Proteins In Parallel

AN ARRAY OF CHOICES

The term protein array includes arrays with two different categories of immobilized proteins and two correspondingly different functions. Protein-expression arrays (also known as capture arrays) are the more familiar type. The direct protein equivalents of DNA arrays, these chips use high-affinity binding reagents with known specificity (usually antibodies) to probe protein levels in a given sample.

The second type is the protein-function array, also known as a protein-interaction array. Composed of grids of proteins or protein domains, these arrays are used to detect novel relationships between proteins and other molecules. A common application is a protein-protein interaction assay, which measures binding between the probe protein and the bound proteins, though function arrays can also be used to assay interactions – for instance, between bound proteins and drugs, enzymes, and DNA sequences.

Henry Hepburne-Scott, a product manager at Maidenhead, UK-based Procognia, prefers an alternative classification of array types. He suggests that DNA and antibody arrays be grouped together as profiling arrays, with protein-function arrays in a category all their own. "The key difference," he explains, "is that with expression arrays a complex probe is used, for example a cell lysate or cDNA population. The aim is to determine the relative levels of the components in the mix. With a protein-interaction array the probe is a single entity of interest – for example, another protein or DNA molecule. The aim is to determine how proteins on the array interact with the particular probe in question."

EXPRESSION/CAPTURE ARRAYS

The success of expression arrays relies on the specificity and affinity of the capture reagents they use. These reagents are usually antibodies, and the term antibody array is often synonymous with this class of chip. In all the protein-expression arrays discussed here, the capture molecules are antibodies. But they don't have to be: specific binding of proteins can also be achieved using high-affinity nonprotein reagents such as aptamers.

While expression arrays can monitor protein expression for hundreds of proteins on a single chip, there is no single proteome equivalent of a whole-genome gene expression chip. Because antibodies require development, purification, and validation, they are less amenable to massively parallel production than nucleic acid probes. Plus, distinguishing protein variants with different posttranslational modifications would require multiple antibodies per gene.

The typical approach to this limitation has been to develop arrays with biologically coherent themes, such as cell signaling, cytokines, apoptosis, or cancer. One exception is the Ab Microarray 500 from BD Biosciences-Clontech of Palo Alto, Calif. It contains 500 antibodies chosen against a wide range of proteins in multiple functional categories and multiple cellular compartments, according to applications product manager Sejal Desai.

Basic and clinical researchers are the typical users of protein expression arrays, but the eventual goal of users is biomarker detection in the clinic. Zimmer man believes that though researchers will remain the dominant users in the near future, eventually "clinical arrays will become very important as diagnostic tool."

Worcester, Mass.-based Hypromatrix manufactures an interesting expression array variant called the Staining Antibody Array, which is designed test for both protein presence and cellular localiza tion. James Wang, the company's chief technology officer, says the technology is based on the ability dissociate the arrayed antibodies from their supports. An antibody array is incubated with cells that, like the antibodies, have been positionally fixed to a support. The antibodies are then separated from their support, after which secondary antibodies are used to detect the antibodies bound to their target proteins in situ. Thus, researchers can extract data on both cellular localization and protein expression in one step.

FUNCTION/INTERACTION ARRAYS

In protein-function arrays, the immobilized element is typically a protein or protein fragment. But two companies, Pepscan, based in the Netherlands, and JPT Peptide Technologies of Berlin, synthesize and immobilize peptides instead of whole proteins or large protein fragments. The two approaches provide very different windows into protein function.

"The strategy [when immobilizing peptides] is to reduce the complexity of one of the binding partners," explains Peter-Paul C. Henze, director of sales and marketing for JPT Peptide Technologies. The company's PepStar peptide microarrays contain sequences that are known substrates for specific enzymes (for example kinases, phosphatases, or proteases) and are designed to elucidate enzyme specificities.

Since biological interactions often depend on folding, several companies ensure that arrayed proteins are also functional. One such array is the Active Protein Array from ProteinOne in College Park, Md., according to Lawrence Charity, director of marketing and business development. Procognia likewise makes sure proteins are properly folded on its Panorama Human Protein Function Microarray. The proteins are tagged with a peptide that is biotinylated only when folded, Hepburne-Scott explains. The biotinylated tag is then bound to a streptavidin-coated glass slide. And according to Eric Gay, product manager at Carlsbad, Calif.-based Invitrogen, data on a subset of the 1,857 proteins that make up its higher-density ProtoArray Human Protein Microarray indicate that up to 90% retain functional activity.

The current density of protein arrays indicates that a biologically active human proteome array is probably not around the corner. But such an array has been developed for yeast. Invitrogen's ProtoArray Yeast Proteome Microarray incorporates 4,088 Saccharomyces cerevisiae proteins, representing about two-thirds of the organism's 6,000 or so genes.

ARRAY FORMATS

Protein arrays come in a number of physical formats, from nitrocellulose membranes, sometimes mounted on glass slides, to small arrays formed in the bottom of 96-well plates. The general trend is towards smaller arrays with lower sample volume requirements. Henze explains that, especially in human experiments, "many samples are scarce, so researchers need to make as many measurements as possible with the samples they have."

Several detection and labeling strategies exist, the most common of which is directly labeling the sample with a fluorescent dye. An alternative scheme is akin to an ELISA, with the dye attached to antibodies that recognize the target proteins. This strategy allows the end-user to skip labeling the sample and helps to positively identify the proteins bound to the immobilized reagents. But it does require the development of a labeled cocktail of antibodies that can recognize the target proteins.

HOMEGROWN AND CUSTOM ARRAYS

While the variety and depth of commercially available arrays is improving, many researchers still prefer the flexibility of homemade arrays. "There are an increasing number of commercially available systems, but there are still lots of homegrown systems as well," says Zimmerman. Which works best for you first depends on whether the elements required are represented on a commercial array.

If they are, then the question is one of cost. But if premade array content doesn't provide the appropriate gene sets, alternatives are available that don't require purchasing expensive arraying robots. Many of the manufacturers listed, as well as a number of array-supply manufacturers, provide custom arraying services using their own catalog of reagents or reagents provided by the researcher.

Many companies also offer array-processing services, taking care of all the steps between a customer-supplied sample and an output data file. These end-to-end services can eliminate the high set-up expense associated with purchasing array-based instrumentation (such as scanners and assay workstations) and reduce the risk of costly operator-related failure. This is a major advantage for researchers inexperienced with processing protein chips, according to Peter Wagner, chief executive officer at Zyomyx. The Hayward, Calif.-based company previously supplied entire protein biochip systems, but has recently changed its business model toward analytical services. As Wagner explains, that move "ensures that a larger customer base has access to the platform with better data-output quality and flexibility for custom arrays."

But whether a researcher opts for a custom or commercial protein array, and whether that array is used to monitor protein expression or interactions, the change from the way protein biology once was done is fundamentally the same: massive parallelism replacing a one-at-a-time mentality. Says Martinsky, "There are lots of ways to detect a protein, but at the end of the day it's just grind-and-bind biology."