Improving the Biomarker Pipeline

Proteomics has evolved over the past 10 years from a method providing mainly qualitative information—protein identification and modification site determination—to a more quantitative technique. Until quite recently, quantitative proteomics was focused on relative quantification for biomarker discovery. Biomarker discovery, by definition, must take a “shotgun” approach, which usually involves a 2-dimensional separation (1). Unfortunately, this approach also means that the techniques used for biomarker discovery are both time-consuming and costly; consequently, few potential biomarkers have progressed beyond the discovery phase. Liquid chromatography–tandem mass spectrometry quantitative proteomics techniques designed for biomarker discovery, such as isobaric tags for relative and absolute quantification (iTRAQ)22), can take 1–2 weeks for a comparison of 8 samples. One bottleneck is the low throughput of “conventional” proteomics techniques, which usually use low-flow-rate nanoscale liquid chromatography and long analysis times for the highest possible detection of biomarkers and exhaustive coverage of the proteome. Today, there is a great deal of interest in the next phase in the biomarker pipeline, biomarker validation and biomarker verification, which must be done before clinical evaluation can be performed (3). The success of mass spectrometry–based comparative proteomics studies with relative quantitative methods has led to the discovery of thousands of potential biomarkers that now need to be verified and validated. Instead of the in-depth analysis of a few samples, however, the verification/validation phases involve the absolute quantification and targeted analysis of a large number of samples. Different methods can be used for quantification because the protein targets are already known. Today, ELISA assays are used for biomarker verification and validation of targeted proteins. Times are changing, however. Recent developments in mass spectrometry–based proteomics have led to the development of higher-throughput targeted analyses that still retain the superior specificity obtained with gas-phase sequence determination. These analyses focus not on intact proteins, but on their characteristic (or “proteotypic”) peptides (4). Several new developments have coincided to make these changes possible. The first was the demonstration in several large projects supported by the Human Proteome Organisation (HUPO) and funded by the National Cancer Institute that multiple reaction monitoring (MRM) mass spectrometry assays that use stable isotope– labeled peptides as internal standards could provide reliable absolute quantitative data in interlaboratory studies (5). The reproducibility, however, was predicated on the reproducibility of all of the steps in the analysis, from digestion to data analysis. This demonstration of the reproducibility of the MRM approach has led to the founding of libraries of MRM transitions (i.e., precursor/product ion pairs characteristic of a particular target protein), which are publicly available to researchers trying to quantify these proteins (6–8). MRM mass spectrometry can be performed with either MALDI or electrospray ionization, both of which can be coupled to HPLC. MALDI certainly has the potential to be a high-throughput platform capable of analyzing >5000 samples/day per instrument, although MALDI requires that all of the separation be done off-line, either by robotic affinity purification or by off-line liquid chromatography–MALDI. In electrospray ionization, as in liquid chromatography– electrospray ionization–MRM mass spectrometry (LC-ESI-MRM-MS), the final separation is done on-line (9). Both techniques can provide “multiplexed” separations. In MALDI, especially immunoMALDI (iMALDI) (Mason DR, Reid JD, Holmes DT, Borchers CH. Multiplexed iMALDI for the detection of angiotensin I and angiotensin II. Submitted for publication), the affinity-purification step produces spectra in which the targeted peptides predominate. In LC-ESI-MRM-MS, the specific target peptides can be programmed to change as a function of retention time (scheduled MRM). More targeted peptides per analysis mean fewer analyses and higher throughput. As the number of proteins for which stable isotope–labeled peptides are commercially available continues to increase and as multiplexed MRM methods for more and more of these proteins are published and become publicly available, multiplexed MRM methods for several hundred proteins per analysis may be developed (9). These methods can then be used for discovery proteomics, at least for the more abundant plasma proteins. These highly multiplexed MRM methods would then be conceptually analogous to microarrays and might actually replace shotgun proteomics for some biomarker-discovery studies. A major bottleneck in clinical proteomics is sample preparation. To provide the high throughput needed for clinical studies requires that sample preparation—a challenging process in its own right—be performed in an automated and highly reproducible manner. We have seen recent advances in robotic sample-preparation systems, specifically through automated liquid-handling systems, systems for solid-phase extraction separations, or robotic systems such as Thermo Scientific's KingFisher, which is designed for the handling of magnetic beads. The third key event is the maturation of various techniques involving affinity purification before mass spectrometric analysis [stable isotope standards and capture by antipeptide antibodies (SISCAPA) (10) and iMALDI (11)], thereby removing interferences and allowing 1-dimensional separation instead of the 2-dimensional methods used for untargeted biomarker-discovery experiments. Along with this progress has come the realization that although depletion is essential for detecting and quantifying low-abundance proteins, it may not always be needed for high-abundance plasma proteins (9). The fourth factor is the development of mass spectrometers that can handle higher liquid chromatography flow rates. This advance not only shortens analysis times but also increases the “robustness” of these newer methods. In addition, these new instruments have higher sensitivities, mainly owing to technological advances for reducing noise (for example, Agilent Technologies' ion funnel), which compensate for the loss of sensitivity that increased flow rates typically produce. In this issue of Clinical Chemistry, Agger et al. describe the development of a method for mass spectrometric quantification of 2 cardiovascular disease– related apolipoproteins (12). Several aspects of this study make it particularly interesting, far beyond the particular assay that the authors have developed. In many ways this report demonstrates the current state of the art, as mass spectrometry–based proteomics finally breaks through from a discovery method to one that can be used for biomarker verification and validation. First, the authors used MRM assays, with absolute quantification based on the use of stable isotope– labeled internal standards. To assist in the selection of the peptides to be used for the quantification, they used one of the libraries of precursor/product ion transitions (PeptideAtlas, one of several libraries designed to assist researchers in the quantitative targeted proteomics methods that use MRM). With this library as a guide, the authors selected 20 peptide candidates specific for the 2 proteins of interest, and in doing so they also considered the properties that would affect a mass spectrometry–based assay (e.g., peptide ionization efficiencies, no cysteine, no methionine). Second, the authors correctly recognized that one of the major challenges in quantitative proteomics is the reproducibility of protein digestion. Adding a standard isotopically labeled peptide to the sample can correct for losses and changes in sensitivity from this point forward in the analysis, but it cannot correct for nonreproducibility in the digestion step. Therefore, the authors optimized this step of their protocol by comparing several different digestion procedures and examining the effects of these digestion protocols on different peptides from the same protein. The peptides selected from PeptideAtlas were validated by evaluating them for reproducibility and quantification accuracy in comparison with an independent technique (in this case nephelometry). The authors also used a “standard plasma” sample to correct for day-to-day variations in digestion efficiency. Third, the authors used “full-flow” HPLC instead of nanoscale HPLC (approximately 250 nL/min) or microbore HPLC (approximately 1 μL/min). This feature is an additional strength of this study and is another aspect of this research that makes it cutting edge. The use of this higher 400-μL/min flow rate allowed the development of a rapid and robust assay that uses a 10-min gradient. The high throughput of this technique (15 samples/day) makes possible the analysis of the larger numbers of samples required for biomarker verification and validation. In conclusion, this report is one of the first fruits of this new age of proteomics, and we hope it will serve as a model for similar studies of this type. isobaric tags for relative absolute quantification Human Proteome Organisation multiple reaction monitoring liquid chromatography–electrospray ionization–MRM mass spectrometry immunoMALDI stable isotope standards and capture by antipeptide antibodies. The authors would like to thank Genome Canada, Genome BC, and BC Proteomics Network (BCPN) for supporting the UVic – Genome BC Proteomics Centre.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.