Peptide Storage: Are You Getting the Best Return on Your Investment? Defining Optimal Storage Conditions for Proteomics Samples Alexandra Kraut, Marle`ne Marcellin, Annie Adrait, Lauriane Kuhn, Mathilde Louwagie, Sylvie Kieffer-Jaquinod, Dorothe´e Lebert, Christophe D. Masselon, Alain Dupuis, Christophe Bruley, Michel Jaquinod, Je´ro ˆ me Garin, and Maighread Gallagher-Gambarelli* CEA, DSV, iRTSV, Laboratoire d’Etude de la Dynamique des Prote´omes, Grenoble, F-38054, France, INSERM, U880, Grenoble, F-38054, France, and Universite´ Joseph Fourier, Grenoble, F-38054, France Received February 6, 2009
Abstract: To comply with current proteomics guidelines, it is often necessary to analyze the same peptide samples several times. Between analyses, the sample must be stored in such a way as to conserve its intrinsic properties, without losing either peptides or signal intensity. This article describes two studies designed to define the optimal storage conditions for peptide samples between analyses. With the use of a label-free strategy, peptide conservation was compared over a 28-day period in three different recipients: standard plastic tubes, glass tubes, and low-adsorption plastic tubes. The results of this study showed that standard plastic tubes are unsuitable for peptide storage over the period studied. Glass tubes were found to perform better than standard plastic, but optimal peptide recovery was achieved using low-adsorption plastic tubes. The peptides showing poor recovery following storage were mainly hydrophobic in nature. The differences in peptide recovery between glass and lowadsorption plastic tubes were further studied using isotopically labeled proteins. This study allowed accurate comparison of peptide recovery between the two tube types within the same LC-MS run. The results of the labelfree study were confirmed. Further, it was possible to demonstrate that peptide recovery in low-adsorption plastic tubes was optimal whatever the peptide concentration stored. Keywords: Peptide Analysis • Recovery • NanoLC-MS • Repeatability • Peptide adsorption • label-free • isotopic label • selected reaction monitoring
Introduction Until recently, most proteomics studies involved generating lists of proteins present in biological samples. Then, the concept of differential or comparative proteomics was introduced: proteins present in one sample but absent from another. Now proteomics has gone quantitative, either relatively or * To whom correspondence should be addressed. Laboratoire d’Etude de la Dynamique des Prote´omes, U880, iRTSV/EDyP, CEA-Grenoble, 17 rue des Martyrs, 38054 Grenoble Cedex 9. Tel: + 33 (0)4 3878 2548. Fax: + 33 (0)4 3878 5051. E-mail:
[email protected].
3778 Journal of Proteome Research 2009, 8, 3778–3785 Published on Web 04/27/2009
absolutely depending on the quantification method used.1-4 To harmonize proteomics data between studies, a number of guidelines have been formulated.5 These recommend both technical and biological replicates in addition to the use of statistical analyses and error estimation. While technical replicates are straightforward to perform, timing is critical to minimize the effect of drifting experimental conditions. Indeed, a recent study demonstrates very clearly that run order is one of the most important features to take into account when comparing proteomes from similar samples.6 In quantitative proteomics studies, it is often necessary to analyze the same peptide samples several times in order to assess the robustness of the method used and hence comply with the current guidelines.5 Between analyses, the sample must be stored in such a way as to conserve its intrinsic properties, without losing either peptides or signal intensity. A state of the art method allowing accurate wide dynamic range quantification would be useless if simply storing a sample as peptides could introduce undefined biases. While numerous studies have been devoted to establishing the optimal conditions for long-term storage of biological samples prior to analysis,7,8 few studies investigate how peptide samples evolve during storage. Bark et al. examined the effect of different types of plastic tube on sample recovery over very short time periods (15 min).9 The authors found that, even over this short time period, tubes treated to minimize peptide binding were preferable to standard plastic tubes. However, plastic may not be the best material to use for peptide sample storage. In effect, another study10 examined the effect of glass or plastic on sample adhesion to injection vials and found glass to be preferable for the recovery of hydrophobic peptides. While both studies demonstrated an effect due to the container on sample recovery, they used single-protein, low-complexity samples, and neither of them examined long-term storage of samples. For quantitative studies, samples are often highly complex and they may be stored for extended periods. In the present study, we analyze the effect of storage conditions over a relatively long period (28-days). In our laboratory, over 80% of samples are processed within this time frame, with an average interval between sample preparation and mass spectrometry analysis of 14 days. A 12-protein standard sample was used to compare peptide elution LC-MS profiles and intensities from peptide samples stored in different 10.1021/pr900095u CCC: $40.75
2009 American Chemical Society
Defining Optimal Storage Conditions for Proteomics Samples
technical notes
Figure 1. (A) Schema of sample preparation and storage for the label-free study. Different container types are indicated by different colors and M, Mµlti; C, Chromacol; E, Eppendorf. (B) Schema of sample preparation for the labeled study. Different container types are indicated by different colors and M, Mµlti; C, Chromacol. Heavy- or light-spiked peptide samples are indicated by the peptide amount in micrograms followed by H or L, respectively, and by different-shaded triangles.
containers either as dried-down peptides or as resuspended peptides. On the basis of preliminary experiments, we selected three different storage containers: (1) Eppendorf tubes, (2) Mµlti low adsorption tubes, and (3) Chromacol glass inserts. A second part to our study was set up to examine whether the amount of stored peptide affected peptide recovery following storage. For this, we spiked isotopically labeled or unlabeled standard proteins into the 12-protein mixture samples. Following storage, differences in peptide conservation between storage vessels could be compared in the same LC-MS run by mixing samples just prior to analysis. Accurate comparison was therefore possible between different storage conditions. Taken together, these results lead to a clear-cut conclusion: peptides should be dried down in Mµlti low adsorption plastic tubes and stored at -20 °C before analysis. These results have redefined peptide storage conditions in our laboratory and considerably improved analytical reproducibility for late-eluting and hydrophobic peptides in our hands.
Experimental Procedures Protein Samples. SigmaMarker Wide Range molecular weight marker (Sigma, France) was used as the protein sample. The 12 proteins making up the sample were rabbit myosin, Escherichia coli β-galactosidase, rabbit phosphorylase b, bovine serum albumin, bovine glutamic dehydrogenase, chicken ovalbumin, rabbit glyceraldehyde-3-phosphate dehydrogenase, bovine carbonic anhydrase, bovine trypsinogen, soybean trypsin inhibitor, bovine R-lactalbumin, and bovine aprotinin. The sample was resuspended in deionized water as described in the manufacturer’s instructions. Both isotopically labeled (Arg-13C615N4, Lys-13C615N2) and unlabeled forms of lactate dehydrogenase (LDH, UniProtKB/ Swiss-Prot P07195), and isotopically labeled human creatine kinase type M (muscle isoform) (CKM, UniProtKB/Swiss-Prot P06732) were prepared as described by Brun et al.4 Light CKM was de facto native human CKMB heterodimer complex purchased from AppliChem GmbH (Germany). For ease of reference, these 4 proteins will be referred to as heavy (labeled) or light (unlabeled) standard proteins hereafter.
CKM/LDH-spiked samples were prepared as follows: for the heavy-spiked sample, the two heavy standard proteins were mixed before spiking into the 12-protein sample in a 1:9 ratio (2.5 µg of isotopic standard per 22.5 µg of 12-proteins). For the light-spiked sample, the light standard proteins were mixed before spiking into the 12-protein sample in a 1:9 ratio (2.5 µg of isotopic standard per 22.5 µg of 12-proteins). Label-Free Study. 1. Peptide Generation. Figure 1A shows the workflow and conditions for the label-free study. Four identical 12-protein samples (6 µg) were loaded on a NuPAGE 4-12% gradient gel (Invitrogen, France). The migration was stopped just after the sample had entered the gel, causing the 12 proteins to be “stacked” as a single band. Gels were fixed in ethanol 30%, acetic acid 7.5% before coloring with Biosafe gel stain (Biorad, France). Gel bands were cut out using a Biopsy punch (LCH, France) and placed in polypropylene 96-well plates for decoloration and digestion. Protein band decoloration, protein oxidation, trypsic digestion and peptide extraction steps were carried out on an EVO150 robot (Tecan, France) as described previously.11 2. Peptide Drying and Storage. Supernatants from peptide extraction for each of the 4 protein bands were split between 9 tubes (triplicate in each of the 3 types of container): glass inserts (Chromacol; Cluzeau, France), 0.5 mL microtube (Eppendorf; Fischer, France), and low adsorption 0.5 mL microtube (Mµlti; Dutscher, France). Samples were then dried under vacuum. Each tube thus contained peptides generated from 0.66 µg of 12-protein mixture. Dried-down peptides extracted from a single gel band were stored immediately at -20 °C. Peptides from the remaining 3 gel bands were resuspended with vigorous vortexing in 10 µL of trifluoroacetic acid 5%; 10 µL of buffer C (acetonitrile 5%, formic acid 0.2% in water) was added and samples were sonicated 5 min. One aliquot of each of the three bands in each type of tube was analyzed immediately by mass spectrometry (D0 analysis). Another aliquot from each type of tube was stored at -20 °C in liquid form and the third aliquot was dried once again under vacuum before storage at -20 °C. After 28 Journal of Proteome Research • Vol. 8, No. 7, 2009 3779
technical notes days’ storage (D 28), peptides were resuspended (where necessary), as described above, and analyzed by mass spectrometry. 3. Mass Spectrometry Analyses. To be as reproducible as possible between triplicate samples and between container conditions, samples were resuspended in the same way for all mass spectrometry analysis runs. Triplicate samples were resuspended at the same time. The first of the triplicate analyses was always injected immediately, the second and the third had always been on the autosampler for 3 and 6 h prior to injection, respectively. The solubilized peptides were analyzed by nanoLiquid Chromatography (LC)-Mass Spectrometry (MS) (CapLC and Q-TOF Ultima, Waters, Milford, MA). The method consisted of a 60 min run at a flow rate of 200-300 nL/min using a binary solvent gradient: solvent A (2% acetonitrile/97.9% water/0.1% formic acid) and solvent B (80% acetonitrile/19.9% water/0.1% formic acid); gradient 10-90% B. The nanoLC system included a 300 µm × 5 mm C18 precolumn (LC-Packings, Dionex, Sunnivale, CA) to concentrate peptides before injection onto a 75 µm × 150 mm C18 PepMap column (LC-Packings) coupled with the mass spectrometer (Q-TOF Ultima) through a liquid junction. To generate a list of specific peptides from the 12-protein mixture, a separate sample was analyzed by tandem mass spectrometry (LC-MS/MS) on the same instrumental setup as that used for LC-MS analysis. Fragmentation was in datadependent mode using up to 5 fragmentation channels, requiring a minimum threshold of 15 to trigger fragmentation. Fragmentation was stopped when total ion current fell below the threshold. For the more intense peptides, repeated fragmentation was allowed for up to 5 s. 4. Data Analysis. LC-MS/MS data were processed using MassLynx 4.0 software (Waters) to generate a peak list file. This file was submitted to Mascot 2.2 (Matrix Science, U.K.) for identification against the combined SwissProt-trEMBL database (Swiss-Prot release 50.1; trEMBL release 33.1; 3 188 856 sequences) without specifying taxonomy. The search parameters were as follows: enzyme, trypsin/P; variable modifications, Acetyl (Protein N-term), Deamidated (NQ), Dioxidation (M), Oxidation (M), Trioxidation (C); 1 missed cleavage allowed; precursor mass tolerance, 0.4 Da; fragment mass tolerance, 0.4 Da. LC-MS data for the different samples were compared using reconstituted chromatograms generated by MassLynx 4.0 software (Waters). Data from this analysis for the different time points and different storage conditions are available in the Supporting Information (Table “raw data”). Figures showing density ellipses were generated using JMP (v7) (SAS software). Density ellipses were drawn to contain 90% of data points. Statistical analyses were performed by one-way Anova plots using JMP. Differences showing a p-value
