HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: M700282-MCP200v1 7/2/326    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Glossary Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Pflieger, D. Articles by Aebersold, R. Search for Related Content PubMed PubMed Citation Articles by Pflieger, D. Articles by Aebersold, R. Social Bookmarking                      What's this?

Molecular & Cellular Proteomics 7:326-346, 2008.
© 2008 by The American Society for Biochemistry and Molecular Biology, Inc.

---

Research

Quantitative Proteomic Analysis of Protein Complexes

Concurrent Identification of Interactors and Their State of Phosphorylation^*,S

Delphine Pflieger^,, Martin A. Jünger, Markus Müller, Oliver Rinner, Hookeun Lee, Peter M. Gehrig^¶, Matthias Gstaiger and Ruedi Aebersold^,||,**,,

From the Institute for Molecular Systems Biology, ETH, 8093 Zürich, Switzerland, ^¶ Functional Genomics Center Zürich, University of Zürich and Eidgenössische Technische Hochschule, 8093 Zürich, Switzerland, ^|| Faculty of Science, University of Zürich, 8093 Zürich, Switzerland, ^** Institute for Systems Biology, Seattle, Washington 98103-8904, and Competence Center for Systems Physiology and Metabolic Diseases, 8093 Zürich, Switzerland

	ABSTRACT

TOP ABSTRACT EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

 
Protein complexes have largely been studied by immunoaffinity purification and (mass spectrometric) analysis. Although this approach has been widely and successfully used it is limited because it has difficulties reliably discriminating true from false protein complex components, identifying post-translational modifications, and detecting quantitative changes in complex composition or state of modification of complex components. We have developed a protocol that enables us to determine, in a single LC-MALDI-TOF/TOF analysis, the true protein constituents of a complex, to detect changes in the complex composition, and to localize phosphorylation sites and estimate their respective stoichiometry. The method is based on the combination of fourplex iTRAQ (isobaric tags for relative and absolute quantification) isobaric labeling and protein phosphatase treatment of substrates. It was evaluated on model peptides and proteins and on the complex Ccl1-Kin28-Tfb3 isolated by tandem affinity purification from yeast cells. The two known phosphosites in Kin28 and Tfb3 could be reproducibly shown to be fully modified. The protocol was then applied to the analysis of samples immunopurified from Drosophila melanogaster cells expressing an epitope-tagged form of the insulin receptor substrate homologue Chico. These experiments allowed us to identify 14-3-3, 14-3-3, and the insulin receptor as specific Chico interactors. In a further experiment, we compared the immunopurified materials obtained from tagged Chico-expressing cells that were either treated with insulin or left unstimulated. This analysis showed that hormone stimulation increases the association of 14-3-3 proteins with Chico and modulates several phosphorylation sites of the bait, some of which are located within predicted recognition motives of 14-3-3 proteins.

Typically protein complexes are analyzed by mass spectrometry after affinity purification (1). Although this approach has been successfully and widely used, it suffers from a number of limitations, some of which are addressed by specific methods. To distinguish true complex components from co-purified, nonspecific interactors the composition of an immunopurified sample of interest has been compared with a mock-purified sample either using differential isotope labeling followed by quantitative MS (2–4) or label-free MS approaches (5). To detect changes in the protein composition of complexes (6), as might be the case if complexes are isolated from cells at different states or subjected to variable stimulations, similar quantitative MS methods have been used (7–9). To gain insights into the dynamics of protein complexes in the context of signal transduction, research studies have been undertaken to delineate the correlation between complex formation and reversible phosphorylation state of the complex components (10–12). However, in these studies, only relative quantification of phosphorylation was measured without information about the stoichiometry of this modification within proteins. Initial attempts to overcome this limitation were reported in "model protein" studies in which the enzymatic digest of a sample was split into two halves. One fraction was dephosphorylated using a phosphatase, and the two samples were differentially isotopically labeled, recombined, and analyzed by LC-MS/MS (13, 14). Absolute quantification of phosphorylation sites has been measured by comparing the MS signal of phosphopeptides in the sample to that of an isotopically labeled synthetic reference peptide (15–17). However, in this targeted approach only anticipated phosphorylation sites are quantified, and the stoichiometry can only be determined if the absolute protein abundance is also obtained in the same experiment. To estimate the level of phosphorylation on a more global scale, several studies compared the MS signals of phosphopeptides and their non-modified counterparts with (18) or without (19–22) compensation for differences in ionization and detection efficiencies. These methods, although individually quite successful, have failed so far to achieve the comprehensive analysis of protein complexes in a single analysis.

Here we present a method to characterize protein complexes that combines the capabilities of some of the above methods in a single analysis. It consists of the following steps. First an affinity-purified sample of interest and a mock-purified control sample are isolated and digested, and the two samples are split into two halves and subjected to four-channel iTRAQ¹ isotope labeling. Second, two of the four iTRAQ-labeled samples are subjected to dephosphorylation via phosphatase treatment. Third, the samples are recombined and analyzed by liquid chromatography-tandem mass spectrometry using a MALDI-TOF/TOF instrument; and fourth, the data are analyzed to distinguish true complex components from nonspecific interactors and to determine the site(s) and stoichiometry of phosphorylation. Therefore, the method provides the potential to subject immunopurified protein complexes to comprehensive analysis in a single experiment. The method was tested on model samples, angiotensin II and its phosphorylated counterpart, as well as caseins. It was also applied to the study of a complex purified from yeast cells by tandem affinity purification (TAP) (23) and to single step-purified complexes built around the insulin receptor substrate, Chico, in Drosophila melanogaster cells (24). The procedure proved efficient in identifying bona fide interactors of Chico and variable phosphorylation sites regulated by insulin.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

 
Materials
The iTRAQ reagent multiplex kit (reference 4352135, Applied Biosystems) contained the necessary reagents for resuspension (Dissolution Buffer: 500 mM triethylammonium bicarbonate, pH 8.5), reduction (Reducing Reagent: 50 mM tris-(2-carboxyethyl)phosphine), alkylation (Cysteine-Blocking Reagent: 200 mM methyl methanethiosulfonate), and differential labeling of protein samples. For all described samples, the protocol from Applied Biosystems was followed except for adjustments of the added volumes of reducing and alkylating reagents that were applied depending on the initial sample volume. Modified porcine trypsin was purchased from Promega (reference V5113), Rapigest detergent was from Waters, calf intestinal phosphatase (CIP) was bought from Fermentas (reference EF0341; 1 unit/µl), and shrimp alkaline phosphatase (SAP) was from Promega (reference 9PIM820; 1 unit/µl). The amino-reactive bifunctional cross-linker dithiobis(succinimidyl propionate) (DSP) was purchased from Pierce (reference 22585).

Analysis of a Peptide Pair: Angiotensin and Its Phosphorylated Counterpart
Six samples consisting of human angiotensin II (designated "A") and Tyr-phosphorylated angiotensin II (designated "pA") (references 05-23-0101 and 05-23-0111, Merck/Calbiochem) were prepared in proportions (A/pA) 0:100, 20:80, 40:60, 60:40, 80:20, and 100:0, respectively. Each sample, containing a total of 600 pmol of peptide in water, was split in two halves, which were completely dried by SpeedVac and then resuspended in 3 µl of Dissolution Buffer. Two iTRAQ reagent tubes (116 and 117) were dissolved in 70 µl of ethanol, and 11 µl of these reagent solutions were used to perform a differential labeling reaction of the two halves of each mixture (A/pA). After 1-h incubation on the bench, samples were dried by SpeedVac and then resuspended in 20 µl of CIP buffer (Fermentas; provided with the enzyme: 10 mM Tris-HCl (pH 7.5 at 37 °C), 10 mM MgCl₂). After 30 min allowing complete hydrolysis of the excess iTRAQ reagents, 1 unit of CIP was added to every 117-labeled sample. All samples (116- and 117-labeled) were incubated at 37 °C for 2.5 h. After heating at 85 °C for 15 min to inactivate the phosphatase, samples were finally pooled by 116-117 pairs to reconstitute the whole (A/pA) samples. Fifteen picomoles of each treated peptide pair A/pA were cleaned on a ZipTip_C18 (Millipore), following the manufacturer's instructions to remove buffers and excess iTRAQ reagents. The eluates were dried by SpeedVac, and peptides were resuspended in 50 µl of 0.1% (v/v) TFA in water. From each sample A/pA, 0.5 µl was spotted on a MALDI plate and mixed with 0.5 µl of matrix solution containing 4 mg/ml of -cyano-4-hydroxycinnamic acid (Fluka, reference 28480; >99% pure) in (ACN/water/TFA, 70:30:0.1, v/v/v). Samples were analyzed on a MALDI-TOF/TOF 4800 instrument (Applied Biosystems) by accumulating 1500 laser shots in MS analysis mode. In MS/MS analysis, 1500 shots were accumulated for fragmentation of angiotensin, and 4500 shots were summed for its phosphorylated counterpart. To test the relevance of applying a correction factor to measured iTRAQ ratios A(117)/A(116) and A(115)/A(114), a sample consisting of pure angiotensin (600 pmol) and 4 µg of β-lactoglobulin A (Sigma, reference L7880) was submitted to the previous protocol described above for A/pA mixtures but using iTRAQ labels 114 and 115.

Yeast TAP Sample
A yeast strain expressing TAP-tagged Kin28 was obtained from Open Biosystems. Cells were grown at 30 °C until the culture reached an OD of 1.15. Cells were harvested and pelleted at 8000 x g for 10 min. Pellets were washed (20% glycerol, 1 mM EDTA, 50 mM Tris-HCl, pH 7.8, 150 mM KCl) and distributed into 50-ml tubes. Cells were centrifuged again at 4000 rpm in a Microfuge T, and the pellets (4–5-ml pellet were obtained from 3 liters of culture) were quick frozen in liquid nitrogen to be stored at –80 °C until use.

Two such pellets were used for tandem affinity purification. The procedure followed was close to that described by Rigaut et al. (25) and is detailed in Gingras et al. (26). Specifically the following modifications were implemented to accommodate the higher cell amount used. Five milliliters of ice-cold lysis buffer were added to each cell pellet, and cells were resuspended and distributed as 500-µl aliquots into 2-ml tubes. After adding 500 µl of 0.5-mm glass beads, cells were lysed by repeating the following cycle three times: 10-min agitation of the tubes on a vortexer (Vortex-Genie 2, Scientific Industries, Inc., reference SI-0276; equipped with a 1.5-ml Snap-top Microtube Holder, reference 0A-0563-010) at maximum speed and 10-min cooling on ice. After discarding the cell debris by centrifugation at 13,000 rpm (Centrifuge 5415 R, Eppendorf), approximately 10 ml of protein extract (350 mg of protein, determined by protein assay (Bio-Rad)) were obtained. This lysate was precleared on 400 µl of packed glutathione 4B beads (Amersham Biosciences, reference 17-0756-01) for 2 h. The extract was then incubated on 300 µl of packed IgG beads (Amersham Biosciences, reference 17-0969-01) for 4.5 h. IgG beads were rinsed with 3 ml of lysis buffer and then three times with 3 ml of TEV buffer containing PMSF. The tagged protein was released by adding 600 µl of TEV buffer containing 30 µl of TEV protease (Invitrogen, reference 10127017). After overnight incubation on a rotating wheel at 4 °C, eluted proteins were collected by pooling the supernatant and three washes of the IgG beads with 300 µl of calmodulin binding buffer (10-min incubation of the IgG beads in that buffer each time). The resulting 1.5 ml of protein sample were incubated with 100 µl of packed calmodulin beads (50 µl from each manufacturer: Amersham Biosciences reference 17-0529-01 and Stratagene reference 214303) for 3.75 h. After rinsing the beads, the retained proteins were eluted with 3 x 100 µl of elution buffer supplemented with 0.2 mg/ml Rapigest. Each time, the 100 µl of eluting buffer were pipetted up and down for 1 min to thoroughly resuspend the beads and improve protein elution. This purified sample was heated at 95 °C to denature proteins and then split in two halves to mimic two independent protein samples. The resulting samples were dried by SpeedVac to eliminate the ammonium bicarbonate present in the final eluting buffer. Samples were sonicated for 10 min in 40 µl of Dissolution Buffer to help resolubilize proteins. Both samples were reduced, alkylated, and digested by addition of 0.3 µg of trypsin and overnight incubation at 37 °C. Each digest was split in two halves (around 20 µl each), and the four resulting tubes were labeled with the four different iTRAQ reagents (114 and 115 for one digest and 116 and 117 for the other) resuspended in 70 µl of ethanol. After 1-h reaction at ambient temperature, samples were concentrated to dryness and resuspended in 40 µl of CIP buffer. Tubes were further left on the bench for 60 min to allow hydrolysis of excess iTRAQ reagents. After addition of 1 M MgCl₂ to reach a final concentration of 100 mM and fully react with EGTA (used in the elution from calmodulin beads and now at around 95 mM in the four tubes), 2 µl of CIP were added to the 115- and 117-labeled samples. The four tubes were placed at 37 °C for 3 h, and then CIP was inactivated by heating at 85 °C for 15 min. The four samples were pooled and redigested by addition of 1 µg of trypsin and incubation at 37 °C for 1.5 h. The detergent Rapigest was cleaved by adding a solution of 1 M HCl until reaching a pH of approximately 2. After incubation at 37 °C for 45 min, the tube was centrifuged at 13,000 rpm for 10 min (Eppendorf Centrifuge 5415 R) to discard the bottom 10 µl containing the insoluble part from Rapigest. The sample was finally cleaned by using consecutively three ZipTip_C18 pipette tips to enhance peptide recovery.

Analysis of Standard Proteins: Caseins
A sample consisting of 100 µg of β-casein (reference C6905 from Sigma; >90% pure) in 50 µl of Dissolution Buffer was treated following the protocol in Fig. 1 using the combination of CIP and SAP. The sample was supplemented with Rapigest at a final concentration of 0.3% (w/v) and then heated at 95 °C for 5 min. After reduction and alkylation of the sample, 50 µl of Dissolution Buffer and 2 µg of trypsin were added to the sample before overnight incubation at 37 °C. The obtained peptide mixture was split in two halves, which were labeled with iTRAQ reagents 114 and 115. The resulting samples were dried by SpeedVac and resuspended in 100 µl of CIP buffer. Sample tubes were left on the bench for 30 min to allow complete hydrolysis of residual iTRAQ reagents. Then 2 units of CIP and 2 units of SAP were added to the 115-labeled sample. Both tubes were incubated at 37 °C for 2 h before being heated at 85 °C for 15 min to inactivate the phosphatases. The pooled sample was finally redigested by addition of 0.8 µg of trypsin and incubation at 37 °C for 2 h. After cleavage of Rapigest performed as for the TAP sample, the sample was cleaned on a ZipTip_C18 and analyzed by MALDI-TOF/TOF MS (deposition of 350 fmol/spot) in the same conditions as the angiotensin samples. In an alternative experiment, casein peptides were treated with 2 units of -phosphatase (Sigma-Aldrich reference P9614) using the buffer delivered with the enzyme. To finally inactivate phosphatase, the tubes were heated at 85 °C for 30 min and supplemented with 6 mM orthovanadate. After pooling the two tube contents, the sample was redigested by addition of 0.8 µg of trypsin and incubation at 37 °C for 2 h.

View larger version (27K): [in this window] [in a new window]   FIG. 1. A, protocol developed to characterize partially purified protein complexes in terms of phosphorylation sites and stoichiometry. Label P stands for a phosphorylation site. B, expected patterns of iTRAQ reporter groups in the MS/MS spectra of peptides produced by the protocol in A. Panel A, peptides of the proteins initially non-phosphorylated; panel B, phosphorylated peptides; panel C, peptides dephosphorylated by phosphatases.

Comparison of Two HA-Chico IPs from Cells Treated or Not Treated with Insulin—
Two independent experiments, aiming at studying the insulin-induced change in the composition of Chico-containing protein complexes and their state of phosphorylation, were carried out following the procedure in Fig. 2A. The corresponding immunopurified samples +INS/noINS are called "Chico3" and "Chico4." Six flasks of HA-Chico-expressing cells were grown at 25 °C in blasticidin-containing full medium (40 ml for sample Chico3 and 45 ml for sample Chico4). When they reached a concentration of 4·10⁶ and 5·10⁶ cells/ml, respectively, overnight expression of HA-Chico was induced by addition of 600 µM CuSO₄. Cells were harvested and distributed in 50-ml tubes in a manner that the content of each flask was split into two tubes, one of which was stimulated with insulin. Half of the tubes were treated with 100 nM bovine insulin (Sigma-Aldrich) for 7 min at room temperature. Cells were then pelleted by centrifugation at 400 x g for 5 min in a centrifuge precooled at 4 °C. Cells were washed with 40 ml of cold PBS and pooled to form two cell samples, which were treated or not treated with insulin. Cells were pelleted again at 400 x g for 6 min at 4 °C. Cells were further treated as previously described for HA-Chico/control cells with one exception: for preparation of sample Chico4, additional phosphatase inhibitors (20 nM calyculin A, 200 nM okadaic acid, and 4.8 µM cypermethrin, all from Merck KGaA) were introduced into the lysis buffer.

View larger version (28K): [in this window] [in a new window]   FIG. 2. A, global protocol allowing comparison of protein components and their phosphorylations between two different samples. B, expected iTRAQ reporter group patterns during LC-MS/MS analysis of the previous sample. phos, phosphorylated; I, intensity.

MALDI-TOF/TOF Analyses
Yeast Samples—
MALDI plates (192-well) spotted with yeast samples were analyzed in automatic mode by a 4700 MALDI-TOF/TOF instrument (Applied Biosystems). Each spot was first analyzed in MS mode in the mass range from 750 to 4000 Da by accumulating signal over 1500 laser shots. Up to 15 ions giving an MS signal with S/N >60 were then selected for further MS/MS analysis performed in order of decreasing precursor signal intensity. The acquisition of an MS/MS spectrum was interrupted when between 3000 shots (60 subspectra accumulated from 50 laser shots each) and 5000 shots (100 subspectra) were accumulated. The accumulation of additional laser shots was halted whenever at least four ions with S/N above 50 were present in the accumulated MS/MS spectrum in the region from 200 Da to 90% of the precursor mass. Source2 air pressure was set to 2 x 10^–6 torr in a first and 5 x 10^–7 torr in a second MS/MS analysis. Mass calibration of the plate was obtained by spotting a mixture of reference peptides (des-Arg¹-bradykinin, MH = 904.468 Da; angiotensin I, MH = 1296.685 Da; Glu¹-fibrinopeptide, MH = 1570.677 Da; ACTH-(1–17), MH = 2093.087 Da; ACTH-(18–39), MH = 2465.199 Da; ACTH-(1–39)) at six places around the metal plate.

Drosophila Samples—
MALDI 2000-well plates spotted with Drosophila peptide samples were analyzed in automatic mode by a 4800 MALDI-TOF/TOF instrument (Applied Biosystems). Each spot was first analyzed in MS mode by accumulating signal with up to 1000 laser shots (20 subspectra of 50 shots) over the mass range 700–4000 Da; MS acquisition was stopped after summing 10 subspectra when the accumulated spectrum contained at least five peaks with S/N >500. Up to 12 ions giving an MS signal with S/N >40 were then candidates for further MS/MS analysis performed in order of increasing precursor intensity. The acquired MS/MS spectrum was obtained by accumulating 1500 laser shots (30 subspectra of 50 shots); acquisition was stopped after summing 20 subspectra when the spectrum contained at least four peaks with S/N >100. The source2 air pressure was set to 2.5 x 10^–6 torr for MS/MS analysis; it was 5 x 10^–7 torr for MS analysis.

Database Searches of MALDI-TOF/TOF Data
Yeast Samples—
Acquired MS/MS spectra were interpreted using Mascot software version 2.1 embedded in the GPS Explorer software version 3.5 (Applied Biosystems). Each MS/MS spectrum was processed as follows: no smoothing or background subtraction was performed, only peaks of S/N above 6 and of masses between 60 and 20 Da below the precursor mass were taken into account, and a maximum of 65 peaks were selected to build the mass list representative of the MS/MS spectrum. An error tolerance on mass measurement in MS mode of 50–75 ppm could be specified in searches; error tolerance was 0.2–0.25 Da in MS/MS mode. The use of trypsin was specified, and one missed cleavage was allowed. To evaluate the efficiency of iTRAQ labeling, searches were performed by considering this modification as partial on lysines and N termini of fragmented peptides; alkylation of cysteines by methyl methanethiosulfonate was considered to be a complete modification. A database of yeast proteins released on August 11, 2007 and containing 5812 entries (ftp.ebi.ac.uk/pub/databases/SPproteomes/fasta/proteomes/) was used to run searches. In the GPS Explorer results, provided as a supplemental Excel file, the following main parameters allow checking data quality: protein Mascot score, number, sequences and scores of peptides identifying the protein, and quantification data; peptides with scores below the Mascot threshold for reliable identifications were left aside to favor reliable peptide identifications and accurate protein quantification. When a peptide sequence was shared by multiple proteins, it was only assigned to the protein identified with the highest score. A protein relative abundance was calculated by averaging iTRAQ ratios measured on its proteolytic peptides; assuming a log-normal distribution of iTRAQ ratios measured on its peptides, we rejected measurements outside the interval mean ± 2 S.D. (p = 0.05) where S.D. stands for the estimated standard deviation.

Drosophila Samples—
Acquired MS/MS spectra were interpreted using Mascot software version 2.1.0 embedded in the GPS Explorer software version 3.6 (Applied Biosystems). Each MS/MS spectrum was processed as follows: no smoothing or background subtraction was performed, only peaks of S/N above 10 and of masses between 60 and 50 Da below the precursor mass were taken into account, a maximum of seven peaks per 200-Da window, and a maximum of 50 peaks in total were selected to build the mass list representative of the MS/MS spectrum. Using a plate calibration obtained by spotting the same peptide mixture as above at eight places around the metal plate, a tolerance on mass measurement in MS mode of 30 ppm could usually be specified in searches; tolerance was 0.3–0.4 Da in MS/MS mode. The use of trypsin was specified while accepting one missed cleavage. iTRAQ labeling of lysines and N termini of fragmented peptides and alkylation of cysteines by methyl methanethiosulfonate were considered to be complete modifications. The Gene Ontology Annotation (GOA) database of D. melanogaster proteins (from the GOA Proteomes set 17.0 released on July 24, 2006 and containing 15,878 entries) was used to run searches.

Statistical Interpretation of Quantitative Data
The iTRAQ area ratios were normalized to compensate for global area variations in the corresponding iTRAQ channels because the total amount of peptides loaded in the two samples could be different, and the median value of all ratios f could therefore be different from 1. Simply dividing all numerator intensities by f resets the median to 1 and guarantees that most peptides show no change, which is essential for the accurate calculation of the derived values (e.g. x values in Equation 4).

The distribution of ratios in the sample provides information about their variability. To find out which ratios corresponded to a significant change in intensity levels, the p value of each ratio was calculated. The p value is equal to the likelihood that a value equal to or higher than the given ratio can be obtained by chance. These p values are calculated by fitting a log-normal distribution to the ratio values using the Matlab software (The Mathworks Inc.) or the R statistical package (The R Project for Statistical Computing). To make the fit robust against outliers, i.e. ratios that are significantly higher or lower than the bulk of the other ratios, the ratio values lower than the 10th percentile or higher than the 90th percentile were not considered. The scripts can be obtained from the corresponding author on request for non-commercial use.

	RESULTS

TOP ABSTRACT EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

 
Principle of the Method
The protocol implemented for characterizing the state of phosphorylation of proteins within a sample of interest is depicted in Fig. 1A. The protein sample is trypsinized, and the resulting peptide mixture is split into two halves. The samples are then labeled with iTRAQ reagents 115 and 114, respectively; one sample is treated with phosphatases; and the samples are recombined. Non-phosphorylated peptides are unaffected by the phosphatase treatment and are therefore expected to provide an iTRAQ area ratio A(115)/A(114) equal to 1 during MS/MS analysis (Fig. 1B, panel A). In contrast, a protein region originally containing a phosphosite will generate two peptide forms, non-modified at mass M and intact phosphorylated at mass M + 80 Da. Here we describe in detail the case of a protein containing a single phosphosite within a tryptic sequence. The treatment of cases of multiply phosphorylated proteolytic peptides is addressed in supplemental Data 1. A pair of phosphorylated/non-phosphorylated peptides possessing the same amino acid sequence provides MS/MS spectra showing iTRAQ fragment ion signals as represented in Fig. 1B from which the following information can be calculated. First, the yield of phosphatase treatment, denoted y, can be determined from the iTRAQ reporter group areas in the MS/MS spectrum of the phosphorylated species (Fig. 1B, panel B) as follows.

(Eq. 1)

The subscript P indicates that the iTRAQ reporter group areas are measured in the spectrum of a phosphorylated peptide.

Second, the level of phosphorylation x of a peptide can be calculated from the yield of phosphatase treatment y and from the iTRAQ reporter group areas measured in the MS/MS spectrum of the dephosphorylated peptide form (Fig. 1B, panel C), marked with the subscript dP, as follows.

(Eq. 2)

where r is the ratio of iTRAQ reporter group areas: A(115)_dP/A(114)_dP. When the phosphatase treatment goes to completion, Equation 2 becomes Equation 3.

(Eq. 3)

This expression implies that the initial peptide sample was perfectly split into two halves before iTRAQ labeling so that the average A(115)/A(114) ratio calculated over non-phosphorylated peptides is equal to 1. Due to experimental variations (e.g. variability of iTRAQ labeling efficiency and of quantitative measurements on the MALDI-TOF/TOF instrument), such peptides should provide a distribution of ratios centered on 1. The main reason why the distribution may be centered around a different value is the possibly inaccurate splitting of the digested samples into two halves before differential iTRAQ 115/114 labeling. The experimentally obtained distribution of iTRAQ ratios A(115)/A(114) is therefore recentered on 1, and the corresponding correction factor is applied to all individual iTRAQ ratios to calculate proper estimates of phosphorylation stoichiometries. Using this factor, f, Equation 2 becomes Equation 4.

(Eq. 4)

Similar equations can be written for samples labeled with iTRAQ reagents 116 and 117.

Finally it has to be noted that we consider that a protein may be either non-modified or phosphorylated at any given Ser, Thr, or Tyr residue; we do not take into account other possible modifications such as glycosylation that might also be present on these residues. Then the estimated stoichiometry of phosphorylation at a given site corresponds to the phosphorylated fraction of the protein as compared with the fraction that is non-modified at that site.

The strategy is extended to four iTRAQ labels to compare two independent protein samples in terms of protein abundances and phosphorylation levels. A sample immunopurified using a tagged bait can be compared with a corresponding control sample. Alternatively two samples obtained in two different conditions of interest (e.g. variable stimuli applied to cells) can be compared. Fig. 2 schematically illustrates this extended strategy and most of the iTRAQ patterns that can be expected. We evaluated the possibility of comparing in detail two protein samples with this protocol by testing and optimizing it on 1) a pair of peptides that differed by a single site of phosphorylation, 2) a yeast protein sample isolated by TAP, 3) model proteins, caseins, known to contain several phosphorylation sites, and 4) two samples prepared by a single step immunopurification from D. melanogaster cell lysates. To limit the search time we performed the database searches without considering possible phosphorylation sites. This is justified because in most instances phosphorylated species are less likely to be selected for fragmentation than non-modified species due to their lower ionization efficiency at least in MALDI analysis mode, and their fragment ion spectra are often more difficult to read in terms of amino acid sequence.

Analysis of a Single Peptide
The possibility to detect phosphorylated peptides and to estimate the stoichiometry of phosphorylation with the described method was first tested on a commercially available peptide, angiotensin II, whose phosphorylated form is modified on a tyrosine. Mixtures of angiotensin (denoted A) and its phosphorylated counterpart (pA) were prepared in different proportions, described as A/pA: 0:100, 20:80, 40:60, 60:40, 80:20, and 100:0, respectively. These samples were treated following the protocol in Fig. 1A (without the digestion step) using iTRAQ reagents 116 and 117 and were analyzed on a MALDI-TOF/TOF instrument. Peptide A was analyzed in MS/MS mode in each spotted sample, from 20:80 to 100:0, to generate five independent spectra. Peptide pA was also fragmented once in all of the samples 0:100 to 80:20.

Fig. 3, A and B, represents the pair of MS/MS spectra obtained on peptides A and pA in sample 40:60. In the MS/MS spectrum of pA, the weak signal detected at 117.1 Da indicated efficient dephosphorylation by the phosphatase. The yield of dephosphorylation was estimated from that spectrum using Equation 1. It ranged from 0.92 to 1 for the different A/pA samples. The ratio r was calculated from the MS/MS spectrum of angiotensin. Pairs of values (y, r) were obtained for each sample except for sample 100:0 to which y(80:20) was attributed. The phosphorylation level of angiotensin in the initial samples was finally estimated using Equation 2. The experimentally measured level of phosphorylation in the different mixtures was plotted against the expected level of phosphorylation in Fig. 3C.

View larger version (26K): [in this window] [in a new window]   FIG. 3. MALDI-TOF/TOF MS/MS spectra of angiotensin II (A) and phosphorylated angiotensin II (B) obtained when analyzing 300 fmol of a mixture A/pA 40:60 submitted to the treatment in Fig. 1a. The spectra are interpreted in terms of b-type fragment ions. C, plot of the experimentally determined level of phosphorylation of angiotensin II in the mixtures A/pA versus the expected level of phosphorylation. Bars indicate the standard deviations calculated on the phosphorylation level x from five measurements of the ratio r obtained from five MS/MS spectra of angiotensin and one measurement of the yield of phosphatase treatment in the MS/MS spectrum of phosphorylated angiotensin. *, phosphorylated residue.

In the next studied samples consisting of digested protein mixtures, the only source of error affecting the measurements of phosphorylation levels will be the inaccurate splitting of the digested sample before differential iTRAQ labeling. As the majority of the detected peptides are not phosphorylated, the quantitative data generated from those peptides was used to readjust the splitting ratio to 1:1 (see "Experimental Procedures").

Evaluating the Method on a Yeast Protein Complex Purified by TAP
We next tested the method for the characterization of a protein complex. We attempted to address several questions from the same analysis. First, we asked whether an efficient iTRAQ labeling could be obtained on immunopurified samples. Second, because the two compared samples are treated separately during several processing steps per our protocol, we wanted to test the resulting dispersion of quantitative measurements. Third, we asked in which limits our method was able to detect and quantify phosphorylations in target proteins.

These questions were addressed using the trimeric TFIIK subcomplex of the basal transcription factor TFIIH that has been known to be composed of the cyclin-dependent kinase Kin28 and its two regulatory subunits, the cyclin Ccl1 and the RNA polymerase II transcription factor B subunit 3, Tfb3. Our goal was not to determine additional interaction partners of the trimer. Rather we chose this sample to test the method for its ability to identify and quantify sites of phosphorylation. Kin28 was shown to be activated by phosphorylation at Thr¹⁶² (27). Tfb3 was also described as phosphorylated at Thr²⁵⁸ (28), but to our knowledge, no information is yet available on the extent of this modification. The sample was prepared from cells expressing TAP-Kin28. To estimate the reproducibility of quantitative measurements, we compared the TAP sample to itself: the isolated sample was split into two halves called TAP1 and TAP2 that were treated following the protocol in Fig. 2a using iTRAQ labels 114 and 115 or 116 and 117, respectively. Aliquots of one-third of the resulting peptide mixture were analyzed in duplicate by capillary LC-MALDI-TOF/TOF, and the following data interpretation was carried out.

Efficiency of iTRAQ Labeling—
The presence of partially iTRAQ-modified peptides might complicate the quantitative conclusions drawn from the data. The acquired MS/MS spectra were therefore searched allowing partial iTRAQ modification at N termini and lysine residues. From the two MALDI-TOF/TOF analyses, 432 and 531 peptides, respectively, were identified by significant Mascot scores (p < 0.05; threshold score, 26). Among these, less than 3% appeared to be partially or non-iTRAQ labeled. Later MS/MS data were therefore interpreted assuming that complete iTRAQ labeling at both peptide N termini and lysine residues had been achieved.

Sequence Coverage of Proteins—
The two analyses of the TAP sample identified 64 and 75 proteins, respectively, based on at least two peptides with Mascot scores above the significance threshold. Most of the proteins identified in addition to the trimer Kin28-Ccl1-Tfb3 (predominantly ribosomal and heat shock proteins) could be repeatedly detected in TAP samples isolated from yeast using different bait proteins and the same protocol as in the present work (26) and were thus assumed to represent nonspecific background proteins. They are listed in the supplemental Excel file. The sequence coverage of Kin28, Ccl1, and Tfb3 with validated peptide identifications (scores above the significance threshold and spectra that allow reading a three-amino acid sequence tag or more) was 51, 51, and 63%, respectively, when the data acquired in both analyses were pooled (see supplemental Fig. S1). In comparison, the highest score and therefore presumably among the most abundant background proteins identified was the heat shock protein SSB1 with sequence coverage of 49% (molecular mass, 69.9 kDa) in the first analysis. These figures demonstrate that it is possible to reach significant sequence coverage of proteins of interest despite the presence of numerous and possibly abundant contaminants. An important prerequisite for characterizing post-translational modifications of target proteins was therefore fulfilled.

Relative Quantification of Proteins—
The relative abundance of a protein in two samples compared with the protocol in Fig. 2a is calculated by averaging the ratios (A(116) + A(117))/(A(114) + A(115)) for the peptides arising from the protein in question. As shown above, many of the identified proteins are nonspecific contaminants when two immunopurified samples are being compared. For these, abundance ratios close to 1 should be detected as they are expected to be present in equal amounts in the sample of interest and the control. For a protein to be regarded as enriched in either sample, it needs to appear as an outlier from the main distribution of protein ratios centered on 1. It is therefore important for assessing the significance of an enrichment factor to define this distribution. It is characterized by the variance _total² = _IP² + _method², the sum of the variances related to the immunopurification and to the analytical method. The first variance may depend on the two compared cell lysates and on the beads used for immunopurification. The second was estimated on our TAP sample. To determine the protein relative abundance TAP2/TAP1, the ratios (A(116) + A(117))/(A(114) + A(115)) measured on all identified peptides were grouped per protein and averaged. When considering the 51 proteins identified with at least three peptides, an average ratio of 1.16 ± 0.15 (S.D.) was calculated. Such a dispersion is close to that obtained when performing the classical iTRAQ procedure, which consists of the comparison of up to four digested protein samples (29). Despite the extra manipulation steps required to analyze immunoprecipitated samples, the procedure therefore provides accurate relative quantification of proteins.

Detection of Sites of Phosphorylation—
To identify sites of phosphorylation, the number of identified peptides measured at a specific iTRAQ ratio was plotted as a function of iTRAQ ratios A(115)/A(114) and A(117)/A(116)_. Plots obtained in the first LC-MALDI-TOF/TOF analysis of the Kin28 complex are shown in Fig. 4 after correction for inaccurate splitting of the peptide samples. The population of non-phosphorylated peptides provides a distribution of iTRAQ ratios centered on 1 (or 0 in log scale) that is characterized by a certain standard deviation. This dispersion of measurements limits the minimum detectable level of phosphorylation: an initially phosphorylated peptide is detected as a dephosphorylated species when the latter gives an iTRAQ ratio appearing as an outlier from the main distribution. In this case ratios A(115)/A(114) and A(117)/A(116) (corrected for inaccurate peptide sample splitting) above 1.22 were indicative of the possible presence of a phosphorylation (p = 0.01). This threshold value corresponds to a phosphorylation level around 18% when y = 1 and 29% when y = 0.5.

View larger version (14K): [in this window] [in a new window]   FIG. 4. Distribution of iTRAQ ratios A(115)/A(114) and A(117)/A(116) measured for peptides identified in the TAP sample. Arrows indicate the measurement appearing as an outlier from the main distribution of ratios, which corresponds to peptide AIPAPHEILTSNVVTR from Kin28. An enlargement leaving aside part of the main distribution is given to highlight the outliers.

Tfb3 was also expected to be phosphorylated specifically at Thr²⁵⁸ within the sequence LKDAVPFTPFNGDR. This peptide was not identified by the initial database search, but closer examination of the MS/MS data identified the peptide in its deamidated form (Asn Asp). This modification is commonly observed with Asn-Gly sequences (30). The fragment ion spectra of the non-modified and corresponding phosphorylated peptides are shown in Fig. 5. The acquisition of an MS/MS spectrum of the phosphopeptide allowed localization of the modification site. The initial phosphorylation level of LKDAVPFT*PFNGDR could be determined to be 98 and 100% from the samples labeled with reagents 115-114 and 117-116, respectively. The same estimations were provided by both LC-MALDI-TOF/TOF analyses (Table I). Again we could conclude that Tfb3 was fully modified at Thr²⁵⁸.

View larger version (24K): [in this window] [in a new window]   FIG. 5. MS/MS spectra acquired on peptide LKDAVPFTPFDGDR from Tfb3, known to be phosphorylated at Thr8 in the intact protein. A, fragmentation of the dephosphorylated species in spot 116; B, fragmentation of the phosphorylated species in spot 112. The whole raw spectrum, an enlargement of the iTRAQ reporter group region, and the interpreted spectrum are given successively. These spectra were obtained in the first LC-MALDI-TOF/TOF analysis of the TAP sample. !!, theoretical residue Asn was detected as Asp because of deamidation.

Improving the Method on Model Phosphoproteins: Casein
Although Kin28 and Tfb3 could be shown to be fully modified at Thr¹⁶² and Thr²⁵⁸, respectively, dephosphorylation of Kin28 was only effective to about 20–30%, a yield that would make the detection of low level phosphorylation sites difficult. The CIP has already been described to only partially dephosphorylate some peptide sequences (33). To improve the level of phosphatase-catalyzed dephosphorylation, we considered using a combination of several phosphatases, CIP, SAP, and -phosphatase. It can indeed be assumed that different phosphatases are somewhat selective for specific substrates, and therefore it is generally advisable to use a mixture of these enzymes. CIP and SAP both require Mg²⁺ cations, whereas -phosphatase is active with Mn²⁺ cations and inhibited by Mg²⁺ (34), which precludes the combined use of all three enzymes. Yet the combination of CIP and SAP could be expected to provide better dephosphorylation of a statistical sample than CIP alone. Both phosphatase conditions were tested on a β-casein sample that also contained minor amounts of _S1-, _S2-, and -caseins following the protocol in Fig. 1A. The resulting sample was deposited as a single MALDI spot. Automated analysis of the most intense ions allowed identification of 11 non-modified peptides from caseins. Their average ratio r was 1.11 ± 0.19 for the sample treated by CIP and SAP versus 1.08 ± 0.09 for the -phosphatase-treated sample. Expected tryptic phosphopeptides were also fragmented in their non-modified form independently of their detection in MS mode. Some fragmented efficiently enough to be conclusively identified; others mainly produced intense iTRAQ reporter ions. In both cases, the ratio r was measured and corrected for inaccurate sample splitting using the average r value determined above on non-modified species. Because the MS/MS spectra of the phosphorylated peptides usually could not be obtained (especially when dealing with multiply modified peptides), the initial levels of phosphorylation were calculated while considering a complete dephosphorylation by phosphatases (y = 1). The calculated phosphorylation levels are presented in Table II: the considered peptide sequences globally appeared to be minimally phosphorylated in the original proteins. Thus, despite probable overestimation of dephosphorylation yield, the estimation of phosphorylation levels was usually underestimated by less than 25%. Because similar estimations of phosphorylation levels were obtained using either CIP + SAP or -phosphatase, both enzyme conditions may be considered for performing the procedure. For a given sample of interest, both phosphatase conditions may be tested to select the one providing the more efficient dephosphorylation. We chose the combination CIP + SAP for the next experiments.

One-third of samples Chico1 and Chico2 was analyzed by LC-MALDI-TOF/TOF, resulting in the identification of 265 and 180 peptides, respectively. Peptide identifications were grouped per protein, and average ratios (A(116) + A(117))/(A(114) + A(115)) were calculated to point to proteins enriched in the sample of interest. The initial protein ratios were corrected by considering the relative protein concentrations measured in the total lysates Chico/control. Eighty-seven and 46 proteins were identified, respectively, from samples Chico1 and Chico2. Table III lists proteins identified on the basis of at least three peptides. The distributions of relative protein abundances were plotted for both samples as described at the end of "Experimental Procedures" (supplemental Fig. S3). In both samples, the control/Chico abundance ratios for 14-3-3, 14-3-3, and the insulin receptor (IR) indicate that they are likely interactors of Chico. Supplemental Fig. S4 shows an MS/MS spectrum identifying a peptide from 14-3-3 and featuring the characteristic iTRAQ group pattern of a protein over-represented in Sample 1. The other proteins identified with at least three peptides, whose quantification was therefore reliable, were associated to ratios spanning over the ranges 0.48–1.78 in Chico1 and 0.38–1.45 in Chico2. Given the thresholds indicative of protein enrichment calculated to be 0.36 and 0.27 for samples Chico1 and Chico2 (p = 0.01), respectively, all of these proteins were contaminants.

View larger version (40K): [in this window] [in a new window]   FIG. 6. MS/MS spectra acquired on peptide LVHSISSEDYTQIK from the protein Chico in sample Chico2. a, fragmentation of the dephosphorylated peptide; b, fragmentation of the phosphorylated counterpart. dhA, dehydroalanine.

View larger version (48K): [in this window] [in a new window]   FIG. 7. Phosphopeptides (underlined) identified in Chico in the analysis of samples Chico3 and Chico4. A color code is used to indicate the effect of insulin on the stoichiometry of phosphorylation: red, PO4 induced by insulin; yellow, PO4 increased by insulin; green, PO4 unaffected by insulin; blue, PO4 decreased by insulin; black, no quantitative information.

	DISCUSSION

TOP ABSTRACT EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Analytical Issues—
Our method allows identification of proteins constitutive of a complex without being limited by a practical maximum number of components. The only critical parameter to identify a protein as a real partner of the bait is its enrichment level in the sample of interest versus the control sample. The main point is that a distribution of non-enriched peptides can be clearly defined from which the enriched peptides statistically emerge. The main distribution of non-enriched species is more solid when the number of background proteins is much higher than that of the complex constituents; this is usually the case in immunopurified samples. The fraction of enriched peptides may become structured in distinct subdistributions, possibly indicative of variably tight interactions with the bait (2). This is not detrimental to identifying the corresponding proteins as real complex constituents as long as the main distribution of non-enriched peptides can be clearly defined. In addition, one prerequisite to identify a protein as a constituent of a complex is to detect it in a possibly large set of background proteins. Automated MALDI-TOF/TOF analysis globally leads to the identification of the more abundant proteins in the samples. Some real protein partners of the bait may not be identified if none of their proteolytic peptides provide sufficient signal in MS mode to be selected for fragmentation. As a global trend, a more comprehensive characterization of a protein complex will be obtained, whatever its complexity, if the proteins of interest represent higher abundance components of the analyzed sample. In our experiments, the preclearing steps of the cell lysates before immunoaffinity must have helped reduce the number and amount of background proteins.

Accurate quantification requires producing intense iTRAQ reporter groups. This can be modulated by the collision cell pressure in the MALDI-TOF/TOF instrument, which influences the fragmentation efficiency of peptides. The two LC-MS/MS analyses of the TAP sample were carried out at a gas pressure of 2 x 10^–6 and 5 x 10^–7 torr, respectively, in the collision cell. The first analysis provided much more intense iTRAQ reporter groups than the second: the sum of the four iTRAQ group areas averaged per identified peptide reached 95,500 for the first analysis versus 42,900 for the second. Quantitative measurements were therefore more reliable in the first analysis. This is also apparent when the number of identified peptides per iTRAQ ratio is plotted: the distribution width (i.e. the standard deviation characterizing the population of peptides theoretically associated to a ratio equal to 1) is significantly lower in the first MS analysis. In other words, the minimum iTRAQ ratio r_(115/114) or r_(117/116) that significantly indicates phosphorylation can be lower when using the higher collision cell pressure. For subsequent analyses, including those of the Chico samples, we used the higher gas pressure.

In addition, measurements of relative protein abundance by our method may also be distorted by the presence of differentially phosphorylated peptides due to the impact of phosphatase on the iTRAQ areas. Nevertheless as discussed in supplemental Data 2, the impact of the several phosphorylations detected in Chico appeared minimal.

Peptides phosphorylated on Ser, Thr, or Tyr residues, respectively, were identified in this study. It is worth mentioning that we successfully quantified phosphorylation on Tyr(P) in angiotensin. Indeed the reaction of tyrosine residues with N-hydroxysuccinimide groups (contained in the structure of the iTRAQ labels) has been documented (38), and the occurrence of this side reaction in the experiments described here would have been detrimental to the correct estimation of phosphorylation stoichiometries on this residue. To eliminate this unwanted reaction product between tyrosines and N-hydroxysuccinimide groups, Munchbach et al. (38) described a final treatment of the sample with 0.5 M hydroxylamine. In our protocol, although partial double iTRAQ labeling of angiotensin (at its N terminus and its tyrosine) could be observed after the labeling step, only the singly modified (at the N terminus) peptide form could be detected in the final samples A/pA. The heating step at 85 °C performed in the CIP buffer (of basic pH) to inactivate phosphatases removed the iTRAQ label from the tyrosine.

Our method proved successful in characterizing in detail singly phosphorylated peptides by providing localization of phosphosites and estimation of their stoichiometries. The study of multiply phosphorylated peptides appears more difficult: determining the number of phosphorylation sites present on a peptide requires detecting it in its fully modified form, whereas the analysis by MALDI-MS of multiply phosphorylated peptides usually requires the use of specific matrices (39) or additives (40, 41) to favor their ionization within the MALDI spot containing a large majority of non-modified peptides. The analysis of IMAC-enriched phosphopeptides recently published by Munton et al. (17) illustrated well the small number of multiply phosphorylated peptides that may be identified by automated LC-MALDI-TOF/TOF analysis of spots prepared in -cyano-4-hydroxycinnamic acid matrix: only 0.4% of all identified phosphopeptides were doubly phosphorylated, and no triply or more phosphorylated species could be reliably identified. Nevertheless our strategy allows pointing to multiply phosphorylated peptides through detection of their dephosphorylated counterparts. This was practically shown by the study of casein for which the dephosphorylated forms of three initially tetraphosphorylated peptides were identified. In all instances, we could conclude that the sequences were fully modified in the initial proteins without being able to determine the number and locations of phosphorylation sites. To characterize these peptides further, we considered that dephosphorylation by phosphatases may not have been 100% at all sites and looked for signals corresponding to residual singly phosphorylated species (still phosphorylated at one of the four phosphosites). From ionic species probably corresponding to (NANEEEYSIGSSSEESAEVATEEVK)* and (ELEELNVPGEIVESLSSSEESITR)* we could obtain MS/MS spectra containing no signal for iTRAQ group 114 (data not shown) as this is expected for initially multiply phosphorylated peptides (see supplemental data for more detailed explanations); we could then conclude that these two peptides were at least doubly phosphorylated in the original protein. However, fragmentation was not efficient enough to allow reading sequence information and to localize any phosphosite. In terms of quantification, estimating the stoichiometry of phosphorylation sites in multiply and heterogeneously phosphorylated peptides (modification level x1 at site Res1, x2 at site Res2, etc. where Res represents residue) requires approximations as explained in Case 2 of the supplemental text.

The proteins analyzed in this study, the trimer Kin28-Ccl1-Tfb3 as well as Chico and the 14-3-3 proteins, were identified with above 50% sequence coverage. To extend the characterization of phosphorylation sites, increased coverage could be obtained by repeated LC-MS/MS analyses, by targeted MS/MS analysis of theoretical masses of their tryptic peptides, and by utilizing a protease different from trypsin. The presented procedure is also consistent with ESI-MS/MS instruments capable of detecting low mass iTRAQ reporter groups. Because they globally favor ionization of different sets of peptides, the two ionization methods are expected to provide complementary sets of data (42). Compared with MALDI, electrospray is likely to favor ionization of phosphospecies, in particular multiply phosphorylated peptides, and thus may allow better characterization of multiply and heterogeneously phosphorylated species. But targeted fragmentation of peptides of interest in LC-ESI-MS/MS has the drawback of requiring injecting new peptidic material, whereas on a MALDI platform the original sample can be reinvestigated.

Phosphopeptides may escape identification for two other reasons. First, a non-modified peptide emerges as a candidate for being phosphorylated in the intact proteins on condition that the yield of phosphatase treatment is sufficiently high to bring the ratio r = A(115)/A(114) (or A(117)/A(116)) above a certain threshold. We estimated on the TAP sample that in the optimal case of complete dephosphorylation by phosphatase a minimum level of 20% phosphorylation could be detected. Second, the presence of a phosphorylation site in the close proximity to a trypsin cleavage site can hamper proteolysis at that site resulting in the complicated scheme depicted in supplemental Fig. S10. In an attempt to find additional phosphopeptides that may have been fragmented in the automated analysis of the plate, we performed a second round of database searches on samples Chico3 and Chico4 while considering potential phosphorylations on Ser, Thr, and Tyr residues. This led to the identification of one additional phosphorylated sequence, which actually corresponded to the second above mentioned issue: peptide GSRES*PPVSACPEDGNTYAK was identified in sample Chico4 (Mascot score, 34). However, as detailed in supplemental Data 3, no estimation of its stoichiometry of phosphorylation could be obtained due to the missed trypsin cleavage. Based on these considerations, we suggest a two-step interpretation of MS/MS data obtained with the procedure. First, a database search performed without considering phosphorylation allows rapid and reliable identification of non-modified peptides and points to the existence of phosphorylated counterparts. Second, to handle the issues of insufficient dephosphorylation of some sequences and inhibited trypsin cleavage, the data can be re-searched, this time taking into account possible phosphorylation. Finally when seeking presumed phosphopeptides, if no fragmentation spectrum can be obtained at a mass 80 Da above that of the non-modified peptide, then look for sequences resulting from a missed cleavage.

Following this procedure, 12 distinct sequences of Chico were proved to be phosphorylated with accurate estimation of the modification stoichiometry down to 40% (see Table VI). Similar results should be generally attainable with immunopurified samples.

Biological Relevance of the Results—
The two 14-3-3 proteins and were identified as interactors of Chico, and their association appeared to increase upon insulin stimulation of cells. The mammalian homologues of Chico, IRS-1 (43) as well as IRS-2 and IRS-4 (44), were also shown to bind to 14-3-3 proteins. IRS-1 was proven to interact with 14-3-3β in 3T3L1 adipocytes, and this binding was shown to increase with insulin treatment (45, 46). In contrast, Ogihara et al. (43) did not observe a significant change of interaction between 14-3-3 and IRS-1 upon hormonal stimulation in HepG2 cells; nevertheless this observation relied on Western blotting, which provides less accurate quantitative data than MS-based approaches and may not have been able to detect changes at or below 2-fold, such as those observed here using mass spectrometry techniques. In NIH-3T3 cells, 14-3-3 was shown to interact with IRS-1 and protein kinase C-, thus modulating insulin signaling and degradation (47). We also observed an increased association of Chico and IR after a 7-min insulin treatment, which reflects activation of the insulin pathway involving tyrosine phosphorylation of Chico by IR: in samples Chico3 and Chico4, the abundance ratio noINS/+INS for the receptor was 0.76 and 0.80, respectively.

We stimulated Kc cells with an insulin concentration and within a time window previously established to give a robust induction of the whole pathway (Ref. 48 and our own experience). As a result, we identified several insulin-dependent phosphosites, mainly phosphoserines, in Chico. The roles of phosphoserines/phosphothreonines in the mammalian homologue IRS-1 have been studied with regard to the regulation of the insulin pathway. Some serine residues, when phosphorylated, participate in the negative control of insulin signaling (49–51), whereas others appear to have a positive regulatory function (52, 53). The homology of the Chico sequence to the mammalian IRS homologues is too weak to allow precise comparison of phosphosites. Nonetheless it is worth mentioning that some serine residues were shown previously to become partially or fully phosphorylated in rat and mouse IRS-1 after 5-min stimulation with 80–100 nM insulin (52, 54), which is in agreement with our observations. Among the phosphorylated residues identified in Chico, several appear to correlate with insulin stimulation either positively or negatively. Most interestingly, five sequences overlap with predicted recognition motives of 14-3-3 proteins. All but one of them were shown to become more highly phosphorylated upon stimulation, which correlates well with an enhanced association of the two 14-3-3 proteins with Chico. The differences of phosphorylation levels measured in samples Chico3 and Chico4 may be, at least in part, due to the different cell densities reached before induction. Despite differences in absolute phosphorylation levels, similar variations of the phosphorylation states (increase or decrease) were observed in the two samples upon insulin stimulation.

Phosphorylations on tyrosine residues were also expected at least upon insulin treatment (55). As described in supplemental Data 4, the presence of phosphotyrosine-containing peptides could not be conclusively established by our MS data. Nonetheless the intact protein Chico could be shown to contain phosphorylated tyrosines: a fraction of the samples Chico3 and Chico4 was analyzed by Western blot using an anti-phosphotyrosine antibody, and signal was detected in both insulin conditions with increased signal in the +INS case as expected (55).

Conclusion—
We have developed a method that provides in a single LC-MALDI-TOF/TOF analysis in-depth characterization of partially purified protein complexes: it allows identifying genuine partners of the bait, detecting variations in complex composition, and pointing to phosphorylation sites and estimating their stoichiometry. As illustrated on the Chico samples, this method should allow deciphering the functioning of protein modules regulated by dynamic phosphorylations and protein associations. This method could be transposed to different enzymes to characterize other modifications than phosphorylations, such as N-acetyl-β-D-glucosaminidase to study O-linked N-acetylglucosamine, which can also modulate protein association.

	ACKNOWLEDGMENTS

 
D. P. is particularly grateful to Anne-Claude Gingras and Jeff Ranish for precious advice on the analysis of the TAP yeast sample and to Bernd Bodenmiller for fruitful discussions on the analysis of phosphopeptides. We also thank Sheng Pan for preliminary MALDI-TOF/TOF experiments.

	FOOTNOTES

 
Received, June 18, 2007, and in revised form, October 2, 2007.

Published, MCP Papers in Press, October 23, 2007, DOI 10.1074/mcp.M700282-MCP200

¹ The abbreviations used are: iTRAQ, isobaric tags for relative and absolute quantification; A(114), area of the iTRAQ reporter group 114; A, angiotensin; pA, phosphorylated angiotensin; CIP, calf intestinal phosphatase; IP, immunoprecipitation; IR, insulin receptor; IRS (Chico), insulin receptor substrate; SAP, shrimp alkaline phosphatase; TAP, tandem affinity purification; DSP, dithiobis(succinimidyl propionate); TEV, tobacco etch virus; HA, hemagglutinin; S/N, signal to noise ratio; ACTH, adrenocorticotropic hormone; INS, insulin.

² Nomenclature for phosphosites: a phosphorylated residue is indicated in bold font and followed by an asterisk. When the precise location of a phosphorylation remains ambiguous, a group of residues, one of which bears the modification, is surrounded by parentheses: AT(STS)*MNK.

^* This work was supported in part by federal funds from the NHLBI, National Institutes of Health under Contract N01-HV-28179, by a grant from the Swiss National Science Foundation (to R. A.), and by internal funds from ETH Zurich. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^S The on-line version of this article (available at http://www.mcponline.org) contains supplemental material.

Received a postdoctoral fellowship from the French Foundation for Medical Research (Fondation pour la Recherche Médicale). Present address: CNRS, UMR 8587, 91025 Evry, France and Laboratoire Analyse et Modélisation pour la Biologie et l’Environnement, Université d'Evry Val d'Essonne, 91025 Evry, France.

To whom correspondence should be addressed. E-mail: aebersold{at}imsb.biol.ethz.ch

	REFERENCES

TOP ABSTRACT EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

 

1. Gingras, A. C., Aebersold, R., and Raught, B. (2005 ) Advances in protein complex analysis using mass spectrometry. J. Physiol. 563, 11 –21[Abstract/Free Full Text]

2. Ranish, J. A., Yi, E. C., Leslie, D. M., Purvine, S. O., Goodlett, D. R., Eng, J., and Aebersold, R. (2003 ) The study of macromolecular complexes by quantitative proteomics. Nat. Genet. 33, 349 –355[CrossRef][Medline]

3. Wang, T., Gu, S., Ronni, T., Du, Y. C., and Chen, X. (2005 ) In vivo dual-tagging proteomic approach in studying signaling pathways in immune response. J. Proteome Res. 4, 941 –949[CrossRef][Medline]

4. Selbach, M., and Mann, M. (2006 ) Protein interaction screening by quantitative immunoprecipitation combined with knockdown (QUICK). Nat. Methods 3, 981 –983[CrossRef][Medline]

5. Rinner, O., Mueller, L. N., Hubalek, M., Muller, M., Gstaiger, M., and Aebersold, R. (2007 ) An integrated mass spectrometric and computational framework for the analysis of protein interaction networks. Nat. Biotechnol. 25, 345 –352[CrossRef][Medline]

6. Paoletti, A. C., Parmely, T. J., Tomomori-Sato, C., Sato, S., Zhu, D., Conaway, R. C., Conaway, J. W., Florens, L., and Washburn, M. P. (2006 ) Quantitative proteomic analysis of distinct mammalian Mediator complexes using normalized spectral abundance factors. Proc. Natl. Acad. Sci. U. S. A. 103, 18928 –18933[Abstract/Free Full Text]

7. Blagoev, B., Kratchmarova, I., Ong, S. E., Nielsen, M., Foster, L. J., and Mann, M. (2003 ) A proteomics strategy to elucidate functional protein-protein interactions applied to EGF signaling. Nat. Biotechnol. 21, 315 –318[CrossRef][Medline]

8. Brand, M., Ranish, J. A., Kummer, N. T., Hamilton, J., Igarashi, K., Francastel, C., Chi, T. H., Crabtree, G. R., Aebersold, R., and Groudine, M. (2004 ) Dynamic changes in transcription factor complexes during erythroid differentiation revealed by quantitative proteomics. Nat. Struct. Mol. Biol. 11, 73 –80[CrossRef][Medline]

9. Corvey, C., Koetter, P., Beckhaus, T., Hack, J., Hofmann, S., Hampel, M., Stein, T., Karas, M., and Entian, K. D. (2005 ) Carbon source-dependent assembly of the Snf1p kinase complex in Candida albicans. J. Biol. Chem. 280, 25323 –25330[Abstract/Free Full Text]

10. Smolka, M. B., Albuquerque, C. P., Chen, S. H., Schmidt, K. H., Wei, X. X., Kolodner, R. D., and Zhou, H. (2005 ) Dynamic changes in protein-protein interaction and protein phosphorylation probed with amine-reactive isotope tag. Mol. Cell. Proteomics 4, 1358 –1369[Abstract/Free Full Text]

11. Blagoev, B., Ong, S. E., Kratchmarova, I., and Mann, M. (2004 ) Temporal analysis of phosphotyrosine-dependent signaling networks by quantitative proteomics. Nat. Biotechnol. 22, 1139 –1145[CrossRef][Medline]

12. Hinsby, A. M., Olsen, J. V., and Mann, M. (2004 ) Tyrosine phosphoproteomics of fibroblast growth factor signaling: a role for insulin receptor substrate-4. J. Biol. Chem. 279, 46438 –46447[Abstract/Free Full Text]

13. Zhang, X., Jin, Q. K., Carr, S. A., and Annan, R. S. (2002 ) N-terminal peptide labeling strategy for incorporation of isotopic tags: a method for the determination of site-specific absolute phosphorylation stoichiometry. Rapid Commun. Mass Spectrom. 16, 2325 –2332[CrossRef][Medline]

14. Hegeman, A. D., Harms, A. C., Sussman, M. R., Bunner, A. E., and Harper, J. F. (2004 ) An isotope labeling strategy for quantifying the degree of phosphorylation at multiple sites in proteins. J. Am. Soc. Mass Spectrom. 15, 647 –653[CrossRef][Medline]

15. Gerber, S. A., Rush, J., Stemman, O., Kirschner, M. W., and Gygi, S. P. (2003 ) Absolute quantification of proteins and phosphoproteins from cell lysates by tandem MS. Proc. Natl. Acad. Sci. U. S. A. 100, 6940 –6945[Abstract/Free Full Text]

16. Ruse, C. I., Tan, F. L., Kinter, M., and Bond, M. (2004 ) Integrated analysis of the human cardiac transcriptome, proteome and phosphoproteome. Proteomics 4, 1505 –1516[CrossRef][Medline]

17. Munton, R. P., Tweedie-Cullen, R., Livingstone-Zatchej, M., Weinandy, F., Waidelich, M., Longo, D., Gehrig, P., Potthast, F., Rutishauser, D., Gerrits, B., Panse, C., Schlapbach, R., and Mansuy, I. M. (2007 ) Qualitative and quantitative analyses of protein phosphorylation in naive and stimulated mouse synaptosomal preparations. Mol. Cell. Proteomics 6, 283 –293[Abstract/Free Full Text]

18. Steen, H., Jebanathirajah, J. A., Springer, M., and Kirschner, M. W. (2005 ) Stable isotope-free relative and absolute quantitation of protein phosphorylation stoichiometry by MS. Proc. Natl. Acad. Sci. U. S. A. 102, 3948 –3953[Abstract/Free Full Text]

19. Carr, S. A., Huddleston, M. J., and Annan, R. S. (1996 ) Selective detection and sequencing of phosphopeptides at the femtomole level by mass spectrometry. Anal. Biochem. 239, 180 –192[CrossRef][Medline]

20. Tsay, Y. G., Wang, Y. H., Chiu, C. M., Shen, B. J., and Lee, S. C. (2000 ) A strategy for identification and quantitation of phosphopeptides by liquid chromatography/tandem mass spectrometry. Anal. Biochem. 287, 55 –64[CrossRef][Medline]

21. Schroeder, M. J., Webb, D. J., Shabanowitz, J., Horwitz, A. F., and Hunt, D. F. (2005 ) Methods for the detection of paxillin post-translational modifications and interacting proteins by mass spectrometry. J. Proteome Res. 4, 1832 –1841[CrossRef][Medline]

22. Wu, S. L., Kim, J., Bandle, R. W., Liotta, L., Petricoin, E., and Karger, B. L. (2006 ) Dynamic profiling of the post-translational modifications and interaction partners of epidermal growth factor receptor signaling after stimulation by epidermal growth factor using Extended Range Proteomic Analysis (ERPA). Mol. Cell. Proteomics 5, 1610 –1627[Abstract/Free Full Text]

23. Puig, O., Caspary, F., Rigaut, G., Rutz, B., Bouveret, E., Bragado-Nilsson, E., Wilm, M., and Seraphin, B. (2001 ) The tandem affinity purification (TAP) method: a general procedure of protein complex purification. Methods 24, 218 –229[CrossRef][Medline]

24. Bohni, R., Riesgo-Escovar, J., Oldham, S., Brogiolo, W., Stocker, H., Andruss, B. F., Beckingham, K., and Hafen, E. (1999 ) Autonomous control of cell and organ size by CHICO, a Drosophila homolog of vertebrate IRS1–4. Cell 97, 865 –875[CrossRef][Medline]

25. Rigaut, G., Shevchenko, A., Rutz, B., Wilm, M., Mann, M., and Seraphin, B. (1999 ) A generic protein purification method for protein complex characterization and proteome exploration. Nat. Biotechnol. 17, 1030 –1032[CrossRef][Medline]

26. Gingras, A. C., Caballero, M., Zarske, M., Sanchez, A., Hazbun, T. R., Fields, S., Sonenberg, N., Hafen, E., Raught, B., and Aebersold, R. (2005 ) A novel, evolutionarily conserved protein phosphatase complex involved in cisplatin sensitivity. Mol. Cell. Proteomics 4, 1725 –1740[Abstract/Free Full Text]

27. Espinoza, F. H., Farrell, A., Nourse, J. L., Chamberlin, H. M., Gileadi, O., and Morgan, D. O. (1998 ) Cak1 is required for Kin28 phosphorylation and activation in vivo. Mol. Cell. Biol. 18, 6365 –6373[Abstract/Free Full Text]

28. Gruhler, A., Olsen, J. V., Mohammed, S., Mortensen, P., Faergeman, N. J., Mann, M., and Jensen, O. N. (2005 ) Quantitative phosphoproteomics applied to the yeast pheromone signaling pathway. Mol. Cell. Proteomics 4, 310 –327[Abstract/Free Full Text]

29. Ross, P. L., Huang, Y. N., Marchese, J. N., Williamson, B., Parker, K., Hattan, S., Khainovski, N., Pillai, S., Dey, S., Daniels, S., Purkayastha, S., Juhasz, P., Martin, S., Bartlet-Jones, M., He, F., Jacobson, A., and Pappin, D. J. (2004 ) Multiplexed protein quantitation in Saccharomyces cerevisiae using amine-reactive isobaric tagging reagents. Mol. Cell. Proteomics 3, 1154 –1169[Abstract/Free Full Text]

30. Krokhin, O. V., Antonovici, M., Ens, W., Wilkins, J. A., and Standing, K. G. (2006 ) Deamidation of -Asn-Gly- sequences during sample preparation for proteomics: consequences for MALDI and HPLC-MALDI analysis. Anal. Chem. 78, 6645 –6650[Medline]

31. Jona, G., Livi, L. L., and Gileadi, O. (2002 ) Mutations in the RING domain of TFB3, a subunit of yeast transcription factor IIH, reveal a role in cell cycle progression. J. Biol. Chem. 277, 39409 –39416[Abstract/Free Full Text]

32. Keogh, M. C., Cho, E. J., Podolny, V., and Buratowski, S. (2002 ) Kin28 is found within TFIIH and a Kin28-Ccl1-Tfb3 trimer complex with differential sensitivities to T-loop phosphorylation. Mol. Cell. Biol. 22, 1288 –1297[Abstract/Free Full Text]

33. Ahmad, Z., and Huang, K. P. (1981 ) Dephosphorylation of rabbit skeletal muscle glycogen synthase (phosphorylated by cyclic AMP-independent synthase kinase 1) by phosphatases. J. Biol. Chem. 256, 757 –760[Abstract/Free Full Text]

34. Zhuo, S., Clemens, J. C., Hakes, D. J., Barford, D., and Dixon, J. E. (1993 ) Expression, purification, crystallization, and biochemical characterization of a recombinant protein phosphatase. J. Biol. Chem. 268, 17754 –17761[Abstract/Free Full Text]

35. Zhang, Y., Wolf-Yadlin, A., Ross, P. L., Pappin, D. J., Rush, J., Lauffenburger, D. A., and White, F. M. (2005 ) Time-resolved mass spectrometry of tyrosine phosphorylation sites in the epidermal growth factor receptor signaling network reveals dynamic modules. Mol. Cell. Proteomics 4, 1240 –1250[Abstract/Free Full Text]

36. Ishihama, Y., Wei, F. Y., Aoshima, K., Sato, T., Kuromitsu, J., and Oda, Y. (2007 ) Enhancement of the efficiency of phosphoproteomic identification by removing phosphates after phosphopeptide enrichment. J. Proteome Res. 6, 1139 –1144[CrossRef][Medline]

37. Torres, M. P., Thapar, R., Marzluff, W. F., and Borchers, C. H. (2005 ) Phosphatase-directed phosphorylation-site determination: a synthesis of methods for the detection and identification of phosphopeptides. J. Proteome Res. 4, 1628 –1635[CrossRef][Medline]

38. Munchbach, M., Quadroni, M., Miotto, G., and James, P. (2000 ) Quantitation and facilitated de novo sequencing of proteins by isotopic N-terminal labeling of peptides with a fragmentation-directing moiety. Anal. Chem. 72, 4047 –4057[Medline]

39. Yang, X., Wu, H., Kobayashi, T., Solaro, R. J., and van Breemen, R. B. (2004 ) Enhanced ionization of phosphorylated peptides during MALDI TOF mass spectrometry. Anal. Chem. 76, 1532 –1536[Medline]

40. Asara, J. M., and Allison, J. (1999 ) Enhanced detection of phosphopeptides in matrix-assisted laser desorption/ionization mass spectrometry using ammonium salts. J. Am. Soc. Mass Spectrom. 10, 35 –44[CrossRef][Medline]

41. Kjellstrom, S., and Jensen, O. N. (2004 ) Phosphoric acid as a matrix additive for MALDI MS analysis of phosphopeptides and phosphoproteins. Anal. Chem. 76, 5109 –5117[Medline]

42. Zhen, Y., Xu, N., Richardson, B., Becklin, R., Savage, J. R., Blake, K., and Peltier, J. M. (2004 ) Development of an LC-MALDI method for the analysis of protein complexes. J. Am. Soc. Mass Spectrom. 15, 803 –822[CrossRef][Medline]

43. Ogihara, T., Isobe, T., Ichimura, T., Taoka, M., Funaki, M., Sakoda, H., Onishi, Y., Inukai, K., Anai, M., Fukushima, Y., Kikuchi, M., Yazaki, Y., Oka, Y., and Asano, T. (1997 ) 14-3-3 protein binds to insulin receptor substrate-1, one of the binding sites of which is in the phosphotyrosine binding domain. J. Biol. Chem. 272, 25267 –25274[Abstract/Free Full Text]

44. Jin, J., Smith, F. D., Stark, C., Wells, C. D., Fawcett, J. P., Kulkarni, S., Metalnikov, P., O'Donnell, P., Taylor, P., Taylor, L., Zougman, A., Woodgett, J. R., Langeberg, L. K., Scott, J. D., and Pawson, T. (2004 ) Proteomic, functional, and domain-based analysis of in vivo 14-3-3 binding proteins involved in cytoskeletal regulation and cellular organization. Curr. Biol. 14, 1436 –1450[CrossRef][Medline]

45. Kosaki, A., Yamada, K., Suga, J., Otaka, A., and Kuzuya, H. (1998 ) 14-3-3β protein associates with insulin receptor substrate 1 and decreases insulin-stimulated phosphatidylinositol 3`-kinase activity in 3T3L1 adipocytes. J. Biol. Chem. 273, 940 –944[Abstract/Free Full Text]

46. Xiang, X., Yuan, M., Song, Y., Ruderman, N., Wen, R., and Luo, Z. (2002 ) 14-3-3 facilitates insulin-stimulated intracellular trafficking of insulin receptor substrate 1. Mol. Endocrinol. 16, 552 –562[Abstract/Free Full Text]

47. Oriente, F., Andreozzi, F., Romano, C., Perruolo, G., Perfetti, A., Fiory, F., Miele, C., Beguinot, F., and Formisano, P. (2005 ) Protein kinase C- regulates insulin action and degradation by interacting with insulin receptor substrate-1 and 14-3-3. J. Biol. Chem. 280, 40642 –40649[Abstract/Free Full Text]

48. Radimerski, T., Montagne, J., Rintelen, F., Stocker, H., van der Kaay, J., Downes, C. P., Hafen, E., and Thomas, G. (2002 ) dS6K-regulated cell growth is dPKB/dPI(3)K-independent, but requires dPDK1. Nat. Cell Biol. 4, 251 –255[CrossRef][Medline]

49. Kim, J. A., Yeh, D. C., Ver, M., Li, Y., Carranza, A., Conrads, T. P., Veenstra, T. D., Harrington, M. A., and Quon, M. J. (2005 ) Phosphorylation of Ser24 in the pleckstrin homology domain of insulin receptor substrate-1 by Mouse Pelle-like kinase/interleukin-1 receptor-associated kinase: cross-talk between inflammatory signaling and insulin signaling that may contribute to insulin resistance. J. Biol. Chem. 280, 23173 –23183[Abstract/Free Full Text]

50. Liu, Y. F., Herschkovitz, A., Boura-Halfon, S., Ronen, D., Paz, K., Leroith, D., and Zick, Y. (2004 ) Serine phosphorylation proximal to its phosphotyrosine binding domain inhibits insulin receptor substrate 1 function and promotes insulin resistance. Mol. Cell. Biol. 24, 9668 –9681[Abstract/Free Full Text]

51. Sommerfeld, M. R., Metzger, S., Stosik, M., Tennagels, N., and Eckel, J. (2004 ) In vitro phosphorylation of insulin receptor substrate 1 by protein kinase C-: functional analysis and identification of novel phosphorylation sites. Biochemistry 43, 5888 –5901[CrossRef][Medline]

52. Giraud, J., Leshan, R., Lee, Y. H., and White, M. F. (2004 ) Nutrient-dependent and insulin-stimulated phosphorylation of insulin receptor substrate-1 on serine 302 correlates with increased insulin signaling. J. Biol. Chem. 279, 3447 –3454[Abstract/Free Full Text]

53. Luo, M., Reyna, S., Wang, L., Yi, Z., Carroll, C., Dong, L. Q., Langlais, P., Weintraub, S. T., and Mandarino, L. J. (2005 ) Identification of insulin receptor substrate 1 serine/threonine phosphorylation sites using mass spectrometry analysis: regulatory role of serine 1223. Endocrinology 146, 4410 –4416[Abstract/Free Full Text]

54. Hers, I., and Tavare, J. M. (2005 ) Mechanism of feedback regulation of insulin receptor substrate-1 phosphorylation in primary adipocytes. Biochem. J. 388, 713 –720[CrossRef][Medline]

55. Poltilove, R. M., Jacobs, A. R., Haft, C. R., Xu, P., and Taylor, S. I. (2000 ) Characterization of Drosophila insulin receptor substrate. J. Biol. Chem. 275, 23346 –23354[Abstract/Free Full Text]

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				K. C. Wilcox, L. Zhou, J. K. Jordon, Y. Huang, Y. Yu, R. L. Redler, X. Chen, M. Caplow, and N. V. Dokholyan Modifications of Superoxide Dismutase (SOD1) in Human Erythrocytes: A POSSIBLE ROLE IN AMYOTROPHIC LATERAL SCLEROSIS J. Biol. Chem., May 15, 2009; 284(20): 13940 - 13947. [Abstract] [Full Text] [PDF]

				A. Doucet, G. S. Butler, D. Rodriguez, A. Prudova, and C. M. Overall Metadegradomics: Toward in Vivo Quantitative Degradomics of Proteolytic Post-translational Modifications of the Cancer Proteome Mol. Cell. Proteomics, October 1, 2008; 7(10): 1925 - 1951. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: M700282-MCP200v1 7/2/326    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Glossary Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Pflieger, D. Articles by Aebersold, R. Search for Related Content PubMed PubMed Citation Articles by Pflieger, D. Articles by Aebersold, R. Social Bookmarking                      What's this?