Quantification of ErbB Network Proteins in Three Cell Types Using

Dec 6, 2013 - ABSTRACT: Relating protein concentration to cell-type-specific responses is one of the remaining challenges for obtaining a quantitative...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/jpr

Quantification of ErbB Network Proteins in Three Cell Types Using Complementary Approaches Identifies Cell-General and Cell-Type-Specific Signaling Proteins Christina Kiel,*,†,∥ H. Alexander Ebhardt,⊥ Julia Burnier,†,∥ Claire Portugal,†,∥ Eduard Sabidó,∥,‡ Timo Zimmermann,∥,§ Ruedi Aebersold,⊥,#,▽ and Luis Serrano†,∥,○ †

EMBL/CRG Systems Biology Research Unit, ‡Proteomics Unit, §Advanced Light Microscopy Core Facility, Centre for Genomic Regulation (CRG), Dr. Aiguader 88, 08003 Barcelona, Spain ∥ Universitat Pompeu Fabra (UPF), Plaça de la Mercè, 10, 08003 Barcelona, Spain ⊥ Department of Biology, Institute of Molecular Systems Biology, Eidgenössische Technische Hochschule (ETH) Zürich, Wolfgang-Pauli-Str. 16, 8093 Zürich, Switzerland # Competence Center for Systems Physiology and Metabolic Diseases, Schafmattstr. 22, HPL G 32.2, CH-8093 Zürich, Switzerland ▽ Faculty of Science, University of Zürich, Winterthurerstr. 190, 8057 Zürich, Switzerland ○ Institució Catalana de Recerca i Estudis Avançats (ICREA), Pg. Lluís Companys 23, 08010 Barcelona, Spain S Supporting Information *

ABSTRACT: Relating protein concentration to cell-type-specific responses is one of the remaining challenges for obtaining a quantitative systems level understanding of mammalian signaling. Here we used mass-spectrometry (MS)- and antibody-based quantitative proteomic approaches to measure protein abundances for 75% of a handcurated reconstructed ErbB network of 198 proteins, in two established cell types (HEK293 and MCF-7) and in primary keratinocyte cells. Comparison with other quantitative studies allowed building a set of ErbB network proteins expressed in all cells and another which are cell-specific and could impart specific properties to the network. As a proof-of-concept of the importance of protein concentration, we generated a small simplified mathematical model encompassing ligand binding, followed by receptor dimerization, activation, and degradation. The model predicts ErbB phosphorylation in HEK293, MCF-7, and keratinocyte cells simply by incorporating cell-type-specific ErbB1, ErbB2, and caveolin-1 abundances but otherwise contains similar rate constants. Altogether, the data provide a resource for protein abundances and localization to be included in larger mathematical models, enabling the generation of cell-type-specific computational models. MS data have been deposited to the ProteomeXchange via PRIDE (with identifier PXD000623) and PASSEL (with identifier PASS00372). KEYWORDS: mass spectrometry, selected reaction monitoring, shot-gun proteomics, stable isotope-labeled peptides , ErbB signaling, quantitative Western blotting, immunofluorescence



which can generate diverse signaling responses by using similar − in all cell types present − effectors.3 The work suggests that not every model for each cell type requires its own in-depth experiments and model training but proposes that cell-general models could be broadly applicable with just some new, cell-type-specific data. Thus, one crucial thing will be to distinguish cell-general from cell-typespecific components. Different cells of higher eukaryotic organisms need to perform similar house-keeping functions and, at the same time, cell-type specific functions. Recent work based on mass spectrometry (MS) has identified a so-called ‘core proteome’, which is defined as those proteins that are expressed in most of the cell

INTRODUCTION Despite the apparent complexity and cross-talk in signal transduction cascades as reported in the literature, not all possible signaling routes are taken in a given cell type, and different cell types can exhibit very specific responses to the same signal.1 Understanding these cell-type-specific signaling responses and including them in quantitative predictive mathematical models is challenging. Considering that around 230 different human cell types exist,2 the essential question is whether one indeed needs to quantify all components and signaling flows with all ligands in all cell types to be able to make predictive mathematical models. Experimental and computational modeling work from Lauffenburger and colleagues proposed that this is may not be needed. In their pioneering work, they proposed a mechanism of ‘common effector processing’, which suggests that cell-type-specific responses emanate from the differential activation of kinases and other upstream transducers, © 2013 American Chemical Society

Special Issue: Chromosome-centric Human Proteome Project Received: August 27, 2013 Published: December 6, 2013 300

dx.doi.org/10.1021/pr400878x | J. Proteome Res. 2014, 13, 300−313

Journal of Proteome Research

Article

types and tissues.4−6 It is noteworthy that many signaling-related proteins were also identified in the core proteome. The complexity and high dynamic range of a eukaryotic proteome makes it challenging to determine protein abundances.7,8 However, estimating protein abundances from transcript levels provides only a rough estimate of protein expression levels,9,10 which is due to differences in translation efficiency, mRNA stability, and protein stability for each component. While nevertheless mRNA levels could be a valid approximation when looking at changes in one cell type, for example, during stimulation or cell differentiation,11 measurements of mRNA levels in different cell types or tissues do not provide a reliable value of protein abundances. In addition, transcriptomic analyses do not capture the post-translational modifications (PTMs), such as phosphorylation, which are especially important in signaling cascades. Mass-spectrometry-based approaches in the ‘shot-gun’ setup, combined with prior extensive chromatographic fractionation, have recently achieved a remarkable detection and quantification acquisition of ∼10 000 proteins in mammalian cell lines.5,12−14 However, as a consequence of frequent gene duplication events in mammals, often similar proteins (e.g., AKT1 and AKT2) cannot be distinguished using the detected peptides, and they can only be assigned to a protein group/family. Furthermore, it is known that signaling proteins are usually expressed at lower levels, which is expected to decrease the discovery percentage and the accuracy of their quantification.5 Recently, proteins were quantified in unfractionated eukaryotic cell lysates using the selected reaction monitoring (SRM) mode, where a predetermined number of target peptides are quantified over a wide range of conditions.15−17 Thus, SRM is a promising alternative when focusing on the analysis of single signaling pathways, but its limiting step is the availability of peptides with high-quality MS signals without complex cleavage patterns (i.e., presence of several Lys/Arg at the N- or C-termini).18 The other possibility for quantifying proteins is using supportive antibodies (band of predicted size ±20% with additional bands present) and purified protein as standards (quantitative Western blot analysis19). Quantitative Western blotting, in principle, should be accurate, but the main obstacle in this approach is the availability of high-quality antibodies and the lengthy purification of pure protein standards. In the case of membrane proteins, quantitative FACS analysis could be a method of choice as it only measures proteins exposed at the cell membrane.20 Comparisons of different protein quantification approaches in yeast S. cerevisiae, including those based on TAP tag and labelfree MS abundances, identified poor correlations, suggesting that currently none of the existing methods can yet be considered a gold standard.21 Moreover, it was shown that there is a strong bias in absolute protein quantitation depending on the protease used, even when performed on the same mass spectrometer.22 These differences were shown to be not restricted to low abundant proteins but can also differ for high abundant up to 1000-fold. To ensure the accuracy of quantification, in this work we have used different complementary MS- and antibody-based quantitative proteomic approaches to quantify a large fraction of 198 proteins involved in ErbB signaling in two established cell lines and in primary cells (see flowchart in Figure 1). We identified both cell-general and cell-type-specific signaling components. However, levels of proteins expressed in all cell types can vary significantly. Caveolin-1, a protein that sustains signaling by preventing EGFR degradation, is highly expressed only in primary keratinocytes but not in the other two cell types. We generated a small simplified cell-type-specific mathematical model for ligand binding and EGFR/ErbB2 receptor activation

and degradation, including caveolin-1, by incorporating similar rate constants in each model but applying cell-type-specific protein abundances. Receptor activation levels predicted by the model qualitatively reproduce the experimentally determined EGFR phosphorylation levels detected by Western blotting.



EXPERIMENTAL SECTION

Cell Culture of HEK293 and MCF-7 Cell Lines and of Primary Human Keratinocytes

The human embryonic kidney cell line HEK293 and the human breast cancer cell line MCF-7 were cultured in Dulbecco’s modified Eagle medium (Gibco) supplemented with L-glutamine and 10% v/v heat-inactivated fetal calf serum (FCS). Keratinocytes were grown in keratinocyte serum-free medium (KSFM) (Gibco) + 1% penicillin-streptomycin (Gibco) supplemented with EGF (0.2 ng/mL) (Gibco) and pituitary extract (30 μg/mL) (Gibco). Cloning, Expression, and Purification of Protein Standards for Western Blotting

Cloning was done using the Gateway (Invitrogen) system following the manufacturer’s protocol. The attB-PCR was done using primers designed for the respective gene of interest with the cDNA of the gene of interest (purchased from Geneservice). The pDONR223 donor vector was used in the BP recombination reaction, and the gene of interest was verified by full-length sequencing. The LR recombination was performed with the pDEST17 vector, which is suitable for bacterial expression and contains a poly-His tag for purification. After transforming into a Rosetta strain, expression screening was performed in a 24-well plate using 2 mL of autoinduction medium. Cells were incubated 4 h at 37 °C and then 20 h at 25 °C at 800 rpm. Proteins were extracted and purified after by breaking the cells in 1 mL of lysis buffer LEW 1(Protino Kit) + lysozyme (1 mg/mL). After 1 h of incubation at room temperature, the cell lysate was centrifuged for 15 min at 4000 rpm. The supernatant was kept (soluble proteins), and the pellet was resuspended in 1 mL of lysis buffer in denaturing conditions (PBS 1×, 6 M GuHCl, 300 mM NaCl, 0.5% Triton X-100). After 1 h of incubation at room temperature, the cell lysate was centrifuged at 13 000 rpm for 30 min. Protino Mulit-96 Ni-IDA (Machery-Nagel) columns were equilibrated with 500 μL of loading buffer (100 mM Tris pH 7.4, 500 mM NaCl, 6 M urea, 5 mM imidazole), and the supernatants were loaded onto the Ni columns and washed twice with 500 μL of loading buffer. Proteins were eluted twice with 250 μL of elution buffer under denaturing conditions (100 mM Tris pH 7.4, 6 M urea, 1.5 M imidazole), and both elutions were collected. Scale-up of expression and purification (where necessary) was done in 50 mL of autoinduction media under the same expression and purification conditions by adjusting the respective volumes of solutions. Protein expression was detected by SDS-PAGE and Western blotting using anti-His antibody (SIGMA). All proteins were analyzed by SDS gel electrophoresis depending on their molecular weight, using 5, 7.5, 10, or 12.3% Ready Gels (BioRad), followed by colloidal blue staining (Invitrogen). In two additional separate SDS gel electrophoresis runs, the gels were subsequently blotted using the iBlot system (Invitrogen) and analyzed by Western blotting using a His-specific antibody and a protein-specific antibody. The overall protein concentration was first determined at OD280 using NanoDrop (Thermo Scientific). The percentage of protein of interest compared with the total protein content was calculated using ImageJ (NIH). OD280 values were then adjusted for purity, and the concentration of 301

dx.doi.org/10.1021/pr400878x | J. Proteome Res. 2014, 13, 300−313

Journal of Proteome Research

Article

Figure 1. Flowchart of experiments, data analysis, and data integration performed in this work.

resuspended in 800 μL of lysis buffer (0.1% SDS, 25 mM Tris pH7.8, 1:1000 protease inhibitor cocktail 1 and 2 [Sigma]). In parallel, cells from another 10 cm dish, grown under equal conditions, were trypsinized and counted using a Neubauer chamber. Cell lysates were loaded together with purified proteins (of different concentrations, for those proteins of purity >10%) on SDS gels (5, 7.5, 10, or 12.5%, depending of the molecular weight of the protein to be detected) and separated by electrophoresis; gels were then transferred onto nitrocellulose membrane using the iBlot system (Invitrogen), and the blots were incubated 1 h at room temperature or overnight at 4 °C in TBS + 5% milk. The primary antibody was incubated at 4 °C overnight (1:1000 dilution), and the HRP-coupled secondary

the protein of interest was calculated using the protein-specific extinction coefficient. Of the proteins from our set of ErbB-related proteins, 98 His-tagged proteins were expressed in the bacterial expression system, with 81 proteins expressed and purified in sufficient amounts to be used as standards for quantitative Western blotting, and two proteins were purchased commercially. Ras was purified as previously described.23 Absolute Protein Concentration Determination by Quantitative Western Blotting

HEK293, MCF-7 cell lines, and primary human keratinocytes (isolated form the epidermis of juvenile foreskin) were grown on 10 cm dishes to 80% confluence, washed twice with PBS, and 302

dx.doi.org/10.1021/pr400878x | J. Proteome Res. 2014, 13, 300−313

Journal of Proteome Research

Article

standard curve dilution series of the stable isotope-labeled peptides. To obtain absolute copies per cell numbers, we also took into account the loss during the workup (see previous). For a tutorial on SRM, please see Lange et al. 2008.24 To quantify proteins, for which peptides and transitions were measured by SRM but no stable isotope-labeled peptides were available, we estimated the amount using a calibration curve by plotting the abundance determined using stable isotope-labeled peptides against the sum of the TOP3 transitions.

antibody (1:10000 dilution) was incubated 1 h at room temperature, both in TBS + 0.5% milk. Blots were developed using high-sensitivity ECL reagent (Thermo) and visualized using the Fujifilm LAS-3000 developer. Bands were analyzed using ImageJ, and the number of molecules per cell was calculated using the calibration curve from the purified proteins that were coloaded. Quantitative FACS

Flow cytometry was used to quantify the absolute surface expression of EGFR, ErbB2, and ErbB3 receptors. HEK293, MCF-7, and primary keratinocyte cells were cultured as previously described. Cells were cultured in serum-free medium 14 h prior to the experiments. Cell were detached with trypsin (Gibco) and labeled for 1 h with saturating amounts of FITCconjugated antibodies against EGFR/ErbB1 (sc-120 FITC, Santa Cruz) and ErbB2/Neu (sc-23864 FITC, Santa Cruz). Calibration beads (Quantum Simply Cellular antimouse IgG, Bangs Laboratories) were incubated with the same antibodies following the manufacturers’ instructions. Cells or beads were washed twice with PBS and then analyzed by FACS (LSR II). The number of surface receptors per cell was calculated using the calibration curve from the beads.

Protein Abundance Quantification by Fractionation Followed by Shot-Gun Mass Spectrometry

Cell lysates coming from aliquots of 2 × 106 cells of HEK293 were reduced, alkylated, and digested to peptide mixes according to the filter-aided sample preparation (Wisniewski et al. 2009) method using sequence-grade trypsin at 1:10 ratio (w/w; enzyme/substrate) at 37 °C overnight. Pure synthetic peptides (stable isotope-labeled peptides) corresponding to 17 proteins of interest were then spiked into the digested sample, and tryptic peptides were desalted using a C18 column. For fractionation, 500 μg of tryptic peptides was loaded on a PolyLC column (PolyWAX LP 200 × 2.1 mm, 5 μm, 300-A) for electrostatic repulsion liquid chromatography (ERLIC) fractionation using an Agilent 1200 HPLC chromatographic system. Buffer A: 90% ACN, 0.1% acetic acid, buffer B: 30% ACN, 0.1% formic acid. Flow rate: 0.5 mL/min. Gradient: 0−30% B in 40 min, 30−100% B in 40 min. One minute fractions were collected in a 96-well plate. Fractions were pooled into 10 samples. 10% of each sample was analyzed by LCMS in Orbitrap LTQ Velos. Fractionated peptide mixes were analyzed using an LTQOrbitrap Velos mass spectrometer (Thermo Fisher Scientific, San Jose, CA) coupled to nano-LC (Proxeon, Odense, Denmark) equipped with a reversed-phase chromatography 12-cm column with an inner diameter of 75 μm, packed with 5 μm C18 particles (Nikkyo Technos, Japan). Chromatographic gradients started at 97% buffer A and 3% buffer B with a flow rate of 300 nL/min and gradually increased to 93% buffer A and 7% buffer B in 1 min and to 65% buffer A and 35% buffer B in 60 min. After each analysis, the column was washed for 10 min with 10% buffer A and 90% buffer B (Buffer A: 0.1% formic acid in water; Buffer B: 0.1% formic acid in acetonitrile). The mass spectrometer was operated in positive ionization mode with nanospray voltage set at 2.2 kV and source temperature at 250 °C. Ultramark 1621 for the FT mass analyzer was used for external calibration prior the analyses. Moreover, an internal calibration was also performed using the background polysiloxane ion signal at m/z 445.1200. The instrument was operated in DDA mode and full MS scans with 1 microscan at resolution of 60 000 were used over a mass range of m/z 250− 2000 with detection in the Orbitrap. Auto gain control (AGC) was set to 1e6, dynamic exclusion (60 s with a repeat duration of 30 s [repeat count = 1], and an exclusion list size of 500 ions), and the charge state filter disqualifying singly charged peptides was activated. Following each survey scan, the top 20 most intense ions with multiple charged ions above a threshold ion count of 5000 were selected for fragmentation at normalized collision energy of 35%. Fragment ion spectra produced via collisioninduced dissociation (CID) were acquired in the linear ion trap, AGC was set to 5e4, isolation window of 2.0 m/z, and activation time of 0.1 ms, and maximum injection time of 100 ms was used. All data were acquired with Xcalibur software v2.2. Acquired data were analyzed using the Proteome Discoverer software suite (v1.3.0.339, Thermo Fisher Scientific), and the

Protein Abundance Quantification by Selected Reaction Monitoring Mass Spectrometry

Keratinocyte stem cells, MCF-7 or HEK293 cells (∼1.5 × 107) were resuspended in 800 μL of lysis buffer (8 M urea, 0.2% RapiGest, 0.1 M ammonium bicarbonate), vortexed for 10 s, and shaken at 1000 rpm at room temperature for 10 min. Samples were sonicated (amplitude 90%, cycle 0.6) for 2 × 2 min at 4 °C and centrifuged at 13 200 rpm for 10 min. The clear lysate was transferred to a new tube, the amount of protein was determined using the BCA assay (Pierce, Thermo Scientific), and the protein concentration was diluted to 5 g/L with lysis buffer. The sample was reduced using TCEP (tris-2-carboxyethylphophine) and alkylated using iodoacetamide, and unreacted iodoacetamide was then neutralized with N-acetyl-cystein. The sample was then diluted with 0.1 M ammonium bicarbonate and 0.2% RapiGest to a final concentration of 5-fold) between cell types (on average around 9-fold comparing HEK293 and keratinocytes, in our work, and 7.5-fold comparing the 11 cell lines analyzed in Geiger et al.5). Quantifications of proteins in different cell lines and tissues using antibody- and MS-based proteomics showed that a high fraction of proteins were expressed in most cells and tissues.4,5,37 However, using hierarchical clustering analyses, cells could still be clustered according to the histological relationships between different cell types, which suggests that cell specificity is achieved by the exact regulation of protein levels in space and time rather than which specific proteins are expressed.37 In agreement with

not in another in the same experiment/blot and is thus a good evidence for low or no expression (28% in our Western blot data set). In our data set, the most drastic example is caveolin-1, which is very high expressed (with >1 × 107 molecules/cell) in primary keratinocytes and not detected in the other two cell types (Supporting Information Figure 15C). The high caveolin-1 expression in keratinocytes is also confirmed by SRM−MS. Our detection coverage (75%) by using several methods exceeds previous published data set (49,12 67,13 68,14 and 68% (eliminating 18 unreliable proteins with copy numbers