Joshua LaBaer on Doing Proteomics Without Mass Specs or 2D Gels

At A Glance

Name: Joshua LaBaer

Age: 43

Position: Director, Institute of Proteomics, Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School

Prior Experience: BA, University of California, Berkeley, 1977-81

MD/PhD, University of California, San Francisco, 1981-90. Studied glucocorticoid receptor-DNA interactions.

Internship and residency in internal medicine, Brigham and Women’s Hospital, 1990-92

Fellowship in medical oncology, Dana-Farber Cancer Institute, Boston, 1992-95

Postdoc, Harvard Medical School, 1994-99. Studied cell cycle proteins with Ed Harlow.

How did you become interested in proteomics?

[After finishing my medical degree and my PhD] I went on to do a postdoctoral fellowship with Ed Harlow at Harvard Medical School, doing cell cycle research. It was in that context that I started to get interested in more global questions. What we used to do in Ed’s lab was screen through libraries of cDNA to find genes that could create certain phenotypes in cells. One of the frustrations of that approach was how pitiful it was to do in mammalian cells. We used to dream about what it would be like if one had the perfect library, a complete collection of all the genes in the human. Also, I was always someone in the lab who liked to try new technologies and new approaches. When Ed Harlow became the chairman of the department here, he had this idea to start an institute to use this more globalized approach to study protein function, and recruited me to run it.

Tell me more about the Harvard Institute of Proteomics.

It’s about three and a half years old now and has a staff of about 40 in four or five major projects. It’s funded through a variety of mechanisms, including NIH grants. We do some work with industry, and we have had some industry and philanthropic support, for example from the Breast Cancer Research Foundation. Pfizer has given us money — they support a postdoc — and Vertex also gave us some. Harvard Medical School has provided us with seed money to get the institute off the ground. We also have a long-term relationship with [BD Biosciences] Clontech regarding the Creator system [for cloning]. We take a slightly broader view of what we would call proteomics than some might take.

Can you explain that?

In many people’s minds, proteomics is somewhat restricted to what I would call abundance-based proteomics. What you are driving at with that approach is to see if you can find differences in protein abundance, which might represent some clue as to a role it might play in disease, or [whether] it may be a marker in disease.

We don’t do any of that. We have no mass spectrometers in the lab, we don’t do any 2D gels, and I don’t think we have any liquid chromatography. In contrast, our interest is in what I would call functional-based proteomics. What we want to do is express the proteins and examine their functions. There are a lot of methods that people have used to study proteins in the past. The difference is that historically, people have focused entirely on one or maybe a couple of proteins. What we saw as our future was to come up with ways to use that same approach, but multiply it by a thousand: Not to express one protein at a time, but to express hundreds or thousands

That sounds similar to what the structural biologists are doing…

Structural biology is just one aspect of that, to crystallize a protein and look at its three-dimensional structure. But you could imagine a couple of dozen others: For example, where is the protein expressed in the cell? What tissues is it in? What proteins does it interact with? What is its catalytic activity? What pathways does it participate in? The idea is to come up with ways of asking that to many proteins at a time.

To do that, you need to have the genes in hand to actually make those proteins. What we quickly realized was that there weren’t large collections of full-length cDNAs available, and where there were, they were not available in a format that was conducive to high-throughput expression. One of the first efforts that we began was to build a collection of clones, for humans and other organisms. We called this the FLEXGene repository, which stands for full-length expression-ready genes. The genes themselves are in a recombination-based vector, which means they can be moved in a single step from one vector to another. We started out with the Gateway technology from Life Technologies, which was bought by Invitrogen. Most currently, we have also been working with the Creator system from BD Biosciences Clontech.

What have you achieved so far?

We have got somewhere around 5,000 human clones, and we are nearing completion of the Saccharomyces cerevisiae collection. We have also built a complete collection for Pseudomonas aeruginosa. Besides Pseudomonas, we are looking now at things like Vibrio cholerae and other potential pathogenic organisms, because we are pretty good at capturing genes and can do it pretty fast. Each of those sets is slightly different in terms of where we are with it. A number of the human clones will be coming out soon and will be available to everybody. Our goal for all these sets is to make them publicly available, and we are right now trying to figure out how to get them out.

What’s the next step after making these clones?

One approach is to develop methods for high-throughput protein expression and purification. A student in the lab attempted to express and purify over 1,000 different human proteins on a microscale in a bacterial system. Under denaturing conditions, he could [purify] about 80 percent of them; I think the number is closer to 60 percent under non-denaturing conditions. You can do some information mining and ask questions like ‘what characteristics of proteins predict whether they will or will not be able to be purified using a high throughput setting?’ This might be a nice front-end for building protein microarrays, which only requires a very small amount of protein for spotting. [The student] has just moved over to Gavin MacBeath’s lab [at Harvard’s Bauer Center for Genomics Research], and we are planning some collaborations.

We are also doing a lot of work right now in cell-free extracts. For example, the Japanese structural genomics initiative is using almost exclusively cell-free expression of proteins as its front end. We have modeled after that approach and can make close to one milligram of protein per milliliter. We see that as a nice way of doing 96-well production of proteins because it obviates the need to lyse the cells and extract the proteins from them. The other nice thing about it is that you can add other things to the mix, for example, that posttranslationally modify the proteins.

What else have you been doing with those clones?

Another approach that we have taken is a more in vivo functional approach. Basically, the approach is to put these genes into retroviruses or other delivery systems that let us get the genes into mammalian cells and ask biological or phenotypic questions. As an example, we have built a collection of genes related to breast cancer. We are hoping that by the end of this year, we will have 1,000 of these genes assembled and sequenced. We then put them into immortalized breast epithelial cells. This is in collaboration with Joan Brugge here at the medical school, whose lab developed ways to grow these cells in culture so that they form structures that look very much like the structures in the normal breast. One of the things we are interested in is what genes are responsible for enabling these cells to acquire the ability to invade the tissue. To do that, we put these genes into cells, put the cells onto a membrane, and ask, ‘can we induce the cells to migrate to the other side of the membrane?’ The goal is to screen as many genes as we can to identify all the possible candidates, and then use other assays to look for other features and try to assemble the importance of these genes. Once we identified genes and their proteins that play a role, then the next step would be to zero in on those proteins and see what they do.

What do you think will be the most immediate impact proteomics will have in medicine?

There is a huge raging search right now for biomarkers, and I think eventually, that search will yield fruit. But I also think that it’s farther off than some people may think [due to] the number of validation studies that need to take place to confirm that these markers are real.

I also think that people are underestimating the power of functional proteomics. I think that as people start to do high-throughput functional studies, they are going to zero in quickly on interesting pathways and proteins that play important roles, and that’s going to get them very quickly to identify proteins that may be important in various biological functions, and that will lead to drug discovery.

I also think you are going see things like protein microarrays appear in the context of validation experiments in the pharmaceutical industry — not just for discovery but also validation, and in comparisons; for example, comparing the binding of a protein to its intended target as well as to a series of all possible secondary targets.

What is your institute planning to do in the near future?

One of our first steps is that we are moving, hopefully in June. We will still be Harvard, but we are going to be in the same building with Eric Lander’s group and Stu Schreiber’s group at the Whitehead Genome Center, and will be almost tripling our space. This is an opportunity for our labs to commingle and interact.

Our hope is that in a couple of years we can wrap up the cloning effort, so that we have a complete collection of the human, and then to start doing these high- throughput experiments.