At A Glance
Name: Shane Burgess
Position: Assistant professor, department of basic sciences, College of Veterinary Medicine, Mississippi State University, since 2002.
Background: PhD, University of Bristol School of Medicine and Institute of Animal Health, United Kingdom, 1999.
BVSc, Massey University, School of Veterinary Medicine, New Zealand, 1989.
As Thermo Electron prepares to launch its new Bioworks 3.2 software which will incorporate the open source format mzData (see story on p.1), ProteoMonitor caught up with one of the software beta testers who faces special challenges in working with poorly annotated, non-traditional genomes.
How did you end up combining proteomics with production animal research?
I got my vet degree in New Zealand, and then after going all over the place, I ended up doing a PhD in Britain. My PhD was conferred by the Bristol Medical School, but I actually was at the Institute for Animal Health in Compton, which is one of the BBSRC. The BBSRC is a large umbrella organization which runs a lot of research. There isn’t really a similar thing in the US — it’s kind of like a cross between the NSF and the USDA. Anyway, I did my PhD there, so I’m a vet. And I did my PhD in immunology, cancer biology, and virology.
We moved over to the US, and at that time I had the feeling that the chicken genome would be done and I thought this would be an ideal time to apply some of these post-genomic techniques to animals. The reason I actually went to Mississippi State was because it turned out in the year 2001, just before I got here, a couple of quite visionary people at Mississippi State invested quite heavily in the Life Sciences Biotechnology Institute. And that Life Sciences Biotechnology Institute was established to take advantage of technology like proteomics, in addition to other functional genomics technology.
So it kind of worked out at the same time when I was looking at what I was going to be doing the future, I thought proteomics had a lot of potential. At that stage I realized the complexity of the proteome, but I didn’t realize the complexity of the data analysis. So basically, I started the lab in the beginning of 2002 trying to apply a number of techniques from humans to chicken, primarily because [chickens] had the most advanced database resource of the non-traditional biomedical animals. I have a background in chicken anyway.
Had you worked with the chicken proteome in the past?
No, not with proteomics. Most of my animal model work during my PhD was in chicken. Up until my PhD, I worked for five or six years as a vet, with cats and dogs and rodents and various pets.
How did you get your start in proteomics?
So what happened was we started off doing 2D gels. Then in November of 2002 I was incredibly lucky. I was able to get into the inaugural proteomics workshop in Cold Spring Harbor Labs, where I met Phil Andrews and Andrew Link. They basically allowed me to make a quantum leap in training me to do this fairly well. At this point we basically have four postdocs, three PhD students and one lab RA, and none of us have backgrounds in chemistry or protein biochemistry. We all come with some form of molecular biology/genetics background, and we find that very, very useful because we are very familiar with gene structure and protein structure, and we don’t get stuck on protein identification just at the level of the mass spectrometery. We also look at peptides in the context of the entire genome. It just helps us understand things like splice variation, post translational modification, gene duplication, etc. So we’re all new to proteomics. We’re all trying to use mass spectrometers. We’ve worked with Missisippi State’s Life Sciences Biotechnology Institute — they have the machines and they run the samples. We do everything at the front end, they run the sample through the machine, they get the data, and then we do everything from the back end from the spectra on. We’re not exactly experts in mass spectrometry, but it’s a tool, and we can use it just like we can use any other tool.
We’re really just applying these techniques and technologies to try to do two things: To use proteomics to answer important questions in production animal species, to try to get some value out of the genome — we’ve looked at the genome sequences. The chicken is done now, the bovine will be done very soon. We look at the genome as one of the many tools we want to be able to use to answer these important production animal questions. Our second goal is to use these animals as biomedical models and to start to answer biomedical questions that can be answered in non-traditional biomedical model species. We don’t have a problem with mice, we just see a need for people to be able to do things with species other than mice.
Thirdly, because we’re trying to do this, we’ve really found that we need to understand better comparative aspects of genomics as it affects our ability to do high-throughput proteomics. So basically now we do some 2D gel work still, but we’re going through a phase where we’re going through a lot of these non-electrophoretic, shot-gun type high-throughput proteomics. We do use mass spectrometry and biochemistry. We do identify single proteins from bands.
The other thing we do is we care about a lot of microorganisms of the production-animal species. Again we have the same problem — a lot of them have a sequenced genome, but the genomes are in various stages of annotation. So we find that somewhat challenging as well.
What kind of questions are you trying to answer using proteomics?
Well, it depends. We have a number of different projects. We do some work with nutrition — poultry is one of the leading animals for understanding nutrition, so we’re doing some work with amino acid nutrition. That’s one end of the spectrum. At the other end of the spectrum, we’re looking at antibiotic resistance in bacterial pathogens that are both pathogens for production animals and also for human disease. We do work on a model of human lymphoma, and we’re also doing some work on B cell development in chickens. The good thing about the chicken bursar is that it’s a discrete organ, whereas the bone marrow is not. So what this allows us to do is to take this discrete organ, and we’ve got it in hand as a nice discrete organ, not a sludge. We’ve got a paper coming out in February in the Journal of Proteome Research where we’ve shown some of our first work in this model system where we’re using B-cell development in the model for looking at a structured, whole organ.
What are the biggest challenges in working with non-traditional animals?
The biggest problem we’ve found with doing proteomics in production animals comes with the database. Because we want to model biological pathways, the biological models that are best done — human and mouse — the trouble is, though there is a lot of homology between species, once you start to move down to fish and birds, you find that the pathways for something like apoptosis or cell proliferation are a little different. So it’s a little difficult sometimes to overlay all of these proteins. A problem we may have is that we haven’t got the sensitivity to identify a protein, or in this particular physiological case, is the protein not there, or is the protein not in the genome. Right now, we’re able in the chicken to simply look at the genome and see if the protein is not there. We’ve certainly found a number of proteins that you may expect to be present in chicken that are just not present in the genome. We’re assuming the chicken doesn’t have that gene. But the other thing we’ve found too is that as we get further down into the production animal species, there’s going to be lower and lower sequence coverage, just simply due to cost. What this means is we find it a little difficult sometimes to be certain that certain genes are missing.
Our work is related to the Gene Ontology people at EMBL in the UK. In terms of functional genome annotation, it’s lagging a long way in production animal species. We’re about to submit 3,000 up to as many as 10,000 new functional annotations to the Gene Ontology group. None of us are computational biologists, so thankfully, the Gene Ontology people are helping us out a lot in dealing with this. So functional annotation is a huge issue for us.
How do you deal with working with species that have poor genome coverage and annotation?
If we’re working with a species that has poor genome coverage, then what we’ll do is we’ll work from EST sequences. When we work with bacteria, some bacteria have sequence done but no annotation, not even genome labeling. So what we’ll do is search open reading frames, and then the putative peptide ID may or may not be the real gene. So what we need to do then is spend a lot of time and effort BLASTing against closely related species that actually have good genome annotation.
Our work is driven by real questions and real biology. We want to try to take proteomics to the next step from model systems into agricultural animals and non-traditional biomedical research organisms.
What kind of things are you working on for the future?
What we’ve found is that we’ve got two aspects to this work. Certainly we’ll continue with the agriculturally important livestock species and their pathogens. We’ll be working on those with questions that are directly related to production and health and welfare issues for these species. In parallel, we’ll be working with these species, where they are, where they provide biomedically relevant models. That’s our more applied work.
Our more fundamental work is we will be continuing to try to more clearly define and annotate the new genomes as they come along.