AT A GLANCE
NAME: Norm Dovichi
POSITION: Professor of Analytical Chemistry, University of Washington
PRIOR EXPERIENCE: Developed flow cytometry and capillary electrophoresis components for sequencing DNA, later licensed by Applied Biosystems for use in the 3700 DNA sequencer
How did you get involved in proteomics?
It’s a very long and roundabout route. My bachelor’s degree is in chemistry and mathematics [from Northern Illinois University], my PhD is in analytical chemistry [from the University of Utah], but my postdoc is really where I started to learn about the biological sciences. I did my postdoc at Los Alamos. I spent half of my time doing physical chemistry, but through a happy series of accidents I spent the other half of my time doing biophysics, and in particular doing flow cytometry.
How were you using flow cytometry?
I am an instrumentation developing kind of person, and so I was using flow cytometry to push its analytical capabilities, to see what it can do as a fluorescence detection technology. I did not have a particular application, but instead I was exploring what that technology can do as a high-sensitivity technology.
In whose lab were you working?
That was with the flow cytometry group in Los Alamos, so Jim Jett was the senior person with whom I worked on that project. The other name was Dick Keller — he hired me and I spent half my time working with Jett.
What did flow cytometry lead to?
I went back and turned into a laser jock for a while, but where it got interesting again, was when we coupled capillary electrophoresis with the fluorescence detector that is at the heart of the flow cytometer. We used it for the analysis of fluorescent versions of PTH (phenylthiohydantoin) amino acids.
What led you to those amino acids?
Because they were easy to synthesize, and because they were an interesting target to look at. This was back in the early to mid 80s, and the idea was to develop an ultra sensitive Edman sequencer. We coupled capillary electrophoresis with the fluorescence part of the flow cytometer.
The next step is a detour. We coupled capillary electrophoresis with the fluorescence detector for sequencing DNA, and developed a multiple capillary DNA sequencer, based on that fluorescence detector. That technology was licensed to Applied Biosystems, and formed in part the basis of their 3700 DNA sequencer, which is what Celera used in their genome sequencing project, and in fact what most labs use.
Simultaneously we were developing technology for ultra-sensitive protein analysis using the capillary electrophoresis and the fluorescence detector.
What’s the difference between those two applications?
Well, DNA is easy and proteins are tough from an analytical point of view. With DNA you can easily put a fluorescent label enzymatically onto your template, and the separation of DNA is straightforward — once you figure out one piece of DNA any other piece of DNA has a similar separation. Sequencing the human genome is really much simpler than doing hard core protein analysis, simply because of the complexity of the proteins. So for the proteins we had to figure out how to do 2D electrophoresis on them in capillaries, we had to figure out appropriate fluorescent labeling technology, and ultimately our current project is to figure out how to do all of this on a single cell.
So what we do today is we take single cancer cells — stem cells, neurons, whatever your favorite cell is — lyse the cell, fluorescently label the proteins, do 2D capillary electrophoresis on the proteins, detect with ultra sensitive laser-induced fluorescence, and resolve a few hundred components.
Why jump to proteins?
Funding in the DNA world dried up about five years ago, and it was boring, frankly. We had the technology in hand, we demonstrated that it worked, and while we could have dotted a few i’s and crossed a few t’s, it was boring. It was clear that the more interesting science was going to be at the protein expression level rather than at the genomic level.
Why choose to work on just one cell?
Because we can. What we put down in proposals is a number of reasons why people should care about it. It goes back to my cytometry experience at Los Alamos though, where I saw the power of characterizing single cells. If you think about flow cytometry, they’re able to characterize [not much more than] five properties of a cell, and we’re able to characterize hundreds of properties, hundreds of components within a single cell. So anything flow cytometry can be of value for, this can offer twenty or thirty or forty times the value.
How do you isolate just one cell?
We do cell cultures, lift the cells off the plates, put a drop of cell suspension on a microscope, find a cell that we wish to study, suck it inside of a capillary, and get to work.
How does the separation occur?
Well, it’s capillary electrophoresis, and we do the capillary version of SDS PAGE on the first capillary, then we transfer fractions to a second capillary, where we do a different separation based on some other properties of the protein. We can’t do bioelectric focusing at the moment, so we explore different types of capillary electrophoresis.
The technical term is micellar electrophoresis, and it relies on the interactions between proteins and surfactants.
How sensitive is the technique?
We get detection limits of a few thousand copies, say 2,000 or 1,000 copies of most proteins.
What determines what cell you want to look at?
The first answer is that at the moment, if we have a cell suspension, we look under the microscope and just take one of the puppies, [because] we’re busy developing the separations. [But] what we have demonstrated is that we can measure the phase of the cell in the cell cycle, by using DNA-intercalating stains.
The idea there, is that we’ll characterize cells as to whether they’re in the G1 or G2M phase, and do our study as a function of the cell cycle. In the future it’s pretty straightforward to imagine, using some sort of fluorescent antibodies, characterizing the cells before plucking one out for analysis. We’re collaborating with oncologists, and it’s also straightforward to assume that a pathologist could sit at the microscope and based on the morphology of the cells, suggest that we analyze one of those, for example some specific cell surface marker.
What allows you to pick out cells at a particular stage in the cell cycle?
We stain the cell with a particular fluorescent reagent that binds to DNA, and use the amount of fluorescence from the cell to determine where the cell is in the cell cycle, using a fluorescent microscope. So, we stare at the cell under the fluorescent microscope and measure the fluorescence of the cell, and use that to decide which cell to select for analysis.
What do you use to identify the different components?
I didn’t say anything about identifying the components. What we do is like old O’Farrell’s electrophoresis — we get a bunch of spots. The identification of the proteins at the moment relies on spiking the sample with standard proteins, and one way that we get standard proteins is to do O’Farrell 2D electrophoresis, cut out a spot, use mass spectrometry to identify that protein, and then use that protein to spike our sample.
Are there other things that you envision?
Well, we have a bunch of students who are getting into applications of the technology: cancer biology, developmental biology. One other project is to perform protein analysis in single cells of lower copy number proteins.
How do you do that?
We take the gene for GFP [green fluorescent protein], fuse it with the gene of interest, and express it in the system — we’re working in yeast. And with the GFP we can detect very near, if not at, the single copy level. The idea is to detect single copies of specific proteins in single cells. We do this with electrophoresis, and what that buys us is the ability to see posttranslational modifications on the target, and we’ve already demonstrated that phosphorylation and proteolysis can be detected this way.
Finally, the whole point in the exercise is to determine what it means to say that a protein is not expressed, or that a gene is not expressed. It might mean that there’s no copies present, or it might mean that the analytical technology is not sufficiently sensitive. We’ll have sensitivity for single copy detection so we’ll be able to answer those questions, and we’ll be able to answer the fate of those very low level expressed proteins.