In-depth analyses of cancer cell proteomes facilitate identification of novel drug targets and diagnostic biomarkers. We describe an experimental workflow for quantitative analysis of (phospho-)proteomes in cancer cell subpopulations derived from liquid and solid tumors. This is achieved by combining cellular enrichment strategies with quantitative Super-SILAC-based mass spectrometry.
In-depth analyses of cancer cell proteomes are needed to elucidate oncogenic pathomechanisms, as well as to identify potential drug targets and diagnostic biomarkers. However, methods for quantitative proteomic characterization of patient-derived tumors and in particular their cellular subpopulations are largely lacking. Here we describe an experimental set-up that allows quantitative analysis of proteomes of cancer cell subpopulations derived from either liquid or solid tumors. This is achieved by combining cellular enrichment strategies with quantitative Super-SILAC-based mass spectrometry followed by bioinformatic data analysis. To enrich specific cellular subsets, liquid tumors are first immunophenotyped by flow cytometry followed by FACS-sorting; for solid tumors, laser-capture microdissection is used to purify specific cellular subpopulations. In a second step, proteins are extracted from the purified cells and subsequently combined with a tumor-specific, SILAC-labeled spike-in standard that enables protein quantification. The resulting protein mixture is subjected to either gel electrophoresis or Filter Aided Sample Preparation (FASP) followed by tryptic digestion. Finally, tryptic peptides are analyzed using a hybrid quadrupole-orbitrap mass spectrometer, and the data obtained are processed with bioinformatic software suites including MaxQuant. By means of the workflow presented here, up to 8,000 proteins can be identified and quantified in patient-derived samples, and the resulting protein expression profiles can be compared among patients to identify diagnostic proteomic signatures or potential drug targets.
Mass spectrometry-based proteomics has emerged and is now a widely used discipline in cell biology and translational biomedical research. Technical advances in the field have made it possible to study complex cellular processes in cell lines and animal models, as thousands of proteins can be identified in a single mass spectrometric experiment1,2. Similar progress has been made for the analysis of many posttranslational modifications such as phosphorylation or ubiquitination, although this usually requires bespoke workflows for enrichment and data analysis for each type of modification2,3. Moreover, the involvement of chemical and metabolic labeling strategies, including SILAC, enables accurate relative quantification of proteins and PTMs, making this method particularly attractive for the discovery of basic cellular processes, diagnostic biomarkers and potential drug targets in human cancer4.
However, several challenges have to be overcome with regard to proteomic analysis of primary human cancers5. First of all, human cancer samples often show a high degree of cellular heterogeneity that is due to the presence of various cell types belonging to the tumor microenvironment, including immune cells and fibroblasts. Secondly, clonal evolution leads to genetic diversity within the tumors themselves, resulting in the existence of several cellular subpopulations with distinct functional properties. According to the current concept of a few cancer stem cells being the driving forces behind cancer development and progression, proteomic analysis of such (functionally highly relevant) cellular subpopulations is expected to be of major importance for a better understanding of oncogenic mechanisms relevant for clinical application6. Thirdly, quantitative protein expression profiles of large sets of samples are often required for the identification of robust clinical biomarkers; this cannot be accomplished by chemical labeling such as iTRAQ7, while metabolic labeling strategies – relying, as they do, on cellular proliferation4– are likewise inapplicable. Fourthly, most available solid tumor samples are formalin-fixed, which complicates mass spectrometric proteome analysis due to the formation of protein crosslinks8. Finally, most existing proteomics workflows require significant amounts of sample processing and data acquisition, rendering the analysis of sample numbers relevant for clinical research difficult, and calling for new workflow paradigms9,10.
To address these obstacles we developed an experimental setup that combines either flow-cytometric cell sorting11 or laser-capture microdissection12,13 for cellular enrichment with a Super-SILAC14 strategy to introduce a global internal standard for comprehensive mass spectrometric analysis. By using the method described here, it is possible to quantify up to 8,000 proteins in single human tumor samples derived from either liquid or solid cancers.
All experiments on human tissue or blood samples must be approved by an ethics committee and conducted according to any guidelines given in the ethics vote.
1. Cellular Enrichment
2. Protein Extraction
3. Establishment of a SILAC Spike-in Quantification Standard (Super-SILAC Standard)
NOTE: The quantification standard consists of a SILAC-labeled protein mixture derived from 4 – 6 cells lines that match the tumor type of interest. To achieve the maximum overlap between the SILAC-labeled reference proteome and the tumor-derived proteome, a principal component analysis needs to be performed before cell lines are selected for the quantification standard14.
4. Measurement of Protein Concentration and Spike-in
5. Sample Separation and Protein Digest
6. Liquid Chromatography and Mass Spectrometric Analysis
7. Data Analysis
Proteomic profiling of liquid and solid tumors from patients is a promising approach for the discovery of new diagnostic and predictive biomarkers. However, the sample preparation procedure and the mass spectrometric analysis are challenging, owing to the complexity of the samples and the need for accurate protein quantification in large sets of samples. The experimental procedure described above starts with the isolation of cells of interest, either by fluorescence-activated cell sorting or by laser-capture microdissection.
For this purpose, cells from patient-derived bone marrow aspirates or blood samples are stained with fluorescence-labeled antibodies against defined surface markers of interest before FACS-based sorting and cell lysis.
To enrich cellular subsets from tissue sections, these first have to be mounted on appropriate slides to allow for laser-capture microdissection. For this purpose mount the tissue sections on appropriate membrane slides and use a laser-capture microdissector according to manufacturer’s instructions. Selecting the areas and cells of interest requires appropriate knowledge about the histomorphology of healthy and neoplastic tissue. The extracted cells are then transferred into an appropriate collection tube. Initial experiments should be performed in order to define the amount of tissue necessary for extraction of sufficient amounts of protein (minimum 10 µg total protein) for every new tissue of interest. Each microdissected sample is treated with 60 µl of tissue lysis buffer. For efficient lysis, cells from tissue samples have to be sonicated for 3 min and after the addition of 15 µl of SDS samples are incubated in a thermomixer at 99 °C for 1 hr to remove formalin-induced protein crosslinks. Centrifugation will finally facilitate the collection of the cleared cellular lysate.
For every disease of interest (liquid and solid tumors) a suitable Super-SILAC quantification standard has to be established14. An appropriate standard should represent more than 90% of the proteins expressed in the samples of interest. For preparation of a quantification standard, cell lines related to the cancer type of interest should first be analyzed by mass spectrometry to determine their protein expression profiles and subsequently principal component analysis should be performed. 4 – 6 different cell lines with differential protein expression patterns should be chosen and SILAC-labeled with “heavy” arginine and lysine. Labeling efficiency should be checked by nanoLC-MS/MS and should be greater than 98%. Afterwards, cell lines should be lysed in the same way as the respective patient-derived sample and, subsequently, equimolar protein amounts of each sample and the SILAC spike-in standard should be mixed. A critical step for accurate protein quantification of clinical samples of interest is the accurate mixture of equimolar protein amounts of the patient-derived samples and the SILAC spike-in quantification standard. To determine the protein concentration of lysates derived from sorted cells, various protein quantification assays can be used. For cellular lysates from tissue-derived cells an assay is needed that can deal with high concentrations of SDS and DTT in the lysate. Protein concentrations of the patient-derived samples and the SILAC spike-in standard should be measured each time prior to combination with the respective clinical samples in order to ensure correct mixture, which is crucial in making even hundreds of samples comparable.
After the mixture of equal amounts of spike-in standard and lysate, samples obtained from liquid tumors are mixed with LDS, heated in a thermomixer at 72 °C for 10 min and subsequently subjected to 1D-PAGE and in-gel digestion of the separated proteins with trypsin. Samples obtained from tissue are freed from SDS and digested with trypsin by using the FASP approach as established by Wisniewski et al.15
The tryptic peptides obtained can be quantified by measuring tryptophan fluorescence, which is of special interest for tissue samples as the amount of peptides released from the filter differs between 15% and 75% compared to the protein amounts initially loaded onto the filter. For this purpose we measure the fluorescence of tryptophan. Comparison with an appropriate tryptophan dilution series allows for the quantification of peptides, as 1.1 µg of tryptophan corresponds to approximately 100 µg of peptide. If necessary, samples derived especially from FASP processing can be pre-purified on either commercial or home-made stage tips packed with reversed phase-C18 (RP-C18) material before loading onto the nanoLC/MS/MS system18.
Mass-spectrometric analysis is performed on a high-resolution, high-sensitivity mass-spectrometry system. In short, peptide samples are desalted and preconcentrated on an RP-C18 precolumn, and separated on an RP-C18 analytical column coupled directly to the mass spectrometer. To achieve sufficient depth of analysis, we employ either a combination of SDS-PAGE protein prefractionation with 40 min RP-C18 gradients (for liquid tumors) or single injections with long 2 – 3 hr RP-C18 gradients (for solid tumors processed by FASP). MS spectra are acquired at a resolution of 70,000 FWHM or better, to allow accurate quantification of SILAC pairs by integration of chromatographic peak profiles. For protein identification, a Top15 data-dependent acquisition method is used to generate a large number of peptide MS/MS spectra for peptide and protein identification.
From the resulting raw data, protein identification and quantification are achieved by database searching with MaxQuant software against a UniProt Knowledgebase Human Complete Proteome sequence database17. In MaxQuant software (current version 1.5.0.25) peptides are identified from the acquired MS/MS spectra by peptide fragment-matching against spectra derived in silico from the protein sequence database. At the same time, precursor ion isotopic profiles are extracted around their chromatographic retention times, and their integrated peak areas are used for relative quantification of the light:heavy peptide pairs generated by SILAC labeling. Peptide identities and relative intensities are then assigned to the corresponding protein properties. Perseus software (current version 1.5.0.15) is then used to perform further downstream statistical evaluation of the MaxQuant processing results, including sample-to-sample comparisons, PCA and hierarchical clustering.
Using the experimental setup described we have identified and quantified up to 8,000 proteins from as little as 30 µg of total protein derived from liquid tumors.
Up to 2,500 proteins from solid-tumor samples can be identified and quantified in a shotgun proteomic approach with an LC gradient of only 2 hr, allowing the analysis of hundreds of clinical samples in a relatively short time.
Figure 1. Experimental workflow. The main steps of cellular-subset enrichment, protein isolation, spike-in of the quantification standard and mass-spectrometric analysis are shown for liquid and solid tumors. Please click here to view a larger version of this figure.
Figure 2. Cellular enrichment strategies. FACS sorting and LCM. (A) Exemplary gating strategy depicting gating on a CD117-stained leukemic cell population acquired by a cell sorter. (B) Laser capture microdissection of solid tumor tissue. Tissue sections of 5 – 10 µm thickness were mounted on film-covered membrane slides before staining. Region of interests was selected manually for laser capture microdissection. Shown are sections before microdissection with the region of interest marked yellow and after microdissection. A scale bar of 50 µm is included in the figure. Please click here to view a larger version of this figure.
Figure 3. Cluster analysis of human cancer-cell proteomes. Unsupervised cluster analysis of protein expression profiles of liquid and solid tumors was performed using the computational platform Perseus. AC – Adenocarcinoma, SCC – Squamous cell carcinoma, SCC-Met – Squamous cell carcinoma metastases from head and neck cancer, R – technical replicate.
Mass spectrometric profiling of patient-derived cancer-cell proteomes is needed for the discovery of new diagnostic/predictive biomarkers as well as to give a better understanding of cancer-cell biology, which might in turn lead to the identification of new potential drug targets. However, such mass spectrometric analyses are highly challenging, in particular because various pre-analytical issues have to be solved, if one is to obtain robust and biologically relevant results.
The experimental workflow described here allows for quantitative proteomic characterization of proteomes derived from cellular subpopulations of both liquid and solid tumors. The initial enrichment of tumor cells by either FACS-based cell-sortingor microdissection is needed to avoid contamination by cells of the tumor microenvironment. Furthermore, these techniques allow one to isolate cellular subpopulations of interest. Recent cell-biological studies have demonstrated that certain cellular subpopulations have tumor-initiating properties and are thus highly relevant for cancer pathogenesis19,20. As mass spectrometry has become more sensitive in recent years, quantitative proteomic analyses are feasible for the small amounts of protein that can be derived from a few thousand cells, making it possible to focus on functionally relevant cell populations.
The set-up presented here can be used to identify and validate novel diagnostic biomarkers in FFPE samples. Hence, it promises to bea useful tool for the improvement of clinical diagnostics, as to date there is still a lack of molecular biomarkers in adequate number and quality for many cancer types. Important examples of difficult differential diagnoses, for which biomarkers are lacking, are the discrimination between primary lung cancer from metastases in the lung, intrapancreatic cholangiocellular carcinoma and pancreatic adenocarcinoma, as well as differentiation of benign neurofibroma from highly malignant peripheral nerve-sheath tumors. Moreover, we and others have shown that quantitative elucidation of proteomic signatures can be useful in studying cancer cell biology in general, and for revealing predictive biomarkers of therapeutic response in cancer patients21.
Two current drawbacks of the method presented here are the requirement for extensive manual sample processing and the demand on nanoLC-MS/MS acquisition time. While the former can be addressed by moving sample preparation to e.g., 96-well formats and using robotic processing, the latter will require a change in mass spectrometric acquisition strategy. Once subsets of target proteins have been identified that can be associated with e.g., tumor classification, we envisage the design of targeted mass spectrometry methods that provide quantitative readouts for these subsets with a greatly reduced separation effort, and therefore with a correspondingly reduced acquisition time. If the required acquisition time required could be reduced from 24 – 36 hr (liquid tumors) or 3 hr (solid tumors) to e.g., 1 hr using targeted mass spectrometry and a simple one-dimensional separation of peptides, then the resulting gain in throughput could be used to increase significantly the numbers of biological and technical replicates examined, with corresponding improvements in the significance of quantification results. Targeted mass spectrometric approaches have already been demonstrated to be a suitable tool for the verification of cancer-associated protein biomarker-candidates22, and have been developed to a point where they show promise for validation or even as a potential tool for routine clinical use23,24.
The authors have nothing to disclose.
The authors thank Uwe Plessmann, Monika Raabe und Silvia Münch for technical support.
Name of Material/ Equipment | Company | Catalog Number | Comments/Description |
660 nm Kit | Thermo scientific | 22662 | |
Cell culture medium depleted of arginine and lysine | Thermo Scientific | 88421 | |
Coomassie Brilliant Blue R-250 staining solution | Bio Rad | 161-0436 | |
Dialyzed fetal calf serum (FCS) | PAA | A15-107 | |
Diffuser caps for microdissection | MMI | 50202 | |
FACS-sorter | BD | FACSAria III | |
Ionic Detergent Compatibility Reagent | Thermo scientific | 22663 | |
Laser-capture microdissector | MMI | cell cut plus | |
LDS buffer | Life Technologies | NP0009 | |
Membrane slides for microdissection | MMI | 50103 | |
Microcon YM-30 | Millipore | MRCF0R030 | |
NuPAGE 4-12% Bis-Tris Mini Gels | Life Technologies | NP0335PK2 | |
Picofrit Self-Pack Columns | New Objective | PF360-75-15-N-5 | Mass Spectrometry Column/Emitter |
Reducing agent | Life Technologies | NP0007 | |
Reprosil-Pur LC/MS/MS Column stationary phase | Dr. Maisch | 120 C18-AQ, 3 µm | |
Reprosil-Pur LC/MS/MS Precolumn stationary phase | Dr. Maisch | 120 C18-AQ, 5 µm | |
SILAC-labeled arginine | Eurisotop | CLM-2265-H-0.1 | |
SILAC-labeled lysine | Eurisotop | DLM-2640-0.25 | |
Trypsin, NB Sequencing Grade | Serva | 3728301 | for in-gel digests |
Trypsin, Sequencing Grade | Promega | V5111 | for in-solution digests |
Buffer and solutions | |||
Cell lysis buffer: 150 mM NaCl, 50 mM Tris/HCl pH 7.8, 5mM NaF, 0,5% NP40, 0.1% laurylmaltoside, Roche complete protease inhibitor, 1 mM Na3VO4 | |||
Tissue lysis buffer: 100 mM Tris/HCl pH 7.8, 0.1 M DTT | |||
Urea: 8 M urea in 0.1 M Tris-HCl, pH 8.5 | for FASP-protocoll | ||
IAA: 0.05 M iodoacetamide, 8 M urea, 0.1 M Tris-HCl, pH 8.5 | for FASP-protocoll | ||
0.05 M NH4HCO3 | |||
10 mM dithiothreitol (DTT) in 0.1 M ammonium bicarbonate | for in-gel digest | ||
55 mM iodoacetamide (IAA) in 0.1 mM ammonium bicarbonate | for in-gel digest | ||
5% aqueous formic acid. |