At A Glance
Name: Philip Andrews
Position: Director, Michigan Proteome Consortium, since 2001
Associate professor, later professor, Biological Chemistry, University of Michigan, since 1990
Background: PhD in Biochemistry, Purdue University, 1974-1977. Characterized pyrophosphatase.
Postdoc, later research scientist, Purdue University, 1978-90. Studied prohormones using mass spectrometry.
How did you get into proteomics?
I really started working with proteomics back in the days before it was called proteomics. I worked on the characterization of prohormones, like proinsulin and proglucagon, asking what their processing products were in tissues. I began to use fast atom bombardment mass spectrometry at that point, in the mid’ 80s. It had just become available, and so had the genes of these hormones. We wrote some software early on that allowed us to take the gene sequences and predict what the masses of all the fragments would be. This is exactly analogous to what we do now with tryptic digests, except that we didn’t have the specificity.
We had some great collaborators back then. The person who helped us the most getting off the ground was Catherine Fenselau. She was incredibly helpful to us in the early days of learning about what fast atom bombardment could do. I actually went out to her laboratory for a few days and we collected a tremendous amount of data in a short period of time. I came back to Purdue University and said, ‘Look, we have got a fast atom bombardment machine here: Let’s use that instrument to get this kind of information.’
I came to Michigan in 1990 as an associate professor and as director of a protein structure facility, which I ran for several years here. My goal then was to bring mass spectrometry online to support not just protein chemists, but all biochemists and molecular biologists. We brought online electrospray mass spectrometry, which was brand new at the time. We got the first model of the Vestec electrospray instrument — in fact, it still said “Thermospray” on the front of the instrument. Then we got a MALDI-TOF, also from Vestec.
Now you are director of the Michigan Proteome Consortium — tell me about it.
The state of Michigan decided with great foresight to take about $1 billion of the tobacco settlement funds and put those into development of biotechnology. They realized that the infrastructure for biotechnology companies in the state was not extremely strong. In order to beef it up somewhat they decided to put funds into something called the Core Technologies Alliance, of which the Michigan Proteome Consortium is one component. Our job is to provide infrastructure for all investigators in the state of Michigan, whether they are academics or in industry. Amongst all the four sites, we have over 20 people at this point. The consortium opened for business at the beginning of April of last year, but in terms of having received the funding, we are starting our second year.
How much funding does the consortium receive?
The budget is $13.6 million for 5 years; that’s split amongst all four of the cores. It allowed us to invest in the new instrumentation and the expert personnel to provide services. For example, in the proteome mapping core, we have two LCQs from Thermo Finnigan, an electrospray Q-TOF and a MALDI-Q-TOF from Waters, and a MALDI-TOF as well as a MALDI-TOF/TOF — the model 4700 — from ABI. We have also just ordered the new Finnigan ion trap FTMS instrument. We think the new [FTMS] designs, not just from Thermo Finnigan, at least promise to be higher throughput.
What kinds of services do you offer?
We have high throughput yeast two-hybrid, which is directed by Russ Finley at Wayne State University. Then there is a protein microarray core that is run by Brian Haab at Van Andel Research Institute. He can offer custom arraying of proteins, and we expect this will continue to increase over this next year. He is using the Hydrogel technology that works extremely well for him.
The third component is of course classic proteome mapping. This is a core that is split between University of Michigan and Michigan State. We have an associate director, Brett Phinney at Michigan State, and I am the director of that core. We run the bulk of the large scale 2D gel projects here at University of Michigan, whereas they do the bulk of the 2D-LC-MS/MS work at Michigan State. Of course there is overlap to some extent.
How many collaborators do you have in the consortium? How many companies have used your services?
At the time of our last progress report, which was about four months ago, we had completed or had in process 60 projects for about 50 investigators. The bulk of those in the beginning have been in the academic community, and that’s primarily because the demand at the moment, in the state of Michigan, is much higher on the academic side. We have also done projects for two large companies in Michigan, including Dow Chemical, and one outside of the state, as well as maybe seven with smaller Michigan companies.
What have been the greatest challenges so far?
We have some basic instrumentation that is very reliable and robust, but we have also committed to higher end instrumentation where we expect more downtime, at least at the beginning. Also, there is training of personnel, and holding on to them once they are trained. But I think the information management has been the biggest challenge for us. You can’t do high throughput proteomics, or even development of methodologies in proteomics, anymore without some information management capabilities. One of our efforts in our first year was to develop an information management system that was focused on proteomics. It was not intended to be a general LIMS but a very focused one that would do everything you need within proteomics. It is called PRIME [Proteome Research Information Management Environment].
The goal of PRIME is to support proteomics efforts of medium scale, and cutting edge research type proteomics where you expect the methodologies and instrumentation to change on a regular basis. It’s intended to be as low maintenance as we can make it, yet flexible. Phase 1 is just being finished up this week: It is to support 2D gels, and also includes client access to the data. Phase 2 is targeted to July 1. It involves bringing the LC-MS/MS and 2D LC-MS/MS capabilities online and to implement the bar-coding capabilities. Phase 3 begins the protein interaction mapping support and is to be complete by the end of the year. What we have done is design PRIME so that any independent developers can write their own Java applications that interface with the PRIME database.
Do you plan to make PRIME commercially available?
There has been a tremendous amount of interest in PRIME, and a number of sites would like to have it installed. One possibility is to spin off a company, another to license this technology to an existing company. We are still exploring our options at this point.
How do you deal with different data formats of mass spec vendors?
That is an issue, and the solution depends on the manufacturer. Some are easier to deal with than others in this regard. There is a need for standard file formats amongst the manufacturers. This is of critical importance if proteomics is to make the kind of advances that genomics has, and this is completely within the control of the mass spectrometry and the robotics companies. They don’t have to give up their proprietary file formats: What they need to provide is the ability to export to preferably a standard file format, or into an ASCII file, and not do it in a manual way. All of those companies have the ability to export in ASCII file format, but not in an automated way.
How many researchers have you trained so far?
We have a postdoctoral training program here in the consortium that is training postdocs for jobs in industry; I don’t know of any other program like this in the country. We have had seven postdocs go through this program, two of which have already found positions in industry. We have three positions open this year and will advertise these next week.
Where do you see proteomics going?
Mass spectrometry, on a per-sample basis, is quite expensive at this stage. The advent of instruments like the TOF-TOF that can give you higher throughput has dramatically decreased cost on a per-sample basis, but if you are talking about clinical proteomics, the cost is still too high. One obvious area for that is protein arrays.
On the computational side, there is a huge amount of information available in all proteomics data that is not being fully utilized. For example, there is a huge need for de novo sequencing algorithms that are more effective. We also need better tools for fully automated image analysis of 2D gels.
Also, we need improved separation technologies. You see actually a resurgence in interest in chromatography, similar to what happened with 2D gels once it became obvious that you could identify proteins from them. One of the things that could have a tremendous impact on proteomics on the mass spec side is FTMS. At the last ASMS meeting, there was a feeling that FTMS was coming into its own in proteomics. It was still not a high throughput methodology then, but the capabilities that people were demonstrating, like top-down sequencing of proteins, are extraordinary.
Posttranslational modifications are perhaps the largest frontier for proteomics. And that’s going to require development of new technologies.