Automation Advances in Proteomics

Courtesy of the Institute for Systems Biology MOVING FORWARD: The LCQ Deca XP, an electrospray ionization/ion trap mass spectrometer from Thermo Finnigan The sheer number of new protein-focused mass spectrometry (MS) instruments introduced last year is a testament to the growing importance of the technique in proteomics research. Coupled to this trend is a growing need for automation of upfront sample preparation to feed these analytical machines. From the specialized academic lab to higher-capacity proteomics core facilities and drug-discovery centers, researchers look to robotics to increase walkaway time, reduce human error, and increase throughput. Many of the protein sample preparation processes have been automated, including gel imaging, spot cutting, in-gel digestion, and MALDI sample plate loading.1 Recently introduced automated devices such as the Advion BioSciences NanoMate™ 100 have made throughput bottlenecks like electrospray ionization (ESI) MS more efficient and less labor-intensive,2 and improvements to existing technologies, such as the new microfractionation capability offered by Agilent's 1100 series capillary HPLC system, increase the precision of MS analysis.3 And the market is expected to continue growing. By 2008, the total market for proteomics automation (including modular and fully integrated systems), is projected to grow to $755 million (US), an increase nearly 10 times that of 2002, says Sinead Igoe, research analyst, drug discovery technologies, for growth consulting firm Frost & Sullivan. Whether a particular laboratory opts for complete automation or chooses to buy the latest robotic toys depends largely on its research needs and, ultimately, its budget. Vendors of automated equipment for proteomics tend to agree that it is impossible to make generalizations about the market. "Because this is a developing market, in many cases you're not going to find that there's one typical customer," says Mary Lopez of PerkinElmer Life and Analytical Sciences, Boston. "If there's anything that's typical ... that's definitely it." Richard Cumming, vice president, proteomics product management, Amersham Biosciences, Piscataway, NJ, agrees. "It's a different model from the one we saw for genomics, with 24-hour sequencing and very high throughput. ... Customers want to use a variety of different techniques: expression, purification analysis, [and] mass spectrometry," he says. TO INTEGRATE OR NOT TO INTEGRATE Ken Miller, proteomics product marketing manager for Thermo Finnigan (San Jose, Calif.), has identified two types of customers for automated proteomics systems: those who are "performance driven" and want to purchase the best possible components, regardless of their compatibility; and those who are looking for efficiency and speed and tend to purchase complete systems from one vendor. Courtesy of Amersham Biosciences ACCELERATING AUTOMATION: The Ettan Spot Handling Workstation automatically handles every step in the process of protein manipulation, from spot picking and digestion to MS analysis. Eliminating the need for hands-on operation, this increases the reliability of results and throughput levels. Thomas Franz, head of the proteomics core facility at the European Molecular Biology Laboratory (EMBL), Heidelberg, Germany, favors the latter approach. His laboratory uses equipment supplied by Micromass in the UK and BioRad of Hercules, Calif., through the two companies' ProteomeWorks™ alliance. Franz says that during the planning stages for his facility three years ago, he and his colleagues specifically sought an integrated platform to streamline the exchange of data between system components. "You don't have to worry about the problems that can occur if the systems are not able to communicate with each other," Franz explains. Other companies offering integrated automated platforms include PerkinElmer, Tecan in Maennedorf, Switzerland, and Beckman Coulter in Fullerton, Calif. Laboratories with fewer samples to process, or conversely those with money to burn, tend to favor the performance-driven approach. David Speicher, proteomics facility director for the Wistar Institute, Philadelphia, explains that low-volume facilities have less of a need for immediate automation and can hold out for the best technologies. "In academia, we have to make good decisions. There's some difficulty in raising funds, and you want to make sure that when you buy the robot you buy the best possible one," he says. Likewise, well-funded facilities tend to seek high-quality instruments, as they often have the personnel, and the financial resources, to custom-integrate modular components from a variety of vendors. "When we set up our facility, the main issue that we had [was] ... to have the best instruments," says Christoph Borchers, faculty director of the proteomics facility at University of North Carolina, Chapel Hill. Recognizing the importance of integration, Borchers and his colleagues first considered a variety of "total solutions," but were unhappy with individual components of the systems. "So we stuck with our philosophy to first buy the best machine and then try to integrate it," he says. The "best machine," in Borchers' case, includes modular components from Leap Technologies, Carrboro, NC; Genomic Solutions, Ann Arbor, Mich.; Agilent Technologies, Palo Alto, Calif.; Applied Biosystems, Foster City, Calif.; and Bruker Daltonics, Billerica, Mass. The glue holding them together: a customized laboratory information management system (LIMS) from Applied Biosystems. Proteomics labs can either customize a LIMS themselves, or outsource the work to a company such as Springfield, NJ-based Hudson Control Group. "If you purchase things from different vendors, they don't come together in a cohesive way. Somebody has to do it," explains Jim Schools, the Group's marketing manager. "We get everything communicating, acting together in a repeatable manner." Other companies have designed software that allows their equipment to readily interface with third-party suppliers and enables their systems to be compatible with a laboratory's existing LIMS platform. "We can integrate and position a wide range of assay devices on our automated liquid-handling platform, and everything is driven by our scheduling software," says Jason Fawcett, strategic product marketing manager, Beckman Coulter. Borchers estimates the total cost of outfitting a fully automated proteomics facility (robotics, mass spectrometers, and LIMS) at $3 million (US). But, he explains, the investment ultimately pays off; a high-throughput automated facility requires fewer employees, which keeps costs low. "From the institution's standpoint, [low cost] means that investigators can submit more samples, get more results, and finally can apply for more grants, and this brings more research and research funding back to the university," he says. Courtesy of Zymark RIDING SHOTGUN One of the reasons proteomics facilities need data management systems is that they generate protein data by the terabytes--a phenomenon largely propelled by shotgun proteomics, often referred to by the creative acronym MudPIT, for multidimensional protein identification technology. In this technique, originally developed in the laboratory of John Yates at the Scripps Research Institute in La Jolla, Calif., proteins are digested and then subjected to two-dimensional HPLC, followed by tandem MS. Christine Wu, postdoctoral researcher in the Yates lab, explains that MudPIT is more robust than two-dimensional (2-D) gel-based methods for identifying membrane proteins, which can comprise up to 30% of the predicted proteins in a genome. "My own experience has been that when I run membrane-enriched fractions on 2-D gels, membrane proteins are hugely under- represented because of poor solubilization, lack of dynamic range of the stain, and dispersion of the spots due to covalent modifications," she says. And, the MudPIT process can be completely automated; The ProteomeX™ Workstation from Thermo Finnigan, for example, automates the separation and analytical steps of multidimensional LC-MS. Wu notes that in the past, one drawback to MudPIT was that it was not quantitative. In the past few years, however, researchers have increasingly tried to rectify that problem by using differential labeling procedures to distinguish multiple protein samples in a single mixture. In one approach, one set of cells is labeled in vivo using metabolically labeled growth media, while another culture is grown in unlabeled growth media. In another method, called the isotope-coded affinity tag (ICAT) approach, the protein samples are labeled in vitro with either a heavy or light form of a specific reagent. In both cases, the two samples are then mixed and analyzed, and the relative abundance of a given peptide is shown by the difference in signal intensities between light and heavy peptides. Such experiments generally employ ESI-MS, but Ruedi Aebersold, cofounder of the Institute for Systems Biology in Seattle, and inventor of the ICAT method, says combining ICAT with MALDI tandem mass spectrometry instead enables researchers to target their analyses to only those proteins that are differentially regulated in the experiment. In a typical multidimensional LC experiment, samples are sent directly to an ESI-based tandem mass spectrometer as they come off the column. The spectrometer automatically selects which peptides to sequence according to signal intensity, and thus the most abundant ions tend to be sequenced, regardless of their relevance to the experiment. In Aebersold's alternative, samples can be sent by the HPLC instrument offline to a MALDI target plate first. "In the MALDI instrument we have the option to first quantify and then to only sequence those peptides that actually change in their abundance. So we can concentrate or train the instruments to focus on those peptides and proteins that actually tell us something about the system that we're studying," Aebersold explains. This in turn increases the number of relevant samples that can be analyzed and reduces data complexity. "It can lead to an increase in efficiency and throughput of sample analysis by at least an order of magnitude," says Aebersold. "Shotgun" approaches such as MudPIT and ICAT-LC-MS have clear advantages over traditional gel-based methods as a means of cataloging expressed proteins, says Aebersold, but don't expect these new techniques to completely replace 2-D gels any time soon. Aebersold notes that ICAT reagent kits, commercially available through Applied Biosystems, are both expensive and cysteine-specific. Therefore, proteins that do not contain a cysteine will not be quantified, and posttranslational modifications are less likely to be detected. Further, because the sample is digested prior to separation and detection, modifications due to proteolysis cannot be detected by LC-MS/MS. "There is no way that any method other than 2-D gels could pick up proteolysis," says Stuart Cordwell, senior research fellow, Australian Proteome Analysis Facility. For low-to-medium-throughput labs, or labs performing targeted proteomics research, 2-D gels offer a straightforward, relatively inexpensive way to differentially analyze samples prior to MS and reduce data generation by eliminating proteins not central to the experiment. "What does it help me if I analyze with my sophisticated mass spectrometry machine albumin 10,000 times in a human serum sample?" remarks Thomas Franz. CREATING A WISH LIST While proteomics facility directors tend to agree that most of what is needed for automating the proteomics laboratory is available, some "wish list" items remain. Stuart Cordwell, for example, sees room for improvement in the introduction of low-volume samples to the mass spectrometer. "The methods that are currently available for sample concentration and desalting don't necessarily work well with the very low-volume samples we want to work with," he says. Thomas Franz would like to see a completely automated system for 2-D gel electrophoresis (such as the one currently under development by Tecan). Mark Roskey of Hopkinton, Mass.-based Zymark comments that his customers are looking for more precision pipetting for MALDI-target spotting. And, David Speicher cites improvements in sample prefractionation methods as central to comprehensive protein profiling of complex samples. "No protein profiling technology is currently capable of reproducibly quantitatively comparing more than about 1,000 to 1,500 proteins in a single experiment, including ICAT and MudPIT," he explains. While data management headaches remain for high throughput proteomics laboratories, most researchers in these laboratories are content with what's currently available. Suppliers of mass spectrometers and robotics systems alike have responded to the proteomics market with surprising speed, says Cordwell. "If people are complaining about things, they probably shouldn't be. That's my point of view." Aileen Constans can be contacted at aconstans@the-scientist.com. References 1. J.M. Perkel, "Technologies vie for dominance," The Scientist, 17[5]:20-2, March 10, 2003. 2. A. Constans, "Better mass spec results off-line," The Scientist, 17[7]:49, April 7, 2003. 3. A. Sreenivasan, "Nano-LC for proteomics," The Scientist, 17[14]:49, July 14, 2003. AUTOMATING YEAST TWO-HYBRID ASSAYS: MYRIAD'S APPROACH Automating protein assays can require a flexible approach, in which existing equipment is modified according to application-specific needs. For example, scientists at Salt Lake City-based Myriad Proteomics chose to customize Zymark's Cyclone 96-channel pipettor instead of buying an off-the-shelf product to automate the company's proprietary HyNet yeast two-hybrid assay for discovery of protein-protein interactions. Making the device more washable to reduce or even eliminate the need for disposable tips permitted Myriad to scale up production while reducing costs. "This has allowed us to ... discover more protein-protein interactions than we would have been able to if we had to use classical laboratory automation equipment," says Tracy Erkkila, director of process development. Added to the pipettor was a specially designed plate handler for performing PCR and DNA sequencing. The combination dramatically increased laboratory throughput, Erkkila says. "A typical yeast two-hybrid laboratory in an academic or small industrial setting is doing on the order of something like 100 of these experiments a day, with minimal automation. And we're able to do on the order of a couple thousand a day," he relates. Erkkila adds that Myriad has worked with Zymark on other automation-related initiatives, such as modifying a plate-handling stacker robotic system. "The stacker that they had ... was a very fixed system, and the modification that we worked with them on was to make it so the storage array or rack of shelves is a removable piece that can be removed from the automation and rolled around the lab" like a baker's cart, he says. This allows lab workers to move plates easily from station to station in the correct order without having to continually load and unload plates. Such modifications seem fairly straightforward, but Erkkila notes that they are key to enhancing productivity in laboratories that take a systems-scale approach to protein research. "In the proteomics field now, and in the life sciences in general, [there is] a definite trend towards doing larger system scale types of experiments to try on a much larger scale to understand some mechanism, biological process, or pathway," he says. "That requires much higher throughput." function sendData() { document.frm.pathName.value = location.pathname; result = false if (document.frm.score[0].checked) result = true; if (document.frm.score[1].checked) result = true; if (document.frm.score[2].checked) result = true; if (document.frm.score[3].checked) result = true; if (document.frm.score[4].checked) result = true; if (!result) alert("Please select") return result; } .myradio{background-color : #94bee5} Please indicate on a 1 - 5 scale how strongly you would recommend this article to your colleagues? Not recommended 1 2 3 4 5 Highly recommended Please register your vote

Automation Advances in Proteomics

Interested in reading more?

Become a Member of