of the symposium was Craig Venter. Talking with Craig, I realized that the model of PE was that they had applied a system, a sequencing machine, and then created Celera to buy their machine, and then [sell a] database. So they win twice because they sell sequencers and the data. If they have ABI and Celera, why don’t we do GeneBio and GeneProt? GeneBio [a nonprofit bioinformatics company that supports Swiss-Prot] can sell bioinformatics to GeneProt, and GeneProt would be a facility to analyze proteins on a large scale.
We had quite a lot of experience, we had many bioinformaticians, and we had known since ‘84 how to separate proteins. On February 29, 2000 — it took some months to think about it — we sat down and it was one of those days you never forget. It was go or no go.
QWhat is GeneProt’s approach to proteomics?
AThere are different ways to look at [proteomics]. It depends on where you want to start, and what type of sample. To me as a physician, one of the most exciting samples is obviously blood, because blood goes everywhere, it carries many things, including hormones, which are potential therapeutic agents. Blood might be the most complex and most difficult sample to work with, but blood has many advantages: you can draw it, so you don’t have to do biopsies, and most of the proteins you find in it are soluble, so they’re a little bit easier to separate.
In one approach, you start with samples in small amounts, and compare 100 patients with 100 control. But [in this approach] you will only see the tip of the iceberg, because you cannot draw 5 L of blood from a patient; you can only draw a few mL and you cannot go that deep into the proteome. Also, if you analyze 100 samples, you will only see phenotypic differences of the most abundant proteins. For GeneProt that''''s not what we are most interested in. So our strategy has been to pool patients’ sera to dilute the phenotypic differences. But at the same time we increase the proteins in low concentration because they are the same in every patient. We took that approach.