Professor of Microbiology
Royal Institute of Technology
At A Glance
Name: Mathias Uhlén
Position: Professor of microbiology, Royal Institute of Technology (KTH), Stockholm, Sweden, since 1988
Background: Postdoctoral studies, EMBL Heidelberg, 1985-1986. Worked with Riccardo Cortese on human liver gene regulation. PhD in biotechnology, Royal Institute of Technology, Stockholm, 1984. Cloning of protein A. Undergraduate degree in chemical engineering, Royal Institute of Technology, Stockholm, 1979.
Mathias Uhlén, a professor at the Royal Institute of Technology in Stockholm, Sweden, boasts a long list of technologies he invented many of them well-known to protein researchers. They include the cloning of protein A and protein G, the first gene fusion system for protein affinity purification, affibodies, and pyrosequencing. He also co-founded a number of companies, among them Affibody, Pyrosequencing, SweTreeGenomics, Magnetic Biosolutions, and Creative Peptides.
On Monday, Mathias Uhlén's research group at the Royal Institute released the first version of the "Human Protein Atlas" (http://www.proteinatlas.org), a freely available and rapidly growing tissue image database showing the expression and localization of human proteins, based on immunohistochemistry.
ProteoMonitor caught up with Uhlén, who also chairs the HUPO antibody initiative, just before the start of HUPO's Fourth Annual World Congress in Munich this week.
Tell me about the Human Protein Atlas how has it developed over the last two-and-a-half years?
We set up an antibody factory, where we are trying to produce validated antibodies to proteins. The current throughput is about five-six new antibodies every day throughout the system. [The first version of the atlas contains] 400,000 images [created from] 718 antibodies. The images are basically high-resolution immunohistochemistry images where you can see where your protein is in different parts of the body. In the first version, we are concentrating on normal tissues, and [tissues from] 216 cancer patients. They have been selected so they represent all the major classes of cancers. In the future, we also want to include other types of diseases.
We are 60 people working full-time on this project, and then there are about another 40 people involved part-time. Out of these 40, 10 are pathologists who are annotating the images. We have taken great care that every image we release to the public has been looked at with the eyes of a pathologist.
How will researchers be able to gain access to this resource?
It's only a click in to www.proteinatlas.org. We will not have any password protection, and it will be open to everyone and free to use, with no restrictions.
How about getting access to the antibodies?
That's a good question. We plan to make the antibodies available. In the first version, about half of the antibodies are commercial antibodies. They have been provided by external collaborators and companies. Those are all available and you can get to them through the protein atlas. Our own antibodies and we hope in the end that there will be much more of our own than the commercial ones will be made available to the scientific community through an antibody provider sometime next spring. But we simply haven't got the logistics ready yet to make them available, [and] we haven't actually selected who should do it yet.
What have been the greatest challenges in creating this database, and the antibodies?
I think there have been two major challenges for our project: one is simply to logistically scale up a project like this, with all the IT and the database handling, and taking care of 200 GB of data, at the same time as you hire 60 people and involve almost 100 people daily. The second is quality assurance, that is, how to decide what antibodies should go into the atlas, [which] are likely to have the correct specificity and selectivity, and which give cross-reactivity. This is an effort that we have started and I think we are quite happy with where we are, but still, there are a lot of things you can do to this, and this will be a continuation for a long time.
Have you encountered proteins that are difficult to raise antibodies against?
We have a way of producing our antigens that seems to be working across the different protein families, everything from transcription factors to kinases and proteases and membrane proteins. The only protein class that is very difficult is a relatively rare class of proteins, the seven-membrane proteins and the 12-membrane proteins, where almost all of the protein is sitting in the membrane, and there are only small loops sticking out. But they are very interesting from a pharmaceutical point of view, so we are pursuing these vigorously.
When is the next version likely to come out?
We have said that we will at least do one version every year, and we hope to have, from now on, about 2,000 new proteins every year. But we might actually do it more frequently. Then, also, we want to expand it into other diseases, but also extend it into blood screening and comparisons with transcript profiles of RNA.
What are your plans for follow-on funding once the funding from the Wallenberg Foundation has run out?
We got $30 million [from the Wallenberg Foundation, starting in 2003]. We have spent about half of that, so we have funding for another two years. But right now, we are in the planning phase for a scale-up of the project to speed up things even more. We are hoping that we can obtain more funding for this, but so far, we don't know yet. We are investigating [possible funding sources].
How does this project tie in with the HUPO antibody initiative?
I am responsible for the antibody initiative at HUPO. We have a meeting [this week] in Munich where we will discuss how our efforts can be coordinated with similar efforts in the United States, Asia, and Europe.
What other antibody projects exist?
I don't think there is any other effort which is doing a similar production, but there are several pilot projects going on, at the Sanger Institute in England, in Germany, and several efforts in China. And then there are more than 100 antibody companies. What I am very keen is to try to not compete with the antibody companies but to collaborate with them to in the end have a comprehensive toolbox of antibodies to all human proteins.
You have invented a number of important technologies which of these are you most proud of, and which are the most important ones for proteomics?
We published already in 1983 a paper on the use of protein A as an affinity tag for purification of proteins. That was the first paper ever published on using affinity tags for the purification of proteins. And obviously, tens of thousands of researchers are now using affinity tags for purification. That is probably what, in a way, I am most proud of.
The discovery and the generation of protein A and protein G for antibody purification has, of course, become an enormous success, and biopharmaceuticals purified with these proteins are now a multi-billion-dollar business.
And then I am quite proud of the affinity reagents that we created, which are artificial antibodies and called affibodies, which have been published in numerous journals. We recently have very, very promising results in pre-clinical cancer applications in animal models.
And then I have to say, this is not proteomics, we developed some years ago a new DNA sequencing technology called pyrosequencing, which was published in Science. [Recently] in Nature, there [was] a landmark paper from a company in the [United States] called 454 that has made a new DNA sequencing instrument that is about 100 times more efficient than the ABI sequencer. They are using the technology from our group, and I am extremely proud, actually.
What do you think have been the most interesting developments in proteomics in the last couple of years?
I think you are asking the wrong person here because I am a great believer in what I call genome-based proteomics, that is to take a gene-by-gene approach to proteomics and catalog the corresponding proteins. I think there is only one way of doing that, and that is to make affinity reagents to every human protein. This is obviously what I have been spending the last maybe five years of my research career trying to perfect.
Obviously there is, then, the "classical" proteomics involving mass spec and electrophoresis and so on. But in my view, I think it's more valuable for the scientific community to have protein atlases that catalog the proteins, and this is what we try to do. But nothing has really come out of that yet, the first sort of stumbling steps will be released on Monday. It remains to be seen how useful it will be, I guess.