AT A GLANCE
NAME: Thomas Conrads
POSITION: Director, Mass Spectrometry Center, Laboratory of Proteomics and Analytical Technologies, SAIC-Frederick, 2002-present.
Associate director, Mass Spectrometry Center, 2001-02.
BACKGROUND: Senior staff scientist, Pacific Northwest National Laboratory, 2000-01.
Post-doc in proteomics with Richard Smith, Pacific Northwest National Laboratory, 1999-2000.
PhD in biochemistry, Ohio State University, 1999.
BS in biochemistry and molecular biology, Washington State University, 1993.
How did you get involved with proteomics?
For me, proteomics came about because in graduate school I did a lot of functional studies on proteins. What I mean by that is the normal type of research, where I studied one protein in excruciating detail using whatever technology might be available. We employed a lot of spectro-scopies, so I got a good foundation in physical biochemistry.
I wanted to get at using the hardcore biophysics that I learned and fuse it with more clinically relevant science. So I got a post-doc in Richard Smith’s lab. That’s where I met Timothy Veenstra. About two years ago he got an offer to be director of proteomics at the NCI-Frederick and offered me the associate directorship of the lab. The opportunity to be at the NCI presented a certain uniqueness to it, because at the NCI you need to be aware every time you walk into a building about how you can apply your science to the study of cancer and our better understanding [of it]. We work extremely closely with physicians.
How does SAIC-Frederick relate to the NCI?
SAIC-Frederick is a part of Science Applications International Corporation. We were awarded seven years ago [the designation] to run the contracts for the NCI. We operate a lot of the core technologies for the NCI. So for example at our laboratory of proteomics and analytical technologies, we work extremely closely with folks like [Emanuel] Petricoin and [Lance] Liotta. What Tim has been charged with is forming a biomedical proteomics program at Frederick. That program is populated by multiple laboratories whose expertise is in different fields of systems biology. So we’ve got our lab of proteomics and analytical technology, a lab of molecular technology that does a lot of microarray and DNA sequencing, we’ve got a protein expression lab that does extremely high throughput expression of proteins, etc. The idea is to bring together a wide variety of biological techniques and be able to use them to accomplish mission-type[s] of problems — so to tackle metastasis or onset of ovarian cancer, [for example].
My job is to apply mass spec to a variety of biological problems. So investigators come to me and say, ‘what can proteomics do to better understand orphan receptors, or better understand mechanisms of cancer?’ My job is to interface with these more clinically inclined folks and try to come up with a strategy to tackle their problem. The technology is moving so fast that it’s difficult for clinicians to keep apprised of all their options.
The best scenario we’ve been involved with is that with Petricoin and Liotta, where they formed a nice basis for a diagnostics [tool], but it had some drawbacks. So we sat down two years ago and proposed some changes that might be made to lead to a better diagnostic — improving things like false positives. That was a beautiful example of how the technology-inclined can bridge the gap with the clinically inclined. The biggest change that we helped with was the application of higher resolution and higher performance mass spec. You can go down a path that seems very fruitful, but you may end up hitting a brick wall. And that brick wall might be the underlying technology.
Tell me more about your ongoing projects at the proteomics lab.
Of course the largest project is the ovarian cancer diagnostic, which is based on the use of SELDI interface with a high performance mass spectrometry technique. Of course we’ve got a lot of spinoff projects that surround the main thrust — we’re making progress on other cancers, with SELDI projects that are covering things like indications [differentiating] benign from malignant prostate cancer. But of course that leads us to other types of problems, like trying to understand the nature of the serum proteome.
We’ve got three thrusts of trying to understand serum. We’ve got a global serum study that we’re doing that is essentially just the cataloguing of the global constituents of serum. That’s led us to the identification of 1,600 proteins in serum. We’re trying to develop techniques where we can actually use that in a higher throughput fashion to start to compare serum versus serum. So we’re looking at labeling serum using a technique that Catherine Fenselau (see story p. 1) has been largely responsible for forwarding, which is using trypsin-mediated O18 labeling. There we’ve been playing off another serum study that we’re doing, which is looking at the low molecular weight proteome. In that study, we filtered serum through a low-pass cut-off filter, for essentially 15 to 16 kDal [proteins]. We recovered the low molecular weight proteome, and we started to enumerate what was in it. So now we’re doing isotope labeling studies using that fraction of serum.
If you read the SELDI literature, all those peaks that seem to be important turn out to be less than 15 kDal. So we think this low molecular weight proteome is a huge archive of histopathological information.
How does this all relate to the albumin sponges you discussed at HUPO (see PM 10-17-03)?
That’s our last thrust. There’s criticism out there that SELDI-based technologies are only looking at the tip of the iceberg, because they’re only seeing the most high abundant proteins. Our argument is that, ‘yes that’s true, but what we’re actually detecting as being diagnostic are species that are bound to these high molecular weight, high abundant proteins that we’re pulling out by SELDI.’ So we’re looking at these species that are binding to albumin or transferrin, which are the ones that are being pulled down by the SELDI surface.
The argument is, you’d need a 5 kg tumor to actually secrete enough species to really see. That’s true if there’s no mechanism of concentrating the species. But what we think is, these species are being secreted into the bloodstream, and partitioning onto these high abundant, sticky carrier proteins. It’s through this partitioning on the backs of these proteins with very long half lives of 15 to 30 days in the bloodstream that these are being concentrated enough so we can see them by SELDI. We have done quite a lot of work to show proof of principle, that in fact if you do selectively pull out albumin or transferrin, you see diagnostic species in that pull-down fraction. A couple of those papers are now submitted for publication. It makes too much sense. That’s the business of these proteins.
All of these technologies are being developed and designed to be implemented to the clinic. So [for] the low molecular weight filtrate, we’re designing a high-throughput fashion in which we can interrogate many different clinical sera. The pull-down of albumin, for example, we’re doing now in array format so we can do very high-throughput studies.
Do you do clinical studies at SAIC-Frederick as well?
Yes, the SAIC has set up a clinical reference laboratory. We’re reproducing the technology in our lab down at the laboratory there. The CRL is charged with taking everything we do on the research scale and taking it through clinical trials. We’ve got three Q-STARs there, robotics, and everything needed to start the clinical trial process is down there. That’s administered by SAIC for the NCI, and that’s their mission.
So what will you be working on in the next year?
One of the major thrusts in the lab is trying to take the SELDI-based diagnostic and move it to a higher-throughput, cheaper platform which would be MALDI-based. You’ve got now MALDI platforms where you’ve got a disposable chip in an extremely high throughput fashion, and you can blast through that in a day. Essentially it’s an order of magnitude higher throughput than anything we’ve got now. One of the important things for any test is to make it approachable to the entire population, and to do that, it has to be cheap enough. One of the ways to do that is to make it higher throughput. But there are a lot of challenges to making a MALDI test. SELDI is uniquely designed to pull out species and clean up the sample at the same time, and doing a MALDI-based type of application, there’s more sample handling involved. That’s what we’re trying to work on — to get the liquid handling down and get it to the point where it’s producing MALDI diagnostics.
Of course another large effort in the lab is to get at the identities of the diagnostic features that we’re finding in the SELDI patterns. That involves going back to the bench and doing a lot of good old biochemistry to try to get at how to purify these to the point where we can now start to identify them. Even though the basis of the test isn’t requiring their identification, of course it makes us all feel better that the formation of our diagnostic is actually based on clinically relevant peptides.