Courtesy of Bio-Rad Laboratories
Thierry Rabilloud has been doing proteomics since long before the word proteome was even coined. For years Rabilloud, currently at the Atomic Energy Commission Research Center in Grenoble, France, used two-dimensional gel electrophoresis (2-DE) to break complex protein mixtures down into their component parts, purified individual proteins that interested him, and then sent these out for identification by mass spectrometry (MS). But at this point his research hit a bottleneck: the MS facility could accept only a few samples at a time, because the sample processing procedure was so labor-intensive. Then the MS lab acquired robots to automate the processing steps, and the bottleneck disappeared. "Before we were sending in maybe 10 spots. It was a lot of work," he recalls. "Now we can send 50 spots. That's a piece of cake."
Rabilloud's experience is not unique. Traditionally, the 2-DE-based workflow is labor-intensive, technically challenging, difficult to reproduce, and low-throughput. Instrument manufacturers have responded with immobilized pH gradient strips, precast gels, and highly parallel electrophoretic chambers; automated gel staining robots; spot pickers; in-gel digestion robots; and MALDI (matrix assisted laser desorption/ionization) target-spotting robots. As such tools have lowered the technological barrier to entering the field, more and more researchers have jumped in. Alex Kurosky, director of the University of Texas Medical Branch biomolecular resource facility, says, "Everybody and his uncle is coming out of the woodwork" to do proteomics research at his core facility, which contains a suite of such equipment. "We're getting inundated."
Courtesy of Genetix
But automation isn't for everyone. For starters, it's expensive; automating a proteomics lab can cost more than $500,000 (US). Further, Mary Ann Gawinowicz, facility manager for the Howard Hughes Medical Institute protein core facility at Columbia University, says automation makes sense only for labs that process large numbers of relatively homogeneous samples.
Some researchers say such considerations simply mask the real issue. No amount of automation, they say, can change the fact that the 2-DE approach is fundamentally flawed. These scientists say the future of proteomics lies in a different technology altogether.
RAPID RESULTS WITH ROBOTICS Vito DelVecchio, director of research, Institute of Molecular Biology and Medicine, University of Scranton, Pa., runs a comparative proteomics lab that studies biological warfare agents. Recently, his lab published a proteomic comparison of two Brucella melitensis strains.1 Such a task done manually would be Herculean and mind-numbingly tedious. Thanks to a battery of robots, however, this work--from 2-DE gel to MS--was almost entirely automated; the only manual intervention was moving items from instrument to instrument. "We're probably as automated as can be," he says.
DelVecchio's lab is stocked with a gel-staining robot, a spot-picking robot, an in-gel digestion robot, and a MALDI target-spotting robot, all from Genomic Solutions of Ann Arbor, Mich. A Fujifilm LAS-1000plus imaging system and an Axima CFR MALDI from Shimadzu Biotech round out the equipment, which DelVecchio's group puts to good use: the lab runs 10 2-D gels per week, picking hundreds of spots per gel, he says.
Frank Witzmann, professor, cell and integrative physiology, Indiana University School of Medicine, has a similarly well-appointed lab, containing a suite of instruments, including imagers, a spot cutter, sample prep workstation, and MS, from ProteomeWorks (a collaboration between Bio-Rad Laboratories of Hercules, Calif., and UK-based Micromass).
Witzmann's group applies 2-DE-based proteomics to toxicology, an approach he calls 'toxicoproteomics.' In a typical application, Witzmann's team runs 20 8 (infinity) 10-inch gels in parallel: five replicates each of a control and three toxicological endpoints. This, he says, provides statistical confidence that his group will neither miss interesting proteins nor chase red herrings.
A physiologist who's been running 2-D gels for 20 years, Witzmann calls himself "a 2-D gel guy." For most of those years, he did proteomics the old-fashioned way: manually. "It was much more difficult, much more time-consuming," he says. Adds Kurosky, "You're not doing proteomics unless you're into high, high-throughput and you're using robotics."
But setting up an automated proteomics lab is not a trivial exercise, says Kurosky. It took him three to six months to iron out all the creases in his lab, particularly problems with software integration. DelVecchio notes, however, that other than some initial problems with the automatic staining instrument, his experience has been positive. "We were getting meaningful preliminary data in a matter of [three or four] months ... I was really surprised and quite elated."
BETTER THAN MANUAL? Robotic instruments offer some obvious advantages. One is improved sample purity. Keratin, a protein found in skin, hair, and dust, represents a major source of contamination in proteomics labs. As a result, many instruments are available, or come standard, in a protective enclosure or hood to minimize this problem. Other benefits include a reduced error rate, increased speed, and a lower risk of repetitive stress injury. "The amount of pipetting is phenomenal ... You'd have to get a thumb transplant," observes Witzmann.
Yet some researchers still process 2-DE samples by hand. Witzmann says that his postdoc--despite all the automation available to him--prefers manual sample preparation. And Columbia's Gawinowicz, whose facility has no robotic equipment, says automation is not really an option for her lab, because the samples she works with are so heterogeneous in terms of protein concentration, gel volume, and staining.
These variables affect the way samples should be processed. John Yates, a proteomicist at the Scripps Research Institute in La Jolla, Calif., explains that where a machine will treat every sample exactly the same, a technician doing manual digestion will account for spot intensity by adjusting variables like amount of enzyme and resuspension volume to improve results.
FUNDAMENTAL FLAWS The debate over manual versus automated processing may be insignificant in the bigger picture. The real problem, says Ruedi Aebersold, is that 2-DE simply doesn't have the technical power to drive full-scale proteomics in an economic manner. Aebersold, professor and cofounder, Institute for Systems Biology in Seattle, says these problems have nothing to do with automation; they are inherent to the 2-DE process itself.
He notes, for example, that low-abundance proteins tend to be systematically excluded in 2-DE analyses, and that seemingly discrete 2-D gel spots often contain multiple proteins, making it difficult to determine which protein's concentration is changing when comparing gels. In addition, membrane proteins are notoriously underrepresented in 2-D gels, and then there's the so-called dynamic-range problem--"The proteome in a given cell can be as much as a million-fold wide with respect to abundance," says Jim LaDine Thermo Electron's director of New Technology Development, Life and Laboratory Science Sector.
Researchers have, over the years, developed workarounds for each of these problems. Last year, for instance, Aebersold described the application of isotope-coded affinity tag (ICAT) technology to 2-DE to help solve the problem of protein comigration.2 New protein extraction methods and detergents are improving the detection of membrane proteins, and sample prefractionation is helping researchers deal with the dynamic-range problem and loss of low-abundance proteins. With each of these methods, Aebersold says, researchers "can see deeper into the proteome." But the amount of work and the complexity of the experiment increases substantially, too.
WHAT THE FUTURE HOLDS Despite these fixes, Aebersold says the future of proteomics lies in liquid chromatography-based approaches, for reasons of both cost and ease of automation. This point-of-view is shared by, among others, Thermo Electron's LaDine.
But many other proteomics manufacturers, while acknowledging a shift toward such strategies, say 2-DE isn't going anywhere soon. Typical among responses is that of Sandra Rasmussen, Proteomics Business Leader, PerkinElmer Life and Analytical Sciences: "Our feeling is that 2-D gels will probably be in the picture at least for the next five years and probably longer, because it is accepted as the gold standard."
This is especially true for small academic labs, says Columbia's Gawinowicz. "There are different levels of labs doing these things. There are people like me, who are in this little lab, doing everyday-type things with a limited budget, limited labor, trying to get an answer for an investigator. Then there are people like Ruedi Aebersold, who's on the cutting edge, and trying to develop methods, who has infinite resources." It might take years, Gawinowicz says, for new technologies to trickle down to small core facilities. In the meantime, such labs rely on what works: 2-DE and MALDI MS.
Jeffrey M. Perkel can be contacted at firstname.lastname@example.org.
1. M. Eschenbrenner et al., "Comparative proteome analysis of Brucella melitensis vaccine strain Rev 1 and a virulent strain, 16M," J Bacteriol, 184:4962-70, September 2002.
2. M. Smolka et al., "Quantitative protein profiling using two-dimensional gel electrophoresis, isotope-coded affinity tag labeling, and mass spectrometry," Mol Cell Proteomics, 1:19-29, 2002.