- © 2011 American Society of Plant Biologists
Abstract
Stable Isotope Labeling by Amino Acids in Cell Culture (SILAC) is a widespread method for metabolic labeling of cells and tissues in quantitative proteomics; however, incomplete incorporation of the label has so far restricted its wider use in plants. Here, we argue that differential labeling by two different versions of the labeled amino acids renders SILAC fully applicable to dark-grown plant cell lines. By comparing Arabidopsis thaliana cell cultures labeled with two versions of heavy Lys (Lys-4 and Lys-8), we show that this simple modification of the SILAC protocol enables similar quantitation accuracy, precision, and reproducibility as conventional SILAC in animal cells.
INTRODUCTION
In a typical Stable Isotope Labeling by Amino Acids in Cell Culture (SILAC) experiment (Ong et al., 2002; Ong and Mann, 2006), cells are grown in a defined medium complemented with certain essential amino acids, usually Arg and Lys, containing naturally occurring atoms (the light culture) or a specific number of their stable isotope counterparts (the heavy culture). After full incorporation of the labeled amino acid, the light and heavy cultures are subjected to different perturbations, harvested, and combined. Peptides resulting from enzymatic digest of the combined cell lysate are then detected by mass spectrometry (MS) in form of ion pairs (doublets), and their ratios reflect the relative changes in protein abundance between the light and the heavy culture (Figure 1A).
The Principle of the Employed SILAC Labeling Strategy.
(A) In a conventional SILAC experiment, light- and heavy-labeled cultures are mixed upon treatment and the relative protein changes are determined from the ratio of heavy and light peptide signals in MS.
(B) Arabidopsis liquid suspension cell cultures were labeled with medium-heavy (Lys-4) and heavy label (Lys-8). Most of the proteins in both cultures were incompletely labeled, with their peptides showing a Lys-0 signal in MS (#). After mixing, peptides were detected as SILAC triplets; the Lys-0 (#) peaks were disregarded in quantitation, whereas the relative protein abundance difference was measured from the ratio of Lys-8/Lys-4 MS peaks (*).
[See online article for color version of this figure.]
Mixing of the light- and heavy-labeled cells for relative quantitation requires full incorporation of the labeled amino acid into the proteome and removal of free amino acids from the medium. In addition, the cell line of interest must be auxotrophic for the amino acid used for labeling. Since plants are autotrophic organisms, these two fundamental requirements have been a serious obstacle to a wider application of SILAC in plants (Thelen and Peck, 2007; Gouw et al., 2010; Schulze and Usadel, 2010). In the only systematic SILAC study of the Arabidopsis thaliana cells, a labeled amino acid incorporation of 80% was reported (Gruhler et al., 2005), which is high but requires elaborate experimental control for accurate quantitation.
Here, we reasoned that the quantification bias introduced by incomplete labeling could be eliminated by a simple modification of the labeling and quantitation strategy. Instead of contrasting two cultures with light (Lys-0) and heavy amino acids (Lys-4 or Lys-8), one culture is labeled with a medium-heavy (Lys-4) and the other with a heavy amino acid (Lys-8). The portion of the proteome that retains endogenously synthesized (unlabeled) amino acids will be the same, allowing for direct quantitative comparison of the two cultures. Mixing of the protein extracts from the two cultures results in detection of peptide triplets in MS (Figure 1B). Peptides containing unlabeled Lys contribute only to the light (Lys-0) member of the triplet, which is disregarded during quantitation, whereas the intensities of Lys-4 and Lys-8 peaks correspond to the relative protein abundances in the two cultures. Therefore, quantitation is performed by measuring the Lys-8/Lys-4 ratio. The underlying principle of this strategy is equal incorporation of labeled amino acids in both analyzed cell cultures, which effectively cancels out the influence of partial labeling.
RESULTS AND DISCUSSION
To test this strategy, we labeled two Arabidopsis cell cultures with Lys-4 and Lys-8, respectively, and showed that both cultures had the same level of Lys incorporation. The incorporation levels were relatively low after only 7 d of culturing on labeled Lys in the dark, where we measured a mean incorporation of 83.6% with a variability of 1.1 percentage points in three biological replicates (see Supplemental Figures 1A to 1C online). After 12 d, we achieved mean incorporation levels of 91% and variability of 1.8 percentage points (see Supplemental Figures 1D to 1F online). This value presented the highest reported incorporation level in plants but still too low for conventional use of SILAC. Interestingly, culturing of the cells in the light led only up to 58% labeling and showed a higher level of variability (see Supplemental Figures 1G and 1H online).
To address the quantitation accuracy, we mixed the protein extracts from Lys-4 and Lys-8 cultures (experiments 7 and 8, both labeled to 93%) in 11 different ratios ranging from 1:10 to 10:1, separated them on a one-dimensional gel, and performed in-gel digestion with endoproteinase Lys-C. We then measured the resulting peptide mixtures on the Orbitrap MS and used the MaxQuant software (Cox and Mann, 2008) to perform SILAC quantitation of the Lys-8/Lys-4 peaks. We achieved a mean absolute deviation from the expected ratios of <11% across all 11 measurements (Figure 2; see Supplemental Data Set 1 online), demonstrating that quantitation accuracy was comparable to that usually reported in SILAC experiments (Ong et al., 2002). To directly compare this approach with the conventional SILAC of animal cells, we performed a similar experiment with a light (Lys-0 and Arg-0) and heavy (Lys-8 and Arg-10) labeled human hapatocyte cell line HuH7. SILAC quantitation of the HuH7 cell line showed a similar quantitation accuracy and precision to that achieved in Arabidopsis (Figure 2), albeit with better accuracy at higher ratios. The results were similar when we used medium-heavy (Lys-4 and Arg-6) and heavy (Lys-8 and Arg-10)-labeled HuH7 cells for comparison (see Supplemental Figure 2 online).
Comparison of the SILAC Quantitation Accuracy in Arabidopsis and HuH7 (Human) Cell Lines for All Measured Samples.
The following mixing ratios of cells labeled with Lys-4 and Lys-8 were measured: 1:10, 1:5, 1:3, 1:2, 1:1.5, 1:1, 1.5:1, 2:1, 3:1, 5:1, and 10:1.
The tight ratio distributions in different Arabidopsis samples confirmed a high precision of quantitation and showed that a reliable quantification of relative protein changes of twofold and up can be routinely achieved (Figure 3; see Supplemental Figure 3 online). We addressed the reproducibility of quantitation by performing a technical replicate, deliberately using Arabidopsis cultures labeled by Lys-4 or Lys-8 to a lower extent (84%) (see Supplemental Figure 4 online). Despite of the different incorporation levels between the experiments, the achieved Pearson correlation coefficient between the replicates was 0.96 (Figure 4), demonstrating the high reproducibility of quantitation for all of the analyzed mixing ratios.
Precision of Quantitation.
Tight distributions of measured SILAC ratios demonstrate a high precision of quantitation and ability to fully resolve differences of twofold and up (only ratios of 1:2, 1:1, and 2:1 are presented; the complete comparison is presented in the Supplemental Figure 3 online).
Reproducibility of Quantitation.
Experiments 1 and 2 (Arabidopsis cells labeled to 84% and mixed in 11 different ratios) are plotted against experiments 7 and 8 (Arabidopsis cells labeled to 93% and mixed in 11 different ratios).
[See online article for color version of this figure.]
Our results demonstrate that the proposed strategy efficiently sidesteps the effects of incomplete SILAC labeling, which so far was a major restriction in the use of SILAC in plant cell lines. However, the strategy also has notable limitations. Since amino acids are used for labeling, this approach cannot be used to label whole plants. Relatively low incorporation levels of the labeled amino acids in the green cell culture currently limit its application to dark-grown cells. Finally, certain biological conditions (e.g., pathogen infection) can potentially alter the uptake and metabolism of amino acids from the medium, which makes the approach less suitable for investigation of such conditions without appropriate controls. Nevertheless, the approach is widely applicable to the analysis of short-term processes (e.g., for analysis of signal transduction) (Macek et al., 2009) or for quantitation of microRNA targets at the proteome level (Selbach et al., 2008). In addition, the approach is simple and also applicable to autotrophic bacterial cell cultures or indeed any cell line or organism where incomplete labeling prevents the use of conventional SILAC.
Similar strategies were proposed for prevention of quantitation bias related to conversion of Arg to Pro in human embryonic stem cells (Van Hoof et al., 2007) and for enabling the use of nondialyzed serum in SILAC culture of human cell lines (Imami et al., 2010). However, the application of this approach was never explored in plant proteomics, where it arguably has the highest application potential.
METHODS
Cell Culture and Protein Extraction for Arabidopsis thaliana Cells
Cell suspension cultures were generated from leaves of Arabidopsis (cv Columbia) plants grown under sterile conditions. Calli obtained on 1% agar plates containing 1.2a medium were transferred to a liquid 1.2a medium (200 mL without agar in 500-mL Erlenmeyer flasks) and grown at 26°C in the dark on a rotary shaker (Infors; 130 rpm). New medium was added every week. This suspension culture was taken as stock for repeated callus formation as described previously (Barjaktarović et al., 2007). The Lys uptake of the cells was tested by adding the amino acid to the medium, monitoring the Lys content daily by HPLC analysis of aliquots, and adjusting it if necessary. For SILAC labeling, the cells of a 6- to 7-d-old stock culture were transferred to a 1.2a medium with 10 mM ammonium nitrate as the sole nitrogen source. This medium was supplemented daily with either 350 μM Lys-4 (4.4.5.5.-2H4) or Lys-8 (13C615N2) (Cambridge Isotope Laboratories). The cells for experiments 1 to 6 were labeled for 7 d and then harvested by filtration and frozen at −80°C. The cells for the experiments 7 to 12 were labeled for 12 d (5 d into the labeling, they were transferred to fresh SILAC media and harvested after additional 7 d of growth). The green cell culture was cultivated under constant light in 1.2a medium with 10 mM ammonium nitrate and 1% Suc.
Protein extraction was based on the method of Niini et al. (1996) and performed according to Barjaktarović et al. (2007) with some modifications. After extraction in lysis buffer, 2.5 mL phenol (BioUltra; Sigma-Aldrich) was added to the sample and shaken for 30 min on ice. After centrifugation, the phenolic phase was washed twice with lysis buffer and proteins precipitated with 5 volumes of acetone. Proteins were pelleted by centrifugation and resuspended in a small volume of denaturation buffer (6 M urea, 2 M thiourea, 10 mM Tris, pH 8.0).
Cell Culture and Protein Preparation for Mammalian Control Cells
Human HuH7 cells were grown under standard conditions using Dulbecco’s modified Eagle’s medium lacking the amino acids Lys and Arg (PAA Laboratories). The medium was supplemented with dialyzed serum (Gibco), l-Gln, antibiotics (penicillin and streptomycin [Gibco], 1% [w/v] each), and the amino acids l-Lys and l-Arg. The unlabeled amino acids (Lys-0 and Arg-0; Sigma-Aldrich) were used for the light SILAC culture, the stable isotope-containing variants Lys-4 and Arg-6 were used for the medium-heavy SILAC culture, and Lys-8 and Arg-10 were used for the heavy SILAC culture. All SILAC amino acids were purchased from Cambridge Isotope Laboratories.
Before lysis, the cells were washed twice with PBS (PAA Laboratories); after complete removal of PBS, denaturation buffer was added to the plate and evenly distributed. After 10 min incubation on ice, the lysed cells were harvested using a cell scraper. After adding Benzonase (Merck) and an additional incubation for 10 min at room temperature and subsequent centrifugation at 13,000g for 15 min at 4°C, the supernatants were transferred into a new tube. Protein concentrations were determined using the Bradford assay (Bio-Rad).
Protein Digestion and MS
Protein extracts of the corresponding sample pairs (Arabidopsis Lys-4 and Lys-8 cultures [experiments 1 and 2 and experiments 7 and 8, respectively, and human cell line HuH7 cultures Lys-4 and Lys-8) were mixed in a series of different ratios (1:10, 1:5, 1:3, 1:2, 1:1.5, 1:1, 1.5:1, 2:1, 3:1, 5:1, and 10:1). Plant samples were then subjected to an additional cleaning step via a short gel run (~1 cm). Gels (NuPAGE 12% precast Bis/Tris gels; Invitrogen) were run and stained using the Novex colloidal blue staining kit (Invitrogen) according to the manufacturer’s instructions. The gel slices were excised, cut into small pieces, and then washed three times in destaining solution (10 mM ammonium bicarbonate [ABC] and acetonitrile [ACN]; 1:1 [v/v]), followed by reduction with 10 mM DTT in 20 mM ABC at 56°C for 45 min and alkylation with 55 mM iodoacetamide in 20 mM ABC at room temperature for 30 min in the dark. After washing in destaining solution, the gel pieces were dehydrated using ACN. The liquid was removed and the gel pieces were swollen at room temperature by adding 13 ng/μL of endoproteinase LysC (Waco) in 20 mM ABC, and digestion was performed overnight at 37°C. Resulting peptides were eluted in three subsequent steps with 30% ACN/3% TFA, 80% ACN/0.5% acetic acid, and 100% ACN. Supernatants were combined, the ACN was evaporated in a vacuum centrifuge, and peptides were prepared for liquid chromatography–MS using C18 StageTips (Ishihama et al., 2006). HuH7 proteins were digested in solution by trypsin as described previously (Borchert et al., 2010).
Samples were analyzed using a Proxeon Easy-LC system (Proxeon Biosystems) coupled to a LTQ-Orbitrap-XL (ThermoFisher) equipped with a nanoelectrospray ion source (Proxeon Biosystems). Chromatographic separation of the peptides was performed using a 15-cm fused silica emitter of 75-μm inner diameter in-house packed with reversed-phase ReproSil-Pur C18-AQ 3 μm resin (Dr. Maisch). The peptide mixture was loaded onto the nano-HPLC column using the IntelliFlow option with a maximum back pressure of 280 bar and subsequently eluted with a segmented gradient of 5 to 80% of solvent B (80% ACN in 0.5% acetic acid) with a constant flow of 200 nL/min over 228 min. Full-scan MS spectra were acquired in a mass range from m/z 300 to 2000 with a resolution of 60,000 in the Orbitrap mass analyzer (after accumulation to a target value of 106 charges in the linear ion trap) using the lock mass option for internal calibration (Olsen et al., 2005). The 10 most intense ions were sequentially isolated for CID fragmentation in the linear ion trap with normalized collision energy of 35% at a target value of 5000. The fragment ions were recorded in the linear ion trap. Up to 500 precursor ion masses selected for MS/MS were dynamically excluded for 90s.
MS Data Processing
Acquired MS spectra were processed using MaxQuant software (Cox and Mann, 2008), version 1.0.14.3. For SILAC quantification, Lys-4 (2H4) and Lys-8 (13C615N2) were specified as medium (M) and heavy labels (H), respectively. Lys-0 (12C614N2) was set as the light label (L). SILAC pairs (H/L) or SILAC triplets (H/M/L) were determined and quantified by the software prior to identification of fragmented peaks. Derived peak lists of MS/MS spectra and accurate precursor masses were searched against a decoy protein database consisting of the IPI databases of human (v3.64), Arabidopsis (v3.62), and 262 commonly observed contaminants using the Mascot search engine (Matrix Science). Initial database search mass tolerances were set to 7 ppm for precursor ions measured in the Orbitrap analyzer and 0.5 D for fragment ions detected in the linear ion trap. Full LysC specificity was required for in silico protein digestion, and up to two missed cleavages were allowed. Carbamidomethylation on Cys was defined as fixed modification; Met oxidation and protein N-terminal acetylation were defined as variable modifications. Identified peptide sequences were further processed by MaxQuant to assemble peptides and protein groups at a specified false discovery rate of 1%. Protein group ratios were calculated from unmodified, Met-oxidized, and N-terminally acetylated unique and razor peptides. At least two quantification events for each protein were required for protein ratio calculation. The requantification option of MaxQuant was disabled leading to fewer quantification events but higher confidence in ratio estimation. To compensate for systematic mixing errors of different SILAC cultures, a normalization factor obtained from the respective 1:1 mixtures was determined and used to normalize the ratio distributions of the different dilution experiments. The accuracy of each dilution experiment was estimated as the distance of the median ratio to the expected ratio; standard deviation of the distribution of ratios was used to determine the precision of experiments.
Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure 1. Incorporation Levels of Lys-4 and Lys-8 Amino Acids in Both Biological Replicates.
Supplemental Figure 2. Comparison of the SILAC Quantitation Accuracy in Arabidopsis and HuH7 (Human) Cell Lines Labeled with Medium-Heavy (Lys-4 and Arg-6) and Heavy (Lys-8 and Arg-10) Amino Acids.
Supplemental Figure 3. Scatterplots Comparing SILAC Ratios Measured in Different Experiments Show That the Ratios of Twofold and up Can Be Fully Resolved at the Peptide and Protein Level.
Supplemental Figure 4. Comparison of the SILAC Quantitation Accuracy in the Arabidopsis Experiments 1 and 2 (Cells Labeled to 84%) and HuH7 Cell Lines for All Measured Samples.
Supplemental Data Set 1. Accuracy and Reproducibility of SILAC Ratio Measurements at the Peptide and Protein Level in All Experiments.
Acknowledgments
We thank Matthias Mann, Detlef Weigel, and Silke Hauf for critically reading the manuscript, H.-P. Fiedler for Lys analysis, K. Harter for kindly providing the green cell culture, and Silke Wahl for excellent technical support. This work is supported by Deutsches Zentrum für Luft- und Raumfahrt e.V. (to R.H.) and Landesstiftung BW (to B.M.).
Footnotes
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: Boris Macek (boris.macek{at}uni-tuebingen.de).
↵1 These authors contributed equally to this work.
↵[C] Some figures in this article are displayed in color online but in black and white in the print edition.
↵[OA] Open Access articles can be viewed online without a subscription.
↵[W] Online version contains Web-only data.
- Received December 8, 2010.
- Revised March 23, 2011.
- Accepted April 18, 2011.
- Published May 3, 2011.
Open Access articles can be viewed online without a subscription.