Institute for Population Proteomics
Director of research and technology development
Name: Dobrin Nedelkov
Position: Founding President, Institute for Population Proteomics; director of research and technology development, Intrinsic Bioprobes, 1999 to present
Background: Postdoc associate in molecular biophysics and biochemistry, Yale School of Medicine, 1998 to 1999; teaching assistant, biochemistry and biophysics, Arizona State University
While the effort to find variations in individual genes across a population has been ongoing since 2002 with the creation of the International HapMap Project, no similar work is being done on a protein level.
Dobrin Nedelkov has been pushing for such work for several years now, though the community has been cool to the idea. In a paper published in the Jan. 21 online edition of Proteomics, he outlines the concept of population proteomics, and the results of two studies he and Intrinsic Bioprobes conducted looking at protein diversity across a healthy population in the United States.
ProteoMonitor spoke with Nedelkov in 2005 about how he got into proteomics [See PM 01/14/05]. Below is an edited version of a recent conversation with ProteoMonitor where he sketched out his ideas for population proteomics.
I’m guessing population proteomics is based on the idea of population genomics.
Exactly. This was termed toward the end of 2003, beginning of 2004, so it’s been four years now. Essentially, it’s a result of some developments in protein assays that would enable us to screen single proteins from many individuals. And yes, it’s drawing on the idea of population genomics, but the whole concept there is enabled by much simpler approaches.
This is a little bit different. I work at Intrinsic Bioprobes [and] we’ve been developing these assays, they’re called mass spectrometric immunoassays, essentially targeting individual proteins. Then instead of fluorescence-based detection, for example, we use mass spectrometry.
Now after developing the assays, and confirming that they’re reproducible and sensitive enough, we said, ‘Well, it’s probably time to start applying them onto a larger number of samples.’ And these samples come from various people. Once we started putting this together, we started realizing that single proteins will have different structural modifications in various individuals.
So, hence the term population proteomics because proteomics, the way we practice it, is applied to [whole] populations. We’re interested in seeing a single protein across many individuals versus the current work or what has been where people analyze hundreds and thousands of proteins from a single sample.
In a way aren’t people doing population proteomics even if they don’t call it that, because there is so much data out there on single proteins across different groups of people? It may just not be organized in the way you’re talking about.
Well, I don’t see that many publications in that regard. Most publications of proteomics, first, have been on technology development and then they put a couple of columns together and they try to see how many proteins can be detected with one experiment which might last for a couple of weeks.
The problem with these experiments is they’re hardly reproducible and there’s really no true answer. Now on the clinical proteomics side, it’s a little bit different where a number of years ago with [Lance Liotta and Emanuel Petricoin’s] Lancet paper, people started using the technology to analyze validated clinical samples and seeing what the differences were.
The issue there is in the detection of proteomic patterns versus no known proteins. This is a little bit different. The whole concept of population proteomics is to study protein diversity in [the] human population.
I don’t know if you read the last issue of Science in 2007. The cover page said the breakthrough of the year was the human genetic variation. And I’m saying now while we’re still doing genetic studies, we need to seriously address the issue of variation at the protein level where post-translational processing will introduce various differences.
And this is a dynamic process.
I don’t really hear a lot of support for this. I don’t hear people talking about it at conferences, for example. Why not?
Because the people don’t have in their labs and their research centers technologies and approaches that will enable this effort. ELISAs are common, but those will produce only quantitative read-outs. In order to do mass spectrometry on the scale that we’re doing, it takes some effort.
It’s a different mindset. Either people are not ready for it yet, or they’re just realizing that proteins can be different. We’re about the only group that constantly publishes on this [issue]. But there are some other people who are referencing our papers and supporting the idea. A lot of people are actually supporting the idea. Why they’re not doing anything about it, well, you should ask them.
But if you look at HUPO, the plasma proteome initiative, the third aim is to study protein diversity. That’s the final aim of that initiative, so we’re doing that, and I think the organization is still focused more on sample preparation, which technique to go with and so on and so forth.
We picked one approach and we’re moving forward with that. My feeling is that it’s going to take some time for people to start accepting this, so they need to get some enabling technologies and work at their place and start analyzing people.
Do you think the field has matured enough that you can talk about such a large effort? People are still having trouble identifying one biomarker, one protein, in one sample, and you’re talking about biomarkers across a very large population.
It is mature enough. More and more people are [doing work] in biomarker studies, but nobody’s doing any systematic effort. These studies are done, for example, with 100 control samples, and 50 diseases.
This needs to be done first, to find out the underlying variation in normal people, in healthy subjects. There’s no effort to study healthy people, let’s put it that way. Everybody’s interested in getting biomarkers because that’s the biggest payout there.
I think people realize the power of mass spectrometry, and now they’re applying it. I think more and more people are doing it. The problem is there is still no conclusive data that has identified and validated specific biomarkers for [any] disease. So, it’s still in development and it’s still undergoing … and I think it will [continue to be] undergoing for the next couple of years. Hopefully, some good positive outlook will come out of it and people will start realizing how important this is and powerful.
What do you think it’s going to take to get people not only interested but actually active in pursuing work in this area?
Good data. Basically, discovery that will be independently validated through standard assays like ELISAs, and then people will say ‘Wow, this can make a difference in the medical field, in clinical diagnosis, or in therapeutic treatment.’
So, until the promise is delivered, people will be skeptical.
So are we looking at a few years down the road?
Absolutely. And I think people have to realize that sometimes the simpler approach is the better approach. Instead of looking at 1,000 things at once using computer algorithms, look at a couple of things, use your common sense and simple programs to find out differences in these things in the proteins.
The more complex the data is, the more unlikely that a meaningful clinical outcome will be produced. Look at the ELISAs, they’re the simplest things possible, and most clinical biomarkers today are assayed by ELISA. So one marker at a time.
Lay out for me how you would go about doing this. What lessons have you learned from the HapMap model?
First, we’re trying to establish the variations among the healthy population. We started with 100 people, and then we went to 1,000. One thousand is not nearly enough because you start seeing these modifications that occurred one in 1,000.
But you need to get maybe up to 10,000 controls … to establish with good statistical validity the occurrence of these modifications in the healthy population. And then you start studying disease cohorts. It can be a specific type of cancer, it could be cardiovascular disease. We’re already starting to do this internally. And you look at the differences between the profiles for specific proteins among the disease cohorts and the healthy individuals.
Of course, the big question is, ‘How do you choose which protein to work with?’ You start by analyzing all these established biomarkers or proteins that are either pathways or have been indicated as potential biomarkers. For example, cardiovascular disease, you start to study BNP [brain natriuretic peptide] or troponin. It’s been shown that these proteins show extensive truncations, so you look at how these truncations can be correlated with the stage of disease. Mass spectrometry is the only way to provide that information.
This has to be applied to disease, of course.
Would we need to identify the entire human proteome before we embark on this?
No, absolutely not, because there’s no such thing as a definite number [of proteins]. There’s a definite number of genes, but at any point in time, the proteome is dynamic. Not all genes are expressed in all tissues. Even among people, it depends on what the specific person ate that morning.
It’s a noble task to identify all proteins expressed, but it’s almost meaningless. Everybody will be right. Somebody will say there are 2,000 proteins, another group will say there are 2,200 expressed, but it’s not going to provide an answer to this question of discovering novel biomarkers.
We can’t wait until that point, so that’s why we have to start working now. And I think people have realized that. Most groups have moved away from the quest for delineating the whole proteome and moved into the clinical proteomics arena.
Is there any commercial interest in what you’re trying to do in population proteomics?
Right now there is not. But population proteomics applied to specific diseases, yes there is. And we’re working on a couple of projects right now. Some people call it biomarker discovery, we call it population-based biomarker discovery.
But for broad, general population screening, there’s no interest from commercial companies.
That’s more of a non-profit proposition, and that’s why we formed the Institute for Population Proteomics. That will pretty much undertake the studies that will be to the public’s benefit.
Who formed this Institute?
I am the founding president. Intrinsic has no stake in it.
The Institute has been incorporated now for two years. It’s a non-profit, now it’s a matter of sending out the message. In the beginning we will disseminate data that was already published. We’ll try to assemble everything that’s known about this protein modification among the general population. It will serve as a depository of samples, maybe, that people will have access to do these types of studies.
And the mission of it is what?
The mission is to study protein diversity essentially in the human population.
Is there any talk of a HapMap-type of project for proteins?
No, not right now. [It’s] too early. With the HapMap, there is a set of genes. With the proteins, so many diseases are related to these proteins, and it’s hard to pick a topic. Every group pushes their own agenda essentially, so there is very little common interest.