Characterization of the Rat Liver Membrane Proteome Using Peptide Immobilized pH Gradient Isoelectric Focusing J. M. Chick,† P. A. Haynes,†,‡ M. P. Molloy,†,‡ B. Bjellqvist,§ M. S. Baker,*,†,‡ and A. C. L. Len†,‡ Department of Chemistry & Biomolecular Sciences, Macquarie University, NSW, 2109, Australia, Australian Proteome Analysis Facility Ltd., Macquarie University, NSW, 2109, Australia, and GE Healthcare, Amersham Biosciences AB, Björkgatan 30, SE-751 84 Uppsala, Sweden Received September 20, 2007
Membrane proteins are of particular interest in proteomics because of their potential therapeutic utility. Past proteomic approaches used to investigate membrane proteins have only been partially successful at providing a comprehensive analysis due to the inherently hydrophobic nature and low abundance for some of these proteins. Recently, these difficulties have been improved by analyzing membrane protein enriched samples using shotgun proteomics. In addition, the recent application of methanolassisted trypsin digestion of membrane proteins has been shown to be a method to improve membrane protein identifications. In this study, a comparison of different concentrations of methanol was assessed for assisting membrane protein digestion with trypsin prior to analysis using a gel-based shotgun proteomics approach called peptide immobilized pH gradient isoelectric focusing (IPG-IEF). We demonstrate the use of peptide IEF on pH 3–10 IPG strips as the first dimension of two-dimensional shotgun proteomics for protein identifications from the membrane fraction of rat liver. Tryptic digestion of proteins was carried out in varying concentrations of methanol in 10 mM ammonium bicarbonate: 0% (v/v), 40% (v/v), and 60% (v/v). A total of 800 proteins were identified from 60% (v/v) methanol, which increased the protein identifications by 17% and 14% compared to 0% (v/v) methanol and 40% (v/v) methanol assisted digestion, respectively. In total, 1549 nonredundant proteins were identified from all three concentrations of methanol including 690 (42%) integral membrane proteins of which 626 of these proteins contained at least one transmembrane domain. Peptide IPG-IEF separation of peptides was successful as the peptides were separated into discrete pI regions with high resolution. The results from this study prove utility of 60% (v/v) methanol assisted digestion in conjunction with peptide IPG-IEF as an optimal shotgun proteomics technique for the separation and identification of previously unreported membrane proteins. Keywords: membrane proteins • shotgun proteomics • isoelectric focusing • protein identifications • mass spectrometry
Introduction Analysis of membrane proteins is an important field in proteomics because membrane proteins are represented by 30% of the genome1 and constitute approximately 70% of all human protein based drug targets.2 Analysis of membrane proteins has been notoriously difficult, which has been demonstrated by their under-representation in 2D gels.3–5 The majority of proteins presented by 2D gels has a tendency to be hydrophilic with grand average hydrophobicity (GRAVY) scores of less than 0.3.3 The problems associated with membrane protein analysis have been attributed to poor extraction and/or poor solubility3 either during sample preparation or * Corresponding author. Professor Mark S. Baker, Australian Proteome Analysis Facility, Building F7B4, Research Park Drive, Macquarie University 2109, Australia. Telephone: +61 2 9850 6209. Fax: +61 2 9850 6200. E-mail:
[email protected]. † Department of Chemistry & Biomolecular Sciences, Macquarie University. ‡ Australian Proteome Analysis Facility Ltd., Macquarie University. § GE Healthcare, Amersham Biosciences AB.
1036 Journal of Proteome Research 2008, 7, 1036–1045 Published on Web 01/23/2008
during protein separation such as the first dimension of twodimensional gel electrophoresis (2D-GE).6 Although sodium dodecyl sulfate (SDS) remains the most efficient choice of solubilizing reagent well suited to 1D gels, this surfactant hampers other techniques such as two-dimensional liquid chromatography and two-dimensional gel electrophoresis (2DGE). Improvements in membrane protein analysis have occurred through increasing protein solubility during sample preparation with the aid of solvents. Trypsin retains enzymatic activity in various solvent environments.7 Methanol, isopropanol, acetonitrile, and acetone have all been shown to be compatible with trypsin enzymatic activity,7 and recently, 60% (v/v) methanol has shown utility for membrane protein digestion.8 Membrane protein digestion studies have also investigated the use of alternative enzymes to or in conjunction with trypsin9 and the use of acid labile salts for increasing protein solubility.10 SDS solubilization of membrane proteins has also been adapted to strong cation exchange (SCX) fractionation but only at very low 10.1021/pr700611w CCC: $40.75
2008 American Chemical Society
research articles
Membrane Protein Analysis Using Peptide IPG-IEF concentrations so that the charge of the detergent has little affect on SCX fractionation.11 For solubility reasons, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of membrane proteins has been successful in providing a large coverage of membrane proteomes because SDS has such high solubilizing power.12–15 Alternative surfactants have also been used for solubilizing membrane proteins for 2D-GE to compensate for the incompatibility of SDS with 2D-GE.4,16–18 For protein identification studies, shotgun proteomics methods have alleviated some of the problems associated with membrane protein analysis. Earlier shotgun proteomics approaches utilized MudPIT19 for the identification of membrane proteins.21 Recently alternative approaches using offline SCX chromatography8,11,20 and SDS-PAGE based techniques12–14 have been employed. For example, Wang and colleagues have recently identified 7800 mouse brain proteins, including 1447 integral membrane proteins by fractionating 5 mg of peptides using SCX fractionation and analysis with a linear ion trap MS/ MS.21 Many shotgun proteomic methods typically use fast scanning, low resolution mass measurements, which require filtering methods to minimize false positive identifications.22–24 Researchers have demonstrated an overall improvement in membrane protein identifications using shotgun proteomics in comparison to the number identified from 2D gels, especially proteins that are inherently hydrophobic.8,10,21,25 The use of SDS-PAGE-based shotgun proteomics is hampered by poor efficiency of in-gel digestion of proteins and the extraction of peptides from the polyacrylamide gel.26 Recently, an alternative to using SCX or SDS-PAGE to separate peptides or proteins prior to LC-MS/MS analysis was described where trypsin digested proteins were separated over an IPG using IEF, and this technique was termed peptide IPG-IEF.22,27 Analysis of tryptic peptides from proteins from Escherichia coli whole cell lysate using peptide IPG-IEF on a pH 3–10 linear IPG strip allowed for the identification of 1223 proteins from LC-MS/MS analysis using a three-dimensional ion trap mass spectrometer.27 Recently, linear ion trap FT-ICR mass spectrometry has identified approximately 1700 proteins from a pH 3–10 nonlinear IPG strip from Drosophila melanogaster nuclear extracts.24 Peptide IPG-IEF was advantageous as peptide pI was demonstrated to be an accurate primary filtering criterion to reduce false positive and false negative identifications.22–24,27,28 In these studies, peptide pI was used as part of a two-step sequential filtering method which was used in addition to either false positive rate analysis24 or using Xcorr cutoff scores.22,23,29 The advantages of peptide pI filtering have not been evaluated for a two-step sequential filtering approach which uses false discovery rates (protFDR) at the protein level for assigning high confidence protein identifications. The distinction between protFDR is that it is calculated at the protein level, where protein scores are ranked and a threshold level is set beyond which assignments are incorrect. However, false positive rates (pepFPR) are calculated at the peptide level in which the following formula applies, 2n(reverse)/n(reverse) + n(forward),24 where reverse refers to reverse database assignments and forward refers to forward database assignments. The aim of this study was to evaluate peptide IPG-IEF as a technique for shotgun proteomics for the characterization of the rat liver membrane proteome. We used this technique to investigate the addition of methanol to the digestion solution for trypsin digestion. Here we also demonstrate the utility of a two-step sequential filtering approach based on orthogonal
peptide and protein characteristics (peptide pI and protFDR) for assigning high confidence protein identifications. This sequential filtering approach reduces the discovery of false positive protein identifications and helps to recover false negative protein identifications. Our results demonstrate that peptide IPG-IEF is a valuable alternative technique to traditional shotgun proteomics techniques for the separation of peptides derived from trypsin digestion of membrane proteins.
Materials and Methods Membrane Protein Isolation and Preparation for Peptide Immobilized pH Gradient Isoelectric Focusing (IPG-IEF). Rat livers obtained from 8 week old Dark Agouti rats (Save Sight Institute, Sydney Eye Hospital, Australia) were perfused with 0.9% (w/v) phosphate buffered saline. Rat liver tissue (1.5 g) was then homogenized in 10 mM HEPES buffer (pH 7.8) supplemented with 2 mM NaCl, 10 mM NaOH, 500 mM EDTA, and protease inhibitors (3 mg of antipain-dihydrochloride, 0.5 mg of aprotinin, 0.5 mg of bestatin, 1 mg of chymostatin, 3 mg of E-64, 10 mg of EDTA-Na2, 0.5 mg of leupeptin, 20 mg of Pefabloc SC, 0.5 mg of pepstatin, 3 mg of phosphoramidon) (Roche Diagnostics, Switzerland) using an Omni TH homogenizer (Omni International Inc., VA, USA). The homogenized tissue was then centrifuged at 13 000g for 15 min at room temperature (RT). Membrane proteins were isolated using a modified sodium carbonate stripping method.4,30 Briefly, the supernatant was collected and diluted to a final volume of 40 mL in 0.1 M sodium carbonate (pH 11) then incubated for 1 h rotating at 4 °C. The carbonate-treated membranes were sedimented by ultracentrifugation at 120 000g for 1 h at 4 °C. The supernatant was removed, and the membrane pellet was washed once with 0.1 M sodium carbonate (pH 11) and resuspended in 10 mM ammonium bicarbonate (NH4HCO3, pH 7.8). Sample was then transferred to a 20 mL glass scintillation vial and pulse sonicated using a Branson 450 Sonifier (Branson, Danbury, USA) using 2 s bursts for 15 intervals on ice. The sonicated sample was then reduced with 10 mM dithiothreitol for 1 h at 37 °C and subsequently alkylated with 55 mM iodoacetamide at RT for 30 min. Protein quantification was conducted by Bradford Assay (Sigma-Aldrich, MO, USA). Trypsin digestion was conducted in 10 mM NH4HCO3 (pH 7.8) with three different concentrations of methanol, which included 0%, (v/ v) 40% (v/v), and 60% (v/v). A 300 µg aliquot of protein was digested for each experimental replicate with a total of 6 µg of trypsin. The three trypsin digestions were carried out in two stages: 3 µg of trypsin was added to each followed by a 20 min water bath sonication and incubation at 37 °C for 1.5 h followed by a further 3 µg addition and incubation at 37 °C for a further 1.5 h. All three digested samples were evaporated to dryness in a vacuum centrifuge and resuspended in 250 µL of 8 M urea supplemented with a trace of bromophenol blue. Peptide IPG-IEF. Digested proteins (300 µg) from each of the three digestion strategies were used to passively rehydrate a linear pH 3–10 (in which the actual pH range of the IPG strip is 3.85-9.36, information available at www.gehealthcare.com) 18 cm IPG strips (GE Healthcare) for 6 h at room temperature. Isoelectric focusing was conducted on an IPGphorII (GE Healthcare) with a current limit of 50 µA per strip at 20 °C with the following focusing program: 300 V for 1 h, a gradient to 1000 V for 1 h, a gradient to 4000 V for 3 h, a gradient to 8000 V for 3 h and 8000 V until 100 kVh was reached. The strips were then cut (with plastic backing still in place) with a scalpel Journal of Proteome Research • Vol. 7, No. 3, 2008 1037
research articles blade into 24 equal length pieces. Peptides were extracted from each fraction by incubation in 100 µL of 0.1% (v/v) formic acid for 1 h at RT. The extraction was repeated twice and subsequently combined with the initial fractions. Combined peptide extracts from each fraction were concentrated in a vacuum centrifuge to approximately 30 µL. Each fraction was desalted using C18 tips (PerfectPure C18 Tips) (Eppendorf, Germany) and the eluate was dried using a vacuum centrifuge followed by resuspension in 0.1% (v/v) formic acid in preparation for nanoLC-MS/MS. Liquid Chromatography-Mass Spectrometry. Each of the 24 fractions from the three trypsin digestion methods was analyzed by nanoLC-MS/MS using a LCQ-Deca ion-trap mass spectrometer (ThermoFinnigan, CA, USA) according to Breci et al.31 Reversed phase columns were packed in-house to approximately 7 cm (100 µm i.d.) using 100 Å, 5 mM Zorbax C18 resin (Agilent Technologies, CA, USA), in a fused silica capillary with an integrated electrospray tip. A 1.8 kV electrospray voltage was applied via a liquid junction upstream of the C18 column. Samples were injected onto the C18 column using a Surveyor autosampler. Each sample was loaded onto the C18 column followed by an initial wash step with buffer A (5% (v/ v) ACN, 0.1% (v/v) formic acid) for 10 min at 1µL/min. Peptides were subsequently eluted from the C18 column with 0–50% buffer B (95% (v/v) ACN, 0.1% (v/v) formic acid) for 58 min at 500 nL/min followed by 50–95% buffer B for 5 min at 500 nL/ min. The column eluate was directed into a nanospray ionization source of the mass spectrometer. Spectra were scanned over the range 400–1500 amu. Automated peak recognition, dynamic exclusion, and tandem MS of the top three most intense precursor ions at 40% normalization collision energy were performed using the Xcalibur software (ThermoFinnigan, CA, USA). Protein and Peptide Identification. Raw files were converted to mzXML format and processed through the global proteome machine (GPM) software, freely available from www.thegpm.org.32,33 For each experiment, the 24 fractions were processed sequentially with output files for each individual fraction, and a merged, nonredundant output file was generated for protein identifications with log(e) values less than -1. These parameters for protein identification represent, and are referred to in this publication as, the minimal filtering criteria. Peptide identification was determined using a 0.4 Da fragment ion tolerance. Carbamidomethyl was considered as a complete modification, and partial modifications were also considered, which included oxidation of methionine and threonine and deamidation of asparagine and glutamine. MS/ MS spectra were searched against the Rattus norvegicus database (Database derived from SwissProt, Ensemble and NCBI), and reverse database searching was used for estimating false discovery rates.34 Protein identifications were validated using two filtering methods. The first filtering method uses 1% protFDR, as assessed by reverse database searching,30 which was applied to nonredundant protein lists (method A). ProtFDR is calculated by the following formula: reverse/(reverse + forward)*100. The second filtering method uses peptide pI filtering as the initial filtering criteria followed by imposing a 1% false discovery rate at the protein level (method B). Peptide pI was calculated through an open source pI calculator, Compute pI from ExPASy (http://au.expasy.org/tools/pi_tool.html), which calculates peptide pI from amino acid values described by Bjellqvist et al.35 Peptide pI filtering was conducted on the data obtained from 1038
Journal of Proteome Research • Vol. 7, No. 3, 2008
Chick et al. each of the fractions by removal of statistical outliers that were defined as peptides with pI values that fall outside the pI boundaries, determined using ( 0.5 of the standard deviation (calculated over the entire fraction) from the median pI value of the fraction. Protein hydrophobicity was calculated through GRandAVerage hydrophobicitY (GRAVY) analysis using the ProtParam algorithm (http://au.expasy.org/tools/protparam.html). Transmembrane segment annotation of the identified proteins was determined using either SwissProt annotation (www.ebi.ac.uk/ swissprot) or the TransMembrane Hidden Markov Model (TMHMM, http://www.cbs.dtu.dk/services/TMHMM-2.0/).36–38 The cellular location of identified proteins was derived from SwissProt annotation. Statistical analysis was conducted with Microsoft Excel using a student t test with two samples assuming unequal variance with P < 0.05 for assignment of statistical confidence.
Results Resolution of Identified Peptides within IEF Experiments. Rat liver membrane proteins were purified using a modified sodium carbonate extraction method4,30 and digested with one of three different trypsin digestion solutions containing varying concentrations of methanol (0%, 40%, and 60%) in 10 mM ammonium bicarbonate (pH 7.8). Tryptic peptides were separated by IPG-IEF with an 18 cm linear pH 3–10 gradient, extracted, and analyzed by nanoLC-MS/MS. To evaluate the separation efficiency of peptide IPG-IEF, three replicate experiments were carried out, and the number of peptides per fraction was calculated as an average of three replicate experiments for each digestion condition (Figure 1). The majority of peptides distributed throughout the 24 fractions before peptide pI filtering was separated into three distinct pI regions (theoretical): pI 3 to 5, 5.5 to 7, and 8 to 10 which corresponded to fractions 2–6, 9–16, and 20–24, respectively (Figure 1A). The variability in the number of peptides per fraction was higher in the basic end of the IPG strip compared to the acidic to neutral fractions. In comparing the number of peptides identified in each fraction from the three concentrations of methanol, there was little observable difference, with exception to fraction 20, which demonstrated large variability. After filtering and validating the data set by using the pI of the peptides and removing statistical outliers, the peptide distribution pattern was unaltered for all three digestion conditions (Figure 1B). In addition, peptide pI filtering demonstrated that between 21 and 23% of identified peptides were statistical outliers over the 24 fractions. The basic region (fractions 17–24) of the IPG strip contained the most statistical outliers with between 41 and 43% of all peptides within this region. To demonstrate that the separation of the peptides follows the linearity of the pH range of the IPG strip, the average peptide pI per fraction was calculated for the triplicates of each methanol assisted digest (Figure 2A). The average experimental peptide pI per fraction before peptide pI filtering shows some correlation between average pI in individual fractions and the actual pH of these fractions. The correlation between theoretical and observed pI for individual fraction shows no correlation in fractions 7–9, the basic region (fractions 17–23), and the ends (fractions 1 and 24) of the IPG strip. The standard deviation of the pI as calculated from the peptides was observed to be in the range 0.18–0.28 pI units. After pI filtering of individual fractions (Figure 2B), average peptide pI per fraction and the actual pH of the IPG strip correlated well within the acidic to
Membrane Protein Analysis Using Peptide IPG-IEF
research articles
Figure 1. IPG-IEF separation of peptides throughout the 24 fractions obtained from the 18 cm, pI 3–10 IPG strip before and after pI filtering. (A and B) The average number of peptides per fraction over the 24 fractions of the IPG strips for each of the methanol concentrations used.
neutral pH regions suggesting that separation of peptides by IPG-IEF is accurate and reproducible. However, the correlation between observed peptide pI and the pH of the IPG strip was not optimal within the basic region. Interestingly, the standard deviation for peptide pI for fractions 21 and 22 in the basic end of the IPG strip was approximately 0.2 pI units, suggesting good reproducibility, but differences in observed and theoretical pI do exist. Fraction 19 demonstrated large variability in average pI after pI filtering. The low number of peptides in this fraction which is consistent with previous publications22,24 may influence the average pI of this fraction between experimental replicates. The resolving capabilities of IPG-IEF of tryptic peptides was assessed by analyzing whether the peptides identified were present in one or more fractions (Table 1). On average, 84.7–86.8% of all peptides were only detected from a single fraction, and another 10.9–11.2% of all peptides were found in two fractions. There were minimal differences in this distribution that could be attributed to the concentrations of methanol used to assist protein digestion. After peptide pI filtering, the percentages of peptides found in one or two fractions slightly increased (1–3%) suggesting the removal of some peptides found in greater than two fractions as statistical outliers. These data provide a strong argument for the use of peptide pI filtering as an additional filtering constraint for assigning high-confidence protein identifications. Protein Identifications. The total number of protein identifications obtained from the membrane preparations was
assessed using two filtering methods as detailed above (Figure 3). Filtering method A (1% protFDR) resulted in 506 proteins (0% [v/v] methanol), 405 proteins, (40% [v/v] methanol), and 652 proteins (60% [v/v] methanol) on average from three experimental replicates. Filtering method A has been shown to be prone to false negative identifications,22–24,27–29,39 so to recover these proteins, filtering method B was used. Filtering method B consisted of peptide pI as an initial filtering criterion before imposing a 1% protFDR at the protein level. This analysis resulted in 623 proteins (0% [v/v] methanol), 687 proteins (40% [v/v] methanol), and 800 proteins (60% [v/v] methanol) on average (Supporting Information Table 1), representing a 19–41% increase in protein identifications compared to filtering method A. These data provide evidence that peptide pI filtering is a useful method for improving the protFDR of protein identifications and to decrease false negative identifications. Comparison between the concentrations of methanol used to assist membrane protein digestion showed that 60% (v/v) methanol allowed increased recovery in total number of proteins compared, 0% (v/v) and 40% (v/v) methanol, for both filtering methods. From high confidence protein identifications using filtering method B, a 17.6% (P < 0.05) and 14% increase in protein identifications was observed for 60% (v/v) methanol compared to 0% (v/v) and 40% (v/v) methanol, respectively. These results demonstrate that 60% (v/v) methanol is the optimal concentration for digestion of proteins with trypsin for increasing the proteome coverage and in combination with Journal of Proteome Research • Vol. 7, No. 3, 2008 1039
research articles
Chick et al.
Figure 2. Average peptide pI per fraction for each concentration of methanol before (A) and after (B) pI filtering and the actual pI values of the IPG strip.
peptide IPG-IEF is a valuable method for membrane protein shotgun proteomics. A total of 1549 unique protein identifications, including 690 integral membrane proteins, of which 626 were transmembrane proteins, were identified from the three concentrations of methanol (Supporting Information Table 2). Integral Membrane Protein Identifications. Of the nonredundant proteins identified from the three digests, between 44 and 49% were classed as integral membrane proteins using either SwissProt or the TransMembrane Hidden Markov Model. Transmembrane proteins were the main type of integral membrane proteins identified with 419, 421, and 513 found when using 0%, 40% (v/v), and 60% (v/v) methanol, respectively (Figure 4). This analysis of transembrane proteins provides compelling evidence for the compatibility of peptide IPG-IEF shotgun proteomics for the analysis of transmembrane proteins. Approximately 50% of the identified transmembrane proteins possessed a single transmembrane domain; moreover, peptides were also recovered from proteins with up to 19 transmembrane domains such as the voltage-dependent R-type 1040
Journal of Proteome Research • Vol. 7, No. 3, 2008
calcium channel subunit alpha-1E. When the number of transmembrane domains for identified transmembrane proteins across the different digestion strategies was compared, it was found that 60% (v/v) methanol demonstrated an increase in each of the categories ranging from a single transmembrane domain up to 12 transmembrane domains. The largest difference was observed in the number of proteins identified with a single transmembrane domain in which 60% (v/v) methanol provided 16% more proteins. Integral membrane proteins were not only made up of transmembrane proteins, but other types of integral membrane proteins from the SWISS-PROT database were also annotated. Four porin proteins were identified, which included voltage dependent anion channel (VDAC) proteins. A subset of proteins were annotated as integral membrane proteins but were not annotated as transmembrane or porin proteins, which included many subunits of the electron transport chain. Lastly, as high as 27 integral membrane proteins were identified as possessing
research articles
Membrane Protein Analysis Using Peptide IPG-IEF Table 1. Determining the Extent of Peptide Distribution across IPG Fractions Before Peptide pI Filtering Number of Fractions
methanol concentrations
1
0% Methanol 40% Methanol 60% Methanol
84.71 ( 0.002% 84.97 ( 0.02% 83.18 ( 0.02%
a
2
3
4
g5
10.88 ( 0.00% 10.90 ( 0.01% 11.30 ( 0.01%
2.46 ( 0.001% 3.18 ( 0.004% 2.71 ( 0.003%
1.06 ( 0.001% 1.41 ( 0.003% 1.37 ( 0.004%
1.30 ( 0.000% 1.70 ( 0.005% 1.28 ( 0.01%
After Peptide pI Filtering
a
Number of Fractions
methanol concentrations
1
2
3
4
g5
0% Methanol 40% Methanol 60% Methanol
86.76 ( 0.005% 87.32 ( 0.006% 84.95 ( 0.019%
11.21 ( 0.002% 11.18 ( 0.01% 11.27 ( 0.008%
1.56 ( 0.003% 1.91 ( 0.006% 2.21 ( 0.007%
0.49 ( 0.001% 0.84 ( 0.003% 1.02 ( 0.004%
0.06 ( 0.00% 0.34 ( 0.003% 0.55 ( 0.003%
Values shown indicate the number of fractions a peptide is identified from, expressed as a percentage of the total number of peptides identified.
Figure 3. Comparison of data filtering methods A and B and a comparison of methanol concentrations for assisting protein digestion with trypsin. * Student t test comparison between 60% methanol assisted digest and 0% methanol assisted digest using filtering method B.
Figure 4. Number of transmembrane proteins identified from the three digestion strategies categorized by their number of transmembrane segments.
lipid anchor modifications such as palmitate, myristate, prenyl, and glycophosphoinositol. In addition to annotating membrane proteins through predicting protein secondary structures, protein identifications provided in the literature have also been presented through hydrophobicity analysis.40 To illustrate the diverse array of proteins identified through peptide IPG-IEF shotgun proteomics, identified proteins were ranked based on GRAVY scores (Figure 5A). Approximately 22% of all proteins identified
displayed a GRAVY score greater than zero. Furthermore, approximately 5% (82) of the nonredundant proteins identified from all three digests had a GRAVY score greater than 0.3, a barrier noted by others,3,5 above which most proteins are refractory to analysis by methods like 2D-GE. Increasing the organic solvent content of digestion did not considerably improve the ability to recover hydrophobic proteins, despite being of value to increase overall numbers of recovered proteins. Nonetheless, GRAVY analysis performed on the identified integral membrane proteins (Figure 5B) demonstrated that 60% methanol led to a slight increase (1.2%) in the number of hydrophobic integral membrane proteins identified. Cellular Location of Proteins Identified from Each Digest. A comparison of the percentage of nonredundant proteins based on cellular locations demonstrated no significant difference between methanol assisted digests (data not shown). As a representation of the percent coverage of proteins based on cellular location, proteins identified through 60% (v/ v) methanol assisted digestion are presented in Figure 6. In each data set, 31% of the protein identifications had no clear cellular location. Cytoplasmic proteins made up 18% of the data set with proteins such as tubulin, actin, and ribosomal proteins identified. Proteins localized to the mitochondria constituted 14% of the data set with many proteins derived from the electron transport chain. Plasma membrane proteins were also well represented, being 10% of the identified proteins. In particular, solute carriers, receptors such as the EGF receptor, and transferrin receptors and transporters were observed. The endoplasmic reticulum made up 9% of the identified proteins with 43 identified cytochrome p450 enzymes. Lastly, the nucleus, secreted/extracellular, microsome, peroxisome, golgi, and miscellaneous locations each represented less than 5% of the entire proteome.
Discussion The aim of this study was to evaluate peptide IPG-IEF as a valuable high coverage shotgun proteomics technique particularly for the analysis of membrane proteins. In addition, this study evaluated the use of a particular organic solvent to assist in trypsin digestion when applied to analyses by peptide IPGIEF. The increase in the number of membrane proteins identified from complex protein mixtures has been shown to be enhanced with the use of organic solvents such as acetoJournal of Proteome Research • Vol. 7, No. 3, 2008 1041
research articles
Chick et al. in the basic end of the pH gradient has been observed previously.24,27 The average peptide pI per fraction in this study correlated well with the pH gradient throughout the acidic to neutral pH regions and in addition contained pI value standard deviations similar to recent reports in the literature.24,27,29 However, the differences between observed and predicted pI in the basic end of the IPG strip can be explained by inaccurate pKa values for the amino acid arginine, where the reactive group pKa value is less than the pKa value of 12 prescribed in the Expasy pI algorithm. In addition, differences between observed and theoretical pI can be attributed to low buffering capacity within the basic region of the IPG strip because only one-third of peptides within fractions 21–23 are unique to a particular fraction. The resolution in peptide separation, as determined by assessing the number of fractions each peptide is detected from, is similar to Krijgsveld et al.,24 in which >94% of all peptides were found in one or at most two fractions. This resolution in peptide separation is slightly higher than studies that use off-gel electrophoresis for the isoelectric focusing of peptides.41 In all, the separation of peptides in this study demonstrates the ability of IPG-IEF to provide a well-separated sample of tryptic peptides derived from membrane proteins, and it also supports the use of peptide pI as an additional filtering constraint for high confidence protein identifications.
Figure 5. Distribution of proteins based on GRAVY scores for the total number of nonredundant proteins (A) and integral membrane proteins (B) identified from the three digestion strategies.
Figure 6. Cellular locations of proteins identified from 60% methanol digestion as a representation of the proteins identified from all concentrations of methanol.
nitrile, acetone, isopropanol, and methanol during trypsin digestion.7 Where methanol at as high a concentration as 60% has been used,8 it has been theorized that this concentration facilitates increased peptide sequence coverage and ion intensities.7 Recently, peptide IPG-IEF has been demonstrated as an alternative shotgun proteomics technique, which is advantageous because peptides are separated based on pI, and the pI of the peptide can then be used as additional filtering criteria for protein identifications.24,27,29 Our data show that peptides were separated into three distinct pH regions, where the peptides were concentrated particularly in the acidic to neutral regions of the IPG strip. This phenomenon has been attributed to the digestion of proteins with trypsin, which produces more acidic peptides.27 Furthermore, the observation in this study of a poorer recovery of peptides and a greater reduction in peptides after pI filtering 1042
Journal of Proteome Research • Vol. 7, No. 3, 2008
Peptide IPG-IEF has been represented as an alternative first dimension shotgun proteomics technique. However, compared to MudPIT19,42,43 it is less applicable to high throughput analysis as several additional manual steps are required. Furthermore, pI filtering of peptides is hindered by either chemical modifications or post-translational modifications. Peptides with disparate pI’s that would otherwise be removed as outliers may in fact be present as modified peptides, through modifications such as acetylation or deamidation or through carbamidomethylation of cysteine. Each of these modifications creates an acidic product that would otherwise not be taken into consideration using the Compute pI/MW algorithm utilized through ExPASy (www.expasy.org). For example, basic to neutral peptides identified from acidic fractions with an acetylated lysine that are outliers based on theoretical pI may be correct identifications since the basic reactive group of the lysine is exchanged for an uncharged acetyl group. Further improvements to calculating peptide pI in relation to peptide modifications would help to recover some of these peptides otherwise lost due to modifications. Peptide pI filtering has been shown to increase the confidence in protein identifications through the use of accurate peptide pI as an additional filtering constraint.22,24,29 In this study, our results demonstrated that peptide pI filtering prior to imposing a 1% protFDR (method B) increases protein identifications by at least 19% compared to only using a 1% protFDR (method A). Peptides from proteins with log(e) scores that lay outside threshold scores of the 1% protFDR imposed by method A still contain false negative identifications. By using peptide pI filtering, false negative identifications are recovered because their pI values should lie within the pH boundaries of the individual fractions, in contrast to false positives which should have random pI values. This recovery of false negative identifications can be illustrated by assessing the log(e) threshold scores of a 1% protFDR with or without the prior use of peptide pI filtering on the data set. For example, for one of the data sets from 0% methanol assisted digest, the log(e) cutoff score is -8 for protein identifications when a 1% protFDR is applied to the data set. However, when peptide pI filtering is
research articles
Membrane Protein Analysis Using Peptide IPG-IEF used first, this log(e) score increases to -2, allowing more protein assignments with high confidence. When comparing methanol concentrations for assisting protein digestion with trypsin, 60% methanol was shown to be the optimal concentration because of a higher average number of protein identifications found between experimental replicates (800 proteins) and a combined total number of nonredundant protein identifications (1233 proteins). The increase in protein identifications can be attributed to increased peptide sequence coverage and fragment ion intensities that have been observed for methanol assisted digestion of proteins with trypsin.7,8 Both peptide sequence coverage and fragment ion intensities are two parameters that are used by protein scoring algorithms such as X!Tandem; therefore, increases in these parameters result in higher peptide scores and, furthermore, protein scores.32,33 Methanol assisted digestion results in proteins that would otherwise not meet the threshold level of 1% protFDR for protein identifications becoming elevated above threshold limits due to an increase in peptide scores. In total, 1549 nonredundant proteins, including 690 integral membrane proteins, were identified from this analysis providing evidence of the utility of this shotgun proteomics methodology. Analysis of the membrane proteins identified in this study demonstrates a diverse range of proteins with various types of secondary structures and post-translational modifications that embed these proteins into the membrane. This study shows that 60% methanol assisted digestion provided the highest number of integral membrane proteins with 574, and in total, 690 nonredundant integral membrane proteins were identified from all three concentrations of methanol. In a previous study on rat natural killer cells, 876 integral membrane proteins were identified from 60% methanol assisted digestion followed by offline SCX fractionation of peptides into 96 fractions.8 However, the 60% methanol assisted digestion method used in this study resulted in slightly higher identifications than results from Wu et al.,25 which used alternative protein cleavage methods like proteinase K and cyanogen bromide treatment with MudPIT analysis for the identification of 454 rat brain membrane proteins. Ruth et al. have compared the digestion of proteins with trypsin in the presence of an acid labile salt and diluted urea, which identified 288 membrane associated and 80 transmembrane proteins from human leukemia cells.10 The analyzed membrane proteins consisted of both integral membrane proteins and membrane associated proteins, hence a greater percentage of membrane proteins.10 Contrary to our results, these authors suggested that a minority of membrane proteins contained transmembrane segments. In our hands, the majority of integral membrane protein identifications were transmembrane proteins, which is also supported by previous studies.8,25 Wang et al. have demonstrated the identification of greater than 1300 integral membrane proteins from mouse brain through high sample loading and use of a linear ion trap mass spectrometry.21 Increasing the sample load and using more sensitive mass spectrometry such as a linear ion trap LC-MS/ MS are avenues that could enhance the number of membrane proteins identified from this study. Nonetheless, the use of methanol assisted digestion in conjunction with peptide IPGIEF shotgun proteomics has been presented here as an alternative method for the analysis of membrane proteins. In this study, increasing methanol concentration increased the number of proteins in total and identified more integral
membrane proteins. In addition, increasing methanol concentration increased the number of transmembrane proteins with between one and twelve transmembrane segments. Increasing the methanol concentration did not recover peptides from more hydrophobic proteins, contrary to a previous publication.8 GRAVY analysis of integral membrane proteins demonstrated that integral membrane proteins can contain a diverse range of hydrophobicity scores from largely hydrophilic to quite hydrophobic proteins. From nonredundant protein lists, 82 proteins were considered highly hydrophobic with GRAVY scores greater than 0.3. The number of identified highly hydrophobic proteins is higher than studies from the bacteria Deinococcus radiodurans and mouse brain homogenate, which used gas phase fractionation and MudPIT analysis for peptide separation and analysis.8,25 Lastly, cellular location annotation of proteins demonstrated no significant differences between the proteins identified using the concentrations of methanol described in this study. Approximately 30% of the proteins were annotated with unclear cellular location. Interestingly, 43 cytochrome p450s were identified from the endoplasmic reticulum providing the largest number of identified cytochrome p450s to date from a rat liver membrane preparation, which demonstrates the utility of methanol assisted digestion and peptide IPG-IEF shotgun proteomics for identification of this subset of proteins. In addition, we observed that golgi and nuclear annotation percentages were quite low, suggesting that subcellular fractionation methods, which have proven successful for enrichment and analysis of proteins from organelles,44–47 could improve the coverage of the rat liver membrane proteome.
Conclusion In summary, this study has demonstrated the utility of peptide IPG-IEF as a shotgun proteomics technique for membrane protein analysis. In addition, the use of 60% methanol assisted digestion with peptide IPG-IEF provided more protein identifications in total and enhanced the number of membrane protein identifications. Peptide IPG-IEF was highly advantageous because of the ability to use pI as an additional filtering criteria for accurate peptide identifications. IPG-IEF was found to be a high-capacity and high resolution analytical tool which also provided the fractionation required for the first dimension of a shotgun proteomics analysis. Furthermore, the pI information provided by the IPG-IEF allowed a two-step sequential result filtering approach to be used, which gives an improved protFDR while minimizing false negative assignments. Abbreviations: IEF, isoelectric focusing; IPG, immobilized pH gradient; IPG-IEF, immobilized pH gradients isoelectric focusing; GRAVY, grand average hydrophobicity; 2D-GE, twodimensional gel electrophoresis; SCX, strong cation exchange; MS, mass spectrometry; MS/MS, tandem mass spectrometry; LC, liquid chromatography; LC-MS/MS, liquid chromatography tandem mass spectrometry; MudPIT, multidimensional protein identifications technology; protFDR, protein level false discovery rate; pepFPR, peptide level false positive rate.
Acknowledgment. The research was supported by the Australian Research Council and GE Healthcare. The authors would like to thank Susanna Lindman, Lena Hornsten, Kristena Uhlen, and Brian Hood. This research was facilitated by access to the Australian Proteome Analysis Facility Ltd., established under the Australian Commonwealth Government Major National Research Facilities Scheme. Journal of Proteome Research • Vol. 7, No. 3, 2008 1043
research articles Supporting Information Available: Tables contain the nonredundant protein identifications from each of three methanol assisted digests including their triplicates. A combined list of nonredundant proteins identified from all three methanol assisted digests and their replicates is also available. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Wallin, E.; von Heijne, G. Genome-wide analysis of integral membrane proteins from eubacterial, archaean, and eukaryotic organisms. Protein Sci. 1998, 7 (4), 1029–38. (2) Hopkins, A. L.; Groom, C. R. The druggable genome. Nat Rev. Drug Discovery 2002, 1 (9), 727–30. (3) Molloy, M. P.; Herbert, B. R.; Walsh, B. J.; Tyler, M. I.; Traini, M.; Sanchez, J. C.; Hochstrasser, D. F.; Williams, K. L.; Gooley, A. A. Extraction of membrane proteins by differential solubilization for separation using two-dimensional gel electrophoresis. Electrophoresis 1998, 19 (5), 837–44. (4) Molloy, M. P.; Herbert, B. R.; Slade, M. B.; Rabilloud, T.; Nouwens, A. S.; Williams, K. L.; Gooley, A. A. Proteomic analysis of the Escherichia coli outer membrane. Eur. J. Biochem. 2000, 267 (10), 2871–81. (5) Santoni, V.; Molloy, M.; Rabilloud, T. Membrane proteins and proteomics: un amour impossible. Electrophoresis 2000, 21 (6), 1054–70. (6) Adessi, C.; Miege, C.; Albrieux, C.; Rabilloud, T. Two-dimensional electrophoresis of membrane proteins: a current challenge for immobilized pH gradients. Electrophoresis 1997, 18 (1), 127–35. (7) Russell, W. K.; Park, Z. Y.; Russell, D. H. Proteolysis in mixed organic-aqueous solvent systems: applications for peptide mass mapping using mass spectrometry. Anal. Chem. 2001, 73 (11), 2682–5. (8) Blonder, J.; Goshe, M. B.; Moore, R. J.; Pasa-Tolic, L.; Masselon, C. D.; Lipton, M. S.; Smith, R. D. Enrichment of integral membrane proteins for proteomic analysis using liquid chromatographytandem mass spectrometry. J. Proteome Res. 2002, 1 (4), 351–60. (9) Wu, C. C.; Yates, J. R., III. The application of mass spectrometry to membrane proteomics. Nat. Biotechnol. 2003, 21 (3), 262–7. (10) Ruth, M. C.; Old, W. M.; Emrick, M. A.; Meyer-Arendt, K.; AvelineWolf, L. D.; Pierce, K. G.; Mendoza, A. M.; Sevinsky, J. R.; Hamady, M.; Knight, R. D.; Resing, K. A.; Ahn, N. G. Analysis of membrane proteins from human chronic myelogenous leukemia cells: comparison of extraction methods for multidimensional LC-MS/MS. J. Proteome Res. 2006, 5 (3), 709–19. (11) Han, D. K.; Eng, J.; Zhou, H.; Aebersold, R. Quantitative profiling of differentiation-induced microsomal proteins using isotopecoded affinity tags and mass spectrometry. Nat. Biotechnol. 2001, 19 (10), 946–51. (12) Shi, R.; Kumar, C.; Zougman, A.; Zhang, Y.; Podtelejnikov, A.; Cox, J.; Wisniewski, J. R.; Mann, M., Analysis of the Mouse Liver Proteome Using Advanced Mass Spectrometry. J. Proteome Res. 2007, 6, 2963-2972. (13) Kanaeva, I. P.; Petushkova, N. A.; Lisitsa, A. V.; Lokhov, P. G.; Zgoda, V. G.; Karuzina, I. I.; Archakov, A. I. Proteomic and biochemical analysis of the mouse liver microsomes. Toxicol. In Vitro 2005, 19 (6), 805–12. (14) Josic, D.; Brown, M. K.; Huang, F.; Callanan, H.; Rucevic, M.; Nicoletti, A.; Clifton, J.; Hixson, D. C. Use of selective extraction and fast chromatographic separation combined with electrophoretic methods for mapping of membrane proteins. Electrophoresis 2005, 26 (14), 2809–22. (15) Andon, N. L.; Hollingworth, S.; Koller, A.; Greenland, A. J.; Yates, J. R.; Haynes, P. A. Proteomic characterization of wheat amyloplasts using identification of proteins by tandem mass spectrometry. Proteomics 2002, 2 (9), 1156–68. (16) Maicher, M. T.; Tiedtke, A. Biochemical analysis of membrane proteins from an early maturation stage of phagosomes. Electrophoresis 1999, 20 (4–5), 1011–6. (17) Chevallet, M.; Santoni, V.; Poinas, A.; Rouquie, D.; Fuchs, A.; Kieffer, S.; Rossignol, M.; Lunardi, J.; Garin, J.; Rabilloud, T. New zwitterionic detergents improve the analysis of membrane proteins by two-dimensional electrophoresis. Electrophoresis 1998, 19 (11), 1901–9. (18) Rouquie, D.; Peltier, J. B.; Marquis-Mansion, M.; Tournaire, C.; Doumas, P.; Rossignol, M. Construction of a directory of tobacco plasma membrane proteins by combined two-dimensional gel electrophoresis and protein sequencing. Electrophoresis 1997, 18 (3–4), 654–60.
1044
Journal of Proteome Research • Vol. 7, No. 3, 2008
Chick et al. (19) Link, A. J.; Eng, J.; Schieltz, D. M.; Carmack, E.; Mize, G. J.; Morris, D. R.; Garvik, B. M.; Yates, J. R. Direct analysis of protein complexes using mass spectrometry. Nat. Biotechnol. 1999, 17 (7), 676–82. (20) Nielsen, P. A.; Olsen, J. V.; Podtelejnikov, A. V.; Andersen, J. R.; Mann, M.; Wisniewski, J. R. Proteomic mapping of brain plasma membrane proteins. Mol. Cell. Proteomics 2005, 4 (4), 402–8. (21) Wang, H.; Qian, W. J.; Chin, M. H.; Petyuk, V. A.; Barry, R. C.; Liu, T.; Gritsenko, M. A.; Mottaz, H. M.; Moore, R. J.; Camp Ii, D. G.; Khan, A. H.; Smith, D. J.; Smith, R. D. Characterization of the mouse brain proteome using global proteomic analysis complemented with cysteinyl-peptide enrichment. J. Proteome Res. 2006, 5 (2), 361–9. (22) Cargile, B. J.; Bundy, J. L.; Freeman, T. W.; Stephenson, J. L., Jr. Gel based isoelectric focusing of peptides and the utility of isoelectric point in protein identification. J. Proteome Res. 2004, 3 (1), 112–9. (23) Cargile, B. J.; Bundy, J. L.; Stephenson, J. L., Jr. Potential for false positive identifications from large databases through tandem mass spectrometry. J. Proteome Res. 2004, 3 (5), 1082–5. (24) Krijgsveld, J.; Gauci, S.; Dormeyer, W.; Heck, A. J. In-gel isoelectric focusing of peptides as a tool for improved protein identification. J. Proteome Res. 2006, 5 (7), 1721–30. (25) Wu, C. C.; MacCoss, M. J.; Howell, K. E.; Yates, J. R. A method for the comprehensive proteomic analysis of membrane proteins. Nat. Biotechnol. 2003, 21 (5), 532–8. (26) Rabilloud, T. Membrane proteins ride shotgun. Nat. Biotechnol. 2003, 21 (5), 508–10. (27) Cargile, B. J.; Talley, D. L.; Stephenson, J. L., Jr. Immobilized pH gradients as a first dimension in shotgun proteomics and analysis of the accuracy of pI predictability of peptides. Electrophoresis 2004, 25 (6), 936–45. (28) Cargile, B. J.; Sevinsky, J. R.; Essader, A. S.; Stephenson, J. L., Jr.; Bundy, J. L. Immobilized pH gradient isoelectric focusing as a firstdimension separation in shotgun proteomics. J. Biomol. Technol. 2005, 16 (3), 181–9. (29) Essader, A. S.; Cargile, B. J.; Bundy, J. L.; Stephenson, J. L., Jr. A comparison of immobilized pH gradient isoelectric focusing and strong-cation-exchange chromatography as a first dimension in shotgun proteomics. Proteomics 2005, 5 (1), 24–34. (30) Fujiki, Y.; Hubbard, A. L.; Fowler, S.; Lazarow, P. B. Isolation of intracellular membranes by means of sodium carbonate treatment: application to endoplasmic reticulum. J. Cell Biol. 1982, 93 (1), 97–102. (31) Breci, L.; Hattrup, E.; Keeler, M.; Letarte, J.; Johnson, R.; Haynes, P. A. Comprehensive proteomics in yeast using chromatographic fractionation, gas phase fractionation, protein gel electrophoresis, and isoelectric focusing. Proteomics 2005, 5 (8), 2018–28. (32) Craig, R.; Beavis, R. C. A method for reducing the time required to match protein sequences with tandem mass spectra. Rapid Commun. Mass Spectrom. 2003, 17 (20), 2310–6. (33) Craig, R.; Beavis, R. C. TANDEM: matching proteins with tandem mass spectra. Bioinformatics 2004, 20 (9), 1466–7. (34) Peng, J.; Elias, J. E.; Thoreen, C. C.; Licklider, L. J.; Gygi, S. P. Evaluation of multidimensional chromatography coupled with tandem mass spectrometry (LC/LC-MS/MS) for large-scale protein analysis: the yeast proteome. J. Proteome Res. 2003, 2 (1), 43–50. (35) Bjellqvist, B.; Hughes, G. J.; Pasquali, C.; Paquet, N.; Ravier, F.; Sanchez, J. C.; Frutiger, S.; Hochstrasser, D. The focusing positions of polypeptides in immobilized pH gradients can be predicted from their amino acid sequences. Electrophoresis 1993, 14 (10), 1023–31. (36) Krogh, A.; Larsson, B.; von Heijne, G.; Sonnhammer, E. L. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J. Mol. Biol. 2001, 305 (3), 567– 80. (37) Sonnhammer, E. L. L.; von Heijne, G.; Krogh, A. A hidden Markov model for predicting transmembrane helices in protein sequences; AAAI Press: Menlo Park, CA, 1998; pp 175–182. (38) Speers, A. E.; Wu, C. C. Proteomics of integral membrane proteins– theory and application. Chem. Rev. 2007, 107 (8), 3687–714. (39) Cateau, E.; Tournier, N.; Dupuis, A.; Le Moal, G.; Venisse, N. Determination of atazanavir in human plasma using solid-phase extraction and high-performance liquid chromatography. J. Pharm. Biomed. Anal. 2005, 39 (3–4), 791–5. (40) Kyte, J.; Doolittle, R. F. A simple method for displaying the hydropathic character of a protein. J. Mol. Biol. 1982, 157 (1), 105– 32.
research articles
Membrane Protein Analysis Using Peptide IPG-IEF (41) Horth, P.; Miller, C. A.; Preckel, T.; Wenz, C. Efficient fractionation and improved protein identification by peptide OFFGEL electrophoresis. Mol. Cell. Proteomics 2006, 5 (10), 1968–74. (42) Washburn, M. P.; Wolters, D.; Yates, J. R. Large-scale analysis of the yeast proteome by multidimensional protein identification technology. Nat. Biotechnol. 2001, 19 (3), 242–7. (43) Wolters, D. A.; Washburn, M. P.; Yates, J. R. An automated multidimensional protein identification technology for shotgun proteomics. Anal. Chem. 2001, 73 (23), 5683–90. (44) Houde, M.; Bertholet, S.; Gagnon, E.; Brunet, S.; Goyette, G.; Laplante, A.; Princiotta, M. F.; Thibault, P.; Sacks, D.; Desjardins, M. Phagosomes are competent organelles for antigen crosspresentation. Nature 2003, 425 (6956), 402–6.
(45) Rout, M. P.; Aitchison, J. D.; Suprapto, A.; Hjertaas, K.; Zhao, Y.; Chait, B. T. The yeast nuclear pore complex: composition, architecture, and transport mechanism. J. Cell Biol. 2000, 148 (4), 635–51. (46) Schirmer, E. C.; Florens, L.; Guan, T.; Yates, J. R.; Gerace, L. Nuclear membrane proteins with potential disease links found by subtractive proteomics. Science 2003, 301 (5638), 1380–2. (47) Wu, C. C.; MacCoss, M. J.; Mardones, G.; Finnigan, C.; Mogelsvang, S.; Yates, J. R.; Howell, K. E. Organellar proteomics reveals Golgi arginine dimethylation. Mol. Biol. Cell 2004, 15 (6), 2907–19.
PR700611W
Journal of Proteome Research • Vol. 7, No. 3, 2008 1045
