Label-free quantification in clinical proteomics

Biochimica et Biophysica Acta 1834 (2013) 1581–1590

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbapap

Review

Label-free quantification in clinical proteomics Dominik A. Megger ⁎, Thilo Bracht, Helmut E. Meyer, Barbara Sitek ⁎⁎ Medizinisches Proteom-Center, Ruhr-Universität Bochum, Bochum, Germany

a r t i c l e

i n f o

Article history: Received 16 November 2012 Received in revised form 26 March 2013 Accepted 1 April 2013 Available online 6 April 2013 Keywords: Quantitative proteomics Label-free proteomics Clinical proteomics Disease biomarker

a b s t r a c t Nowadays, proteomic studies no longer focus only on identifying as many proteins as possible in a given sample, but aiming for an accurate quantification of them. Especially in clinical proteomics, the investigation of variable protein expression profiles can yield useful information on pathological pathways or biomarkers and drug targets related to a particular disease. Over the time, many quantitative proteomic approaches have been established allowing researchers in the field of proteomics to refer to a comprehensive toolbox of different methodologies. In this review we will give an overview of different methods of quantitative proteomics with focus on label-free proteomics and its use in clinical proteomics. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Accompanied by rapid technical developments in the field of mass spectrometry, proteomics has evolved into a very powerful bioanalytical platform for answering multidisciplinary scientific questions from medicine, biology, and biochemistry. This widespread applicability of proteomics necessarily implies the need of customized techniques and workflows depending on the scientific question itself, the kind of proteome under investigation (e.g. soluble or membrane proteins, post-translationally modified proteins, protein isoforms) as well as the sample types to be analyzed (e.g. tissue, cultured cells, body fluids, plants, bacteria). To meet all of these demands, a comprehensive repertoire of experimental techniques for isolation, separation, digestion, enrichment, depletion, identification as well as absolute and relative quantification of proteins has been developed over the years and further enhancements are still part of ongoing research. In particular, label-free proteomics has emerged as a high-throughput method for quantitative clinical proteomics studies. In this review we will give an overview about label-free proteomics and its use in the investigation of scientific questions with clinical relevance and a translational intent, widely referred to as clinical proteomics. We will discuss different approaches of label-free proteomics (except MALDI-MS-based strategies like MALDI imaging or quantitative LC-MALDI-MS/MS) in comparison to each other and labeling-based methods in order to shed light on the advantages, disadvantages and limitations of the different techniques. Furthermore, several experimental aspects ranging from sample

preparation to data acquisition will be reviewed. Apart from these, software solutions for data analyses of label-free proteomics experiments and further data interpretation will be presented and selected examples from recent clinical proteomics studies will be discussed. 2. Quantitative proteomics 2.1. 2D gel electrophoresis Since its development almost 40 years ago, the two-dimensional gel electrophoresis is still one of the methods of choice for protein separation and quantification. Using an isoelectric focussing in the first dimension and a separation via SDS-PAGE in the second dimension, thousands of protein spots can be separated, visualized and quantified in a single 2D gel [1,2]. The isolated protein spots of interest are then digested, extracted from the gel and identified via mass spectrometry. Even if the quantification is very accurate and sensitive in this gel-based approach, the relative high amount of protein sample necessary for protein identification as well as multiple experimental steps are the major disadvantages of this technique. Due to these drawbacks and as a consequence of the technical improvements in the fields of chromatography and mass spectrometry, novel mass-spectrometrybased quantification strategies have been developed that allow highthroughput proteome analyses and are complementary to gel-based approaches leading to a higher proteome coverage. 2.2. Labeling-based quantification

⁎ Corresponding author at: Medizinisches Proteom-Center, Ruhr-Universität Bochum, 44801 Bochum, Germany. Tel.: +49 234/32 26119. ⁎⁎ Corresponding author at: Medizinisches Proteom-Center, Ruhr-Universität Bochum, 44801 Bochum, Germany. Tel.: +49 234/32 24362. E-mail addresses: [email protected] (D.A. Megger), [email protected] (B. Sitek). 1570-9639/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbapap.2013.04.001

Over the years, several mass-spectrometry-based quantitative proteomic strategies utilizing different labeling strategies have been published. Most of these techniques rely on the labeling of samples from different conditions with stable isotopes ( 2H, 13C, 15N, 18O) and a

1582

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

following quantitative analysis in a mass spectrometer. The introduction of the isotopic label can be performed by metabolic, chemical or enzymatic labeling. Approaches utilizing stable-isotope labeling are: stable isotope labeling by amino acids in cell culture (SILAC) [3], stable isotope labeling of mammals (SILAM) [4], isotope-coded affinity tags (ICAT) [5], isotope-coded protein labeling (ICPL) [6], isobaric tags for relative and absolute quantification (iTRAQ) [7], tandem mass tags (TMT) [8,9], isobaric peptide termini labeling (IPTL) [10,11], dimethyl labeling [12] as well as several variants of these techniques. Apart from SILAC and SILAM which are introduced by metabolic labeling, the above-mentioned quantification strategies are directly applicable for proteomic studies of clinical samples. However, to overcome the limitation of SILAC to cell culture models a Super-SILAC approach has been developed. Here, quantitative changes of the proteome in different clinical samples (e.g. tumor tissue samples) can be determined by the comparison to an internal standard consisting of an isotopically-labeled pool of cancer cell lines [13]. Aside from the labeling with stable isotopes, a labeling strategy based on the attachment of a metal complex to peptides or proteins has been reported on. This approach is known as metal-coded affinity tag labeling (MeCAT) and enables absolute quantification with high sensitivity and a wide linear dynamic range via inductively coupled plasma mass spectrometry (ICP-MS) [14,15]. However, its potential applicability for clinical samples has not been tested yet. 2.3. Label-free quantification Approaches of label-free quantitative proteomics can be divided into two different quantification strategies that are briefly described in the following. A schematic representation of both approaches is shown in Fig. 1. The first approach is termed spectral counting and implies a counting and a comparison of the number of fragment-ion spectra (MS/MS) acquired for peptides of a given protein. Due to the empirical observation that the number of tandem mass spectra of a particular peptide increases with an increasing amount of the corresponding protein, a relative quantification of proteins between different samples is possible [16]. However, as in this method the quantification relies on a simple counting of acquired spectra rather than on measuring physical data, the spectral counting method is controversial [17]. Nevertheless, spectral counting is widely used and was further developed over the years. For example, modified approaches of spectral counting have been reported that take into account aspects influencing the number of spectral counts, like physicochemical properties of peptides as well as the lengths of the corresponding proteins. These approaches are known as absolute protein expression (APEX) [18] and normalized spectral abundance factor (NSAF) [19,20]. More recently, normalized spectral index (SIN) was introduced which combines three MS abundance features, namely peptide count, spectral count and fragmention intensity. This approach has shown to eliminate variances between replicate measurements and allows quantitative reproducibility and significant quantification in replicate MS measurements [21]. For more detailed methodological reviews of spectral counting, see [22,23]. The second approach of label-free quantitative proteomics implies the measurement of chromatographic peak areas (also termed mass spectrometric signal intensities) of peptide precursor ions. Depending on the chromatographic method (e.g. reversed-phase liquid chromatography) the peptides are separated according to their particular physical properties (e.g. hydrophobicity, charge), subsequently ionized in an ion source and finally detected in a mass spectrometer. In the acquired mass spectrum each peptide of a particular charge and mass generates one mono-isotopic mass peak. The intensity of this peak as a function of the retention time can be visualized in an extracted ion chromatogram (XIC) and the area under the curve (AUC) can be determined. The areas of chromatographic peaks have been shown to correlate linearly in a wide range with the protein abundance which makes their

measurement feasible for quantitative studies [24,25]. At a first glance, this approach looks straightforward and very convenient, but to obtain reliable results several experimental and technical aspects have to be considered (see: Section 3). Furthermore, raw LC–MS data generated in the experiments have to be post-processed (e.g. feature detection, alignment of retention times, normalization of MS intensities, peak picking, noise reduction) in the course of a quantitative analysis (see: Section 4). 2.4. Labeling-based versus label-free quantification The first question arising prior to a quantitative proteome analysis refers to the quantification method itself. In principle, it is beneficial to use more than one technique for quantification, as the complementarity of various approaches implies a greater proteome coverage if they are used in combination. Apart from this aspect, different approaches have their particular advantages and limitations. For example, a clear advantage of labeling-based strategies over label-free approaches is the possibility of a multiplexed analysis, allowing the simultaneous measurement of differentially labeled samples in a single experiment. In particular, multiplexing capacities of 2-plex, 4-plex and 8-plex can be achieved with commercially available iTRAQ and TMT reagents. However, one should keep in mind that different quantification techniques imply variable requirements to the sample type and amount as well as the mass spectrometer used for the analysis. As mentioned before, metabolic labeling strategies are for example not applicable for proteome analysis of clinical samples and therefore limited with respect to the sample type. Chemical labeling strategies like iTRAQ or TMT on the other hand have special requirements concerning ion trap mass spectrometers. Contrary to label-free approaches, iTRAQ or TMT requires alternative fragmentation methods like HCD (higher-energy collisional dissociation) or ETD (electron transfer dissociation) that should be used instead of CID (collision induced dissociation), which is not compatible with the low mass range reporter ions. For label-free analysis via peptide ion intensities a high-resolution mass spectrometer is recommended, because the mass of the precursor ions needs to be determined very accurately. Contrarily, label-free analysis by spectral counting can also be performed on low-resolution mass spectrometers and was shown to give more accurate quantitative results than the ion-intensity-based approach in such a case [21]. To assess the performance of variable quantification approaches several comparative studies were carried out. A comprehensive study, for example, was performed in 2006 by the Association of Biomolecular Resource Facilities aiming the quantification of eight known proteins in different sample mixtures. The methods used in this study included gel-based approaches as well as MS-based techniques, either labelfree or labeling-based. In case of label-free methods, the monitored protein ratios were close to the expected values, especially for the protein with the lowest abundance in the investigated mixture [26]. More recent comparative studies were published by Li et al. as well as Merl et al. [27,28]. In the former study, a comprehensive systematic comparison of label-free quantification based on spectral counting with SILAC, iTRAQ and TMT was performed. The authors were able to show that among these techniques the label-free approach has the largest dynamic range and the highest proteome coverage for identification. However, the quantification accuracy and reproducibility are worse in comparison to the investigated labeling-based strategies [27]. In the latter study, the authors used a combination of label-free quantification based on peptide ion intensities and SILAC for proteomic profiling of primary retinal Müller cells. Here, a significant complementarity concerning quantified and identified proteins was shown. In agreement with earlier studies, the labelfree approach was also found to yield a higher proteome coverage. Within the label-free approach itself significant differences were also monitored depending on the software used for the data analysis [28]. In the literature, many more studies can be found that focused

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

1583

Fig. 1. Schematic representation of the two different label-free approaches. In the spectral-counting approach (left) peptide and protein abundances can be estimated based on the number of acquired peptide spectrum matches. In the ion-intensity-based approach (right) the changes of peptide abundances are determined by measuring and comparing the chromatographic peak areas of the corresponding peptides. In this example, both results would indicate a higher peptide abundance in State A.

on the comparison of different mass-spectrometry-based quantification strategies [29–34]. 3. Experimental aspects 3.1. Sample preparation, purification and separation The robustness and reproducibility of sample preparation are two of the most important aspects for a successful quantitative proteome analysis, especially for label-free quantification where each sample is handled separately starting from sample acquisition to the final measurement. Hence, every step within the experimental pipeline is a potential source of errors and can introduce several biases that might produce misleading results. An overview of a typical proteomics pipeline including the different experimental steps that generally can influence the outcome of a proteome analysis is shown in Fig. 2. The first step of a quantitative proteome analysis of clinical samples that determines the ongoing workflow is the choice of a sample type and its acquisition. Principally, clinical proteomics studies can be performed with numerous materials ranging from different body fluids (e.g. serum, plasma, urine, bile) to cell lines or isolated cell types and tissue samples, but specific workflows for different sample types need to be followed. In the case of tissue-based quantitative proteomics, microdissection of cells or tissue regions of interest might be necessary, depending on the heterogeneity of the investigated tissue. It has been shown that microdissection can resolve the problems of sample heterogeneity and contaminations in tissue samples [35,36]. Over the years, several proteomic workflows including microdissection have been established and published. Optimized workflows for quantitative 2D-DIGE analyses of microdissected cells derived from a broad variety of different tissue specimen have been developed and successfully applied in several biomarker discovery studies [37–45]. Concerning mass-spectrometry-based quantitative proteomics, Umar et al. reported on a proteomics pipeline for biomarker discovery of laser capture microdissected breast cancer tissue, very recently [46]. At the same time, this research group systematically evaluated the impact of several sample preparation steps on the results of a proteomic study of the same tissue type [47]. Apart from these two studies, many others are reported in literature that focused on the improvement of sample

preparation steps for quantitative proteomics analyses of diverse clinical samples. The subsequent steps, namely cell lysis, protein isolation and digestion, can also significantly influence the results of a proteomic study. Different cell lysis buffers and conditions need to be chosen depending on the part of proteome under investigation (e.g. membrane proteins, organelle proteins, cytosolic proteins) or the sample type itself. Many examples of fine-tuned workflows for protein isolation from different sample types and cellular compartments can be found in the literature. For example, Bergquist et al. evaluated the applicability and performance of different lysis buffers for the extraction of plasma membrane proteins from mouse brain tissue, very recently [48]. Earlier, Mann et al. reported a method known as filter-aided sample preparation (FASP) that allows protein extraction from formalin-fixed tissue samples [49]. In addition to lysis protocols, digestion protocols have to be carefully chosen as well. Even if the proteolytic digestion with trypsin or other proteases is the most-widely used method in proteomics, digestion methods relying on chemical or physical methods might be useful and necessary in special cases. Concerning proteolysis, one has furthermore to decide whether to perform an in-gel or in-solution digestion. If detergents that might influence the performance of the liquid chromatography were used during cell lysis, the former one should be used, as it implies an intrinsic purification step of the proteins and peptides in a gel. However, this method also implies a significant sample loss due to an incomplete extraction of the peptides from a gel piece. Alternatively, spin columns are increasingly used for sample purification [50] and corresponding on-filter digestion protocols are available as well [49]. For minute sample amounts in-solution digestion should be used. In this regard, improved in-solution digestion protocols for low sample amounts were established by Vékey and co-workers [51]. Comparative studies of different digestion methods can be found in literature as well. Fenselau et al., for example, presented a comparative study of different digestion methods for plasma membrane proteins and found in-gel digestion to be the most advantageous one for this particular type of proteins [52]. Apart from the abovementioned steps, the determination of protein concentration in a given sample is a very important experimental aspect, especially in the case of label-free quantification. A determination directly after protein extraction from a sample is required to confirm the successful isolation of proteins and to adjust the amount of protease

1584

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

Fig. 2. Generalized workflow of a label-free quantitative proteome analysis. The workflow contains experimental and data analysis steps that can influence the outcome of a label-free quantitative proteomic study and potentially can introduce several biases.

used in the following digestion step. However, if in-gel digestion is used, an additional determination of the peptide concentration after the extraction from the gel is beneficial, as a variable extraction efficiency might lead to artificial peptide regulations in the subsequent labelfree analysis. For the determination of protein concentration several widely used colorimetric methods like Bradford [53], bicinchoninic acid (BCA) [54], biuret, Lowry [55] or Popov assay [56] are available. Alternatively, protein concentration can be determined by amino acid analysis very accurately [57]. Each of the listed methods has its particular advantages and disadvantages and should be chosen with care. Some aspects that should be considered for the selection of an appropriate assay are: working range and detection limit of the assay, tolerance of detergents, chaotropic or reducing agents and the question whether peptides or proteins should be investigated. In the general workflow of a MS-based proteome analysis, peptides obtained from a tryptic digestion are separated via reversed-phase liquid chromatography (RP-LC) prior to the mass-spectrometric analysis. For label-free quantitative analysis by peptide ion intensities, the reproducibility of this step is extremely crucial, as the measured peptide precursor masses are matched to their corresponding retention times. To overcome the technical variance between several chromatographic runs within a label-free study algorithms for chromatographic alignment have been developed and implemented into several software packages (see: Section 4). To ensure chromatographic reproducibility within a label-free study the use of an internal standard consisting of stable-isotope coded peptides is advantageous. The retention times of these standard peptides can be used to monitor the LC–MS performance over a period of time and can act as landmarks for the chromatographic alignment during the following data analysis. In this context, Sickmann et al. presented an internal standard of synthetic stable-isotope coded

peptides that can be used to evaluate several parameters on LC and MS level [58]. If desired or necessary, samples can be pre-fractionated (e.g. via 1D-PAGE, isoelectric focussing, ion exchange chromatography, high-pH reversed-phase liquid chromatography) prior to RP-LC and each fraction can then be analyzed in a single LC–MS/MS experiment. A clear advantage of such a two-dimensional setup is a decomplexation of the sample leading to a higher proteome coverage as well as the possibility to select particular fractions of interest for the subsequent analysis [59–62]. However, to obtain reliable quantitative results and to avoid any bias each step of a pre-fractionation has to be carried out in a highly-reproducible manner. Furthermore, one should keep in mind that the number of LC–MS/MS experiments significantly increases with the number of fractions previously separated. 3.2. Mass spectrometry Electrospray ionization (ESI) and matrix-assisted laser desorption/ ionization (MALDI) are the ionization methods of choice for the mass-spectrometric investigation of biological macromolecules. An ESI source can directly be coupled to a liquid chromatography set-up and therefore can be perfectly integrated into an automated proteomic workflow. Hence, it is not surprising that ESI has become the most widely used ionization method in MS-based proteomics applications. Nevertheless, MALDI-MS has also its particular advantageous features regarding dynamic range, sensitivity or stability of ionization. Furthermore, the use of MALDI-MS as an imaging technology is an emerging and promising approach in the field of clinical proteomics. As a detailed explanation of MALDI-MS-based quantitative proteomics would exceed the scope of this review, we would like to refer to in-depth reviews at

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

this point for further reading [63–65]. Apart from different ionization methods, instruments with various high-resolution mass analyzer types can be used for label-free proteomics analyses including time-of-flight (TOF), Fourier transform ion cyclotron resonance (FT-ICR) or Orbitrap analyzers [66,67]. Depending on the mass spectrometer used for a label-free analysis, different acquisition modes can be used. Generally, mass spectrometric analyses for label-free quantification can be performed in two different manners, namely a data-dependent acquisition (DDA) and a data-independent acquisition (DIA). The former one includes the acquisition of a survey scan and a subsequent fragmentation of selected precursor peptide ions. Here, one should keep in mind that the right balance between acquired survey and fragment ion spectra needs to be found. For example, an increased number of acquired fragment ion spectra leads to a higher proteome coverage, but concomitantly increases the cycle time of the whole MS/MS acquisition. This in turn results in less acquired MS survey spectra which are needed to describe the chromatographic peptide ion peak for quantification. Hence, mass spectrometric parameters like the number of acquired fragment ion spectra, MS acquisition time, fragmentation method and many more have to be carefully adjusted and optimized with respect to the chromatographic behavior of the sample in order to produce reliable and reproducible quantitative results. This type of acquisition allows quantification via spectral counting as well as peptide ion intensity measurement. In case of DIA, the mass spectrometer constantly operates in an MS/MS mode. In contrast to DDA, no precursor-ion selection is performed during an MS survey scan. Instead, mass data are acquired by alternating the collision energy between a low and an elevated energy state. This method is termed LC–MSE (E for elevated) and shows increased signal to noise ratios and can lead to better identification rates of low-abundant peptides that might be missed during a data-dependent analysis. 4. Data analysis Bioinformatic and biostatistical tools have become indispensable to handle and interpret the vast amount of data generated in a single proteomic analysis. Therefore, a variety of commercial and open-source software solutions for the analyses of different proteomics data types (e.g. 2D gels, LC–MS/MS data sets or data from different quantification techniques) and subsequent statistical tests have been developed. 4.1. Protein identification For the identification of proteins in a given analyzed sample, theoretical peptide spectra of proteins deposited in a protein database are matched to the acquired tandem mass spectra. For this purpose, several protein databases as well as search engines with different scoring algorithms are available. Widely used search engines are MASCOT [68], SEQUEST [69], OMSSA [70] and X!TANDEM [71]. To minimize the false discovery rate (FDR) of the identifications, decoy strategies that imply a search against the true database set and a consciously incorrect one, containing reversed or randomly shuffled peptide sequences, are mostly followed [72]. Using this approach one is able to set a score threshold that corresponds to a desired FDR. 4.2. Software packages for quantitative proteomics Detailed comparisons of different software packages for label-free quantitative proteomics have been reported and reviewed earlier [23,73,74]. General differences between these are the quantification method itself (spectral counting and/or measurement of peptide ion intensity), statistical tests used by the software and the supported mass spectrometric data types. In case of computer programs for quantification via peptide ion intensities, chromatographic alignment, normalization and peak detection algorithms are further important differences that have already been discussed in previous in-depth

1585

reviews [75,76]. Examples for widely used commercial and opensource software packages for label-free quantification are summarized in Table 1. 4.3. Data interpretation Data sets generated in quantitative proteomics experiments contain protein identifications and corresponding regulation factors between at least two experimental groups. However, to extract the biological relevance from the vast amount of proteomics data, the proteins identified and quantified in the experiments need to be functionally annotated and mapped into biological processes. Furthermore, pathway analysis, construction of interaction networks, and final visualization of the data can be useful. Especially for clinical biomarker discovery studies the integration of clinico-pathological data is crucial and it is of interest whether the observed protein regulations have been reported earlier in context to a particular disease. To meet all these demand, several computer programs and databases have been developed over the years. Several software packages easing the interpretation of proteomics data in a biological context were recently reviewed by Malik et al. [80]. 5. Verification and validation In quantitative proteomics experiments with clinical samples mostly small and well-characterized cohorts are investigated to discover protein alteration related to a particular disease. In following methodological verification experiments with the same sample set, the used quantitative approach itself can be verified by alternative, orthogonal techniques. These can include mass-spectrometry-based methods using labeling techniques (for examples of techniques: see Section 2.2) or targeted approaches like selected reaction monitoring (SRM) also referred to as multiple reaction monitoring (MRM) [81,82]. Immunological approaches like Western blots, immunohistochemistry or ELISA are also commonly used. In following validation experiments, biomarker candidates need to be further validated by investigating a large patient cohort that is independent from the sample set used in the discovery and verification phase. Whereas immunological methods like tissue micro arrays or ELISA represent the traditional way of validation, targeted MS-based approaches like MRM are emerging as additional alternatives. Selected examples of recent clinical proteomics studies utilizing label-free approaches during the discovery phase and orthogonal MS-based or immunological methods during the verification/validation phase are presented in the following section. 6. Label-free clinical proteomics studies Since label-free proteomics was introduced, a broad variety of studies aiming for discovery of biomarkers or drug targets have been performed using label-free approaches. Here only a few recent examples are described, illustrating the broad range of different sample types and sample treatments used as well as the variety of clinical questions addressed with label-free proteomics (Table 2). We limited the

Table 1 Selected examples of software packages for label-free quantitative proteomics. Software

MS-level of quantification

Freeware

URL/reference

DeCyder MS ProteinLynx Global Server Progenesis LC–MS Scaffold SIEVE MaxQuant SuperHirn MSight

MS MS

– –

www.gelifesciences.com www.waters.com

MS MS/MS MS MS MS MS

– – – Yes Yes Yes

www.nonlinear.com www.proteomesoftware.com www.thermo.com [77] [78] [79]

1586

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

Table 2 Overview of recent clinical proteomics studies utilizing label-free quantitative approaches. Disease

Sample type

No. of patients/samples

Enrichment/ fractionation

Non-small cell lung cancer Colon cancer

Frozen tissue

15 patients

Frozen tissue

Colon cancer

Frozen tissue

Hepatocellular carcinoma Inflammatory dilated cardiomyopathy Breast cancer

Frozen tissue

6 patients, 6 1D-PAGE-gel slices each 13 patients pooled in 2 samples, 10 1D-PAGE-gel slices each 7 patients

Biopsies

17 patients

PIMAC, LC–MS/MS LTQ-Orbitrap XL 1D-PAGE, LTQ-Orbitrap LC–MS/MS XL Insoluble fraction, QTOF E Premier 1D-PAGE, LC–MS LC–MS/MS LTQ Orbitrap Velos LC–MS/MS LTQ-FT

Cells from LCM tissue

10 patients

LC–MS/MS

Breast cancer (invasive ductal carcinoma) Colon cancer

Cells from LCM tissue

10 sample pairs

LC–MS/MS

3 sample pairs, 6 SAX fractions each

SAX, LC–MS/MS

8 sample pairs + 7 additional samples, 6 SAX factions each 15 patients, 3 samples each

SAX, LC–MS/MS

LTQ-Orbitrap MaxQuant Velos

Breast cancer

Cells from LCM tissue (formalin-fixed and paraffin-embed) Cells from LCM tissue (formalin-fixed and paraffin-embed) Serum

LTQ-FT

Chronic hepatitis C

Serum

96 patients

Isolation of N-Glycopeptides, LC–MS/MS Immunodepletion, LC–MS/MS CE-MS/MS

Colorectal cancer

Cholangiocarcinoma Urine Cholangiocarcinoma Bile Prostate cancer Head and neck cancer Colon cancer

Expressed prostatic secretion Cell lines

Primary cells

148 patients (42 in training set, 106 in validation set) 107 patients (50 in training set, 57 in validation set) 16 patients 34 cell lines, 2 replicates each

3 clones, colonospheres and differentiated cells each, 10 1D-PAGE-gel slices each

CE-MS/MS SCX, LC–MS/MS Affinity purification, LC–MS/MS 1D-PAGE, LC–MS/MS

Mass analyzer

Software

Verification/ validation approach

Reference

SIEVE

WB, IHC

[83]

Progenesis LC–MS Expression

TMA

[84]

WB, IHC

[85]

WB, IHC

[86]

IHC, Comparison to Transcriptomics data

[87]

–

[46]

WB, IHC

[88]

Progenesis LC–MS Rosetta Elucidator

LTQ-Orbitrap MaxQuant XL QTOF Ultima In-house version of MatchRX LTQ-Orbitrap MaxQuant Velos

Not specified

[89] Identification of known colon cancer markers IHC [90]

WB

[91]

Rosetta Elucidator MosaiquesVisu, MosaCluster Micro-TOF MosaiquesVisu, MosaCluster LTQ-Orbitrap Qspec XL LTQ-Orbitrap Progenesis XL LC–MS

Meta-protein based regression model Peptide marker model Peptide marker model WB, SRM-MS

[92]

Functional studies, WB, IHC

[96]

LTQ-FT

Functional studies, WB

[97]

QTOF Premier Micro-TOF

Not specified

[93] [94] [95]

Abbreviations used in this table: LCM: laser capture microdissected, SAX: strong anion exchange, SCX: strong cation exchange, PIMAC: parallel immobilized metal affinity chromatography, CE: capillary electrophoresis, IHC: immunohistochemistry, WB: Western blot, TMA: tissue micro array.

presented studies to those using samples of human origin, thereby not mentioning those using animal models for clinical proteomics studies. As the availability of appropriate sample material is always a bottleneck in clinical proteomics, fresh frozen tissue gained from resections is a material frequently used in comparative studies. Using resected tissue one has to consider that often an experienced pathologist is needed to examine and evaluate the properties of the samples (e.g. discrimination of tumor tissue and non-tumorous tissue, identification of necrotic or inflammatory areas, tumor grading and staging). In the following a variety of studies utilizing resected tissue samples is presented. Recently a multidisciplinary group from Madrid described two novel biomarker candidates for non-small cell lung cancer revealed by a label-free proteomics study. Samples of lung adenocarcinoma and non-squamous cell carcinoma were compared with normal lung tissue samples. An immobilized metal affinity chromatography based phosphopeptide enrichment strategy was applied in combination with LC–MS/MS leading to the identification of two significantly differentially expressed proteins. PTRF/Cavin-1 and MIF were subsequently verified by Western blot analysis and immunohistochemistry [83]. Meding and coworkers described a label-free approach combined with a MALDI imaging approach to study lymph node metastasis in colon cancer. They were able to reveal a panel of proteins associated with lymph node metastasis with respect to the nodal status of the patients. The group described three new biomarker candidates and subsequently verified them in an independent patient cohort using immunohistochemistry [84]. Yang et al. performed a comparative

label-free analysis of colorectal cancer patients focusing on cytoskeletal proteins and their interaction partners as well as extracellular matrix proteins. The authors specifically investigated the sample fraction insoluble in aqueous solution. They found 56 proteins to be differentially expressed between tissue samples from healthy individuals and tumor-samples. Five potential biomarker candidates and/or drug targets were further verified in an independent cohort using Western blot analysis and immunohistochemistry [85]. Very recently, our own group described a label-free study combined with a 2D-DIGE approach, investigating hepatocellular carcinoma (HCC). We compared HCC-tumor samples from seven patients with non-tumorous liver tissue and were able to identify 573 regulated proteins, 476 of them identified in the label-free approach. Six candidate proteins were chosen and verified with immunological methods using the same sample set as used for detection. Subsequently the candidates were tested in an independent patient cohort of 33 samples (16 HCC tissue samples and 17 non-tumorous liver tissue samples). Two proteins were found to be significantly differentially expressed and therefore represent new biomarker candidates for HCC [86]. As many proteomic studies focus on neoplastic diseases it is important to note that also various other clinical questions have been addressed with label-free proteomic studies. Hammer and coworkers investigated endomyocardial biopsies to identify proteins related to inflammatory dilated cardiomyopathy in a comparative study. The group found 174 proteins differentially expressed between patients and control samples from patients with normal heart function and histology and verified

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

them by immunohistochemistry and comparison with transcriptomics data [87]. Since tissues and also tumors consist of various cell types and compounds of the extracellular matrix, laser capture microdissection is often used to isolate the appropriate cell types of interest. By doing so, bias from irrelevant sample components can be ruled out and analysis of specific sample parts is enabled. Again one has to be aware that often pathological experience may be needed for selection and microdissection of the specific sample compound under investigation. Liu and coworkers established a biomarker discovery pipeline for breast cancer based on the analysis of laser capture microdissected epithelial cells. The group described a workflow enabling the identification of about 1800 proteins in samples containing an average of 4000 cells on routine basis. The robustness and reproducibility of the workflow were extensively tested and demonstrated by the analysis of downstream regulated proteins of estrogen receptor α (ER) in ER+ compared to ER− breast tumors [46]. Hill et al. designed a study to investigate tumor vascularization in breast cancer. The group isolated endothelial cells from tumor blood vessels to compare them with blood vessels from non-tumorous tissue in a differential label-free study. The authors described 86 proteins to be over-expressed in tumor vessels, among them several known tumor markers. 40 proteins were found to be under-expressed, including basement-membrane proteins which are also known to show aberrant abundance in tumor vessels. To verify the results eight proteins were further studied by immunohistochemistry and Western blot analysis [88]. Although most described tissue-based studies utilize fresh frozen tissue samples for label-free analysis and frozen tissue remains the gold standard for proteomic studies, the usage of formalin-fixed samples for label-free analysis has also been reported. Since Ostasiewicz and coworkers described a method allowing the proteomic analysis of formalin-fixed and paraffin-embedded (FFPE) tissue the group also reported label-free studies based on FFPE samples. After optimizing the analysis of small numbers of laser capture microdissected colorectal cancer cells and thereby showing the feasibility of such studies [89] a proteomic study of archival FFPE tissue samples was reported very recently. The authors compared laser capture microdissected cells from normal mucosa, primary carcinoma and nodal metastases and demonstrated the relative quantification of 7576 proteins. Expression levels of 1808 proteins were found to be significantly differential between normal and cancer tissue and three potential biomarkers were verified by immunohistochemistry [90]. Albeit the usage of formalin-fixed and paraffin-embedded samples in clinical studies is possible, the benefit for discovery of biomarkers or drug targets remains to be demonstrated. One advantage could be the availability of extensive patient follow up data in samples from historical collections. If no such benefit derives from usage of FFPE samples, utilizing fresh frozen tissue appears to be more obvious, as any bias from fixation can be ruled out. Human serum or plasma samples are generally difficult to analyze because the protein abundances of the plasma proteome span more than ten orders of magnitude and are dominated by a small number of typical plasma proteins. However, serum and plasma samples are easily available and therefore very attractive clinical samples that also have been used in label-free studies. Hyung et al. reported a differential label-free proteomics study using human serum samples. They fractionated the serum samples by isolation of N-glycosylated proteins using hydrazide chemistry thereby avoiding depletion of high abundant proteins prior to analysis. The group studied serum samples of breast cancer patients comparing responders and non-responders to neoadjuvant chemotherapy and was able to identify thirteen novel biomarker candidates. Six selected biomarker candidates were tested in an independent set of patients and allowed prediction of responders and non-responders [91]. Patel et al. described a study in which serum samples of patients with chronic Hepatitis C infection were used in a label-free study to identify potential prognostic markers that allow

1587

discrimination of responders and non-responders to standard-of-care therapy. Samples were immunodepleted for 14 of the most abundant serum proteins prior to analysis by LC–MS/MS. Sera of 55 patients were used in a training set and 41 in an independent validation cohort. Based on 105 identified proteins of interest the authors were able to develop a regression model which allowed them to predict treatment response in the validation set [92]. Although immunodepletion is widely used, one should bring to mind the disadvantages of the procedure when working with serum or plasma samples. Immunodepletion is hardly reproducible and very likely to cause bias during sample preparation. The same holds true for blood coagulation, but moreover several proteases are involved in blood coagulation as well. For this reason immunodepletion should be avoided if ever possible and plasma is the sample of choice when blood samples are investigated using quantitative proteomics studies. Like serum or plasma, urine is also a source of samples which are easily available and can be obtained non-invasively. Metzger and coworkers investigated urine samples of 42 patients possessing cholangiocarcinoma (CC), primary sclerosing cholangitis (PSC) or benign biliary disorder (BBD). The group established a peptide marker model based on ion intensity and the capillary electrophoresis migration time. The model was applied to an independent test cohort of 106 patients where it could discriminate patients with CC from patients with PSC or BBD with 83% sensitivity and 79% specificity [93]. Earlier, the same group already reported a similar approach utilizing bile for proteomic analysis, a sample that can be obtained less invasive than tissue samples during endoscopic examination of patients. A training set of 50 patients with CC, PSC or choledocholithiasis was used to identify disease-specific peptide-patterns. The model was subsequently tested with an independent cohort of 57 patients where it distinguished CC from PSC and choledocholithiasis with 84% sensitivity and 78% specificity [94]. Very recently, a label-free approach utilizing expressed prostatic secretions (EPS) as source of samples was published by Kim et al. Samples from individuals with extracapsular or organ-confined prostate cancer were compared in order to identify prognostic markers which allow discrimination of indolent tumors from aggressive ones. Direct-EPS samples were used for the label-free study in which 133 differentially expressed proteins could be identified. Verification of biomarker candidates was carried out with EPS-urine from an independent patient cohort. 7 selected biomarker candidates were verified by Western blotting and relative quantification via SRM-MS [95]. Cell culture models present an ideal way to study biological questions in a well defined system with reduced biological variation. On the other hand this also means a deficient representation of the complex biological variety one has to consider when addressing questions of clinical relevance. However, several label-free studies utilizing cell culture models to address such questions have been reported. Wu et al. were able to reveal new potential drug targets for squamous head and neck cancer by studying 34 cell lines in a kinase centric label-free approach. They performed affinity purification via kinobeads to enrich kinases which were subsequently quantified in a label-free ion-intensity-based fashion. 146 kinases were quantified among which 42 kinases were shown to be significantly differentially expressed between different cell lines. Loss of function experiments were used for validation of several kinases as potential drug targets for the treatment of squamous head and neck cancer [96]. Van Houdt and coworkers reported a study in which colon cancer cells derived from resections of liver metastasis were cultivated in colonosphere cultures. Cancer stem cell enriched colonospheres were compared with differentiated cells from the same clone in order to identify cancer stem cell specific features. A total of 3048 proteins were identified among which 32 were found to be significantly up-regulated in cancer stem cells. One of the over-expressed proteins was BIRC6, an inhibitor of apoptosis protein, which was further investigated in a functional manner. Knock down of BIRC6 restored sensitivity of cancer stem cells to two tested chemotherapeutic drugs which marks BIRC6 as a potential drug target in treatment of colon carcinoma [97].

1588

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

7. Conclusion In this review we gave an overview of label-free quantitative proteomics ranging from the definition of different approaches over experimental aspects to the final data analysis. The applicability of label-free proteomic approaches in clinical proteomics was depicted by highlighting very recent examples of quantitative proteomic studies with clinical samples or disease-related questions. In general, we conclude that label-free proteomics is a versatile tool to estimate changes of protein abundances between different samples. Like any other quantification method, label-free approaches have also disadvantages. For example, an excellent reproducibility of every experimental step starting from the sample acquisition to LC–MS/MS analysis is required for reliable quantification. This demand of reproducibility during the whole proteomic workflow still poses one of the greatest challenges in label-free proteomics. With regard to clinical proteomics, label-free proteomics turned out to be a promising strategy to detect disease-related protein regulations. However, especially in these types of biomarker studies, the use of microdissected samples is highly recommended and the regulation patterns detected in a label-free quantitative proteomics approach necessarily need to be validated in large and independent patient cohorts with orthogonal techniques. Acknowledgements D. A. Megger and T. Bracht acknowledge the financial support from the PROFILE-consortium Ruhr. The project has been selected under the operational program co-financed by the European Regional Development Fund (ERDF) Objective 2 “Regional Competitiveness and Employment” 2007–2013, North Rhine-Westphalia (Germany). A part of this study was funded from P.U.R.E. (Protein Unit for Research in Europe), a project of North Rhine-Westphalia as well. References [1] J. Klose, Protein mapping by combined isoelectric focusing and electrophoresis of mouse tissues. A novel approach to testing for induced point mutations in mammals, Humangenetik 26 (1975) 231–243. [2] P.H. O'Farrell, High resolution two-dimensional electrophoresis of proteins, J. Biol. Chem. 250 (1975) 4007–4021. [3] S.E. Ong, B. Blagoev, I. Kratchmarova, D.B. Kristensen, H. Steen, A. Pandey, M. Mann, Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics, Mol. Cell. Proteomics 1 (2002) 376–386. [4] C.C. Wu, M.J. MacCoss, K.E. Howell, D.E. Matthews, J.R. Yates 3rd, Metabolic labeling of mammalian organisms with stable isotopes for quantitative proteomic analysis, Anal. Chem. 76 (2004) 4951–4959. [5] S.P. Gygi, B. Rist, S.A. Gerber, F. Turecek, M.H. Gelb, R. Aebersold, Quantitative analysis of complex protein mixtures using isotope-coded affinity tags, Nat. Biotechnol. 17 (1999) 994–999. [6] A. Schmidt, J. Kellermann, F. Lottspeich, A novel strategy for quantitative proteomics using isotope-coded protein labels, Proteomics 5 (2005) 4–15. [7] P.L. Ross, Y.N. Huang, J.N. Marchese, B. Williamson, K. Parker, S. Hattan, N. Khainovski, S. Pillai, S. Dey, S. Daniels, S. Purkayastha, P. Juhasz, S. Martin, M. Bartlet-Jones, F. He, A. Jacobson, D.J. Pappin, Multiplexed protein quantitation in Saccharomyces cerevisiae using amine-reactive isobaric tagging reagents, Mol. Cell. Proteomics 3 (2004) 1154–1169. [8] A. Thompson, J. Schafer, K. Kuhn, S. Kienle, J. Schwarz, G. Schmidt, T. Neumann, R. Johnstone, A.K. Mohammed, C. Hamon, Tandem mass tags: a novel quantification strategy for comparative analysis of complex protein mixtures by MS/MS, Anal. Chem. 75 (2003) 1895–1904. [9] L. Dayon, A. Hainard, V. Licker, N. Turck, K. Kuhn, D.F. Hochstrasser, P.R. Burkhard, J.C. Sanchez, Relative quantification of proteins in human cerebrospinal fluids by MS/MS using 6-plex isobaric tags, Anal. Chem. 80 (2008) 2921–2931. [10] C.J. Koehler, M. Strozynski, F. Kozielski, A. Treumann, B. Thiede, Isobaric peptide termini labeling for MS/MS-based quantitative proteomics, J. Proteome Res. 8 (2009) 4333–4341. [11] C.J. Koehler, M.O. Arntzen, M. Strozynski, A. Treumann, B. Thiede, Isobaric peptide termini labeling utilizing site-specific N-terminal succinylation, Anal. Chem. 83 (2011) 4775–4781. [12] P.J. Boersema, R. Raijmakers, S. Lemeer, S. Mohammed, A.J. Heck, Multiplex peptide stable isotope dimethyl labeling for quantitative proteomics, Nat. Protoc. 4 (2009) 484–494. [13] T. Geiger, J. Cox, P. Ostasiewicz, J.R. Wisniewski, M. Mann, Super-SILAC mix for quantitative proteomics of human tumor tissue, Nat. Methods 7 (2010) 383–385.

[14] R. Ahrends, S. Pieper, A. Kuhn, H. Weisshoff, M. Hamester, T. Lindemann, C. Scheler, K. Lehmann, K. Taubner, M.W. Linscheid, A metal-coded affinity tag approach to quantitative proteomics, Mol. Cell. Proteomics 6 (2007) 1907–1916. [15] D. Esteban-Fernandez, C. Scheler, M.W. Linscheid, Absolute protein quantification by LC-ICP-MS using MeCAT peptide labeling, Anal. Bioanal. Chem. 401 (2011) 657–666. [16] H. Liu, R.G. Sadygov, J.R. Yates 3rd, A model for random sampling and estimation of relative protein abundance in shotgun proteomics, Anal. Chem. 76 (2004) 4193–4201. [17] M. Bantscheff, M. Schirle, G. Sweetman, J. Rick, B. Kuster, Quantitative mass spectrometry in proteomics: a critical review, Anal. Bioanal. Chem. 389 (2007) 1017–1031. [18] P. Lu, C. Vogel, R. Wang, X. Yao, E.M. Marcotte, Absolute protein expression profiling estimates the relative contributions of transcriptional and translational regulation, Nat. Biotechnol. 25 (2007) 117–124. [19] L. Florens, M.J. Carozza, S.K. Swanson, M. Fournier, M.K. Coleman, J.L. Workman, M.P. Washburn, Analyzing chromatin remodeling complexes using shotgun proteomics and normalized spectral abundance factors, Methods 40 (2006) 303–311. [20] B. Zybailov, A.L. Mosley, M.E. Sardiu, M.K. Coleman, L. Florens, M.P. Washburn, Statistical analysis of membrane proteome expression changes in Saccharomyces cerevisiae, J. Proteome Res. 5 (2006) 2339–2347. [21] N.M. Griffin, J. Yu, F. Long, P. Oh, S. Shore, Y. Li, J.A. Koziol, J.E. Schnitzer, Label-free, normalized quantification of complex mass spectrometry data for proteomic analysis, Nat. Biotechnol. 28 (2010) 83–89. [22] D.H. Lundgren, S.I. Hwang, L. Wu, D.K. Han, Role of spectral counting in quantitative proteomics, Expert Rev. Proteomics 7 (2010) 39–53. [23] K.A. Neilson, N.A. Ali, S. Muralidharan, M. Mirzaei, M. Mariani, G. Assadourian, A. Lee, S.C. van Sluyter, P.A. Haynes, Less label, more free: approaches in label-free quantitative mass spectrometry, Proteomics 11 (2011) 535–553. [24] P.V. Bondarenko, D. Chelius, T.A. Shaler, Identification and relative quantitation of protein mixtures by enzymatic digestion followed by capillary reversed-phase liquid chromatography–tandem mass spectrometry, Anal. Chem. 74 (2002) 4741–4749. [25] D. Chelius, P.V. Bondarenko, Quantitative profiling of proteins in complex mixtures using liquid chromatography and mass spectrometry, J. Proteome Res. 1 (2002) 317–323. [26] C.W. Turck, A.M. Falick, J.A. Kowalak, W.S. Lane, K.S. Lilley, B.S. Phinney, S.T. Weintraub, H.E. Witkowska, N.A. Yates, The Association of Biomolecular Resource Facilities Proteomics Research Group 2006 study: relative protein quantitation, Mol. Cell. Proteomics 6 (2007) 1291–1298. [27] Z. Li, R.M. Adams, K. Chourey, G.B. Hurst, R.L. Hettich, C. Pan, Systematic comparison of label-free, metabolic labeling, and isobaric chemical labeling for quantitative proteomics on LTQ Orbitrap Velos, J. Proteome Res. 11 (2012) 1582–1590. [28] J. Merl, M. Ueffing, S.M. Hauck, C. von Toerne, Direct comparison of MS-based label-free and SILAC quantitative proteome profiling strategies in primary retinal Muller cells, Proteomics 12 (2012) 1902–1911. [29] T.S. Collier, S.M. Randall, P. Sarkar, B.M. Rao, R.A. Dean, D.C. Muddiman, Comparison of stable-isotope labeling with amino acids in cell culture and spectral counting for relative quantification of protein expression, Rapid Commun. Mass Spectrom. 25 (2011) 2524–2532. [30] T.S. Collier, P. Sarkar, W.L. Franck, B.M. Rao, R.A. Dean, D.C. Muddiman, Direct comparison of stable isotope labeling by amino acids in cell culture and spectral counting for quantitative proteomics, Anal. Chem. 82 (2010) 8696–8702. [31] V.J. Patel, K. Thalassinos, S.E. Slade, J.B. Connolly, A. Crombie, J.C. Murrell, J.H. Scrivens, A comparison of labeling and label-free mass spectrometry-based proteomics approaches, J. Proteome Res. 8 (2009) 3752–3759. [32] H. Wang, S. Alvarez, L.M. Hicks, Comprehensive comparison of iTRAQ and label-free LC-based quantitative proteomics approaches using two Chlamydomonas reinhardtii strains of interest for biofuels engineering, J. Proteome Res. 11 (2012) 487–501. [33] Y. Tian, A.C. Tan, X. Sun, M.T. Olson, Z. Xie, N. Jinawath, D.W. Chan, M. Shih Ie, Z. Zhang, H. Zhang, Quantitative proteomic analysis of ovarian cancer cells identified mitochondrial proteins associated with Paclitaxel resistance, Proteomics Clin. Appl. 3 (2009) 1288–1295. [34] L. Ijsselstijn, M.P. Stoop, C. Stingl, P.A. Sillevis Smitt, T.M. Luider, L.J. Dekker, Comparative study of targeted and label-free mass spectrometry methods for protein quantification, J. Proteome Res. 12 (2013) 2005–2011. [35] R.E. Banks, M.J. Dunn, M.A. Forbes, A. Stanley, D. Pappin, T. Naven, M. Gough, P. Harnden, P.J. Selby, The potential use of laser capture microdissection to selectively obtain distinct populations of cells for proteomic analysis—preliminary findings, Electrophoresis 20 (1999) 689–700. [36] J. Ai, Y. Tan, W. Ying, Y. Hong, S. Liu, M. Wu, X. Qian, H. Wang, Proteome analysis of hepatocellular carcinoma by laser capture microdissection, Proteomics 6 (2006) 538–546. [37] C. Molleken, B. Sitek, C. Henkel, G. Poschmann, B. Sipos, S. Wiese, B. Warscheid, C. Broelsch, M. Reiser, S.L. Friedman, I. Tornoe, A. Schlosser, G. Kloppel, W. Schmiegel, H.E. Meyer, U. Holmskov, K. Stuhler, Detection of novel biomarkers of liver cirrhosis by proteomic analysis, Hepatology 49 (2009) 1257–1266. [38] G. Poschmann, B. Sitek, B. Sipos, K. Stuhler, Application of saturation labeling in lung cancer proteomics, Methods Mol. Biol. 854 (2012) 253–267. [39] G. Poschmann, B. Sitek, B. Sipos, A. Ulrich, S. Wiese, C. Stephan, B. Warscheid, G. Kloppel, A. Vander Borght, F.C. Ramaekers, H.E. Meyer, K. Stuhler, Identification of proteomic differences between squamous cell carcinoma of the lung and bronchial epithelium, Mol. Cell. Proteomics 8 (2009) 1105–1116. [40] B. Sitek, J. Luttges, K. Marcus, G. Kloppel, W. Schmiegel, H.E. Meyer, S.A. Hahn, K. Stuhler, Application of fluorescence difference gel electrophoresis saturation labelling

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49] [50]

[51] [52]

[53]

[54]

[55] [56]

[57] [58] [59]

[60]

[61]

[62]

[63]

[64] [65]

[66]

[67] [68]

for the analysis of microdissected precursor lesions of pancreatic ductal adenocarcinoma, Proteomics 5 (2005) 2665–2679. B. Sitek, S. Potthoff, T. Schulenborg, J. Stegbauer, T. Vinke, L.C. Rump, H.E. Meyer, O. Vonend, K. Stuhler, Novel approaches to analyse glomerular proteins from smallest scale murine and human samples using DIGE saturation labelling, Proteomics 6 (2006) 4337–4345. B. Sitek, B. Sipos, I. Alkatout, G. Poschmann, C. Stephan, T. Schulenborg, K. Marcus, J. Luttges, D.D. Dittert, G. Baretton, W. Schmiegel, S.A. Hahn, G. Kloppel, H.E. Meyer, K. Stuhler, Analysis of the pancreatic tumor progression by a quantitative proteomic approach and immunohistochemical validation, J. Proteome Res. 8 (2009) 1647–1656. M. Hosako, T. Muto, Y. Nakamura, K. Tsuta, N. Tochigi, H. Tsuda, H. Asamura, T. Tomonaga, A. Kawai, T. Kondo, Proteomic study of malignant pleural mesothelioma by laser microdissection and two-dimensional difference gel electrophoresis identified cathepsin D as a novel candidate for a differential diagnosis biomarker, J. Proteomics 75 (2012) 833–844. T. Kondo, S. Hirohashi, Application of highly sensitive fluorescent dyes (CyDye DIGE Fluor saturation dyes) to laser microdissection and two-dimensional difference gel electrophoresis (2D-DIGE) for cancer proteomics, Nat. Protoc. 1 (2006) 2940–2956. T. Orimo, H. Ojima, N. Hiraoka, S. Saito, T. Kosuge, T. Kakisaka, H. Yokoo, K. Nakanishi, T. Kamiyama, S. Todo, S. Hirohashi, T. Kondo, Proteomic profiling reveals the prognostic value of adenomatous polyposis coli-end-binding protein 1 in hepatocellular carcinoma, Hepatology 48 (2008) 1851–1863. N.Q. Liu, R.B. Braakman, C. Stingl, T.M. Luider, J.W. Martens, J.A. Foekens, A. Umar, Proteomics pipeline for biomarker discovery of laser capture microdissected breast cancer tissue, J. Mammary Gland Biol. Neoplasia 17 (2012) 155–164. R.B. Braakman, M.M. Tilanus-Linthorst, N.Q. Liu, C. Stingl, L.J. Dekker, T.M. Luider, J.W. Martens, J.A. Foekens, A. Umar, Optimized nLC–MS workflow for laser capture microdissected breast cancer tissue, J. Proteomics 75 (2012) 2844–2854. G. Shevchenko, S. Musunuri, M. Wetterhall, J. Bergquist, Comparison of extraction methods for the comprehensive analysis of mouse brain proteome using shotgun-based mass spectrometry, J. Proteome Res. 11 (2012) 2441–2451. J.R. Wisniewski, A. Zougman, N. Nagaraj, M. Mann, Universal sample preparation method for proteome analysis, Nat. Methods 6 (2009) 359–362. B.S. Antharavally, K.A. Mallia, M.M. Rosenblatt, A.M. Salunkhe, J.C. Rogers, P. Haney, N. Haghdoost, Efficient removal of detergents from proteins and peptides in a spin column format, Anal. Biochem. 416 (2011) 39–44. L. Turiak, O. Ozohanics, F. Marino, L. Drahos, K. Vekey, Digestion protocol for small protein amounts for nano-HPLC-MS(MS) analysis, J. Proteomics 74 (2011) 942–947. W. Choksawangkarn, N. Edwards, Y. Wang, P. Gutierrez, C. Fenselau, Comparative study of workflows optimized for in-gel, in-solution, and on-filter proteolysis in the analysis of plasma membrane proteins, J. Proteome Res. 11 (2012) 3030–3034. M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein–dye binding, Anal. Biochem. 72 (1976) 248–254. P.K. Smith, R.I. Krohn, G.T. Hermanson, A.K. Mallia, F.H. Gartner, M.D. Provenzano, E.K. Fujimoto, N.M. Goeke, B.J. Olson, D.C. Klenk, Measurement of protein using bicinchoninic acid, Anal. Biochem. 150 (1985) 76–85. O.H. Lowry, N.J. Rosebrough, A.L. Farr, R.J. Randall, Protein measurement with the Folin phenol reagent, J. Biol. Chem. 193 (1951) 265–275. N. Popov, M. Schmitt, S. Schulzeck, H. Matthies, Reliable micromethod for determination of the protein content in tissue homogenates, Acta Biol. Med. Ger. 34 (1975) 1441–1446. M.I. Tyler, Amino acid analysis. An overview, Methods Mol. Biol. 159 (2000) 1–7. J.M. Burkhart, T. Premsler, A. Sickmann, Quality control of nano-LC–MS systems using stable isotope-coded peptides, Proteomics 11 (2011) 1049–1057. N.A. Patel, A. Crombie, S.E. Slade, K. Thalassinos, C. Hughes, J.B. Connolly, J. Langridge, J.C. Murrell, J.H. Scrivens, Comparison of one- and two-dimensional liquid chromatography approaches in the label-free quantitative analysis of Methylocella silvestris, J. Proteome Res. 11 (2012) 4755–4763. V. Gautier, E. Mouton-Barbosa, D. Bouyssie, N. Delcourt, M. Beau, J.P. Girard, C. Cayrol, O. Burlet-Schiltz, B. Monsarrat, A. Gonzalez de Peredo, Label-free quantification and shotgun analysis of complex proteomes by one-dimensional SDS-PAGE/NanoLC– MS: evaluation for the large scale analysis of inflammatory human endothelial cells, Mol. Cell. Proteomics 11 (2012) 527–539. B.B. Gao, L. Stuart, E.P. Feener, Label-free quantitative analysis of one-dimensional PAGE LC/MS/MS proteome: application on angiotensin II-stimulated smooth muscle cells secretome, Mol. Cell. Proteomics 7 (2008) 2399–2409. S. Fraterman, U. Zeiger, T.S. Khurana, N.A. Rubinstein, M. Wilm, Combination of peptide OFFGEL fractionation and label-free quantitation facilitated proteomics profiling of extraocular muscle, Proteomics 7 (2007) 3404–3416. J. Franck, K. Arafah, M. Elayed, D. Bonnel, D. Vergara, A. Jacquet, D. Vinatier, M. Wisztorski, R. Day, I. Fournier, M. Salzet, MALDI imaging mass spectrometry: state of the art technology in clinical proteomics, Mol. Cell. Proteomics 8 (2009) 2023–2033. A.S. Benk, C. Roesli, Label-free quantification using MALDI mass spectrometry: considerations and perspectives, Anal. Bioanal. Chem. 404 (2012) 1039–1056. B. Balluff, C. Schone, H. Hofler, A. Walch, MALDI imaging mass spectrometry for direct tissue analysis: technological advancements and recent applications, Histochem. Cell Biol. 136 (2011) 227–244. A. Panchaud, M. Affolter, P. Moreillon, M. Kussmann, Experimental and computational approaches to quantitative proteomics: status quo and outlook, J. Proteomics 71 (2008) 19–33. R. Aebersold, M. Mann, Mass spectrometry-based proteomics, Nature 422 (2003) 198–207. D.N. Perkins, D.J. Pappin, D.M. Creasy, J.S. Cottrell, Probability-based protein identification by searching sequence databases using mass spectrometry data, Electrophoresis 20 (1999) 3551–3567.

1589

[69] J.K. Eng, A.L. Mccormack, J.R. Yates, An approach to correlate tandem massspectral data of peptides with amino-acid-sequences in a protein database, J. Am. Soc. Mass Spectr. 5 (1994) 976–989. [70] L.Y. Geer, S.P. Markey, J.A. Kowalak, L. Wagner, M. Xu, D.M. Maynard, X. Yang, W. Shi, S.H. Bryant, Open mass spectrometry search algorithm, J. Proteome Res. 3 (2004) 958–964. [71] R. Craig, R.C. Beavis, TANDEM: matching proteins with tandem mass spectra, Bioinformatics 20 (2004) 1466–1467. [72] J.E. Elias, S.P. Gygi, Target-decoy search strategy for increased confidence in large-scale protein identifications by mass spectrometry, Nat. Methods 4 (2007) 207–214. [73] L.N. Mueller, M.Y. Brusniak, D.R. Mani, R. Aebersold, An assessment of software solutions for the analysis of mass spectrometry based quantitative proteomics data, J. Proteome Res. 7 (2008) 51–61. [74] R. Matthiesen, A.S. Carvalho, Methods and algorithms for relative quantitative proteomics by mass spectrometry, Methods Mol. Biol. 593 (2010) 187–204. [75] K. Podwojski, M. Eisenacher, M. Kohl, M. Turewicz, H.E. Meyer, J. Rahnenfuhrer, C. Stephan, Peek a peak: a glance at statistics for quantitative label-free proteomics, Expert Rev. Proteomics 7 (2010) 249–261. [76] S. Nahnsen, C. Bielow, K. Reinert, O. Kohlbacher, Tools for label-free peptide quantification, Mol. Cell. Proteomics 12 (2013) 549–556. [77] J. Cox, M. Mann, MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification, Nat. Biotechnol. 26 (2008) 1367–1372. [78] L.N. Mueller, O. Rinner, A. Schmidt, S. Letarte, B. Bodenmiller, M.Y. Brusniak, O. Vitek, R. Aebersold, M. Muller, SuperHirn — a novel tool for high resolution LC–MS-based peptide/protein profiling, Proteomics 7 (2007) 3470–3480. [79] P.M. Palagi, D. Walther, M. Quadroni, S. Catherinet, J. Burgess, C.G. Zimmermann-Ivol, J.C. Sanchez, P.A. Binz, D.F. Hochstrasser, R.D. Appel, MSight: an image analysis software for liquid chromatography–mass spectrometry, Proteomics 5 (2005) 2381–2384. [80] R. Malik, K. Dulla, E.A. Nigg, R. Korner, From proteome lists to biological impact— tools and strategies for the analysis of large MS data sets, Proteomics 10 (2010) 1270–1283. [81] P. Picotti, O. Rinner, R. Stallmach, F. Dautel, T. Farrah, B. Domon, H. Wenschuh, R. Aebersold, High-throughput generation of selected reaction-monitoring assays for proteins and proteomes, Nat. Methods 7 (2010) 43–46. [82] L. Anderson, C.L. Hunter, Quantitative mass spectrometric multiple reaction monitoring assays for major plasma proteins, Mol. Cell. Proteomics 5 (2006) 573–588. [83] A. Gamez-Pozo, I. Sanchez-Navarro, E. Calvo, M.T. Agullo-Ortuno, R. Lopez-Vacas, E. Diaz, E. Camafeita, M. Nistal, R. Madero, E. Espinosa, J.A. Lopez, J.A. Fresno Vara, PTRF/cavin-1 and MIF proteins are identified as non-small cell lung cancer biomarkers by label-free proteomics, PLoS One 7 (2012) e33752. [84] S. Meding, B. Balluff, M. Elsner, C. Schone, S. Rauser, U. Nitsche, M. Maak, A. Schafer, S.M. Hauck, M. Ueffing, R. Langer, H. Hofler, H. Friess, R. Rosenberg, A. Walch, Tissue-based proteomics reveals FXYD3, S100A11 and GSTM3 as novel markers for regional lymph node metastasis in colon cancer, J. Pathol. 228 (2012) 459–470. [85] H.Y. Yang, J. Kwon, H.R. Park, S.O. Kwon, Y.K. Park, H.S. Kim, Y.J. Chung, Y.J. Chang, H.I. Choi, K.J. Chung, D.S. Lee, B.J. Park, S.H. Jeong, T.H. Lee, Comparative proteomic analysis for the insoluble fractions of colorectal cancer patients, J. Proteomics 75 (2012) 3639–3653. [86] D. A. Megger, T. Bracht, M. Kohl, M. Ahrens, W. Naboulsi, F. Weber, A. C. Hoffmann, C. Stephan, K. Kuhlmann, M. Eisenacher, J. F. Schlaak, H. A. Baba, H. E. Meyer, and B. Sitek, Proteomic differences between hepatocellular carcinoma and non-tumorous liver tissue investigated by a combined 2D-DIGE and label-free quantitative proteomics study, Mol. Cell. Proteomics (in press) (http://dx.doi.org/10.1074/mcp.M113. 028027). [87] E. Hammer, M. Goritzka, S. Ameling, K. Darm, L. Steil, K. Klingel, C. Trimpert, L.R. Herda, M. Dorr, H.K. Kroemer, R. Kandolf, A. Staudt, S.B. Felix, U. Volker, Characterization of the human myocardial proteome in inflammatory dilated cardiomyopathy by label-free quantitative shotgun proteomics of heart biopsies, J. Proteome Res. 10 (2011) 2161–2171. [88] J.J. Hill, T.L. Tremblay, A. Pen, J. Li, A.C. Robotham, A.E. Lenferink, E. Wang, M. O'Connor-McCourt, J.F. Kelly, Identification of vascular breast tumor markers by laser capture microdissection and label-free LC–MS, J. Proteome Res. 10 (2011) 2479–2493. [89] J.R. Wisniewski, P. Ostasiewicz, M. Mann, High recovery FASP applied to the proteomic analysis of microdissected formalin fixed paraffin embedded cancer tissues retrieves known colon cancer markers, J. Proteome Res. 10 (2011) 3040–3049. [90] J.R. Wisniewski, P. Ostasiewicz, K. Dus, D.F. Zielinska, F. Gnad, M. Mann, Extensive quantitative remodeling of the proteome between normal colon tissue and adenocarcinoma, Mol. Syst. Biol. 8 (2012) 611. [91] S.W. Hyung, M.Y. Lee, J.H. Yu, B. Shin, H.J. Jung, J.M. Park, W. Han, K.M. Lee, H.G. Moon, H. Zhang, R. Aebersold, D. Hwang, S.W. Lee, M.H. Yu, D.Y. Noh, A serum protein profile predictive of the resistance to neoadjuvant chemotherapy in advanced breast cancers, Mol. Cell. Proteomics 10 (2011), M111 011023. [92] K. Patel, J.E. Lucas, J.W. Thompson, L.G. Dubois, H.L. Tillmann, A.J. Thompson, D. Uzarski, R.M. Califf, M.A. Moseley, G.S. Ginsburg, J.G. McHutchison, J.J. McCarthy, High predictive accuracy of an unbiased proteomic profile for sustained virologic response in chronic hepatitis C patients, Hepatology 53 (2011) 1809–1818. [93] J. Metzger, A.A. Negm, R.R. Plentz, T.J. Weismuller, J. Wedemeyer, T.H. Karlsen, M. Dakna, W. Mullen, H. Mischak, M.P. Manns, T.O. Lankisch, Urine proteomic analysis differentiates cholangiocarcinoma from primary sclerosing cholangitis and other benign biliary disorders, Gut 62 (2012) 122–130. [94] T.O. Lankisch, J. Metzger, A.A. Negm, K. Vosskuhl, E. Schiffer, J. Siwy, T.J. Weismuller, A.S. Schneider, K. Thedieck, R. Baumeister, P. Zurbig, E.M.

1590

D.A. Megger et al. / Biochimica et Biophysica Acta 1834 (2013) 1581–1590

Weissinger, M.P. Manns, H. Mischak, J. Wedemeyer, Bile proteomic profiles differentiate cholangiocarcinoma from primary sclerosing cholangitis and choledocholithiasis, Hepatology 53 (2011) 875–884. [95] Y. Kim, V. Ignatchenko, C.Q. Yao, I. Kalatskaya, J.O. Nyalwidhe, R.S. Lance, A.O. Gramolini, D.A. Troyer, L.D. Stein, P.C. Boutros, J.A. Medin, O.J. Semmes, R.R. Drake, T. Kislinger, Identification of differentially expressed proteins in direct expressed prostatic secretions of men with organ-confined versus extracapsular prostate cancer, Mol. Cell. Proteomics 11 (2012) 1870–1884.

[96] Z. Wu, J.B. Doondeea, A.M. Gholami, M.C. Janning, S. Lemeer, K. Kramer, S.A. Eccles, S.M. Gollin, R. Grenman, A. Walch, S.M. Feller, B. Kuster, Quantitative chemical proteomics reveals new potential drug targets in head and neck cancer, Mol. Cell. Proteomics 10 (2011), M111 011635. [97] W.J. Van Houdt, B.L. Emmink, T.V. Pham, S.R. Piersma, A. Verheem, R.G. Vries, S.A. Fratantoni, A. Pronk, H. Clevers, I.H. Borel Rinkes, C.R. Jimenez, O. Kranenburg, Comparative proteomics of colon cancer stem cells and differentiated tumor cells identifies BIRC6 as a potential therapeutic target, Mol. Cell. Proteomics 10 (2011), M111 011353.