At A Glance
Name: Hiroyuki Matsumoto
Position: Professor of biochemistry and molecular biology, University of Oklahoma Health Sciences Center, since 1985.
Director, NCRR Oklahoma COBRE Vision Research Proteomics Facility
Background: Assistant research scientist in biological sciences, Purdue University, 1980-85.
Junior researcher in organic chemistry, University of Hawaii, 1977-79.
PhD in biophysics, Kyoto University, Japan, 1977.
BS in biochemistry, Kyoto University, Japan, 1972.
How did you get involved in proteomics?
We were trying to determine the phosphorylation site of Drosophila photoreceptor protein that undergoes reversible light-dependent phosphorylation. We knew it was a phosphoprotein because, if you seed the flies with P32 and then use a 2D gel [you detect the phosphorylation].
There are two schools of thought about the display of proteins. In order to study proteins, you have to display the proteins. That means separation, purification, that kind of thing. There are many ways of displaying proteins. One way is 2D gels. There’s also 1D SDS PAGE. Even any kind of column chromatography can possibl[y] display your protein at very low resolution of course.
After awhile, around [the year] 2000, people started saying the age of 2D gels is over as a way to display proteins because, as everybody knows, a 2D gel is not almighty. It cannot display all of the proteins. [A] 2D gel usually cannot display membrane proteins, it cannot, probably, display minor proteins. So now they are saying you have to use a 2D or 3D HPLC. But we are still a 2D gel group.
Around 1980 I started using Drosophila looking at the proteins expressed in the photoreceptor. As you know, Drosophila is a very good model system to study almost any kind of biological subject. We had been using Drosophila as a model for photoreceptor signal transduction.
The original idea was that if you look at the proteins of the Drosophila photoreceptor preparation, you need a reference. The reference is before you activate the photoreceptor signaling — that is, in the dark — and then after you activate the photoreceptor — that is, when you shine a light. The idea is, you might be able to observe a change in the proteins in a short amount of time, like in a second. Particularly, phosphorylation would change the isoelectric point, [and] the 2D gel is supposed to be able to identify the pI shift.
So that’s where I started this project, which still is going on. It’s very slow!
So you started working on your retina phosphoprotein project in the 1980s?
That’s correct. We identified maybe five or six different phosphorylations induced in vivo. We had a series of mutants in which the visual signal transduction [was] defective. The flies [were] blind because of the mutation. And with blind flies you don’t see any light-induced phosphorylation at all, suggesting that probably [the] reversible light-induced phosphorylation of proteins — especially photoreceptor-specific proteins — must be playing a crucial role in the visual signal transduction. By displaying proteins you can possibly pick up the target — the proteins that look very important. From my point of view, that’s an early attempt at proteomics. And actually, many of the proteomics projects start like that: let’s compare cancer cells with normal cells, for example.
So we identified a phosphoprotein that undergoes reversible phosphorylation in vivo in the living fly eyes. But in the ‘80s, to know what this protein is, is very difficult — almost impossible. In the mid- to late-‘80s, they invented the gas phase sequencer and a PVDF blot to identify the proteins on the gel. If you’re lucky you can possibly sequence the N-terminal 10 amino acids. That’s when you had no genome information, no mass spectrometer, no proteomics. So we did that, and also during the 1980s there was a vector called gt11. This vector expresses your protein in a library. So if you have a library hooked to the gt11 vector, you can express your protein and then display the expressed proteins from the library. Then, to screen the expression library you have to have an antibody. So we raised the antibody against the protein spot separated on the 2D gel. And then we used that antibody to screen the gt11 library. We identified the protein to be a homologue to the already-cloned gene product restin — that’s in 1990. Finding the photoreceptor-specific ligand used for phosphorylation was published in Science. And then cloning of that gene was published in Science again in 1990. So it was going slowly, but going somewhere!
So in 1990 we cloned the gene that turned out to be already cloned in vertebrates — restin — but the difference was that it was a phoshorylated form. Then the question is which [residue] is phosphorylated in vivo. In 1990 is around when John Fenn invented ESI and Koichi Tanaka [developed] the ionization method that became MALDI. Before that, you could not ionize peptides at all. You could ionize anything that is volatile, but not peptides. The problem was, in 1990 if you wanted to use a mass spectrometer you had to be a physicist [and] build your own mass spec. But 1990 was around the time that commercial mass spec companies started building and selling it. So I became very interested in using mass spec. Somehow I raised money to buy an electrospray tandem quadrupole Sciex API 3. On this campus, in collaboration with Ken Jackson, we interfaced a micro-HPLC — that is a very early [HPLC] from BioMicro — it’s a very slow flow. So we interfaced this HPLC from BioMicro.
We digested the target protein in-gel. I think we did the first in-gel digestion of 2D gels followed by mass spectrometry. And we published that in 1994, confirming the phosphorylation site in vivo using HPLC ESI quadrupole. In 1994, I think probably two or three groups published something very similar. That was a very early attempt. That’s when we hooked up to this technology.
So you then started looking at larger quantities of proteins at a time?
Yes. At that time, since we had already been running 2D gels, we could look up other proteins. And now there was genomic information. Because in 1995, if you do in-gel digestion and get the peptide mass fingerprint, there’s no way you could ask the question, ‘what is this?’ Because the genome was not complete. Consider the MS-Fit [PMF] program. In the beginning, since gene entry was not great, you’d never be able to find the target, compared to these days. If someone had an output from MS-Fit five years ago, you would search and not hit, simply because there’s no gene registered. Then that situation improved little by little and now you can hit almost all of the target proteins.
Anyway, we started using a vertebrate retina — first a big one, the bovine. We’d use a purified retina from bovine, run a 2D gel; maybe you can compare it to some other tissue if you want, and then do PMF. The problem with the bovine is still that genomic information is scarce. There’s no [completed] bovine genome project.
What mass spec system were you using at this point?
We had a Voyager Elite [from Applied Biosystems]. That’s a simple, reflectron-type MALDI, which is actually very good. We purchased that in 1995 and it’s still running. But ABI decided to retire that model of the Voyager and not to support [repairs]. Fortunately we had the money to buy the Shimadzu Kratos Axima QIT. This is a MALDI quadrupole ion trap TOF. So you trap the ion in the quadrupole and then kick it out and measure the TOF. Koichi Tanaka, who I know very well, was part of the team that developed this in Manchester, UK. I think this is a great machine, because you can trap the ion and then do multiple MS/MS. This is very similar to LCQ from Thermo, but QIT has the second mass analyzer — the TOF — and the QIT is a higher-end machine.
You can [also] easily identify glycopeptides within the machine, because the sugar units start falling off automatically if you increase the laser power. So what we discovered is, using this machine, we can do the complete analysis of sugar — polysaccharide structure analysis. That means you can determine the site of the sugar. Traditionally, what you have to do is remove the sugar by glycosidase, and then run HPLC. But with this new type of machine, you don’t have to remove the sugar unit by glycosidase. You can just analyze everything from the beginning. We’re writing a paper about this. Of course, there are some other types of machines that could do something similar, but it has to be high-end. The choice of mass spec is the most crucial, and it depends on how much you have [to spend]. If you have lots of money and lots of mass spec skills, then FT-ICR is a good choice. [But] we have never been a mass spec specialist [lab].
So you run the glycopeptide through the mass spec and then take off sugars?
Yes, then you take off the sugars one by one. This is a very new technique. QIT can do that.
So are you still looking at retina?
Yes, we are looking at the proteins expressed in the retina of the developing mouse. Proteomics involves lots of disciplines, but you should never forget that you have to have a project. My future projection is probably everybody will have their own mass spectrometer. So I think lots of PIs who have their own projects will start using proteomics but hopefully within their groups. If you really want to do good science, you probably want to have your own mass spectrometer. But it’s difficult, because even among the faculty in my department — biochemistry — there’s a lot of activation energy necessary. That is, people get intimidated by the machine.
What’s been lacking is education in that area. If you do a web search for ‘proteomics graduate program,’ you don’t find that many worldwide. Somebody has to do it. The education of the graduate students in this area is very important.