At A Glance
Name: Joel Pounds
Position: Staff scientist, protein function group, Pacific Northwest National Laboratory, Richland, Wash. (since 2000)
Background: PhD, University of Wisconsin, Madison, 1977. Studied a protozoan insect pathogen.
FDA National Center for Toxicological Research, Jefferson, Ark. (1977-1985)
Brookhaven National Laboratory, NY (1985-1990)
Professor of Toxicology, Wayne State University, Mich. (1990-1999)
Recently published a paper in Molecular & Cellular Proteomics entitled “Toward a Human Blood Serum Proteome: Analysis by Multidimensional Separation Coupled with Mass Spectrometry.”
How did you get involved with proteomics, and what is your main area of interest?
My training is in toxicology. I have been interested in the biology and toxicology of trace elements, which interact with proteins. This laboratory, the Environmental Molecular Sciences Laboratory, has a great deal of expertise in mass spectrometry-based proteomics and bioinformatics. I was looking for ways to apply this technology and expertise to problems that interested me, such as biomarkers of chemically induced diseases, how an organism responds to disease, biomarkers for exposure to some agent, and biomarkers that would define whether an animal or human is susceptible. My goal is to try to understand the biology of a chemically induced disease in an experimental animal, and then have a much better scientific basis to extrapolate the susceptibility to that disease to humans using proteomic characterization.
For example, lead poisoning is a problem in many parts of this country and of the world. It has persistent and long-term effects on learning, memory, and behavior. What we don’t understand is how some individuals are more susceptible to the effects than others.
For your recent MCP paper, why did you decide to study serum proteins from a healthy individual?
We wanted to work with serum for a couple of reasons. First of all, it’s readily available: I can get blood from an experimental animal or from a human a lot more easily than I can get a piece of brain tissue. Also, proteins in plasma are already used for disease diagnosis. Because we can get the exact same sample both from animals and humans, that gives you a better opportunity for direct cross-species comparisons than trying to look at tissues.
There are other investigators here at the laboratory [Karin Rodland and David Springer] who are working on understanding ovarian cancer, which is a disease for which there are no good diagnostic tools for early detection, and late detection of ovarian cancer has a poor prognosis. There is an immediate need for useful, sensitive, and early markers for ovarian cancer. Through them I could get clinical samples. But the first stage was to determine what is found in normal adult human female serum.
How did you go about your analysis?
We pretty much used existing technologies, but we worked harder at separating out the major components of serum. [After removing immunoglobulins and digesting the remaining proteins by trypsin] we used strong cation exchange chromatography to separate the peptides and analyzed them on a [reverse phase] HPLC tandem mass spectrometer. Furthermore, we did m/z segmentation. Our goal at this point was not to quantitate the proteins but to drill down and find as many different proteins as we could. We identified 490 proteins; about 290 had been described in serum before.
What were the most interesting or surprising results?
There were about 75 proteins that haven’t been annotated yet. Nothing was really surprising, because anything that’s in any cell, in any tissue, could eventually end up in plasma. However, we were surprised by the number of cellular proteins, including transcription factors and proteins related to DNA repair, that we observed. For example, we found G protein coupled kinase, rhodopsin kinase, and retinal outer segment membrane protein in serum. Because the tissue mass of photoreceptors is quite small and they are part of the central nervous system, which is separated from direct contact with blood, I would not have predicted we would observe these proteins in serum. It is conceivable that these proteins originated in some other tissue or that our anonymous, healthy donor has higher than expected levels of these proteins in serum. The functional significance of these and many other proteins we identified awaits further study.
What was the dynamic range of proteins you could detect?
We didn’t measure dynamic range specifically, because we were not measuring the concentration. We know we have a good dynamic range because we have been able to measure proteins that we know from biochemical studies are there in very low abundance. Our fractionation procedure has probably given us a dynamic range of seven orders of magnitude, but we ran the sample many times.
Why did you use serum, not plasma?
We had a debate amongst ourselves when we started. If you have serum, you have a greater chance of getting hemoglobin and other red blood cell proteins from hemolysis. We didn’t really want that, but we went with serum anyway, because it seemed that there were many more opportunities for us to get serum from collaborators, tissue archives, and cancer centers than there were to get plasma. Many cancer centers, for example, have archives of tissue samples, tumor tissue and adjacent normal tissue, and serum. They don’t as often have plasma.
What are you doing now?
Looking for money (laughs). We are trying to get research support to do the ovarian cancer biomarker discovery project. We are also looking at another body fluid from mice, bronchoalveolar lavage, lung washes. We are using Richard Smith’s high-resolution, high sensitivity FT-ICR mass spectrometer [at PNNL] for that project, and we are doing quantitation with that. That instrument gives us, compared to what we did on the plasma, more high-throughput, more sensitivity, and a better dynamic range, as well as quantitation. We are analyzing the data right now, and I hope we will have a manuscript prepared early next year. We will be using that instrument, and that expertise, when we do the ovarian cancer biomarker study.
When will you go ahead with the ovarian cancer study, and how will it differ from the Lancet study Lance Liotta and Chip Petricoin published this year?
I hope we will start early next year. The approach that they used is really excellent at finding the proteomics fingerprint, but what it does not tell them is what proteins are useful for predicting the cancer. What we will be doing is identifying the proteins.
Why do you think that’s necessary?
Because you can’t put a mass spectrometer into hospitals to do proteomics. That is a bit of an exaggeration, but the technology is not there, the cost is not there, the validation is not there. What we will be doing is identifying the proteins that are putative biomarkers by using the FT-ICR based on the database that we just published. Then we will be using protein chips to do high-throughput screening and quantitative measurements. The protein chip studies can be done in a much more high throughput way, much more economical and more quantitative, than mass spectrometry. Protein chips are developed here by Richard Zangar and Susan Varnum.
In your study, you used several human databases — why did you do that?
We were trying to find any differences in the completeness of the databases that would be manifested in protein identifications. It didn’t make a big difference. They weren’t all equally non-redundant though, by which I mean that a gene or a protein may be identified under four different names. Getting rid of that redundancy was one of the things that we worked on to give us confidence in our 490 identifications.
Are you participating in the plasma proteome projects initiated by HUPO and by Leigh Anderson’s Plasma Proteome Institute?
Yes. Those are the kind of efforts that benefit everyone, and we are very happy to be doing that.
You found 490 serum proteins using small amounts of serum from one donor. What did you miss?
I don’t remember the exact numbers, but of proteins that are found in serum or plasma in the zeptomolar range, there are only going to be a few hundred molecules in a body. Your chances of finding enough of those proteins in a milliliter are vanishingly small. There may be fractionation or immunoprecipitation procedures that would allow these proteins to be enriched from very low concentration into some sample that can be analyzed more readily. But the arithmetic supports that you can’t find everything in one milliliter-size sample.
How many proteins are there likely to be in serum?
It depends on how you define the proteins, if you consider all the posttranslational modification and all the splice variants. We really don’t know, because we haven’t been able to get down to the zeptomolar concentrations, even to the attomolar concentrations; we are still discovering what’s there. There can also be foreign proteins from microorganisms, for example viruses or bacteria. Any disease state may produce proteins that end up or are secreted into serum. It undoubtedly will be a very large number, probably much larger than the number of proteins found in any given cell, because any of those proteins from that cell could end up in plasma.