Laboratory automation and robotics range from simple liquid handling to complete systems.

Geoff Smith

Roboteers and automation specialists seek to augment or replace the boring, repetitive, difficult, or dangerous tasks that make up much laboratory practice in protein isolation, analysis, and production. Ideally, this shift to automation will free the human mind and hands for more cost-effective and creative labors while saving time, material costs, salaries, and contract research expenses.

To effectively pyramid sample-handling devices with automated assays, analytical instruments, and the like in a fluid whole, a complex juggling act with equipment and supplies is required. Not only must equipment be physically integrated, but material supplies must be designed to optimize pass-through, maintain sample integrity, and dispense reagents in a coordinated time and space as complex as any Broadway musical number choreographed by a Ziegfeld or a Tommy Tune.

The stakes are high. According to an assessment by Front Line Strategic Management Consulting, Inc. (Foster City, CA), “The proteomics market is expected to grow rapidly throughout the next five years, reaching a value of nearly $3 billion by 2005.” To fully capitalize on this growth, new automation tools will inevitably be required. The model of the Human Genome Project points in the obvious direction—automation was perhaps the most critical factor in its completion well ahead of schedule. Many companies are banking much of their future success on the increasing need for automation and robotics in the brave new world of proteomics.

Juggling choices
On the basis of a complex calculation of cost, space, efficiency, and availability, most laboratories must compromise on the extent to which they use robotic or automated systems for any particular protein project. No one company has all the answers for every laboratory application, and the number of competitors is vast. An article such as this can only scratch the surface of the products and developing technologies that are rapidly becoming available.

Currently, researchers tend to automate individual, repetitive laboratory tasks such as liquid dispensing, plate handling, and sample prep and analysis, using human labor before and after each stage. For proteomics research, especially compared with genomics, complete systems dedicated to protein work that automate a major series of operations are only now becoming routinely available.

The evolution of automation
In their 1999 review of proteomics and automation, M. Quadroni and P. James point out that “proteins, in contrast to nucleic acids, are vastly more diverse, and a universal handling method is unlikely to be found” (1). Still, techniques that can be adapted from genomics—especially for liquid dispensing and plate handling—have been borrowed wholesale, instantly transforming gene-manipulation companies into protein specialists.

Such modular developments, although of tremendous utility to speeding up individual steps in longer processes, can have limited benefits compared with integrated systems, often merely creating bottlenecks when their success demands more input at the beginning than the individual laboratory can efficiently produce, or more output at the end than it can analyze or use in a timely fashion.

The ultimate goal, of course, is to combine automation modules in a mix-and-match fashion to develop systems uniquely tailored to individual laboratory requirements in order to automate whole processes—a much more difficult task. According to M. F. Lopez, “The operating paradigm in proteomic analysis today is a combination of 2-D gel electrophoresis (for protein resolution) with mass spectrometry (for protein identification). All the intermediary steps in the procedure, including gel-staining, image analysis, protein spot excision, digestion, and mass spectrometry can be automated to increase efficiency and save time” (2).

But a variety of other steps can also be considered for automation, especially with the introduction of chromatographic techniques to enhance or replace 2-D electrophoresis (see “Pathways to the proteome”, Modern Drug Discovery, October 2001, pp 32–39). These steps include sample isolation and preparation techniques as well as various liquid purification methods —especially with regard to high-pressure liquid chromatography (HPLC) and capillary zone electrophoresis (CZE). Automation is also being applied to the ancillary but critical areas of protein cloning and protein/peptide synthesis.

As with all automation techniques, individual companies develop their own solutions based on widely varying criteria. As discussed in “On the right track” (Modern Drug Discovery, November 2001, pp 47–49), even deciding whether to use a robotic pick-and-place arm or a conveyor track system for plate handling can be complicated.

Sample prep
The first words in the bible of many a chemical roboteer are, “In the beginning was the sample.” The purity and concentration of the initial sample are critical to successful automation as each step depends on its reproducible behavior in a variety of chemical systems, from gels to spot-excision to cleanup for mass spectrometry. Too little of the desired protein may mean the inability to find it using a particular staining protocol. The wrong pH or too many impurities may interfere with the protein’s localization on a 2-D gel or with subsequent enzymatic digestion. Many companies provide automated solid-phase extraction cleanup systems that prepare samples for various analytical applications.

Also critical is post-gel sample prep, whether it is used to remove stains or gel contaminants such as salts and detergents. Post-gel cleanup has also been automated, and several companies provide commercially available robotic workstations that are capable of desalting, concentrating the peptides, and depositing samples onto targets for analysis by MALDI-TOF MS (matrix-assisted laser desorption and ionization, time-of-flight mass spectometry).

Getting things to gel
Because 2-D gels are still the standard means for beginning protein analysis—and because they are time-consuming and subject to variability—they have received significant attention in the quest for automation. Automated systems for most of the steps, including loading, running, and staining the gels, are available from various companies.

Staining the gel appropriately is critical. Coomasie Blue has historically been the most popular stain, but the use of silver stain or the fluorescent ruthenium-based Sypro Ruby stain has become common as the need for greater sensitivity has developed. Silver stain has some compatibility problems with subsequent in-gel digestion required for most mass spectrometry analysis, which Sypro Ruby does not suffer from. Several companies have designed automated gel-staining systems that handle 1–10 gels simultaneously for various staining protocols.

One of the key problems in using 2-D gels is image analysis. Obviously, if spot excision is to be properly automated, finding the spot to be excised in a reproducible fashion is critical. This process depends on automated image-analysis programs that can scan the gel and locate individual silver-stained or fluorescent spots. In a recent Sites and Software column (Modern Drug Discovery, October 2001, pp 63–66), a discussion of various image analysis software pointed out the difficulties inherent in trying to automate systems when each gel is subject to wide variations in reproducibility even independent of the idiosyncrasies arising from the initial sample purification and preparation steps.

But once spot identification is possible, robot spot-excisors can be used. Automation at this step not only speeds up the process but also minimizes the contamination of spots by slough-off from technicians’ skin and hair as well as laboratory dust particles.

To prepare the proteins for mass spectrometry, spots are usually digested with proteolytic enzymes such as trypsin to produce usable peptides. Automated in-gel digestion instruments designed for use with mass spectrometry are commercially available from companies such as Micromass, which has a MALDI family of 2-D gel analyzers.

Sequencing
Automated instrument systems have been designed to deal with the reality that most protein users are still committed to gel-based research. But direct HPLC-MS/MS systems have been developed that eliminate the need for 2-D gels. These systems often use CZE as their “front end” and rely heavily on automation for routine use.

In another nuance to improving HPLC-based analysis, automated peak-parking systems (in which the HPLC run is “parked” so that multiple mass-spectrometric analyses can be performed on the same sample) have been devised to allow feedback from the mass spectrometer to determine subsequent partitioning and analysis of the peaks.

Cloning and cell systems
One of the greatest benefits of automation in the area of nucleic acids is the production of vast quantities of individualized DNA probes on demand for use in research and high-throughput screening. Because there is no polymerase chain reaction for proteins, one of the most important needs in protein research is to speed up the use of appropriate vectors that can clone the functional protein efficiently, with any necessary posttranslational processing correctly performed.

Perhaps one of the most ambitious areas of automation in this area of proteomics is the attempt of a Harvard University FlexGene consortium to create a robot-accessible library of all the human proteins derivable from genomic sequences, and to “file” them, as it were, in a series of clonal vectors, each with its benefits and disadvantages (www.hip.harvard.edu).

Proteomics also requires cell-based systems other than those used for cloning. In an example of the use of magnetic-bead technology detailed at the CRS Robotics Web site (www.crsrobotics.com/crs/home.shtml), the Samuel Lunenfeld Research Institute at Mount Sinai Hospital in Toronto uses an automated system for high-throughput protein expression and cell-based analysis of protein–protein interactions. The platform includes a cell-culture system for protein expression, a magnetic-bead-based purification system, and an optical detection system.

Such systems can provide the kind of analytical power required to research protein function and not simply structure.

Synthesis
With the explosion of sequence knowledge derived from genomics, the demand for rapid synthesis of specific peptides has increased tremendously. Fast synthetic techniques can be used to develop new peptides for screening in drug development. At the University of Illinois at Urbana–Champaign’s Protein Science Facility, solid-phase automated peptide synthesis is offered using the Rainin Symphony system (Protein Technologies, Inc). Peptides are produced according to researcher design using several custom amino acids, and they can be obtained with customized terminal ends; the peptides are purified after synthesis by using a Beckman HPLC system. Many other companies produce automated synthesizer systems.

Applied automation
The ultimate goal is to apply the information obtained by automating proteomics research to drug discovery, such as by developing high-throughput screening using newly unearthed protein targets. But whatever the automated technology, the quality of the information obtained is only as good as the software that runs and monitors processes, keeping track of performance standards and defined criteria for success.

The programs that integrate and maintain the automated instruments, that take the data and actually use it, are thus the real heart of automation. To be fully useful, automated systems for protein analysis, like their DNA counterparts, must produce results compatible with the growing databases that maintain the vast assemblage of protein sequence and structural information that is the core of proteomics today. Whether the proteins derived from automation techniques are used as affinity probes in microarrays or in Microtiter multiwell plate assays, they will be useful only if they can be identified and characterized with regard to the wealth of information that is already available. (For a review of bioinformatics software, visit www.the-scientist.com/yr2001/oct/profile1_011015.html). Proteomics must take the genomics pathway to its end—and that means automation. From cell to sequence, from separation to synthesis.

Ultimately, from hardware bits to software bytes, whether any particular research protocol in proteomics can be fully or partially automated today depends on juggling a host of factors, from equipment to reagents to software (not to mention budget and workforce). But as costs fall and efficiency in components and systems improves, and the imagination of roboteers takes off, proteomics will likely see the same breakthroughs that automation made possible for genomics. The potential payoff to medical research and the development of new drugs seems, one might say, “automatic”.

References

1. Quadroni, M.; James, P. Proteomics and automation. Electrophoresis 1999, 20, 664–677.
2. Lopez, M. F. Better approaches to finding the needle in a haystack: Optimizing proteome analysis through automation. Electrophoresis 2000, 21, 1082–1093.

Mark S. Lesney is a senior editor of Modern Drug Discovery. Send your comments or questions about this article to mdd@acs.org or the Editorial Office by fax at 202-776-8166 or by post at 1155 16th Street, NW; Washington, DC 20036..