At A Glance
Name: Daniel Liebler
Position: Beginning of May 2003: Professor of biochemistry and director of proteomics, Vanderbilt University Medical Center
Currently: Professor, department of pharmacology and toxicology, University of Arizona, since 1987; Director, Southwest Environmental Health Sciences Center, since 1999
Prior Experience: Postdoc, Oregon State University, 1984-87. Worked with Donald Reed on cellular protection mechanisms and antioxidants.
PhD in pharmacology, Vanderbilt University, 1984. Worked with Fred Guengerich on cytochrome p450 mechanisms and reactive intermediate chemistry.
What got you into proteomics?
When I did my PhD at Vanderbilt in pharmacology in the early 1980s, one of the areas of interest in toxicology was: What are the protein targets of reactive intermediates from toxic chemicals? We have known about DNA damage and its role in cancer for a long time, but we know that reactive intermediates also modify proteins. But back then, there were no useful tools to be able to identify protein targets of reactive intermediates. So I spent the next 20 years working on other things. When I took my position here [at the University of Arizona] in 1987, I started working on antioxidant chemistry, and I did that until about 1998. I did a lot of mass spectrom-etry, but it was of small molecules. In January of 1998, I heard a talk by John Yates about mining proteomes and genomes with mass spectrometry. I [then] quit doing the antioxidant chemistry: I almost made a decision overnight to quit and change my program to proteomics.
The focus of my program has been on developing and adapting new mass spectrometry-based methods to detecting protein targets of reactive chemicals, and to mapping the adducts and understanding the significance of protein modifications in toxicity. This is related to the general problem of protein modifications, which is of great interest in proteomics. Many of the tools and approaches that we develop are relevant to areas outside toxicology.
What approach do you use in your lab?
We work primarily using shotgun analysis of the type that was developed by John Yates and his colleagues, so we do multi-dimensional LC/tandem-MS. We do use a little bit of MALDI in a core laboratory here, but our primary instrumentation has been Thermo Finnigan LC-Qs. We have one in my laboratory and another couple in a proteomics core that we have used here. Right now, I have about 10 people, and about seven of them are moving to Vanderbilt. At Vanderbilt, we are going to have a significant increase in available instrumentation.
What have you been able to achieve so far?
It took us about a year and a half just to reorient the lab towards an entirely new discipline, being able to do certain pretty common proteomics-related techniques. Once we had that going, our focus has been on understanding the tandem-MS fragmentation characteristics of different types of modified peptides. What we wanted to do is to be able to acquire data on complex mixtures of peptides that are modified or unmodified, and then be able to mine the data with new tools and software to identify those spectra that correspond to modified variants of peptides, and then use the MS/MS information in these spectra to map the modifications at the level of the amino acid sequence.
The tools that are commonly used for correlations of spectra to database sequences, such as Sequest, have some utility for mapping modifications, but only when you can anticipate the mass of the modification, and what type of amino acid it is going to be on. But we knew that many modifications produce characteristics in the spectra that Sequest would not be able to use in correlating to sequences in databases. And in many cases, we don’t know the amino acid targets, it could be a number of possibilities, but that yields too many combinatorial possibilities to use Sequest for. And sometimes, the chemical modifications are of unexpected masses because of secondary chemistries that the adducts undergo. We needed a tool to be able to go through data and find spectra that correspond to peptides with specific modifications that yield adduct-specific fragmentation characteristics. So we developed a new algorithm that we call SALSA, that stands for scoring algorithm for spectral analysis. SALSA allows the user to identify spectra that display adduct-specific fragmentation characteristics.
How does SALSA work?
It simply scores all the spectra based on their correspondence to the user-specified set of characteristics. Spectra of peptides that have the adducts score very highly, and spectra of peptides that don’t have the adducts usually don’t score very highly. So it’s a sorting mechanism for a large dataset. What we have extended SALSA to do is look for a series of ions in a spectrum that correspond to a peptide sequence motif. If you have a spectrum of the unmodified peptide, SALSA establishes a virtual ruler, where the tick-marks on the ruler correspond to the spacings between the signals in an MS/MS spectrum that would correspond to a particular sequence. If you had a particular amino acid sequence, you would expect to have a series of ions with specified distances between them on the m/z axis. If you had a modification somewhere on those peptides, some parts or all of that series might be preserved intact, but they might be shifted by the mass of the adduct to different positions on the m/z axis. What SALSA can do is detect spectra that correspond not only to the unmodified target sequence but modified and even sequence-variant versions of the target sequence.
How is SALSA available to researchers?
The University of Arizona licensed it to Thermo Finnigan. It has just been released as a part of their BioWorks 3.1 software. It is an exclusive license; it has been licensed for proteomics applications, and the license has recently been amended for small-molecule analysis. We are continuing to work with Thermo Finnigan on updating it. We have already improved on it [to provide it as] a multiprocessor version for more rapid evaluation of larger datasets. We also have a new similar type of program that’s in development, called P-Mod. It is a program that does something like SALSA but it provides a statistical evaluation of the probability of a match between a sequence and a spectrum, and it exactly locates the site of modification. [Whether it will be commercialized by Thermo has] to be determined by negotiations.
What are your most important scientific results?
We have published work recently on hemoglobin adducts of aliphatic epoxides. We were able to use LC/tandem-MS and SALSA to map over a dozen different adducts, and to document the concentration dependence and selectivity in hemoglobin. We have done this with other proteins, some of which is published. We would like to be able to take a proteome-wide view of susceptibility to chemical injury.
The other thing that we are really interested in is, how do covalent modifications affect protein-protein interactions? We hypothesize that adduct formation will distort protein structures and change their binding partners. This could be a means by which protein modification disrupts cellular signaling mechanisms and leads to gene activation and stress responses.
We are also interested in how protein damage by covalent modification affects protein ubiquitination and SUMOylation. SUMO is a ubiquitin-like protein that attaches to other proteins. It doesn’t target those proteins for degradation, but it seems to affect their subcellular localization and stability. We have found in work that we are going to publish soon that there is a significant shift in SUMO target proteins with chemical stress.
What practical implications could your work have?
One of the most problematic features of some drugs is that they are converted to reactive intermediates that bind to proteins. There is a great deal of evidence that this type of reaction is associated not only with liver toxicities, but also with immune-mediated toxicities, so-called idiosyncratic reactions, that cause drugs to later on be removed from the market. These problems aren’t usually detected until you are in phase IV of clinical trials. Idiosyncratic reactions are very strongly associated with covalent binding of proteins, and it’s thought that that’s a trigger for immune recognition of the damaged protein and for the toxicity. In the pharmaceutical industry right now, there aren’t really good surrogate markers for drugs that can bind to proteins. If we could identify commonly targeted domains that might serve as markers for bioactivation of drugs in vivo, or even in cellular model systems, that would enable the development of some relatively quick assays.
What technical challenges still need to be overcome?
For mapping protein modifications, the biggest problem is getting high-quality datasets that provide spectra of all the peptides in a mixture. You need a high degree of so-called sequence coverage. This is certainly made easier by tandem-LC approaches, but we still have a long way to go in that area in being able to acquire spectra of not only unmodified peptides but the modified forms. New instruments will certainly help us. The new generation of tandem mass spectrometers, whether it’s the MALDI-TOF/TOF, Q-TOFs, and the new ion trap FT instrument that Finnigan has just released, will be the most useful for doing work on protein modifications. We also need to develop better affinity enrichment approaches to capture a subset of proteins that might be modified, for example. That will require some very creative chemistry and biochemistry. We have a paper in press in the Journal of Proteome Research on the use of an N-terminal stable isotope tag, phenyl isocyanate, that tags all peptides that have a free N-terminus, so we can use that to quantify modified peptides, which is what something like the ICAT approach does not allow you to do.