At A Glance
Name: Richard Johnson
Position: Research Scientist 4, Amgen (until 2002: Immunex), since 1995
Prior Experience: PhD, MIT, 1983-88. Sequenced thioredoxins by tandem mass spectrometry in Klaus Biemann’s group.
Postdoc, Harvard School of Public Health, 1988-89
Postdoc, Biomembrane Institute, Seattle, 1990
Postdoc, University of Washington, Seattle, 1990-1995. Studied proteins and their posttranslational modifications by mass spectrometry with Ken Walsh.
Wrote de novo protein sequencing program Lutefisk.
How did you get into protein mass spectrometry?
I got a PhD with Klaus Biemann [at MIT]. At the time, there was no such thing as proteomics, and the way you made ions was with fast atom bombardment. You needed somewhere around a nanomole of a peptide to see FAB-induced fragment ions. In contrast, gas phase Edman sequencing only used picomoles, so there was this difference of three orders of magnitude. Then Biemann managed to get money to buy a four-sector instrument, which was this huge thing that was very complicated to work on. I was originally assigned the task of sequencing small proteins, thioredoxins, and the related glutaredoxins, so I would collect a bunch of MS/MS spectra using the [instrument], which generated high-energy collisions. For a long time, I would just interpret spectra by hand, and try to deduce the sequence, and then piece it all together. I had some Fortran experience from college, so I wrote a Fortran program to assist me. It was just a puzzle, adding and subtracting numbers.
Are you the first person ever to sequence a protein by tandem mass spectrometry?
It was in 1986 when [John] Yates and [Don] Hunt came out with their paper in PNAS that everybody cites now [entitled “Protein sequencing by tandem mass spectrometry”], and they did it on a known protein. I have at least the distinction of saying that I [first sequenced an] unknown protein, a thioredoxin [from Chromatium vinosum. This was published in 1987].
What did you do after leaving Biemann’s lab?
[My wife] was finishing up her residency at Mass General, so I had to find something to do for a year and a half. I was allowed to work at Vernon Reinhold’s lab at the Harvard School of Public Health, who was doing mass spectrometry of carbohydrates. I did some lipid A work; there was no particular big project. Then I very briefly went to Senitiroh Hakomori’s lab at the Biomembrane Institute in Seattle. Right after I got there, he started losing his funding, and it was starting to look pretty bleak. I thought [that] maybe I [should] just go learn protein chemistry [in Ken Walsh’s lab at the University of Washington]. [When] I first called him — it was a cold call — he [kept saying] ‘I am busy right now, call me in a few minutes.’ So I kept calling him. Finally he was willing to talk to me, and it turns out the reason he kept putting me off was, he was busy on the phone negotiating with Sciex for buying a mass spectrometer for the first time. He was quite happy then. It was mutually agreed upon that I should help them learn mass spectrometry, and he would help me learn protein chemistry. It worked out pretty well, and I was there for just about five years. The most exciting thing that came out of that was when we would find that recoverin and transducin, these eye proteins, had this heterogeneous acylation.
[After finishing my postdoc] I tried to find an academic position that was within an hour’s drive of a ski area, but I was not successful in my search. So I went to Immunex [in Seattle] as a staff scientist. Amgen bought Immunex last July.
Tell me about Lutefisk, your de novo sequencing program.
I was reading about how other people were doing de novo sequencing using this graph theory approach. I thought I would take a stab at it, so I wrote it, elaborated on it, and spent a lot of time working at it, and I convinced Immunex to let me give it away. Anything that I do to Lutefisk is freely available, including source code. [If] someone e-mails me, I send it back as an attachment.
I think Alex Taylor and I were probably the first ones to try a combination of de novo sequencing results with a homology search, which is different from the usual pattern matching from a database search. That idea has been picked up by Micromass, and [Andrej] Shevchenko has done the same thing. We did it with FASTA. [For example], if you don’t have a fragment ion between two glycines, then the combination of those masses adds up to asparagine, so your de novo sequencing result would put an asparagine at that site, and when you do a homology match, you would be off by one amino acid. Alex informed FASTA how to take into account these problems with MS/MS sequencing. Other things we sometimes see is, a dipeptide is inverted, and Alex’s modified FASTA will [also] take that into account.
What else have you been doing at Immunex?
[Around the time that] I started, Matthias Mann came out with his nanospray [ionization], and I thought, that’s really neat. Mostly what I did with that technology was help [people identify] bands from gels. The first really interesting thing was finding a receptor for TRAIL, a TNF receptor family member, using affinity columns made with TRAIL. [After many trials] finally we hit on something that wasn’t in the database; we actually deduced a sequence using mass spectrometry. That was kind of my bread and butter for a long time, working with biologists who were doing affinity captures and helping them identify proteins.
How is proteomics organized at Amgen?
At Amgen, they used to have a huge project run by Scott Patterson. They would create culture supernatants [from cell lines], and try to identify as many ESTs as possible, using some sort of glycoprotein capture method. I believe that they were very successful and they did a pretty good job, but in the end, I can’t say that there was any drug developed from that, so Amgen subsequently shut most of the proteomics down. When they bought Immunex, oddly, the Immunex proteomics group was twice the size of the Amgen group — there are four of us doing this [here at Immunex], including me. We have got first generation Q-TOFs, LC-Qs, and MALDI-TOFs, pretty standard [equipment].
Where do you see the greatest utility for proteomics in drug development?
For discovery, my bias is towards affinity capture, assisting a hypothesis-driven approach. The global proteomics stuff, I don’t know that that is going to fly, just gathering more and more information until someone writes a program that figures it out. A lot of pharmaceutical companies have switched their proteomics efforts to biomarker discovery. Amgen is [also] thinking that might be an interesting thing to start looking into. Medically, if you did find biomarkers, it would be very interesting. If you have an interest in biology, I would say biomarkers are not particularly interesting because what does it mean if you see a piece of haptoglobin clipped in some disease state? The likelihood of ever figuring out the chain of events that led to that is pretty slim.
There is also possibly some interest in finding off-target proteins that interact with a drug, by putting a tether on that drug and then doing affinity capture and seeing what you find.
The other big thing would be mechanism-of-action studies. For example, Amgen/Immunex is still is working on an antibody to EGF receptor for an oncology therapeutic. After Iressa did not exhibit the anticipated efficacy, oncologists were thinking they should spend more time figuring out how this works. There is some interest in looking at phosphorylation changes in EGF receptor or other receptor tyrosine kinases. That may be a big effort that we have in the next year or two, looking at phosphorylation changes. I would say [that] this is the farthest along [in drug development] that I have ever been — I am usually involved way at the beginning.
Where do you see a need for new developments?
I think the big problem is dynamic range. If you are looking at serum, for example, there is a 109-fold difference in amounts. I think mass spectrometry [operates] at a 104 range. In the end, it all boils down to peak capacity, which is either chromatographic, where you just separate out everything as best as you can, or you could think of having higher resolution of the mass spectrometer. [For that] you have the FTMS, but I am personally afraid of it, and possibly a lot of other people are, too. The perception is, you have to hire a postdoc from Dick Smith’s lab to get it to run properly. I think the companies that make FTMS’s have to work hard at making them simple to use.
How about quantitation?
I like isotope dilution, ICAT-like things, but it’s usually pairwise. You can probably devise reagents where you are comparing three things at once, but eventually you cannot multiplex too much, or else you have filled up your mass spectrum. I wonder if there are other ways to quantitate, if you compare an LC/MS run of a protein under one state versus another, and if you were able to look at the ion intensity and somehow normalize to some internal standard to very roughly get some quantitative information. In principle that should work, but commercial software is not quite there.
Where is proteomics going?
I remember going to some meeting where Ruedi [Aebersold] was going on about ‘How can we poor little academic labs ever compete [with industry], and what will our role be in proteomics?’ As things have evolved, I am thinking that proteomics is going right back to academics. The global proteomic studies are not going to be done in a for-profit company, because there is no immediate value seen to that sort of approach.