Seldi Tof

ARTICLE IN PRESS

Pathology – Research and Practice 200 (2004) 83–94

www.elsevier.de/prp

REVIEW

Surface-enhanced laser desorption ionization time-of-flight mass spectrometry (SELDI TOF-MS) and ProteinChips technology in proteomics research . Meuera Volker Seiberta,*, Andreas Wiesnerb, Thomas Buschmanna, Jorn a b

EUROPROTEOME AG, Neuendorfstrasse 24b, 16761 Hennigsdorf, Germany Ciphergen Biosystems GmbH, Hannah-Vogt-Strasse 1, 37085 Gottingen, Germany .

Received 16 December 2003; accepted 5 January 2004

Abstract In this review article, we describe some of the studies that have been performed using the surface-enhanced laser desorption ionization (SELDI) time-of-flight mass spectrometry and ProteinChips technology over the past few years, and highlight both their findings as well as limitations. Proteomic applications, such as target or marker identification and target validation or toxicology, will be addressed. We will also provide an examination of SELDI technology and go into the question of where possible future research may lead us. r 2004 Elsevier GmbH. All rights reserved. Keywords: SELDI; ProteinChipu. s System; Oncology; Toxicology; Immunology

New technologies for old questions Examination of cell or tissue samples by histopathologists is an integral component in the diagnosis and management of cancer patients. The histopathologist provides information regarding tumor type, grade, stage, completeness of excision, and expression of certain receptors (e.g. estrogen and progesterone receptor, HER-2neu, c-kit), which is used by the surgeon, oncologist, and radiologist to select a treatment scheme. Usually, histopathologists apply light microscopic investigations to a cell or tissue sample. This wellestablished approach to diagnosis and patient treatment is challenged and accomplished by new technologies in proteomics that might have dramatic impacts on the early detection of cancer, cancer treatment, and patient surveillance after treatment has started. *Corresponding author. Tel.: +49-(0)3302-202-3276. E-mail address: [email protected] (V. Seibert). 0344-0338/$ - see front matter r 2004 Elsevier GmbH. All rights reserved. doi:10.1016/j.prp.2004.01.010

Proteins, rather than DNA or RNA, carry out most of the cellular functions. Therefore, the direct measurement of protein levels and activity within the cell is the best determinant of overall cellular function. Moreover, since there is often a poor correlation between transcript and protein levels, an accurate conclusion regarding protein function based on mRNA levels cannot be made [2]. Here, proteomic analysis becomes a valuable tool in determining the presence of protein within a sample. Furthermore, it can also play a pivotal role in mapping protein profiles in different sample groups, e.g. healthy and diseased individuals to look for differential protein expression patterns. In the latter case, typically, the final result of such an experiment is a list of proteins that are up- or down-regulated between healthy and diseased individuals. Currently, one of the driving forces in proteomics is the discovery of such biomarkers. The

ARTICLE IN PRESS 84

V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

determination of changes in a relative or absolute concentration is fundamental to the discovery of valid biomarkers. In the last few years, proteomics profiling experiments were performed by applying samples to two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) coupled to mass spectrometry (MS) to perform profiling experiments [32]. Unfortunately, this technique is not only laborious, but also limits the isolation of low-abundant or highly hydrophobic membrane proteins. At present, it requires large amounts of samples. This is particularly problematic for clinical samples. Such samples are obtained through invasive techniques, such as biopsies, and are therefore available in limited amounts. Despite all the possible improvements that can be brought about to such a technique in the future, 2D gel electrophoresis will most likely remain rather a low-throughput approach to proteomic analysis. As the range of protein expression and modification is dynamic, the identification of differences in the proteome is a far larger and more complex challenge than the differential display of DNA or RNA by cDNA arrays. Thus, there is a clear need for high-throughput assays in proteomics. In the past year, the field of proteomics experienced a rapid evolution of innovative technologies. Large-scale quantitative research, comparable to that achieved for gene expression, is thus becoming reality at the protein level. Testing proteins in a high-throughput fashion requires the ability to use arrays with low sample requirement. Although microarrays with immobilized antibodies, variations thereof, or other types of bait proteins have been developed, they are currently only partially robust for high-throughput studies [17]. A potential alternative to the 2D PAGE approach for the multidimensional separation of proteins is the novel analytical technique referred to as surface-enhanced laser desorption ionization (SELDI) technology. This high-throughput, array-based technology can bring us closer to a better understanding of cellular functions at the protein level. Furthermore, this technology can also be successfully applied to future drug discovery. Some research groups [7,16,29] have reported on the development of this novel technology based on MS. The SELDI technology is designed to perform MS analysis of protein mixtures retained on chromatographic array surfaces (reviewed in [31]). It produces spectra of complex protein mixtures based on the mass-to-charge ratio of the proteins and on their binding affinity to the chip surface. Differentially expressed proteins may be determined from these protein profiles by comparing peak intensities. Comparisons of the protein peak patterns obtained from samples representing different states are expected to provide detailed diagnostic patterns classifying cellular or pathological states, and can consecutively provide new insights into the function

and control of biological processes. This method has already yielded biologically and clinically significant results. The SELDI technology is currently implemented in the ProteinChip System produced by Ciphergen Biosystems Inc. (Fremont, CA, USA). The ProteinChip System from Ciphergen Biosystems, Inc. can be regarded as an analogue to the DNA chip strategy, although it is much more complex because of the varying nature of proteins as compared to DNA. One great advantage of this technique lies in the small sample requirement. Clinical material is commonly limited to tissue biopsies or sampled biofluids. Consequently, this microanalytical technique will be of notable importance, enabling the optimal extraction of biological information from low amounts of sample. Also, this technique needs no protein tagging and can be run automatically. The successful use of the ProteinChip System described herein relies entirely on the protein fingerprint pattern of the masses. As these masses were found to be detected both reproducibly and reliably, only the mass values are required to make a correct classification or diagnosis. As it is not necessary to know the identities of the masses for the purpose of differential diagnosis, this technology provides an alternative platform for the differential display of multiple biomarkers. However, in order to identify possible novel therapeutic targets, the biological role of such proteins needs to be examined. For this, the exact identity of each identified biomarker, as well as their cellular localization and biological function, needs to be determined. Furthermore, knowing their identities will be essential for producing antibodies for the development of rather classical diagnostic tools, such as the ever popular ELISAs. Despite this, the potential of MS to yield comprehensive profiles of peptides and proteins in tissue and biological fluids without the need to first carry out protein separation has become a matter of interest.

Technology: the ProteinChip Biomarker System at a glance The basic procedure for Expression Difference Mappingt applications with the ProteinChip Biomarker System is straightforward. Virtually any type of protein-containing solutions can be directly applied to the spots of ProteinChip Arrays. These spots present either chromatographic surfaces with certain physicochemical characteristics (hydrophobic, cationic, anionic, metal ion presenting, or hydrophilic), or are preactivated for the coupling of capture molecules (protein, DNA or RNA) prior to sample loading (Fig. 1). Typically, for Expression Difference Mapping experiments, the chromatographic surfaces are used. Sample requirement is low (1–10 mg total protein per spot), and

ARTICLE IN PRESS V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

85

Fig. 1. The different types of ProteinChip Arrays. The chromatographic ProteinChip Arrays incorporate hydrophobic, cationic, anionic, metal ions or hydrophilic spots. These ‘‘chemical surfaces’’ are best suited for protein expression profiling studies. Another series of ProteinChip Arrays have pre-activated ‘‘biological surfaces’’ designed for coupling of biomolecules with applications in antibody–antigen assays, receptor–ligand interaction studies, and DNA–protein binding experiments.

intensities on the y-axis. ProteinChip Software enables the user to control and influence the automated detection process and incorporates a wide range of software tools for the comparative analysis of higher numbers of samples. Thus, the entire Expression Difference Mapping technique can be achieved in a fast few-step procedure (Fig. 2).

Distinctions of the SELDI process

Fig. 2. Biomarker discovery in patient samples by SELDIbased ProteinChip technology.

sample volume can be freely chosen from 0.5 ml up to around 400 ml. Following a short incubation period, proteins and contaminants that do not bind to the spot surface are removed by washing. Subsequently, a solution of energy-absorbing molecules (EAMs) is applied to each of the spots, and the ProteinChip Array is ready for the analysis in the ProteinChip Reader, a highly sensitive laser desorption/ionization TOF-MS (time-of-flight MS). Results are initially visualized in a graph, with the mass-to-charge ratio of the sample components on the x-axis and the corresponding signal

One of the more obvious advantages of this surfaceenhanced process is that components, such as salts or detergents that commonly cause problems with other analytical tools, are removed prior to analysis. Furthermore, as described in the previous section, only proteins actively interacting with the spot surfaces are analyzed in the ProteinChip Reader, because all other nonspecific components are removed in advance. In addition, as each analysis is automatically linked to an on-spot fractionation step, the complexity of the samples is reduced. This, in turn, results in the increased probability of detecting markers that are present in lower concentrations. Normally, a number of different array types are combined with washing steps of varying stringencies to retain different protein subsets from the original samples. Equivalent array–buffer-combinations of the different sample groups can then be comparatively analyzed to decipher formerly unknown markers. With complex samples, such as serum, pre-fractionation is preferable, as it enhances the total number of signals resolved. For this purpose, ready-to-use kits are provided from Ciphergen Biosystems, Inc. Subsequent biomarker purification is easier to achieve, because in addition to the exact molecular

ARTICLE IN PRESS 86

V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

Table 1. Features of the SELDI-based ProteinChip Biomarker System Sample Analysis Quantification Comparative analysis Protein purification On-spot process Protein identification Time Handling

No pretreatment necessary. Sample can be directly applied. Sample requirement is low. Peptide sensitivity in the femtomolar range Fully automated, no manual adjustment of laser position Signal intensity correlated with protein concentration. Standard curves permit absolute quantification Wide variety of software tools for comparative expression profiling and marker discovery. Multivariate decision tree analysis by the Biomarker Patternst Software On-spot development of purification protocols with subsequent up-scaling to mini-spin or preparative columns No sample loss. Enzymatic reactions can be conducted directly on-spot prior to analysis By on-spot peptide mapping with the ProteinChip Reader. Additional sequencing capability by Tandem-MS-interface About 100 samples per day in manual mode, close to 400 samples per day with automation Professional use after 1 week of training

weight, the basic physico-chemical characteristics of the protein of interest can be elucidated from its binding behavior under certain washing conditions. Optimized separation material mimicking the array surfaces has been developed by BioSepras, the Process Division of Ciphergen Biosystems, Inc. Moreover, the active interaction of the analyte constitutes another important feature of the ProteinChip technology, namely its ability to quantify individual protein levels in a given sample, a prerequisite for the successful use of any technology in comparative profiling. This ability is due to the fact that the proteins are equally distributed on the spot surface so that the signal intensities obtained correspond well to the concentration of proteins, with low femtomolar sensitivities in the best-resolved peptide range. More technical details are given in a recently published review [31]. The unique characteristics of the SELDI-principle (Table 1) are the base of the ProteinChip technology, enabling a wide variety of applications on a single integrated platform.

Automated biomarker discovery and validation With the basic version of the ProteinChip Biomarker System, one can easily prepare and analyze up to 100 samples a day. To further enhance the throughput to more than 380 samples in less than a day, sample preparation can be automated with a customized Biomeks 2000 (Beckman Coulter, Inc., USA) liquidhandling robot and the use of the ProteinChip AutoBiomarker System (Fig. 3), composed of an AutoLoader-equipped ProteinChip Reader, barcode-labeled arrays, and CiphergenExpresst Data Manager Software for automated sample tracking and advanced data analysis. Furthermore, ready-to-use kits for implementing the Expression Difference Mapping protocol are provided for even more convenient sample handling and consistency.

Fig. 3. The ProteinChip AutoBiomarker System. Automated sample fractionation and application is done with a specially configured version of the Biomek 2000 laboratory workstation adapted for use with the ProteinChip Biomarker System. For subsequent analysis in the ProteinChip Reader, the barcodelabeled arrays are inserted into the ProteinChip AutoLoader and automatically fed into the ProteinChip Reader. The basic ProteinChip Software can be extended by adding the CiphergenExpress Data Manager, the Biomarker Wizard application module and Biomarker Patterns Software, altogether enabling the highest level of biomarker discovery and throughput data flow.

When starting a biomarker discovery project, the first runs of each group may contain 10–30 samples, resulting in multiple spectra depending on the number of array type/washing buffer combinations applied. Thus, even with considerably small data sets, the user will be able to generate hundreds of spectra in a short time. During the subsequent validation period, the data output will be much higher. To meet these demands, special software tools are provided to ensure a fast and reliable evaluation process.

ARTICLE IN PRESS V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

After an automated normalization of the profiles to minimize the influence of inherent experimental variations, the ‘‘Biomarker Wizard’’ can be used for the fast detection of single biomarkers. Here, the program summarizes signals of equal masses into clusters and compares all the different clusters of a given range of profiles, returning p-value calculations and graphical presentations. From the results, it can be easily seen which cluster intensities are statistically different between the sample groups. In other words: which proteins are up- or down-regulated in one of the groups, and which of these can be considered as potential markers for a special type of disease or as a reaction to changes of certain experimental parameters? As can be expected from the complexity of life, in many cases, no single marker will be present to distinguish between the sample groups. It is rather a complex pattern of multiple markers that will be required. However, elucidation of such patterns of multimarkers is not trivial, and for these kinds of data sets, Biomarker PatternsTM Software has been used effectively. In contrast to the Biomarker Wizard, the Biomarker Patterns program performs multivariate analyses and is able to classify the processed data by specially adapted algorithms [12]. It creates tree-like structured decision diagrams by splitting the original data set (parent node) in two subgroups (child nodes) of highest possible purity. Each child node then becomes a parent node at the time of creation and can be the origin of a new split. The program calculates which signals, i.e. proteins, are best suited to act as splitters, whereby the splitting rules define what intensity, i.e. protein abundance, a signal must have to be assigned to one of the groups. After the initial learning period where the tree building takes place, the prediction success of the resulting model can be tested with new sets of data. Thus, Biomarker Patterns Software is a true prediction tool enabling the user to conduct multivariate analysis at a professional level without the need to be a specialist in statistical evaluations. In contrast to other classifiers, such as neural networks or nearest neighbor calculations, the models are easier to understand, and each sample can be tracked down the tree [1].

87

unnecessary effort made when identifying common, non-differentially regulated proteins. In most cases, each protein of interest needs to be purified or at least enriched for subsequent identification experiments. One option is the use of antibodies in combination with pre-activated ProteinChip Arrays for the selective capture of the protein. In other cases, minispin columns, microplate-formatted chromatographic devices, or traditional preparative-scale columns are suitable for such a separation. One of the major advantages is that the best suited chromatographic material and the proper elution conditions can be predicted from the binding behavior of the protein on the ProteinChip Arrays. Optimized separation material has been developed by BioSepra, the Process Division of Ciphergen Biosystems, Inc. [30]. For identification, the purified or enriched protein is proteolytically cleaved, and the masses of the resulting peptide fragments are determined using the ProteinChip Reader. Subsequently, searches in public databases are used for the identification by peptide fingerprinting. For unequivocal identifications and partial sequencing of the protein, the ProteinChip Arrays can be read in highresolution tandem-MS instruments [23]. For this purpose, an exchangeable UV laser desorption ion source, the ProteinChip Interface PCI 1000, is directly linked to a tandem-MS system (Tandem MS QSTARs from Sciex-MDS/Applied Biosystems). Very recently, a new type of ProteinChip Array was developed to facilitate the analysis of peptide fingerprints by surface-enhanced neat desorption (SEND) technology for protein identification (ID). In SEND ID, the EAMs are incorporated in the surface chemistry of the array. Thus, adding EAM separately is not needed, and the chemical noise caused by the application of EAM is significantly reduced. It is then possible to detect low molecular weight species with a greatly improved signal-to-noise ratio, while also minimizing the variability associated with matrix addition. The spot surface contains a hydrophobic interaction functional group, which enables on-spot clean-up of samples. Most other approaches to this application require a clean-up of the sample in a small chromatography column and premixing with the matrix. By avoiding these steps with SEND, workflow is simplified, and losses of sticky peptides are minimized. In addition, the suppression of interfering matrix signal results in superior sensitivity and improved sequence coverage.

Protein identification Protein identification [e.g., 9,26,27,33] is the final step in the SELDI-based Expression Difference Mapping technique. Normally, this is not performed before the importance of a marker protein has been recognized. This strategy of analysis incorporated with ProteinChip technology has the advantage of eliminating the

Current applications for the ProteinChip System As for all studies, the successful use of the very sensitive ProteinChip System depends on a good

ARTICLE IN PRESS 88

V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

experimental design and on high-quality biological samples. Sample preparation factors appear to affect quality. Fundamental issues involving biological variability, pre-analytical factors, and analytical reproducibility remain to be resolved before expression profiling becomes a common tool in the pathologist’s laboratory. Expression profiling also relies on the preservation of, e.g. the tissue of interest. Traditional formalin fixation, which is used in routine histopathology, leads to crosslinking of proteins; therefore, alternatively prepared material is required. In addition, the introduction of robotics/laboratory automation will continue to improve the reproducibility of the system, particularly in studies with hundreds of samples. Analysis of clinical samples by the ProteinChip System is complicated through the above-mentioned tissue or body fluid heterogeneity. In addition, the system is expensive and requires trained labor. In the case of tissue heterogeneity, various approaches are beneficial to reduce this problem, but they also further reduce the amount of samples available. In particular, the use of laser-capture microdissection (LCM; [5,11,21]), which allows defined cell types to be isolated from tissues, yields amounts of proteins that are difficult to reconcile with the need for greater amounts of samples. However, also in the field of sample preparation, substantial progress is being made. For example in the field of cancer proteomics, Europroteome has developed a highly standardized method for efficient and reproducible epithelial tumor cell isolation. This method, described by Reymond et al. [24], overcomes the quantitative limitations of potentially alternative preparation methods, such as LCM, and has already been successfully used in combination with the ProteinChip System. To date, the ProteinChip System has found its most widespread clinical applications principally through the comparison of serum from healthy and diseased individuals. In addition to major constituents, such as albumin or immunoglobulins, serum contains many other proteins that are synthesized and secreted, shed, or lost from cells and tissues throughout the body. Serum screens provide us with a useful, less-invasive approach for disease-monitoring. Furthermore, serum protein profiling holds enormous promise as a tool for the early detection and stratified diagnosis of disease, as well as for the assessment of success of a given therapeutic regimen. In such biomarker studies, hundreds of samples need to be profiled by the ProteinChip System, but the clear necessity to identify new serological single biomarkers/biomarker patterns to diagnose diseases justifies such enormous efforts. A series of recent articles have demonstrated the usefulness of this technology in protein profiling in different fields. Some of the applications are described below in more detail.

Oncology Implementation of routine cancer screening accompanied by treatment of early stage disease is of paramount importance for achieving reductions in cancer mortality rates. While morphological assessment of relevant cells has traditionally been used to identify individuals with cancer, this approach cannot be easily used for identifying cancers at less accessible sites, such as the ovary or pancreas. For these and many other cancers, mortality remains high, as sensitive and specific screening assays detecting in situ or early stage disease are still unavailable. Identification of novel markers for the early detection of cancer is one of the major applications of the ProteinChip System. Several studies to identify potential cancer biomarkers have already been performed. As discussed above, the ProteinChip System is very useful for identifying potential cancer biomarkers, as it facilitates the isolation of proteins present at significantly higher levels in malignant than in healthy tissues or bodyfluids. The development of an appropriate blood-based cancer-screening assay would be of significant value in cancer prevention. The most exciting results were yielded from blood-based cancerscreening research, resulting in patterns of protein expression that can distinguish between tumors of different anatomical origin and define new subgroups of cancer with similar histological appearance, but distinct molecular profiles. Some of these new molecular subclasses of tumor appear to correlate with clinical behavior. If substantiated in larger studies, this may form a basis for stratifying patients so that they receive optimal therapeutic treatment and follow-up. Many companies have started to identify new serum markers for cancer by using the ProteinChip System. For example Europroteome’s strategy to fight cancer is shown in Fig. 4. It is advantageous to supplement studies at the serum level with analyses of protein profiles at the patient’s tumor cell level. Such studies of molecules reflecting the metabolism of a tumor could be of value regarding the diagnosis of a tumor (e.g. a serum level above a defined threshold) or the monitoring of therapy response (changes in serum levels under therapy). In the last year, many SELDI-based studies dealing with various tumor types were published, among which two of the most notable ones were conducted by Petricoin et al. [22] and Shiwa et al. [26]. In a study of ovarian cancer, Petricoin et al. [22] identified proteomics patterns in serum that distinguish neoplastic from nonneoplastic diseases within the ovary. For the serological diagnosis of patients with ovarian cancer, different tumor markers have been assessed, including CA125 as the most widely used one. However, the sensitivity and specificity of these markers appeared to be insufficient for the diagnosis of stage I ovarian cancer. The study

ARTICLE IN PRESS V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

89

Fig. 4. Europroteome’s strategy for the detection of cancer biomarkers for diagnosis, prognosis and therapy.

was performed by adding serum samples from 50 unaffected women and from 50 patients with ovarian cancer to a C16 hydrophobic interaction ProteinChip Array. Afterwards, the arrays were analyzed in the ProteinChip Reader. Peptides and proteins below the 20,000m/z range were selected. Positives and controls were run concurrently, intermingled on the same and multiple arrays; the operators were unaware of which was which. After cluster analysis, they recognized a discriminatory pattern of 5–20 key proteins (mass-tocharge values) that distinguished cancer from noncancer best. Analysis of spectra yielded 100% sensitivity and 95% specificity. The positive predictive value for this set of samples was 94%, compared with 35% for CA125 for the same samples. Using this approach, the pattern itself, independent of the identity of the proteins or peptides, is the discriminator and might represent a new diagnostic paradigm. Another study conducted by Shiwa et al. [26] discovered and identified a tissue-specific tumor biomarker from 39 human cancer cell lines using the ProteinChip System. A protein biomarker candidate of 12 kDa was found in colon cancer cells. The peak intensity of the 12 kDa protein in colon cancer cell lines was remarkably higher than that in other cancer cell lines. Samples were analyzed using strong anionexchange, weak cation-exchange, reversed phase, normal phase, and immobilized affinity capture ProteinChip Arrays from Ciphergen Biosystems, Inc. This biomarker was purified and identified by using retentate chromatography MS and ProteinChip-Tandem MS systems. The optimized purification conditions developed ‘‘on-chip’’ were directly transferred to conventional chromatography to purify the biomarker, which was identified as prothymosin-a. The relative expression level of prothymosin-a between colon cancer cells and normal colon mucosal cells was evaluated on the same ProteinChip platform. Prothymosin-a expression in

colon cancer cells was clearly higher than that in normal colon cells. These results indicate that prothymosin-a could be a potential biomarker for colon cancer, and that the ProteinChip System could perform the whole process of biomarker discovery from screening to evaluation of the identified marker. In addition, the ProteinChip System has recently been shown to be useful in discovering cancer biomarkers for the diagnosis of, e.g. head and neck squamous cell carcinoma [19], bladder [28], prostate [1], pancreas [25], and breast [18].

Toxicology/responder profiling The ProteinChip System is a highly sensitive means of screening for toxicity and probing toxic mechanisms. The concept of examining and comparing expression data at the proteomic levels for toxicity prediction and responder profiling screens has become a matter of interest, particularly in the pharmaceutical industry. Potential opportunities to emerge from such studies could be: (i) prediction of a patient’s response to drugs/ treatment followed by a preventive strategy towards drug side effects; (ii) the improvement of the accuracy of determining appropriate dosages of drugs and, therefore, to achieve personalized medical treatment. It is possible to identify biomarkers that are specifically modulated following drug treatment. These changes, caused by alterations, may lead to the biochemical pathways involved in therapy response. The understanding of the biology of drug action will dramatically accelerate the realization of truly personalized medicine. It will increase the efficiency and reduce the cost not only of target and lead discovery, but also of clinical trials. Therefore, the identification of the biomarkers that can predict response to drugs would be helpful in providing insights into the mechanism of the

ARTICLE IN PRESS 90

V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

molecularly targeted chemopreventive agent, drug, and in better designing future clinical trials based on patients’ response profiles. A productive approach for the identification of toxicologically relevant biomarkers in the field of heart disease was performed by Dunn and colleagues [10]. In 2003, they reported on plasma biomarkers that can be exploited as toxicological indicators for detecting both acute and chronic rejection after cardiac transplantation. In acute rejection experiments, they found significantly increased synthesis of three proteins: aB-crystallin (CRAB), myosin light chain 1 (MLC1), and a-tropomyosin. Subsequently, they used ELISA to determine whether these proteins were found in plasma, and whether the levels correlated with acute rejection. There was a significant increase in peak concentrations of CRAB associated with acute rejection. Although a-tropomyosin was also increased, this did not reach statistical significance. MLC1 was not detected in human plasma. In the case of chronic rejection (Tx-CAD), they adopted an alternative approach in which they screened plasma samples for the presence of tissue-specific antibodies. They found a strong association between the development of circulating antibodies reactive with endothelial proteins and the development of Tx-CAD. The major immunogenic polypeptide was identified as the intermediate filament protein, vimentin. Again, biomarkers were confirmed by an ELISA for a-vimentin antibodies in plasma. This assay identified patients at risk of developing Tx-CAD. It also showed that a-vimentin antibodies precede angiographically detectable disease by 3–24 months and may indicate Tx-CAD at an early and potentially treatable stage. In a longitudinal ELISA study of 880 sequential serum samples obtained from 109 post-transplant patients collected over 5 years, the group found that the development of a-vimentin antibodies is an independent predictor of Tx-CAD (sensitivity 63%, specificity 82%) and can be used to identify some of the patients at risk of developing this complication. Other toxicology studies were performed by He et al. [15], dealing with the analysis of arsenic-induced cell transformation, and Dare et al. [8], going into the identification of a biomarker for compound-induced skeletal muscle toxicity in rats.

Immunology The ProteinChip System can also be used with antibody-coupled arrays or as an antibody capture tool. The assay readout allows for the determination of single or multiple antibody–antigen interactions. Mass identity of the antigens enables the detection of post-translational modifications, such as phosphorylation of single

proteins, thus making the system in this respect superior to, e.g. ELISA analysis. Grus et al. [14] performed a study that aimed at detecting autoantibodies for diagnostic purposes and at discovering early autoimmune diseases. To overcome a technical limitation from Western blotting, they used the ProteinChip System approach for the analysis of autoantibodies, as well as arrays with biologically activated surfaces that permit antibody capture studies. Protein-A-coupled arrays were incubated with sera of patients (n ¼ 12). In that study, they showed complex on-chip antibody–antigen reactions. At higher molecular weights (>30 kDa), the detection sensitivity of this on-spot method was comparable to conventional Western blotting. With a lower molecular mass, Western blotting is easily exceeded by the on-spot method. Considering that this on-chip procedure is quite easy to use, is much less time-consuming than Western blotting, and is much more sensitive at least in the low molecular weight range, the ProteinChip System is a very promising approach to the screening of autoantibodies in autoimmune diseases. Owing to its versatility, this onchip technology could allow for the large-scale screening for complex autoantibody distributions for diagnostic purposes, and early detection of autoimmune diseases might be possible.

Neurology papers Several papers describe potential biomarkers for the diagnosis of Alzheimer’s disease. The diagnosis of Alzheimer’s disease, the most common form of dementia in the general population, usually relies upon the presence of typical clinical features and structural changes on brain magnetic resonance imaging. The study of Goldstein et al. [13] describes the cytosolic betaamyloid deposition and supranuclear cataracts in lenses from people with Alzheimer’s disease. Pathological hallmarks of Alzheimer’s disease include cerebral betaamyloid (Abeta) deposition, amyloid accumulation, and neuritic plaque formation. They obtained postmortem specimens of eyes and brain from nine individuals with Alzheimer’s disease and eight controls without this disorder, as well as samples of primary aqueous humour from three people without this disorder and who were undergoing cataract surgery. Aqueous humour was analyzed by anti-Abeta arrays with the ProteinChip System. They identified Abeta1-40 and Abeta1-42 in lenses from people with and without Alzheimer’s disease at concentrations comparable with those of the brain, and Abeta1-40 in primary aqueous humour at concentrations comparable with those of the cerebrospinal fluid. In lenses from individuals with Alzheimer’s disease, Abeta accumulated as electron-dense deposits

ARTICLE IN PRESS V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

located exclusively in the cytoplasm of supranuclear/ deep cortical lens fiber cells (n ¼ 4). Another study dealing with Alzheimer’s disease was performed by Beher et al. [4]. This group used a combination of the ProteinChip System and a specific inhibitor of gamma-secretase to investigate whether the production of all amyloid-beta peptide species requires the action of gamma-secretase. Amyloid-beta peptides aberrantly produced by the processing of beta-amyloid precursor protein leads to the formation of characteristic extracellular protein deposits, which are thought to be the cause of Alzheimer’s disease. Beher et al. demonstrated that the production of all truncated amyloid-beta peptides, except those released by the action of the nonamyloidogenic alpha-secretase enzyme or potentially by beta-site betaAPP-cleaving enzyme 2, depends on gamma-secretase activity. Carrette et al. [6] recently reported on the successful search for new biomarkers to detect Alzheimer’s disease. The combination of five polypeptides for the diagnosis of Alzheimer’s disease showed a specificity and sensitivity of 100% and 66%, respectively.

Signal transduction and post-translational modifications, e.g. phosphorylation Signaling systems regulate almost all aspects of cell behavior, e.g. replication and differentiation. Intracellular protein phosphorylation and other post-translational modifications are important for transmitting signals, and each of those proteins involved can be present in many different states, making this an extremely complicated process. As the SELDI ProteinChip platform delivers the exact molecular weights of the molecules present in a given sample, a wide variety of covalent modifications, e.g. phosphorylation can be analyzed (Fig. 5). Mortier et al. [20] identified the putative binding partner for the FAT sequence, when they analyzed protein binding to GST-FAT by SELDI MS. Briefly, GST or GST-FAT were covalently immobilized on a

91

chip and exposed to primary chicken embryo cell lysate, and bound proteins were analyzed by SELDI MS. A 62kD protein was found to bind specifically to GST-FAT. This protein was not detected when GST-FAT was analyzed without prior exposure to CE cell lysate or when the array was coated with GST alone.

Protein interactions—the Interaction Discovery Mappingt Platform The worldwide applied research efforts will soon result in a sharp increase in identified biomarkers. In addition to their direct application in the diagnosis of cancers and other diseases, these markers will be of special interest in basic cancer research and molecular cell biology. Learning more about the interaction partners and physiological functions of marker proteins will be extremely helpful in understanding the regulatory events in cancer, and will pave the way for the detection of new target proteins in therapeutics. With the aim to deliver the appropriate SELDI-based technology in time, new methodological tools have already been developed. Pre-activated affinity beads are coated with bait molecules or antibodies and are used to capture the interaction partners. Binding proteins or antigens are eluted from the bead surface and recaptured onto ProteinChip Arrays for final quantitative SELDI analysis. This combination of bead-based chromatography and SELDI strongly increases sensitivity, as the chromatographic beads exhibit much higher surface areas than the spot surfaces so that even multiplexing of assays is easy to achieve. A wide range of applications, such as disease biomarker interaction discovery, cytokine and betaamyloid peptide detection and quantification [3,13], as well as capture of HIS-tagged proteins have already been successfully tried. In addition, functional assays can be conducted on this platform. The one-step detection of multiple kinase activities may serve as an example. Importantly, tags and labels, which can interfere with subsequent binding events, are not necessary.

Conclusions/future directions

Fig. 5. Method for detecting protein phosphorylation by SELDI. The mass shift of 80 Da is equivalent to the loss or gain of a phosphate group.

In recent years, new technologies have become available that provide opportunities to address the old questions of proteomics with new approaches. It is clear that these new technologies will lead to some new insights. It is also clear that these new technologies are used at an early stage. Despite the exceptional analytical power of the ProteinChip System, systematic limitations of the approach at the present state of technology have

ARTICLE IN PRESS 92

V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

become apparent. First, as with any other analytical technique, not all proteins can be visualized equally well. Whereas the range below 30 kDa is particularly well resolved, sensitivity for higher molecular weight proteins is clearly lower, resulting in a fewer number of signals in the higher range. Second, there are also limitations regarding the dynamic range of proteins that can be displayed. For example the serum proteome is highly dynamic, and protein concentrations can vary by 10 or more orders of magnitude. The presence of higherabundance proteins interferes with the identification and quantification of lower-abundance proteins, a challenge that can be faced by pre-fractionation. The removal or separation of high-abundance proteins by efficient prefractionation enables improved detection of lowerabundance proteins. Third, the ProteinChip technology is directed towards the more proximal goal of investigating protein differential expression (protein patterns). However, it will also be important to identify proteins. By identifying molecular weights and tagging these with protein names, other high-throughput assays (e.g. ELISAs) can be developed. At the current state of technology, the purification of marker proteins (for subsequent digest and identification) can be timeconsuming and might little biochemical background knowledge and experience. Fourth, detection of high intensities of protein peaks of a specific molecular weight may not necessarily mean that low levels of the corresponding protein product will be present because signals can be suppressed by other sample components, thereby hampering quantitation. However, as long as similar biological fluids are compared, this is not an issue for biomarker discovery. It might become a problem when an attempt is made to directly compare the quantity of a protein in, e.g. urine and serum. Fifth, a limit of most profiling experiments lies in the consensus that low-magnitude changes in expression are not meaningful or perhaps immeasurable. On the other hand, this ensures that confirmed biomarkers must have a certain level of intensity difference before recognized as reliable, and will therefore be more trustworthy predictors than those exhibiting lower differences only. However, despite all the limitations, there is no other experimental platform to systematically measure the diverse properties of proteins at a high-throughput level. Undoubtedly, tackling the numerous facets of disease, proteomics requires implementation of multiple strategies and technology platforms. One of the new promising platforms is the ProteinChip System. The potential of this system in the clinical arena seems almost boundless (s. Fig. 6) and carries with it huge expectations, as the proteins correlating with any given disease state can now be readily identified, provided that sample preparation is adequate and the problem is studied on an appropriate scale. The technology offers a helping

Fig. 6. Possible applications of the ProteinChip System during the course of patient treatment.

hand in the systematic identification and characterization of proteins for diagnostic and prognostic markers in tissue, blood serum, and other body fluids. It greatly increases the speed at which targets for pharmaceutical drugs are discovered, and it improves the knowledge of fundamental biological processes. To draw a conclusion, the discovery of multiple protein biomarkers represents a tremendous progress in proteomics. This highthroughput technology, in conjunction with bioinformatic tools, has the capability of not only detecting biomarkers, but also of predicting response outcomes for specific treatments. The ProteinChip System has several advantages for high-throughput screening: it is rapid, versatile, reproducible, and highly sensitive (detection limit in the femtomolar range). Through the use of protein binding sites, the ProteinChip Arrays provide a layer of specificity when one is probing for biomarkers along defined signaling pathways or specific post-translationally modified protein species. Pathological assessment of tissues has remained the diagnostic method for many years. It has become the core of clinical medical practice, providing data for clinical management and a framework for future correlations of new markers and new therapies. In light of the new technologies, could arrays produce a unique fingerprint from biopsies and will patients receive individualized treatment based on expression analysis alone? Many publications dealing with the ProteinChip System indicate that in the future, we might be able to use protein information to obtain a unique fingerprint of every disease state. This would allow for better diagnosis and prognosis of diseases, as well as for individualized treatments, and will empower physicians to prevent and better treat diseases. SELDI-based ProteinChip technology will undoubtedly add to current pathological classifications, but in the near future, there will be coexistence of both technologies.

ARTICLE IN PRESS V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

Acknowledgements We thank K. Stedronsky for critically reading this manuscript.

References [1] B.L. Adam, Y. Qu, J.W. Davis, M.D. Ward, M.A. Clements, L.H. Cazares, O.J. Semmens, P.F. Schellhammer, Y. Yasui, F. Ziding, G. Wright, Serum protein fingerprinting coupled with a pattern-matching algorithm distinguishes prostate cancer from benign prostate hyperplasia and healthy men, Cancer Res. 62 (2002) 3609–3614. [2] N.L. Anderson, N.G. Anderson, Proteome and proteomics: new technologies, new concepts, and new words, Electrophoresis 19 (1998) 1853–1861. [3] H. Bai, R.M.L. Buller, N. Chen, M.D.P. Boyle, Viral binding proteins as antibody surrogates in immunoassays of cytokines, BioTechniques 32 (2002) 160–171. [4] D. Beher, D.J. Wrigley, A.P. Owens, M.S. Shearman, Generation of C-terminally truncated amyloid-beta peptides is dependent on gamma-secretase activity, J. Neurochem. 82 (2002) 563–575. [5] R.F. Bonner, M. Emmert-Buck, K. Cole, T. Pohida, R. Chuaqui, S. Goldstein, L.A. Liotta, Laser capture dissection: molecular analysis of tissue, Science 278 (1997) 1481–1482. [6] O. Carrette, I. Demalte, A. Scherl, O. Yalkinoglu, G. Corthals, P. Burkhard, D.F. Hochstrasser, J.C. Sanchez, A panel of cerebrospinal fluid potential biomarkers for the diagnosis of Alzheimer’s disease, Proteomics 3 (2003) 1486–1494. [7] J. Ching, K.I. Voivodov, T.W. Hutchens, Surface chemistries enabling photoinduced uncoupling/desorption of covalently tethered biomolecules, J. Org. Chem. 61 (1996) 3582–3583. [8] T.O. Dare, H.A. Davies, J.A. Turton, L. Lomas, T.C. Williams, M.J. York, Application of surface-enhanced laser desorption/ionization technology to the detection and identification of urinary parvalbumin-a: a biomarker of compound-induced skeletal muscle toxicity in the rat, Electrophoresis 23 (2002) 3241–3251. [9] D.L. Diamond, Y. Zhang, A. Gaiger, M. Smithgall, T.S. Vedvick, D. Carter, Use of ProteinChip array surface enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF MS) to identify thymosin beta-4, a differentially secreted protein from lymphoblastoid cell lines, J. Am. Soc. Mass Spectrom. 14 (2003) 760–765. [10] M.J. Dunn, Proteomics of plasma biomarkers: a case study of markers of heart transplant rejection, Proteomic Forum Munich 2003 (2003) 82. [11] M.R. Emmert-Buck, R.F. Bonner, P.D. Smith, R.F. Chuaqui, Z. Zhuang, S.R. Goldstein, R.A. Weiss, L.A. Liotta, Laser capture micro-dissection, Science 274 (1996) 998–1001. [12] E.T. Fung, C. Enderwick, ProteinChip clinical proteomics: computational challenges and solutions, Biotechniques (Suppl. Comput. Proteomics) 32 (2002) 34–41.

93

[13] L.E. Goldstein, J.A. Muffat, R.A. Cherny, R.D. Moir, M.H. Ericsson, X. Huang, C. Mavros, J.A. Coccia, K.Y. Faget, K.A. Fitch, C.L. Masters, R.E. Tanzi, L.T. Chylack, A.I. Bush, Cytosolic beta-amyloid deposition and supranuclear cataracts in lenses from people with Alzheimer’s disease, The Lancet 361 (2003) 1258–1265. [14] F.H. Grus, S.C. Joachim, N. Pfeiffer, Analysis of complex autoantibody repertoires by surface-enhanced laser desorption/ionization-time of flight mass spectrometry, Proteomics 3 (2003) 957–961. [15] Q.Y. He, T.T. Yip, M. Li, J.F. Chiu, Proteomic analyses of arsenic-induced cell transformation with SELDI-TOF ProteinChips technology, J. Cell Biochem. 88 (2003) 1–8. [16] T.W. Hutchens, T.T. Yip, New desorption strategies for the mass spectrometric analysis of macromolecules, Rapid. Commun. Mass Spectrom. 7 (1993) 576–580. [17] T. Kodadek, Development of protein-detecting microarrays and related devices, Trends Biochem. Sci. 27 (2002) 295–300. [18] J. Li, Z. Zhang, J. Rosenzweig, Y.Y. Wang, D.W. Chan, Proteomics and bioinformatics approaches for identification of serum biomarkers to detect breast cancer, Clin. Chem. J. 48 (2002) 1296–1304. [19] C. Melle, G. Ernst, B. Schimmel, A. Bleul, S. Koscielny, . A. Wiesner, R. Bogumil, U. Moller, D. Osterloh, K.J. Halbhuber, F. von Eggeling, Biomarker discovery and identification in laser microdissected head and neck squamous cell carcinoma with ProteinChips technology, two-dimensional gel electrophoresis, tandem mass spectrometry, and immunohistochemistry, Mol. Cell Proteomics 2 (2003) 443–452. [20] E. Mortier, F. Cornelissen, C. van Hove, L. Dillen, A. Richardson, The focal adhesion targeting sequence is the major inhibitory moiety of Fak-related non-kinase, Cell Signal. 13 (2001) 901–909. [21] C.P. Paweletz, L.A. Liotta, E.F. Petricoin 3rd, New technologies for biomarker analysis of prostate cancer progression: laser capture microdissection and tissue proteomics, Urology 57 (2001) 160–163. [22] E.F. Petricoin, A.M. Ardekani, B.A. Hitt, P.J. Levine, V.A. Fusaro, S.M. Steinberg, G.B. Mills, C. Simone, D.A. Fishman, E.C. Kohn, L.A. Liotta, Use of proteomic patterns in serum to identify ovarian cancer, The Lancet 359 (2002) 572–577. [23] G. Reid, B.S. Gan, Y.M. She, W. Ens, S. Weinberger, J.C. Howard, Rapid identification of probiotic lactobacillus biosurfactant proteins by ProteinChip tandem mass spectrometry tryptic peptide sequencing, Appl. Environ. Microbiol. 68 (2002) 977–980. [24] M.A. Reymond, J.C. Sanchez, C. Schneider, P. Rohwer, S. Tortola, W. Hohenberger, T. Kirchner, D.F. Hochstrasser, F. Kockerling, Specific sample preparation in colorectal cancer, Electrophoresis 18 (1997) 622–624. [25] V. Seibert, P. von Hoegen, M. Taeger, H.U. Schulz, W. Halangk, U. Kellner, New approaches for molecular diagnostic of pancreatic cancer, Pancreatology 3 (2003) 243. [26] M. Shiwa, Y. Nishimura, R. Wakatabe, A. Fukawa, H. Arikuni, H. Ota, Y. Kato, T. Yamori, Rapid discovery and identification of a tissue-specific tumor biomarker

ARTICLE IN PRESS 94

V. Seibert et al. / Pathology – Research and Practice 200 (2004) 83–94

from 39 human cancer cell lines using the SELDI ProteinChip platform, Biochem. Biophys. Res. Commun. 309 (2003) 18–25. [27] T. Uchida, A. Fukawa, M. Uchida, K. Fujita, K. Saito, Application of a novel protein biochip technology for detection and identification of rheumatoid arthritis biomarkers in synovial fluid, J. Proteome Res. 1 (2002) 495–499. [28] A. Vlahou, P.F. Schellhammer, S. Medrinos, F.I. Kondylis, L. Gong, S. Nasim, G.L. Wright Jr., Development of a novel proteomic approach for the detection of transitional cell carcinoma of the bladder in urine, Am. J. Pathol. 158 (2001) 1491–1501. [29] K.I. Voivodov, J. Ching, T.W. Hutchens, Surface arrays of energy absorbing polymers enabling covalent attachment of biomolecules for subsequent laser-induced uncoupling/desorption, Tetrahedron Lett. 37 (1996) 5669–5672.

[30] S.R. Weinberger, E. Boschetti, P. Santambien, V. Brenac, Surface-enhanced laser desorption-ionization retentate chromatography mass spectrometry (SELDI-RC-MS): a new method for rapid development of process chromatography conditions, J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 782 (2002) 307–316. [31] S.R. Weinberger, E.A. Dalmasso, E.T. Fung, Current achievements using ProteinChips array technology, Curr. Opin. Chem. Biol. 6 (2001) 86–91. [32] M.R. Wilkins, C. Pasquali, R.D. Appel, K. Ou, From proteins to proteomes: large scale protein identification by two-dimensional electrophoresis and amino acid analysis, Biotechnology 14 (1996) 61–65. [33] L. Zhang, W. Yu, T. He, J. Yu, R.E. Caffrey, E.A. Dalmasso, S. Fu, T. Pham, J. Mei, J.J. Ho, W. Zhang, P. Lopez, D.D. Ho, Contribution of human alpha-defensin 1, 2, and 3 to the anti-HIV-1 activity of CD8 antiviral factor, Science 298 (2002) 995–1000.