Page 1 of 20 Articles in PresS. Physiol Genomics (December 27, 2007). doi:10.1152/physiolgenomics.00282.2007 Invited review Biomarker Discovery: Proteome Fractionation and Separation in Biological Samples Peter Matt 1,2, Zongming Fu 1, Qin Fu 1, Jennifer E. Van Eyk 1 Johns Hopkins Proteomics Center 1 and Howard Hughes Medical Institute 2, Johns Hopkins Medicine, Baltimore, USA Running head: Proteomic Biomarker Discovery Corresponding author: Peter Matt, MD 601 Mason F. Lord Building 5200 Eastern Avenue 21224 Baltimore, MD Phone: 0114107336577 Fax: 0114105508512 Email: pmatt@uhbs.ch Copyright 2007 by the American Physiological Society.
Page 2 of 20 Abstract Proteomics, analogous with genomics, is the analysis of the protein complement present in a cell, organ or organism at any given time. While the genome provides information about the theoretical status of the cellular proteins, the proteome describes the actual content, which ultimately determines the phenotype. The broad application of proteomic technologies in basic science and clinical medicine has the potential to accelerate our understanding of the molecular mechanisms underlying disease and may facilitate the discovery of new drug targets and diagnostic disease markers. Proteomics is a rapidly developing and changing scientific discipline, and the last 5 years have seen major advances in the underlying techniques as well as expansion into new applications. Core technologies for the separation of proteins and/or peptides are one- and two-dimensional gel electrophoresis, and one- and two-dimensional liquid chromatography, and these are coupled almost exclusively with mass spectrometry. Proteomic studies have shown that the most effective analysis of even simple biological samples requires subfractionation and/or enrichment before protein identification by mass spectrometry. Selection of the appropriate technology or combination of technologies to match the biological questions is essential for maximum coverage of the selected subproteome and to ensure both the full interpretation and the downstream utility of the data. In this review, we describe the current technologies for proteome fractionation and separation of biological samples, based on our lab workflow for biomarker discovery and validation. Word count: 230
Page 3 of 20 Keywords Proteomics; Biomarker; Discovery; 2-dimensional gel electrophoresis; Liquid chromatograpy
Page 4 of 20 Introduction While the genome provides information about the theoretical status of the cellular proteins, the proteome describes the actual content, which ultimately determines the phenotype. Not every gene is transcribed and translated into a single protein, due to complex and regulated processes such as mrna splicing (producing protein isoforms), RNA editing, co- and posttranslational modifications (PTMs, e.g. phosphorylation, glycosylation, and acetylation), and protein processing, which extends the number of protein species that are possible within a cell [17]. Single-nucleotide polymorphism (SNP) and polymorphism inherent within the biological variation of the gene lead to additional variations between individuals. Interactions between modified proteins and the dynamics of protein expression under specific biological circumstances lead to an enormous molecular complexity, which exceeds that indicated by the genome sequence alone. Gene expression can be studied either at the mrna level or at the protein level, but full protein characterization is required to truly understand the proteome. Powerful techniques make the rapid screening of mrna expression possible; however, there is often a poor correlation between mrna abundance and the quantity of corresponding functional proteins [16]. Proteomics is an emerging scientific field that involves identification, characterization, and quantification of proteins in a cell, tissue or body fluid. The broad application of proteomics in basic science and clinical medicine, with its range of tools, will accelerate our understanding of disease processes, and may facilitate the discovery of new drug targets and diagnostic markers [7]. Clinical proteomics, defined as the application of proteomics in the field of medicine, has the potential to influence daily clinical practice in providing tools for diagnosis or prognosis, defining disease states, assessing risk profiles and outcomes, and setting up individual therapeutic strategies. With this objective, most clinical applications of proteomics focus on blood (serum or plasma) biomarkers although other body fluids can be used. Unlike cells or tissues, blood does not have a genome, making proteomic techniques one of the few options for discovering biomarkers. Effective biomarker discovery, even in simple biological samples, requires a combination of subfractionation and separation, or targeted protein or peptide
Page 5 of 20 enrichments, before identification and characterization of the markers by mass spectrometry (MS) [4]. The appropriate technology or combination of technologies to match the biological questions to be answered must be identified to allow for maximum coverage of the selected subproteome and to maximize the interpretation and downstream utility of the data. In this review, we describe the current technologies for proteome fractionation and separation from biological samples, based on our laboratory work in the discovery and validation of serum biomarkers. Biomarker Discovery In general, two different strategies are being used to discover biomarkers using proteomic technologies (Figure 1). The first strategy is a targeted approach based on the more traditional hypothesis-driven evaluation of specific biomarker candidates, selected either on the basis of a biological rationale or from analysis of candidates derived from other sources. The second strategy is a de nova discovery approach that uses different proteomic techniques and finally validates potential biomarker candidates. Both strategies are complementary, have advantages and disadvantages, and may be performed in parallel [4]. Regardless of the strategy used, we believe that the discovery and development of a robust biomarker candidate, using proteomics, demands a systematic in-depth approach in which discovery and validation are coupled. Figure 2 illustrates such an approach, as adopted in our laboratory for serum/plasma and tissue samples. Sample Preparation and Fractionation Before any proteomics analysis, blood samples must be collected and processed according to a standardized tightly regulated protocol, which is of great importance in obtaining reproducible proteomic data [3]. Whether plasma or serum should be collected for proteomic analyses is controversial. One study revealed that a large number of highly abundant peptides are detectable only in serum samples, and not in plasma [13]. The authors took the view that those peptides must be produced by ex vivo degeneration during the clotting
Page 6 of 20 procedure, and therefore, suggested that plasma may be superior to serum, in particular for low-molecular-weight proteins. However, there are considerable technical issues and many in the field use serum samples, collected and processed with proper control, for biomarker discovery. After all, the majority of clinical immunoassays are performed on serum. Because of the limitations in the dynamic range of proteomic technology, it is widely accepted that in order to detect proteins that are low in abundance, blood samples must be depleted of highly abundant proteins. The most common depletion method is affinity chromatography. Several companies market affinity columns that remove up to twelve of the most highly abundant proteins, e.g. the column from Beckman Coulter, Inc. (CA, USA) removes albumin, immunoglobulin G (IgG), IgA, IgM, transferrin, fibrinogen, apolipoprotein A- I and A-II, haptoglobin, alpha-1 antitrypsin, alpha-1 acid glycoprotein and alpha-2 macroglobulin. Although these columns are thought to have little non-specific binding, caution is required to prevent carryover (protein stuck to the column and then leaking) between sequential runs on the same column and the loss of proteins present in low abundance during the chromatography or downstream concentration step. Furthermore, these multiple antibody affinity columns are expensive for routine usage in academic laboratories. As an alternative to affinity-based depletion methods, our laboratory has developed a simple, reproducible and inexpensive chemical method for the depletion of lipids, immunoglobulins and albumin [2]. Following a high-speed centrifugation step to deplete lipids, and depletion of IgG with a protein A resin, the serum or plasma is fractionated into albuminenriched and depleted fractions by ethanol precipitation. Both fractions can be used for biomarker discovery using proteomic techniques. Although the observable proteome is significantly enhanced by such depletion procedures, the remaining proteins and peptides are still present in a wide range of concentrations. Therefore, depletion of serum or plasma samples serves only as a starting point for further proteomic analysis.
Page 7 of 20 Sample Separation Because of its complexity, the serum or plasma proteome cannot be resolved completely using a single proteomic technology. Multiple proteomics techniques for protein separation must be combined to analyse and cover a large spectrum of the proteome (Figure 3). Which methods are needed depends on the underlying biological and clinical questions to be answered. Core technologies for protein separation are one- and two-dimensional gel electrophoresis (1-DE, 2-DE) and, for protein or peptide separation, one- and twodimensional liquid chromatography (1DLC, 2DLC), all coupled with mass spectrometry (MS). Two-Dimensional Gel Electrophoresis The first technology to be used in proteomics was 2-DE, which was developed independently in the laboratories of O`Farrell and Klose more than three decades ago [6, 10]. In standard 2- DE, proteins are separated in the first dimension, known as isoelectric focusing, by their molecular charge (pi). The second dimension separates the proteins according to their molecular mass (or molecular weight, MW). The MW separation is done in a polyacrylamide matrix in a sodium dodecylsulfate (SDS) milieu; the most common procedure utilizes an acrylamide gradient of 10 to 20%. Proteins can be visualized in 2-D gels using different detection methods. The more common protein staining methods include Coomassie blue and silver staining, use of fluorescence dye (e.g. Cy dyes, LAVAPurple, Sypro dyes), radiolabeling, and immunodetection. Using standard format SDS-gels for 2-DE, it is possible to routinely separate up to 2000 protein spots from serum/plasma or tissue extracts, which reflects ~100 300 different proteins, depending on the ph gradient used in the first dimension. Although 2-DE is an important and popular protein separation technique, it is limited by the solubility and mass of the separated proteins. Differential in-gel electrophoresis (2D-DIGE) is a recent improvement of the 2-DE technology. It improves gel reproducibility, minimizes alignment issues and allows better quantitative comparison between samples. In 2D-DIGE, proteins from different disease states are separately labeled with different fluorescent dyes,
Page 8 of 20 and an internal pooled standard is labeled with another dye. The labeled samples are then combined and subjected to 2-DE, and the gel is scanned at different emission wavelengths generating multiple images that can be overlaid. Figure 4 shows an example of a 2D-DIGE, which allows the differentially regulated proteins to be viewed as changed in color. 2-D gel images are evaluated and analyzed using specialized software packages. The software stores all of the relevant information on each and all of the spots of a 2-D gel in a database, compares gel patterns using complex algorithms, and highlights differences between gel images. 2-D image analysis can be time-consuming and difficult, particularly if there are marked differences between samples. Software packages can be purchased and used in-house for analysis, or companies will now provide image analysis on a contract basis. However, by using strict inclusion and exclusion criteria one can sieve out the high probability markers (or protein spot changes). Multidimensional Liquid Chromatography Recently, other methods for separating proteins have been exploited in proteomic analysis. Many involve liquid chromatography technologies that use solid- and liquid-phase media to separate proteins and/or their peptide fragments. The basic principle is a soluble sample that is separated in a liquid-phase through a column, which is usually a tube packed with small particles of specific surface chemistry [9, 15]. The sample is resolved as it traverses the length of the column based on protein- or peptide-specific chemical or physical interactions with the solid-phase. The time when the separated sample is detected at the end of the column (e.g. by UV absorbance at 210 nm, which essentially measures the number and quantity of peptide bonds) is the retention time and is quantitative if the peak contains a single protein/peptide (which in proteomics is rare and therefore, peak volume or intensity in this case is semi-quantitative). One-dimensional liquid chromatography (1DLC) can be used to separate proteins according to their molecular mass, isoelectric point, or hydrophobicity, which are the three chemical characteristics that define any given protein. The most commonly used 1DLC is reversed
Page 9 of 20 phase chromatography, in which proteins are separated based on hydrophobicity. Reversed phase chromatography can also be used to concentrate and/or desalt samples. In 2DLC, proteins are separated in the first dimension by chromatographic focusing (pi) and in the second dimension by reversed phase chromatography (hydrophobicity). Thus, 2DLC increases the extent of protein fractionation, which facilitates analysis of a larger spectrum of the proteome, including specific isoforms, PTMs and low-abundance proteins. As with 1DLC, this method has been used in proteomics primarily for peptide separation before MS analysis (due to its compatibility with ESI instruments); however it is increasingly used to separate complex intact protein mixtures, which are then enzymatically digested for LC or MALDI (matrix-assisted laser desorption/ionization) MS/MS analysis. 2DLC requires a larger quantity of sample for a single run (>2.5 ml) as compared to 1DLC (50-100 µl), which can be a difficulty if available sample volumes are small (e.g. from mouse models). It is important both to quantify and to identify proteins present in fractions generated by 1DLC or 2DLC. One strategy is to normalize, overlay and compare elution profiles between different samples using specialized software packages (for which there is currently a need especially when analyzing a large number of samples) and analyze, using MS, only the fraction that varies between samples. Current data suggests that using multiple proteomic technologies dramatically increases the number of proteins detected, especially of those present in the sample at very low abundance [4]. 2-DE, 1DLC and 2DLC are synergistic separation techniques that, coupled with MS identification, expand the observable proteome and will provide a large dynamic protein spectrum for biomarker discovery. In fact, we recently compared 2-DE and 2DLC by creating a large database for serum and isolated inner mitochondrial subproteome, and revealed that only about 12% of identified proteins were common to both platforms [8, 14].
Page 10 of 20 Protein Identification Mass spectrometry Mass spectrometry techniques have greatly advanced proteomics and proteomics-based biomarker discovery in recent years. 2-DE coupled with MS is widely used for tissue and serum analyses. Spots from 2-D gels are excised, the proteins subjected to in-gel digestion and the resulting peptide fragments identified by MS. For identification of 2-D gel spots matrix-assisted laser desorption/ionization time of flight (MALDI-TOF) MS is commonly used [4]. If information on protein isoforms, PTMs or absolute identification of proteins in complex mixtures such as fractions generated by 1DLC and 2DLC are needed, then tandem MS (commonly referred as MS/MS) is required. MS/MS spectra are usually generated by an ion trap or quadrupole time-of-flight mass spectrometer, which allows to generate de novo sequencing and exact localization of PTMs. For protein quantitation in MS analysis, several isotopic labeling techniques (e.g. itraq, 16 O/ 18 O, SILCA) and, recently, label-free methods have been developed, though these are not further described in this review. Biomarker Validation Biomarker candidates have traditionally been evaluated with quantitative immunoassays (e.g. ELISA) that are unique for one analyte [3]. With the rapid development of new potential biomarkers, it is important to develop quantitative assay platforms that can simultaneously measure many proteins in many samples at a small sample volume. A variety of multiplex immunoassays have been developed in recent years, which offer some advantages over traditional quantitative assays [11]. Multiplex immunoassays are essentially the same as an ELISA except that multiple analytes are quantified simultaneously. Thus, many biomarkers can be evaluated at one time under the same standardized conditions, quantitative information can be obtained in a highly parallel analysis, and reagent costs are substantially reduced. The most common multiplex assay used is an array of antibodies printed on slides/or plates at high density. It is now possible to print hundreds of antibodies, although issues with analyte and antibody cross reactivity and matrix affects make smaller numbers
Page 11 of 20 (<20) the preferred choice of many. The current issues with multiplex arrays is their inter- and intra-assay reproducibility, matrix affects, background limits and the specificity and sensitivity of the antibody assay. There are many other quantitative and semiquantitative multiplex immunoassays, such as miniature sandwich immunoassays, bead-based multiplex immunoassays and assays for specific signaling pathways, but investigators must take care to ensure the specificity and reproducibility of each assay within the multiplex [5, 12, 13]. The ultimate success of a multiplex assay depends upon its ability to quantitatively detect proteins at concentrations likely to be present in serum samples, which range from less than 1 pg/ml to over 1 mg/ml. Multiplex assays can be used as powerful validation tool for candidate biomarkers identified by a de nova proteomic discovery approach. In addition, multiplex assays are often used for evaluating a variety of candidate biomarkers in a targeted approach. In either case, the multiplex assay requires the added flexibility of allowing the investigator to mount their own analytes. To test whether or not a newly discovered biomarker is of clinical utility, we recommend evaluating all candidates in relation to existing biomarkers if such exist. Multiplex immunoassays again are a desirable platform for this approach as it provides quantitative information in a higher-throughput format [3]. Perspectives Proteomic technologies applied in basic science will complement genomic-based and physiological approaches. Proteomics will not only reveal new insights into complex molecular processes underlying diseases, but will provide tools to develop novel diagnostic and prognostic biomarker(s) that include unique information about the patient. Such biomarkers could have tremendous benefits for patient management, and may accelerate the development of new therapeutic strategies. In this context, it may be important to integrate proteomic biomarker information with that available from genetic biomarkers, which could provide a powerful integrated risk stratification [1]. Proteomics is a rapidly changing field because of extensive advances in the underlying technologies including the fractionation, separation and identification of proteins in biological samples. Although proteomics is
Page 12 of 20 evolving quickly and providing extensive protein databases with potential biomarkers, the translation of promising disease markers from bench to bedside is another challenge. This requires both close collaboration between basic scientists and clinicians, and well-designed studies with appropriate statistical power, blinding and validation. With the application of such an endeavor proteomics could lead to an optimized and more personalized medicine.
Page 13 of 20 Acknowledgments Peter Matt thanks the Hippocrate Foundation Basel and the Howard Hughes Medical Institute Johns Hopkins Medicine, Baltimore, for financial support. Disclosures Peter Matt is supported by the Hippocrate Foundation Basel and the Howard Hughes Medical Institute Johns Hopkins Medicine, Baltimore. Jennifer Van Eyk is supported by grants from the National Heart, Lung, and Blood Institute Proteomic Initiative (contract NO- HV-28120) and the Daniel P. Amos Family Foundation.
Page 14 of 20 References 1. Arab S, Gramolini AO, Ping P, Kislinger T, Stanley B, van Eyk J, Ouzounian M, MacLennan DH, Emili A and Liu PP. Cardiovascular proteomics: tools to develop novel biomarkers and potential applications. J Am Coll Cardiol. 48: 1733-41, 2006. 2. Fu Q, Bovenkamp DE and Van Eyk JE. A rapid, economical, and reproducible method for human serum delipidation and albumin and IgG removal for proteomic analysis. Methods Mol Biol. 357: 365-71, 2007. 3. Fu Q and Van Eyk JE. Proteomics and heart disease: identifying biomarkers of clinical utility. Expert Rev Proteomics. 3: 237-49, 2006. 4. Guo Y, Fu Z and Van Eyk JE. A proteomic primer for the clinician. Proc Am Thorac Soc. 4: 9-17, 2007. 5. Khan SS, Smith MS, Reda D, Suffredini AF and McCoy JP. Multiplex bead array assays for detection of soluble cytokines: comparisons of sensitivity and quantitative values among kits from multiple manufacturers. Cytometry B Clin Cytom. 61: 35-9, 2004. 6. Klose J. Protein mapping by combined isoelectric focusing and electrophoresis of mouse tissues. A novel approach to testing for induced point mutations in mammals. Humangenetik. 26: 231-43, 1975. 7. Matt P, Carrel T, White M, Lefkovits I and Van Eyk J. Proteomics in cardiovascular surgery. J Thorac Cardiovasc Surg. 133: 210-4, 2007. 8. McDonald T, Sheng S, Stanley B, Chen D, Ko Y, Cole RN, Pedersen P and Van Eyk JE. Expanding the subproteome of the inner mitochondria using protein separation technologies: one- and two-dimensional liquid chromatography and twodimensional gel electrophoresis. Mol Cell Proteomics. 5: 2392-411, 2006. 9. Neverova I and Van Eyk JE. Role of chromatographic techniques in proteomic analysis. J Chromatogr B Analyt Technol Biomed Life Sci. 815: 51-63, 2005. 10. O`Farrel P. High resolution two-dimensional electrophoresis of proteins. J Biol Chem. 250: 4007-21, 1975.
Page 15 of 20 11. Pang S, Smith J, Onley D, Reeve J, Walker M and Foy C. A comparability study of the emerging protein array platforms with established ELISA procedures. J Immunol Methods. 302: 1-12, 2005. 12. Perlee L, Christiansen J, Dondero R, Grimwade B, Lejnine S, Mullenix M, Shao W, Sorette M, Tchernev V, Patel D and Kingsmore S. Development and standardization of multiplexed antibody microarrays for use in quantitative proteomics. Proteome Sci. 2: 9, 2004. 13. Schweitzer B, Roberts S, Grimwade B, Shao W, Wang M, Fu Q, Shu Q, Laroche I, Zhou Z, Tchernev VT, Christiansen J, Velleca M and Kingsmore SF. Multiplexed protein profiling on microarrays by rolling-circle amplification. Nat Biotechnol. 20: 359-65, 2002. 14. Sheng S, Chen D and Van Eyk JE. Multidimensional liquid chromatography separation of intact proteins by chromatographic focusing and reversed phase of the human serum proteome: optimization and protein database. Mol Cell Proteomics. 5: 26-34, 2006. 15. Stastna M and van Eyk J. Liquid Chromatography. In: Clinical Proteomics: From Diagnosis to Therapy. WiLEY-VCH Verlag GmbH&Co. KGaA, Weinheim, 2007. 16. Van Eyk J and Dunn M. Proteomics, a step beyond Genomics. In: Proteomic and Genomic Analysis of Cardiovascular Disease. WILEY-VCH Verlag GmbH&Co. KGaA, Weinheim, 2003. 17. Zerkowski HR, Grussenmeyer T, Matt P, Grapow M, Engelhardt S and Lefkovits I. Proteomics strategies in cardiovascular research. J Proteome Res. 3: 200-8, 2004.
Page 16 of 20 Figure legends Figure 1 Two strategies for biomarker discovery and validation: a targeted approach, based on hypothesis-driven evaluation of specific biomarker candidates, and a de nova discovery approach using different proteomic technologies followed by validation of potential biomarker candidates. Figure 2 The platform for biomarker discovery in serum and tissue samples at our institution combines multiple synergistic protein fractionation and separation methods including one- and twodimensional gel electrophoresis (1-DE, 2-DE), differential in-gel electrophoresis (2D-DIGE), and one- and two-dimensional liquid chromatography (1DLC, 2DLC), coupled with mass spectrometry (MS) identification and validation methods, using multiplex arrays. Figure 3 2-DE, 1DLC and 2DLC are synergistic proteomic separation technologies with minimal overlap that, coupled with MS identification, expand the observable proteome. The figure illustrates the dynamic protein spectrum of the three technologies. Figure 4 2D-DIGE analysis of two different serum samples. One serum sample is labeled with Cy3 (green color in this example), whilst the other is labeled with Cy5 (blue), and equal concentrations of both samples are labeled with Cy2 (red). All three labeled samples are then combined, separated on the same 2-D gel, and scanned at different emission wavelengths which allows the differentially expressed proteins to be viewed as changed in color, see arrows for green or blue spots in enlarged gel area. Proteins that are equally expressed in both samples appear as white spots.
Page 17 of 20 361x277mm (72 x 72 DPI)
374x277mm (72 x 72 DPI) Page 18 of 20
Page 19 of 20 320x216mm (72 x 72 DPI)
353x261mm (72 x 72 DPI) Page 20 of 20