At A Glance
Name: William Lane
Position: Director, Harvard Micochemistry and Proteomics Analysis Facility, Department of Molecular and Cellular Biology, Harvard University, Cambridge, since 1987 (at facility since 1982)
Prior Experience: Research Assistant, Protein Sequencing Core Lab, University of Massachusetts Medical School, 1981
Protein Sequencing Research Assistant, Brandeis University, 1979-81
BA, Cornell University, Biochemistry, 1979
What led you to protein analysis?
I was lucky enough to have two passions, science and music. People often don’t see the connection, [but] when you are doing something very well, the creative part of that has always been as much an art as a science to me.
I took a five-year leave of absence from my undergraduate degree at Cornell University, for music, and then returned. As I was finishing up in 1977, I came to Boston, completing my degree in absentia here at Harvard as a Special Student in Immunology, and also at Northeastern and BU. In 1979, I started Edman sequencing with Ed Cannon at Brandeis University, whom I consider to be a mentor. He taught me everything I know about these technologies. He was doing traditional Edman sequencing, and he was a very good mix of being an incredibly gentle yet intelligent advisor, especially in the sense of analytical particularities. We would identify a single amino acid cycle of Edman chemistry in four different ways: GC, HPLC, TLC, and back hydrolysis/AAA. That approach met well with my analytical compulsiveness. I was also still a professional musician and wanted to do things faster, so that I could get home to perform. That drove me to both the things that have always been critical in what we do today, and that is sensitivity, speed, and quality. You are only as sensitive as the best current technology can be, but in a core facility it doesn’t matter how sensitive you are: if you are not giving robust, defensible answers, then you are dead in a core facility. It was in 1982 that I arrived at the Harvard Microchemistry Facility after a brief stint at the University of Massachusetts Medical School protein sequencing lab, and I have been here ever since.
When did you start using mass spectrometry?
I was starting to watch the field, and how it might impact us, in the late 80s. In 1990 I went to my first MPSA, Methods in Protein Sequence Analysis, now called Methods in Protein Structure Analysis, a biannual meeting. This one was above the Arctic Circle, in Kiruna [in Sweden]. I went to see Don Hunt speak, and it was career altering. In that talk he stated that we would all have ion traps on our desks two years hence. I didn’t know him other than through his work, and I went into the hallway and introduced myself, and he said, ‘I give this class periodically, it’s three days of intense hell.’ And I went [to his course] on de novo sequencing by mass spectrometry, and I just loved it.
By ‘91, I had bought my first triple quadrupole, and about six months later, our first MALDI-TOF. I do remember this being early enough for a typical Edman sequencing core facility that a few faculty members said something like, ‘What the heck does Bill Lane need a mass spectrometer for?’ Of course shortly afterwards, Don [Hunt] was starting to come out with his MHC peptide work by microcapillary HPLC tandem mass spectrometry, and it was not long before people asked, ‘Gee, do we have this?’ Fortunately, I was already moving in that direction.
In the early 1990s, we knew the approach of LC-tandem MS really had an advantage over a MALDI-TOF mass searching-only approach to protein identification because of the specificity and discrimination of a peptide sequence. It was fairly non-trivial for a non-mass spectrometry group, which we were, to learn how to do what is very commonplace today.
By 1995 we placed the very first order for a Finnigan LCQ ion trap. MS/MS search algorithms were starting to become more widely disseminated, and I was collaborating at a fairly simple level with John Yates and Jimmy Eng. They had developed Sequest, but only a preliminary web page for presenting the data. At this time I met a math undergraduate who was taking a year off, his name was Martin Baker, and in a three-month period, we developed a core of an intranet that provided a web-based interface to Sequest to analyze and present the data [which Thermo Finnigan licensed]. Since then, we have been carrying somewhere between three and six part-time programmers, largely drawn from talented undergraduates here. Their talent is often non-biology: they are mathematicians or computer science majors.
If someone were to ask me what the three most important things are in modern proteomics, I would say ‘software, software, software.’
How are you equipped?
There is an Applied Biosystems Voyager-DE STR, a Micromass Q-Tof Ultima API, and [there are] three Thermo Finnigan LCQ Deca XP Plus ion traps. Then there are two ABI Procise Edman sequencers, still cranking. There is nothing quite like [Edman sequencing] as far as true de novo sequencing goes when material exists, and there are projects where material does exist. For example, we work with the CDC on lots of organisms that may not have their genome characterized. We also have an ABI 420 amino acid analyzer, and a Tecan robot for any type of automation of chemistries, spotting, HPLC tube delivery, etc.
What services do you most frequently perform?
At most cores, [including ours], the bulk [of work] is protein identification. This lab is strongly biased towards LC-MS/MS because of [its ability for] tremendous discrimination. In my view, there just is no more powerful set of technologies than a micro-HPLC coupled through nano-electrospray to a high sensitivity tandem mass spectrometer — our favorite for that is an ion trap mass spectrometer. What is rather unsung is that no matter whether you are the best interpreter in the world with the greatest experience, there are MS/MS spectra that algorithms can reveal to you that are correct that you would never have been able to interpret manually. You still have to manually review data, though, if you are going to [operate with] high sensitivity.
Determining covalent modifications is a subset of protein identification to me. You are going through the same process, but it requires much more manual interaction to be able to evaluate and leave no stone unturned in what may be very weak data. Borrowing that joke again, I will almost always start a discussion with a researcher who is calling to set up a phosphorylation project by saying that there are three things that matter in such a project, and they are stoichiometry, stoichiometry, and stoichiometry. People initially imagine that if I have 40 theoretical tryptic peptides, I will see a mass shift, and one of them will tell me what my answer is. In fact, if it is a one percent phosphorylation level, the signal-to-noise will be one in 4,000, rather than in one in 40. I usually emphasize that it’s much less trivial than people assume. The ABRF proteomics research group this year came out with exactly that conclusion in their study: much more work needs to be done.
Where do you see the technology going? People talk a lot about FTMS these days?
There are always three simple drivers to mass spec technology: greater sensitivity, speed, and higher mass accuracy and resolution, and they are often at penalty of one another. There are things that we are going to answer on one instrument that we are not going to answer on another. FT gets attention, especially the incarnation of a linear ion trap FTMS. That has some of the promise of exquisite throughput and sensitivity that the ion traps have in tandem MS mode with a readout that has been the domain of higher resolution mass spectrometers, including the hybrid quadrupole TOF instruments. It remains to be seen, but there is a lot of promise in that and I believe that it will be one of the key instruments.
Are you involved in any large-scale proteomics projects or mostly in smaller ones?
It’s certainly mostly [smaller projects], because there are so many labs needing to identify [proteins]. The most common extension of that is [identifying] associated proteins in complexes, and those complexes can be quite large. By design, those are more mixture analyses rather than individual protein identifications. We also participate in looking at multiple states and continuing to look at variation in those states, like a progression of phosphorylation through the cell cycle or after stimulation.
What makes a successful facility?
My best friend is an MD, his father was a rather famous MD who dealt in back pain on Park Avenue in New York City. I learned early on from [him] that if you resolve somebody’s back pain, they are your friend for life. And there is a similarity to doing a very, very good job here. You need to set the expectations of the researchers in the right place. So much of it is prep-oriented, requires good advice and a good walk-through of what the technologies really can do and what they cannot do. In this lab, there is an insistence that we have extremely thorough discussions on all projects, even what might be considered to be the most trivial of projects — a molecular weight determination, or an amino acid analysis. Researchers must talk to me before a sample is submitted. And that has clarified expectation levels and [ensured] that they understood where and when they get their answers. And that it is expensive to do that. Our lab is 100 percent supported by service fees. We are known as being a slightly more expensive laboratory for that reason, but we have no grant funding subsidizing us. It comes back to my work in music and to doing something meaningful. The funny line we walk is between doing good science, with all of the wonderful things about “giving back” that is inherent in science, and at the same time, having a business component to it. The priority has to be the science, and the business can only follow from that.