Proteome chips: a future perspective

In the last few years, protein chip technology has followed a trajectory of growth similar to that of DNA microarray technology, emerging as a tool for functional studies of proteins. Since the first description of the microarray approximately 15 years ago [1], the advantages of such assays have caused DNA arrays to be omnipresent in research. Protein chips retain most of these advantages, including the planar surface amenable to many spectroscopic detection methods, the use of small amounts of reagents and probes, and the high-throughput screening ability [2–4]. Proteome chips, which have as large a coverage of the proteins within an organism’s genome as possible, are a form of protein chips that allow for studies at the omic level [5].

Two main advantages enable proteome chips to stand out and will drive their usage as primary screens. First, the coverage of proteins allows for finding interactions that would be unpredictable – novel interactions that would be missed with protein chips that have defined subsets of proteins in the array. Second, the near-complete coverage of proteins allows for the mapping of all interactions within a proteome, such that interactions among individual proteins can be classified into networks within a wider system-level view.

What is the cost of these benefits, which has currently limited their availability to researchers? To make proteome chips, the gene for each protein must first be cloned into an expression vector, then the protein must then be expressed in cell cultures (preferably from the native organism), purified and arrayed onto a chip. The initial production of chips is a start-up cost that is correlated with the number of proteins printed, so there is quite a large barrier to making proteome chips that has to be overcome.

However, each of the steps making up this initial barrier is becoming less of a hurdle. As automation and high-throughput technologies become increasingly commonplace, the costs should continue to fall, making entry into proteome chip assays easier. Large collections of cloned genes and better genome annotations are being created, allowing for more complete arrays of complex organisms to be made, and building community resources with wider access by scientists. Improvement in protein purification methods are leading to arrays with proteins of higher quality in greater quantity. Advances in robotics and automation should lead to higher-density arrays, and new surface chemistries will facilitate an even broader spectrum of assays. Other technologies are being developed to circumvent the protein chip start-up cost, for example, by using in vitro protein translation directly on the chip [6], but these methods have not yet been demonstrably scaleable to proteomic scales. The growing sophistication of proteome arrays, as the technology for manufacturing them improves, should quickly increase their usage as screening tools.

Improvements in the arrays

The first hurdle to proteome chip production is the availability of open reading frame (ORF) collections from the organisms of interest. The first proteome chips were constructed with yeast proteins [7], which required the cloning of each ORF from the genome of Saccharomyces cerevisiae – approximately 6000 ORFs. Currently, the cloning requirement has become alleviated for many model organisms, which now have most of their ORFs available as collections. For example, 12,000 of approximately 17,000 Caenorhabditis elegans[8] and 10,000 of approximately 13,000 Drosophila melanogaster ORFs have been cloned [101]. Mammalian systems are close behind, with 13,000 human genes and over 12,000 mouse genes cloned [102]. It has become feasible to make proteome chips for a wider variety of organisms that have larger numbers of genes.

In addition, the production of proteome chips has been commercialized by Invitrogen (CA, USA) [103], allowing any individual researcher to bypass the need for any of the equipment, reagents or time to produce proteome chips, as long as the cost can be afforded. Invitrogen currently sells a yeast proteome chip with over 4000 of the approximately 6000 yeast proteins. Their production of a human proteome chip has included increasing numbers of proteins, with the current generation topping 8000 of the estimated 25,000–30,000 human genes.

One of the current rate-limiting steps to proteome chip production is the production of the content – the proteins – to be printed onto the arrays. While small cultures of a few milliliters can be grown in high-throughput 96-well plate fashion, larger cultures, which produce higher quantities of proteins, improve the signal from the arrays greatly. The host expression organism is also an important consideration. Bacterial expression gives the highest yields at the highest speeds with the simplest workflow, but is lacking in requirements for proteins from more complex organisms, including proper splicing and post-translational modifications, such as glycosylation. For example, Invitrogen has balanced these factors by producing the protein for their human proteome arrays in insect cells as N-terminal fusion proteins, which may not be processed correctly for post-translational modifications but are much more likely to fold than if expressed in bacteria. The development of high-throughput protein purification workflows in more complex organisms should greatly increase the production of proteome chips [9].

One of the most critical factors in proteome chip production is the slide surface chemistry. Since each surface type provides a different constraint on the preparation of the proteins to be arrayed, the density of features on the array, and the types of imaging and detection methods allowable, test assays should first be performed to finalize the slide surface to be used. However, this is one area that is not a real limitation for the use of proteome chips, except in the possible need to prepare different protein batches for various surface chemistries. Otherwise, the current diversity of slide chemstries that can be used for coupling proteins to the surface covalently or noncovalently allows for optimization of the surface for a wide variety of assays [10].

Improvements in the assays

Many assays have been performed already with proteome chips, demonstrating their potential in diverse applications. The arrays have been probed to study protein interactions with many other types of molecules, such as other proteins, nucleic acids, small molecules and lipids. The chips have also been used to characterize the proteins in the arrays in terms of their post-translational modifications and also as substrates for enzymatic reactions.

Now that diverse probes have been used in various proteome chip screens, the sophistication of such screens is increasing. The follow-up validation experiments and secondary assays of the previous screens allow for insight into the design and optimization of the next generation of assays. Background interference from nonspecific interactions can be reduced with blocking proteins and use of different slide surface chemistry. In cases where it is difficult to achieve good signal:noise ratio, differential binding assays and other pseudo-‘two color’ experiments can be used. For example, with proteins where binding depends on a ligand, chips can be probed both with and without the ligand. The ligand-dependent binding interactions of the probe can then be determined by comparing the two sets of assays, thus reducing the background from nonspecific interactions.

In addition, the paradigm of using protein chip screens to call ‘hits’ has been challenged with the efforts to make quantitative measurements on the chips. Leveraging the spectroscopic friendliness of an (optically transparent) planar surface, it is possible to study the binding of probes to microarrays in realtime (e.g., with total internal reflection fluorescence microscopy). Such advancements make it possible to quantify with binding constants the interactions between a probe and proteins on the array [11,12], removing the artificial need for cut-offs in hit calling.

Conclusion

Future assays with proteome chips can be targeted to examine organisms on a broad scale. The mapping of phosphorylation targets of each kinase in S. cerevisiae shows that the place of interactions in networks can be analyzed [13]. With the ability to use each protein of a pathway as a probe, the complete mapping of signaling, regulatory or enzymatic pathways is possible. In addition, this information can be incorporated and integrated with other omic datasets to build a broad view of the compsition of a cell at the systems level. The identification of glycoproteins demonstrates that large classes of proteins can be identified to characterize and sort proteins of unknown function [14]. The large coverage of proteins from an organism means that antibody profiling or identification of biomarkers for diseases can be conducted easily and rapidly [15].

DNA microarray experiments have become a ubiquitous tool in the research scientist’s toolbox now that they have become commercialized as kits or services and matured as somewhat standardized experiments. They are routinely used as complements to other experiments to measure transcriptional activity of genes of interest or for genotyping in the study of diseases. With the commercialization of proteome chips and the increasing ease of their manufacture, proteome chip assays are becoming more common. It seems probable that proteome arrays will soon be used in a widespread manner to find the interactions of proteins of interest or in the search for biomarkers of disease.