MCP Agilent Technologies
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/mcp.M600105-MCP200 on June 23, 2006.
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
M600105-MCP200v1
5/9/1610    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Glossary
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Wu, S.-L.
Right arrow Articles by Karger, B. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Wu, S.-L.
Right arrow Articles by Karger, B. L.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Molecular & Cellular Proteomics 5:1610-1627, 2006.
© 2006 by The American Society for Biochemistry and Molecular Biology, Inc.

Research

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)*,S

Shiaw-Lin Wu{ddagger},§, Jeongkwon Kim{ddagger},§, Russell W. Bandle, Lance Liotta||, Emanuel Petricoin|| and Barry L. Karger{ddagger},**

From the {ddagger} Barnett Institute, Northeastern University, Boston, Massachusetts 01225, Laboratory of Pathology, NCI, National Institutes of Health, Bethesda, Maryland 20892, and || Center for Applied Proteomics and Molecular Medicine, George Mason University, Manassas, Virginia 20110


   ABSTRACT
TOP
 ABSTRACT
EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
In a recent report, we introduced Extended Range Proteomic Analysis (ERPA), an intermediate approach between top-down and bottom-up proteomics, for the comprehensive characterization at the trace level (fmol level) of large and complex proteins. In this study, we extended ERPA to determine quantitatively the temporal changes that occur in the tyrosine kinase receptor, epidermal growth factor receptor (EGFR), upon stimulation. Specifically A 431 cells were stimulated with epidermal growth factor after which EGFR was immunoprecipitated at stimulation times of 0, 0.5, 2, and 10 min as well as 4 h. High sequence coverage was obtained (96%), and methods were developed for label-free quantitation of phosphorylation and glycosylation. A total of 13 phosphorylation sites were identified, and the estimated stoichiometry was determined over the stimulation time points, including Thr(P) and Ser(P) sites in addition to Tyr(P) sites. A total of 10 extracellular domain N-glycan sites were also identified, and major glycoforms at each site were quantitated. No change in the extent of glycosylation with stimulation was observed as expected. Finally potential binding partners to EGFR were identified based on changes in the amount of protein pulled down with EGFR as a function of time of stimulation. Many of the 19 proteins identified are known binding partners of EGFR. This work demonstrates that comprehensive characterization provides a powerful tool to aid in the study of important therapeutic targets. The detailed molecular information will prove useful in future studies in tissue.

Tyrosine kinase receptors, such as the epidermal growth factor receptor (EGFR),1 are important therapeutic targets. There is a growing list of Food and Drug Administration-approved drugs that specifically interact and modulate disorders in the EGFR family. Ligand binding to the extracellular domain of such receptors induces the formation of receptor homo- and heterodimers and the activation of the intrinsic kinase domain resulting in phosphorylation on specific residues within the intracellular tail (1). The autophosphorylation sites inside the intracellular tail often serve as docking positions for a range of proteins, initiating cascades of separate, and functionally distinct, downstream signal pathways. Overexpression of EGFR can lead to uncontrolled growth of cells and tumor formation (2). Thus, the detailed study of tyrosine kinase receptors, particularly upon stimulation by specific ligands, is important in the understanding of disease processes and the identification of new drug targets.

Previous studies of post-translational modifications of EGFR, which have focused mainly on the phosphorylation in the cytoplasmic tail, have indicated that phosphorylation events are transient, rapidly fluctuating over time (2). Indeed the temporal fluctuation of EGFR is thought to be the major mechanism by which a small family of receptors can influence a wide diversity of cellular functions, such as regulation, differentiation, and even apoptosis (3, 4). Importantly the spatial and temporal fluctuations of phosphorylation of EGFR could be causal determinants in diseases such as cancer (5) as these autophosphorylation sites, as noted, are the initiation points for various signaling pathways that control the above cellular functions (6). Thus, it is important to identify patterns of binding to the epidermal growth factor (EGF) receptor that are the first steps in signaling. Another factor known to affect phosphorylation and the resultant intracellular protein binding is sequence mutation in the cytoplasmic tail. Indeed an important therapeutic agent for lung cancer is only successful for that level of the population that has a specific mutation (7–9).

Kinase-driven phosphorylation is only one of several classes of functionally important post-translational modifications. The extracellular domains of EGFR and other receptors are highly glycosylated. Although much less studied, because of their complexity, glycosylation is significant in the control of ligand binding and signaling. For example, elimination of glycosylation at Asn420 and Asn579 on EGFR has been shown to alter ligand binding to the receptor as well as to affect membrane receptor dimerization and subsequent phosphorylation (10, 11). Although the study of glycosylation patterns and their role in signaling is a relatively new area of exploration, it is clear that such study will be a necessary component in the characterization of tyrosine kinase receptors.

To elucidate the molecular basis of signal transduction, i.e. interaction partners and the kinetics of post-translational modifications, new technologies are needed for global characterization of signaling network components. Moreover to extend this analysis ultimately to in vivo systems and clinical material, such as tissue, such technologies must be sufficiently sensitive for analysis at the trace level. The ideal technology would provide a suite of receptor information including high sequence coverage, post-translational analysis, and quantitation of various phospho- and glyco-isoforms and identification of its binding partners. Although methodologies have been implemented for portions of the above, a full analysis has heretofore not been developed. It was the purpose of this study to demonstrate a methodology, Extended Range Proteomic Analysis (ERPA), that can be utilized for comprehensive characterization of the above components of signaling. In the present study we applied this technology to analyze comprehensively EGFR/ERBB1 signal transduction with a particular emphasis on kinetic analysis that captures the critical dynamic nature of cellular signaling.

With respect to protein characterization, high sequence coverage has been a focus of top-down proteomics using FTMS (12). Although some impressive results have been obtained, the method is relatively insensitive (pmol or higher), is not readily useable above protein sizes of 50 kDa, has not been applied to date for glycosylation analysis, and requires relatively pure proteins for analysis. With respect to phosphorylation site identification, LC-MS approaches using proteolytic products of proteins, often obtained by immunoprecipitation, have recently been actively pursued (13). Because of the low stoichiometry of phosphorylation, enrichment approaches using IMAC (14), titanium oxide (15), ion exchange (16), or phosphotyrosine antibodies (17) are often used. Although these approaches represent a means for low level phosphorylation identification, a concern is the bias induced by the enrichment approach (18, 19) and the neglect of the glycosylated peptides. With regard to glycosylation, a typical strategy has often involved the enrichment of the glycopeptides from a proteolytic digest (20) or the affinity capture and separation of oligosaccharides released after glycosidase treatment (21) Again the phosphorylated peptides and non-glycosylated peptides are often neglected by these approaches. To overcome the limits of these approaches, we describe a comprehensive strategy to discover, map, and quantitatively characterize protein phosphorylation sites, glycosylation sites, and protein binding partners.

In a previous study, we introduced the on-line LC-MS method ERPA for the analysis of post-translationally modified EGFR (22) in which the determination of phosphorylation and glycosylation in one experiment was demonstrated. This intermediate approach between bottom-up (23) and top-down (24) proteomics uses proteolytic enzymes such as Lys-C (C-terminal Lys) to cut proteins less frequently than trypsin (C-terminal Arg and Lys). As a consequence, the molecular weight distribution of peptide fragments is generally wider than with tryptic digests, leading to larger fragments and simpler mixtures relative to trypsin (average of 2–3 times larger in size with 2–3 times fewer peptide fragments). Using a hybrid LTQ-FT mass spectrometer and a protocol of MS2 followed by MS3 of the most intense MS2 fragment, high sequence coverage (~95%) as well as the sites and structures of phosphorylation and glycosylation were obtained at the hundreds of fmol to 1 pmol level. In this approach, enzymatically digested fragments up to 10 kDa could be analyzed in which the mass resolution of the FT (100,000) allowed determination of the high charge state of the large peptide ion. Importantly a 10–50-fold increase in signal intensity of post-translationally modified Lys-C fragments (relative to tryptic fragments) was observed as a consequence of additional basic amino acid residues in the peptide backbone as well as an increase in chromatographic retention. In a separate study, we achieved high sequence coverage for EGFR with an order of magnitude decrease in detection limits (low fmol) by using a narrow bore monolithic columns (20-µm inner diameter).2

In addition to comprehensive structural characterization, the ERPA approach can significantly enhance LC-MS as a quantitative method, an area of current active interest (26, 27). Many approaches for quantitation involve the use of stable isotopes either incorporated metabolically (28, 29) or chemically (30–32). Recently label-free methods of quantitation using LC-MS have been introduced (33–36). Because one of our ultimate goals is to conduct quantitative proteomics in tissue, we have focused on label-free approaches. As a result of the high sequence coverage obtained by the ERPA method, normalization for the amount of EGFR present in each time point of stimulation using non-post-translationally modified peptides of EGFR or intrinsic internal standards (i.e. non-phosphorylated counterparts) can be readily achieved. Thus, in addition to the ability to quantify accurately the level of phosphorylation and glycosylation and the abundance of the binding partners, the stoichiometry of phosphorylation and glycosylation can be estimated.

In this study, we measured the temporal changes of the EGFR upon stimulation by EGF in an A 431 cancer cell line. Using an antibody that binds to the intracellular C-terminal region but not to the autophosphorylation sites, EGFR was immunoprecipitated after EGF stimulation for 0, 0.5, 2, and 10 min as well as 4 h. Sequence coverage of 96% for EGFR was obtained at each time point by use of ERPA methodology. A total of 13 phosphorylation sites were identified without the need of enrichment. Direct analysis of EGFR allowed quantitation and estimation of stoichiometry of each site as a function of time of EGF stimulation. Furthermore 10 extracellular glycosylation sites were identified with the major glycosylation patterns characterized, and the relative contribution of individual major glycoforms at a given site was determined. Finally binding partners and their temporal changes with EGF stimulation were determined from quantitative analysis of proteins pulled down in the immunoprecipitation of EGFR. Correction for nonspecific binding proteins and normalization to the amount of EGFR precipitated was also incorporated into the analysis. Importantly because quantitation was achieved by a label-free strategy, the comprehensive analysis of targets in clinical tissue samples will be possible.


   EXPERIMENTAL PROCEDURES
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
RESULTS
 DISCUSSION
 REFERENCES
 
Reagents—
A 431 epidermoid carcinoma cells were obtained from ATCC (Manassas, VA). Dulbecco’s modified Eagle’s medium (DMEM) with or without fetal bovine serum (FBS) was from Gemini Bio (Woodland, CA). Normal rabbit IgG, rabbit anti-EGFR, rabbit anti-lamin A/C, rabbit anti-CD98, rabbit anti-Lyn, rabbit anti-caveolin-1, rabbit anti-clathrin light chain A, rabbit anti-calgranulin B, mouse anti-GRB2, and goat anti-stratifin/14-3-3{sigma} were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit phosphospecific EGFR (Tyr(P)992, Tyr(P)1068, Tyr(P)1148, and Tyr(P)1173) antibodies were from Cell Signaling Technology (Beverly, MA), rabbit anti-Tyr(P)1148 EGFR was from BioSource (Camarillo, CA), and mouse anti-GAPDH was from Ambion (Austin, TX). Recombinant human EGF, Nonidet P-40, Tween 20, and protease inhibitor mixture I were obtained from EMD Biosciences (La Jolla, CA). Dimethyl pimelimidate dihydrochloride (DMP), Micro BCA protein assay, horseradish peroxidase-conjugated goat anti-mouse, goat anti-rabbit, and rabbit anti-goat were from Pierce. BSA, phosphatase inhibitor mixtures I and II, sodium fluoride, sodium orthovanadate, citrate, ß-mercaptoethanol, DTT, iodoacetamide, guanidine hydrochloride, and ammonium bicarbonate were obtained from Sigma. Paramagnetic protein A Dynabeads, NuPAGE protein gel reagents (4x lithium dodecyl sulfate sample buffer, Novex Bis-Tris protein gels, MOPS and MES SDS running buffers, transfer buffer, and antioxidant), Invitrolon 0.45-µm PVDF, and SYPRO Ruby gel stain were purchased from Invitrogen. ECL Plus enhanced chemiluminescence kit was from GE Healthcare. Carnation nonfat dry milk was from Nestle USA (Solon, OH). Achromobacter protease I (Lys-C) was obtained from Wako (Richmond, VA). Formic acid, acetone, and acetonitrile were purchased from Fisher, and the HPLC grade water used in all experiments was from J. T. Baker Inc.

Cell Culture—
Cultures of the A 431 human epidermoid carcinoma cell line were seeded at 1.5 x 107 cells/150-mm dish in complete growth medium (0.22-µm filtered DMEM with 10% FBS, 2 mM L-glutamine, 1.5 g/liter sodium bicarbonate, 4.5 g/liter glucose, 100 units/ml penicillin, and 100 µg/ml streptomycin) and grown at 37 °C with 5% CO2. After 2 days, the growth medium was removed, the cells were washed twice with TBS (0.22-µm filtered 50 mM Tris-HCl, pH 7.4, 150 mM NaCl) and then serum-starved by continuing cell culture overnight in serum-free medium (DMEM without FBS or BSA).

A 431 cells were either untreated (controls) or treated by removing the serum-free medium and replacing with warmed serum-free medium containing 25 ng/ml EGF and then incubated for 30 s, 2 min, 10 min, or 4 h at 37 °C. Each time point produced 1.5 x 108 cells (four 150-mm dishes at 80% confluency).

Cell Lysis—
Immediately following EGF stimulation, the A 431 cells were washed three times with ice-cold TBS and lysed on ice with 5 ml of modified radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 1x protease inhibitor mixture set I, and 1x phosphatase inhibitor mixture sets I and II) per plate for 5 min, scraped, and then transferred to a 50-ml tube. Pooled lysates from each time point were adjusted to 50 ml with additional RIPA buffer followed by five passages through a 21-gauge needle, incubation on ice for 30 min, and centrifugation at 10,000 x g for 10 min. The clarified supernatants were transferred to new 50-ml tubes and normalized by total protein concentration (Micro BCA assay).

EGFR Immunoprecipitation—
The clarified A 431 lysates were precleared with 75 µg of normal rabbit IgG (cross-linked to 750 µl of paramagnetic protein A bead slurry with 20 mM DMP) by tumbling end-over-end for 4 h at 4 °C. Each rabbit IgG-protein A precleared complex was isolated by magnetic capture and washed six times with chilled RIPA buffer and then twice with chilled TBS (containing protease and phosphatase inhibitors). Precleared products were eluted in four washes of 100 mM citrate, pH 2.8, pooled, and neutralized with 1 M Tris-HCl, pH 9.5. For EGFR immunoprecipitation, the precleared A 431 lysates from above were incubated with 75 µg of anti-EGFR (cross-linked to 750 µl of paramagnetic protein A bead slurry with 20 mM DMP) overnight at 4 °C with end-over-end tumbling. The anti-EGFR-protein A immunoprecipitated complex was isolated, washed, and eluted as above. Eluates were aliquoted and prepared for Western blotting or enzymatic digestion and LC-MS analysis.

Western Blotting—
Protein samples were mixed with 4x lithium dodecyl sulfate buffer and ß-mercaptoethanol (2% final concentration) and then heated at 70 °C for 10 min. Cell lysates and immunoprecipitation products were separated by SDS-PAGE (NuPAGE Novex Bis-Tris gels) and transferred onto 0.45-µm Invitrolon PDVF membranes. After overnight incubation at 4 °C in blocking buffer (7.5% glycine, 5% nonfat dry milk, 0.1% Tween 20, 1 mM activated sodium orthovanadate, 50 mM sodium fluoride), the membranes were incubated with primary antibody diluted in wash buffer (TBS, 0.1% Tween 20, 0.1% BSA, 1 mM activated sodium orthovanadate, 50 mM sodium fluoride) containing 1% BSA for 2 h at room temperature. Membranes were then washed five times (3 min each) and incubated with secondary antibody (horseradish peroxidase-conjugated goat anti-mouse, goat anti-rabbit, or rabbit anti-goat) for 1 h at room temperature followed by five washes (3 min each). Blots were developed with ECL Plus reagents (GE Healthcare), and the resulting fluorescent signals were detected using an Eastman Kodak Co. 4000MM unit equipped with 400 nm excitation and 535 nm emission filters. Blot data were quantified using Kodak molecular imaging software. For publication, image brightness and contrast were adjusted with Adobe Photoshop software, and control bands/lanes within each blot were repositioned for continuity.

Enzymatic Digestion—
After immunoprecipitation, proteins (mainly EGFR) eluted from the cross-linked antibody/protein A beads were buffer-exchanged with 6 M guanidine hydrochloride using a Microcon spin column (10-kDa molecular mass cutoff; Millipore, Bedford, MA). The exchanged protein solution (200 µl) was reduced with 20 mM DTT for 30 min at 37 °C and alkylated with 50 mM iodoacetamide in the dark at room temperature for 1.5 h. After buffer exchange (desalting) over a Microcon spin column (10-kDa molecular mass cutoff), endoproteinase Lys-C (1:100, w/w) was added to digest the protein for 4 h at 37 °C. Digestion was stopped by addition of 1% formic acid, and then the solution was concentrated down to ~60 µl. Approximately 300 nl of the digest was directly analyzed by LC-MS, and the rest was stored for repeat use. For the analysis of the 10-kDa peptide, acetonitrile was added to the digest solution to make a 5% acetonitrile solution prior to LC injection.

LC-MS—
LC-MS experiments were performed on an LTQ-FTMS instrument (Thermo Electron, San Jose, CA) with an Ultimate nano-LC pump (Dionex, Mountain View, CA) using a monolithic column (polystyrene divinylbenzene, 50-µm inner diameter x 10 cm), which was prepared in-house.2 Mobile phase A was 0.1% formic acid in water, and mobile phase B was 0.1% formic acid in acetonitrile. A shallow gradient was used for all analyses: (i) 40 min at 0% B for sample loading, (ii) linear gradient to 35% B over 30 min, then (iii) to 80% B over 10 min, and finally (iv) constant 80% B for 10 min. The flow rate of the column (at the initial mobile phase condition) was measured as ~100 nl/min.

The parameters of the LTQ-FT mass spectrometer and the acquisition modes for MS2 and MS3 were similar to those in our previous study (22). Briefly the mass spectrometer was operated in the data-dependent mode to switch automatically between MS, MS2, and MS3 acquisition. Survey full scan MS spectra with two microscans (m/z 400–2000) were acquired in the FTICR cell with mass resolution of 100,000 at m/z 400 (after accumulation to a target value of 2 x 106 ions in the linear ion trap) followed by four pairs of sequential MS2 and MS3 scans. Dynamic exclusion was utilized with an exclusion duration of 30 s and no repeat counts. The total cycle time (one FTICR survey scan with two microscans plus four pairs of sequential linear ion trap MS2 and MS3 scans) was ~2.7 s.

Peptide Assignment—
The assignments of peptides (for charge state ≤3+), large peptides (for charge state ≥4+), phosphopeptides, and glycopeptides were similar to our previous report (22). Briefly the Sequest algorithm in the BioWorks software (Version 3.2, Thermo Electron Corp.) was used to search all MS2 and MS3 spectra against spectra of theoretical fragmentations of a human Swiss-Prot annotated database downloaded in January 2006 containing 14,094 protein entries with a mass tolerance of ±1.4 Da. Peptide ions (≤3+ ions) were assigned by BioWorks 3.2 software, which identified the peptides with a double filter: 1) peptide probability greater than 95% confidence and 2) peptide Xcorr scores above the following thresholds: ≥2.5 for 3+ and higher charge state ions, ≥2.0 for 2+ ions, and ≥ 1.5 for 1+ ions, with semi-Lys-C specificity and up to three internal missed cleavages. The individual identifications were further confirmed by the high mass accuracy (<5 ppm). For larger ions (≥4+ charge), no rigorous statistics are available correlating the probability of correct identification to the Xcorr scores, therefore Sequest was used to select and rank the most probable peptides, and the top assignment was further confirmed by (i) the mass accuracy (<5 ppm) and (ii) preferred fragmentation patterns in the observed MS2 and MS3 spectra (22). For EGFR phosphopeptides, the data were searched against a single database corresponding to the sequence of EGFR with the parameter of differential STY equal to +80 Da. The locations of the phosphorylated sites in the identified phosphopeptides were further confirmed by manual inspection of related b and y ions. For EGFR glycopeptides, the likely glycan structure in a glycopeptide was initially assigned by applying the mass obtained from the difference of a glycopeptide and its deglycosylated counterpart to match against the masses of the glycans in the Glycosuite database (Proteome Systems, Sydney, Australia). Among these likely glycostructure candidates, the best assignment was then selected from the preferred fragmentation patterns obtained in the related MS2 and MS3 spectra.

Quantitation—
The charge state with the highest intensity of a given peptide was used for quantitation. In this charge state, the chromatographic peak area derived from the three highest isotopic ions extracted from the precursor ion in the FTICR scans was calculated by an in-house algorithm to automate the selection of chromatographic peaks for quantitation. In addition to the use of the MS scan of the FTICR for quantitation, the mass spectrometer was also operated in the selected ion monitoring (SIM) mode using the linear ion trap for increased sensitivity when necessary. In this mode, a survey full scan MS spectrum was acquired in the FTICR cell followed by several sequential SIM scans in the linear ion trap. Each SIM scan was the average of at least 10 microscans with the selected mass window of ±2.5 m/z. The SIM scan was operated by segments (or retention time windows). For example, six SIM scans were used in the time window between 20 and 25 min for six specific peptides eluting in that time window. The peak areas of the ions in the SIM scans were used for quantitation, and the accurate masses of the ions obtained in the FTICR full scans were used for confirmation of the specific ions and for alignment of the chromatographic peaks from run to run. In addition, selected reaction monitoring (SRM) or multiple reaction monitoring, which measured the specific fragment ions in the MS/MS mode, was used for identification and quantitation of a specific ion, particularly if the given peptide (e.g. a 10-kDa phosphopeptide) had multiple phosphorylation sites with the same mass (e.g. phosphorylated at either Tyr992 or Ser1002). In this mode, the mass spectrometer was set to acquire the FTICR MS scan followed by a targeted MS/MS scan on the particular m/z ion (average of two microscans) in the linear ion trap within a given retention window. The specific ions in the MS/MS spectra were used to differentiate the phosphorylation sites (e.g. b47 ion for Tyr(P)992 site and y43 ion for Ser(P)1002 site). Similarly a targeted MS3 scan was applied to further confirm the site of phosphorylation. The peak areas of these specific ions from the MS/MS spectra (SRM) were extracted for quantitation (i.e. phosphopeptides) or normalization (i.e. non-phosphopeptides). All chromatographic peak areas derived from the average of all isotopic peaks, obtained either from the SIM or SRM scans in the linear ion trap, were calculated manually using the "add peaks" tool in Xcalibur Qual Browsers Version 1.4 SR1 program.


   RESULTS
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
DISCUSSION
 REFERENCES
 
In this study, an A 431 cell line that overexpresses EGFR was stimulated with EGF (25 ng/ml) for 0 (control), 30-s, 2-min, 10-min, and 4-h intervals. The work flow for analysis is summarized in Fig. 1. As detailed under "Experimental Procedures," ~108 cells at each time point were lysed, immunoextracted, and analyzed by the ERPA approach. Separately immunoextracted EGFR was also analyzed by SDS-PAGE with anti-EGFR blotting or SYPRO staining (for total protein determination and normalization) as shown in Fig. 2. The data of Fig. 2 support the absence of IgG interference and demonstrate the success of the immunoextraction procedure using antibodies cross-linked to beads. Furthermore the specificity of the anti-EGFR can be seen in the single band resulting from the blotting of either crude cell lysate or immunoextracted EGFR samples. Enrichment can be observed in the immunoextracted samples (lanes C–G, darker bands) as compared with the lysate sample (lane B, lighter band).


Figure 1
View larger version (17K):
[in this window]
[in a new window]
  		FIG. 1. The work flow for EGFR analysis by ERPA. A, the A 431 cells are first lysed. B, EGFR is then immunoextracted from the cells. C, extracted EGFR (including the associated proteins) is enzymatically digested with Lys-C. D, Lys-C-digested fragments are separated using capillary LC (monolithic polystyrene divinylbenzene column, 50-µm inner diameter x 10 cm) on-line coupled to an LTQ-FT mass spectrometer for analysis (using ERPA for identification and quantitation). IT, ion trap.

 

Figure 2
View larger version (68K):
[in this window]
[in a new window]
  		FIG. 2. Comparison of the immunoextraction of EGFR at different time points of EGF stimulation. Aliquots of immunoextracted EGFR from the A 431 cell line, stimulated with EGF for 0, 30 s, 2 min, 10 min, and 4 h, were separated by SDS-PAGE and visualized using anti-EFGR blotting (lanes C–G). The lysate sample prior to the immunoextraction (time 0) was also analyzed using SYPRO staining (lane A) or anti-EFGR blotting (lane B). Approximately 40% of the immunoextracted EGFR was used for gel analysis, and the remainder of the sample was further digested by Lys-C for subsequent LC-MS (ERPA) analysis as detailed in Fig. 1. IP, immunoprecipitate.

 
After immunoprecipitation, the released sample was digested with Lys-C to obtain peptide fragments up to 10.3 kDa for analysis. In our previous work (22), we used a 75-µm-inner diameter column with a 300-Å pore C4 bonded phase packing (3-µm particle diameter). We changed in this study to a 50-µm inner diameter polystyrene divinylbenzene monolithic column. The combination of the higher efficiency and narrower bore resulted in a 3–4-fold decrease (relative to the previous work) in the amount of material that was injected in each run, i.e. from 200 fmol of EGFR down to 50–75 fmol per injection. A second important feature of the monolithic column was the open pore structure, allowing high efficiency separation of long peptide fragments up to at least 10 kDa without detectable, on-column peptide loss. These and other features of nanoflow monolithic columns are discussed in detail in a separate study.2

We determined 96% sequence coverage for EGFR in the immunoprecipitated fraction, a level similar to our previous work with a commercial source of the protein (22). For the present study, all results were obtained with three injections or a total of ~200 fmol for each time point of EGF stimulation. With 200 fmol, we were also able to achieve extensive post-translational analysis for phosphorylation and glycosylation. The complete list of phosphorylated and glycosylated peptides are shown in Tables I and II, respectively. Representative mass spectra are provided in the supplemental material (Supplemental Figs. A, B, and C), and further details on spectral interpretation can be found elsewhere (22). Based on these results, we next sought to analyze dynamic EGFR-mediated events and characterize relative changes in post-translational modification as a function of time after stimulation with physiologic levels of EGF. Our label-free strategy for quantitation and the subsequent results are detailed below.


View this table:
[in this window]
[in a new window]
  		TABLE I List of phosphopeptides of EGFR identified upon stimulation of an A 431 cell line with EGF

 

View this table:
[in this window]
[in a new window]
  		TABLE II List of glycopeptides of EGFR identified upon stimulation of an A 431 cell line with EGF

 
Quantitation of EGFR Phosphorylation—
Table I lists a total of 13 phosphorylated sites, seven of which are tyrosine phosphorylated, from the five time points of EGF stimulation with each injection of 50–70 fmol of EGFR. All sites have been reported in the literature; however, Ser(P)1057 and Tyr(P)1114 have not been observed previously in the A 431 cell line. The predominant form in this overexpressed EGFR cell line is single phosphorylation. Double phosphorylation is seen for several of the peptides, but the relative amount (see later) was small. As can be further seen from the peptide sequences in Table I, each of the Lys-C-digested peptides, except 1137–1155, has additional arginine residues (underlined) that yield increased response factors and retention times relative to the shorter tryptic fragments. Without enrichment of phosphopeptides, 13 sites is the largest number revealed in EGFR in any reported experiment. Of course, enrichment could increase the number of low level phosphorylation sites, but label-free quantitation would likely not be accurate.

Fig. 3A presents the extracted ion chromatogram (XIC) of Lys-C fragment 1038–1075 in which four different phosphorylation sites were separated. The high resolution of the monolithic column is important because all four monophosphorylated peptides are isobaric. It can be seen that the corresponding non-phosphorylated peptide eluted close to the phosphorylated forms. Fig. 3B shows similar behavior for peptide fragment 1076–1136 where again the non-phosphorylated peptide eluted in a region of the gradient similar to that of the two phosphorylated forms.


Figure 3
View larger version (30K):
[in this window]
[in a new window]
  		FIG. 3. XICs of phosphorylated isoforms of Lys-C fragments of EGFR. A, base peak XIC of fragment 1038–1075. I, separation of peptide isoforms with phosphorylation at sites Tyr(P)1045, Ser(P)1046, Ser(P)1057, and Tyr(P)1068. II, separation of the non-phosphorylated peptide counterpart. The high resolution MS insets in I and II indicate the use of the three highest isotopes of the most intense charge state (3+) for quantitation. B, base peak XIC of fragment 1076–1136. I, separation of peptide isoforms with phosphorylation at sites Tyr(P)1114 and Tyr(P)1086. II, separation of the non-phosphorylated peptide counterpart. The high resolution MS insets in I and II indicate the use of the three highest isotopes of the most intense charge state (6+) for quantitation.

 
Because the various forms of a given peptide fragment can be seen in one run, we used as a quantitative measure the estimated stoichiometry (ES) at a given phosphorylation site as

Formula 1(Eq. 1)

where SA = chromatographic peak area of peptide A, pSA1 = the peak area of phosphorylated peptide A at site 1, pSA2 = the peak area of phosphorylated peptide A at site 2, etc. The peak area of the XIC, as described under "Experimental Procedures," was determined from the sum of the peak areas of the three most intense isotopes of the most intense charge state of the peptide (see Fig. 3, A and B, insets). Importantly because the phospho- and non-phosphocounterparts eluted close to one another, quantitative variations in ESI due to solvent changes were minimized. In addition, the sum of the chromatographic peak areas of the various forms of the given peptide minimizes variation in quantitation between experiments. Also it should be noted that we examined the use of other peptide fragments of EGFR in the same LC-MS runs as internal standards (36, 37). The trends in change of phosphorylation with time of stimulation were similar using %ES or other peptide fragments of EGFR as internal standards; however, the variation in quantitation from run to run was found to be greater with the latter approach.

It should be noted that the stoichiometry as measured by Equation 1 is estimated because the response factors of Ser(P) and Ser are not identical (13). However, a recent study notes that the response factor differences between the single phosphorylated and non-phosphorylated peptides are generally not great (38). It can be further noted that the large Lys-C-digested peptides (2.5–10 kDa), along with the additional arginines present, should minimize the differences in response factors between the phospho- and non-phosphomoieties further.

We searched for multiply phosphorylated peptides using, as described under "Experimental Procedures," SIM in the ion trap to enhance detection sensitivity. Only three examples of doubly phosphorylated peptides were found, all with relatively low signal even at the maximum level (see Table I). The assumption is made that the response factors of the doubly phosphorylated peptides do not differ in a major way from the corresponding monophosphorylated peptides given the length of the peptides and the acid mobile phase buffer conditions. Therefore, as a result of the relatively low levels observed for the doubly phosphorylated forms and the expected relatively small differences in response factors between the monophospho- and non-phosphopeptides, we concluded that the ES value should be a reasonable estimate of stoichiometry.

Using Equation 1, the phosphorylation changes at all 13 sites on EGFR were quantitated as a function of the time of EGF stimulation. As an example, Fig. 4 shows the temporal changes that occur in phosphorylation for Tyr1068 as measured by both Western blotting (Fig. 4A) and mass spectrometry (Fig. 4B). The blotting revealed an increase in phosphorylation over time with the total amount of immunoprecipitated EGFR remaining approximately constant. The MS results showed the same trend while providing more details on the quantitation. Phosphorylation at this tyrosine site was seen to rise rapidly within 30 s of EGF stimulation, remained roughly constant to 10 min, and finally increased during the extended time to 4 h. The MS results also exhibited variation in %ES for three runs at each time point, demonstrating the reproducibility of the approach.


Figure 4
View larger version (22K):
[in this window]
[in a new window]
  		FIG. 4. Comparison of the temporal changes at the Tyr(P)1068 phosphorylation site of EGFR by Western blotting (A) and mass spectrometry (B). Aliquots of the immunoextracted EGFR at different time points of EGF stimulation were blotted with the Tyr(P)1068 antibody of EGFR, and aliquots of the same immunoextracted EGFR were quantitated by LC-MS analysis (ERPA). m, minutes.

 
Fig. 5 presents the temporal changes for the additional 12 phosphorylation sites that were quantitated in a manner similar to that for Tyr(P)1068. As expected, distinctly different phosphorylation behavior was observed for each site with several continuing to rise in %ES with time of EGF stimulation, e.g. Tyr(P)1148, whereas others reached a maximum %ES at 2 min of stimulation, e.g. Tyr(P)1086, Ser(P)1002, and Tyr(P)1173, or a maximum %ES at 10 min of stimulation, e.g. Tyr(P)1045, Ser(P)1046, and Tyr(P)1114. Other examples demonstrated a continued decrease in phosphorylation with time of stimulation, e.g. Ser(P)967 and Tyr(P)992, or little relative change in phosphorylation with time, e.g. Ser(P)1057 and Ser(P)1142. In addition, several of the phosphotyrosine sites were analyzed with available antibodies by Western blotting, and again, as seen in Fig. 5, the results were in good agreement with the MS trends. It should be noted that two monophosphorylated forms of the same Lys-C peptide were not well separated by LC (Tyr(P)992 and Ser(P)1002 in the Lys-C peptide with the molecular mass of 10.3 kDa). SRM, which measures the specific ions in the MS/MS spectra to differentiate the phosphorylation sites, was used in this case for quantitation (see the details under "Experimental Procedures"). However, we used XIC in all other cases because XIC was found to lead to greater accuracy and less variation than SRM.


Figure 5
Figure 5
View larger version (43K):
[in this window]
[in a new window]
  		FIG. 5. The temporal changes of phosphorylation of 12 additional sites on EGFR as a function of time of stimulation by EGF (A–L). Quantitative analysis by LC-MS (ERPA) is shown. Results of Western blotting using specific phosphotyrosine antibodies are included with Tyr(P)992 (C), Tyr(P)1148 (I), and Tyr(P)1173 (K). IP, immunoprecipitate; ", seconds; ', minutes.

 
Because of the available antibodies, tyrosine phosphorylation has been well studied and understood as the starting point of various signaling cascades (39). With our approach, we also measured the temporal profiles of Thr(P) and Ser(P) sites of EGFR. Such profiles have not been readily determined in the past because of the unavailability of antibodies against specific Thr(P) or Ser(P) sites. It should also be noted that the ES varied significantly at the maximum time point of stimulation from 2.5% (Tyr(P)1045) to 40% (Thr(P)669). These large variations may reflect distinct differences in the role of signaling of individual sites. The methodology introduced in this work can thus serve as an aid in the more complete elucidation of how EGFR responds to different stimulants, drugs, or indeed the role of mutation of the receptor itself (7–9).

Quantitation of EGFR Glycosylation—
As reported in our previous study (22), EGFR has 12 potential N-linked glycosylation sites; 10 of these sites are glycosylated with either high mannose or complex-type structures. As expected, these glycosylation sites are located on the outer membrane surface where ligand binding can occur. Table II lists all the peptide fragments and the glycosylation patterns observed. Note again that in many cases there are additional basic amino acid residues in the Lys-C fragments, leading to an order of magnitude or greater enhanced ionization intensity than the related tryptic fragments. We mapped six sites that were fully glycosylated (Asn151, Asn328, Asn337, Asn420, Asn504, and Asn599) and four sites that were partially glycosylated (Asn32, Asn389, Asn544, and Asn579). All of these EFGR sites were previously known. The soluble variant of EGFR with one additional glycosylation site was not identified because the anti-EGFR used here was against only the C-terminal end of EGFR (10).

As a result of the high sequence coverage obtained by ERPA, the glycostructures of EGFR and the relative amounts of specific glycoforms at individual sites could be determined. The likely glycan structure was initially assigned by comparing the mass obtained from the difference of the glycopeptide and its deglycosylated counterpart against the masses of the glycans in the Glycosuite database (22). Among these likely glycostructure candidates, the most probable assignment was then selected from the preferred fragmentation patterns obtained in the MS2 and MS3 spectra. As noted, the major forms of the glycans (such as high mannose, sialylated complex type, fucosylation, and branching) attached to specific sites were determined but not the specific linkage (such as {alpha}1–6 or ß1–4 linkage) or isoform (such as galactose or mannose). The latter, however, are usually well conserved in human cells and can be extrapolated from the mass assignment in a human glycostructure database (40).

The glycoforms attached to the same peptide generally eluted close to one another in the chromatogram, and the relative percentage at a given site could then be readily estimated. As with phosphorylation, the measurement of the relative percentage can minimize the bias against the immunoextraction and LC-MS loading procedures. For simplification, the response factors of different glycopeptide isoforms at a given site are assumed to be similar because the major contribution of the response factors will most likely be from the peptide moieties, particularly as the peptide size is increased. In addition, for partially glycosylated sites, because of potential differences in ionization efficiencies between glycosylated and non-glycosylated peptides, the relative level of individual glycan forms at a given site would represent an estimate of the actual percentage. The deglycosylated peptide could in principle be quantitated, leading to an estimate of the relative amount of the partially glycosylated forms. As with phosphorylation, we thus report the relative amounts of the different glycoforms at a given site as percent estimated stoichiometry where the amount of a specific glycopeptide is divided by all forms of the given peptide.

All quantitation followed the procedure used for phosphorylation. Specifically the XIC peak area for a given form was divided by the XIC peak areas of all forms of the proteolytic peptide fragment. All Lys-C peptide fragments that were fully or partially glycosylated were found with only one modified site (see Table II).

An example for the calculation of a fully glycosylated high mannose site, namely Asn337, is presented in Fig. 6. The base ion chromatogram is shown in Fig. 6A where it can be seen that the glycopeptide eluted late in the chromatogram. Elution under high organic content, as well as an additional basic amino acid (see Table II), led to a significant increase in signal relative to the corresponding tryptic peptide fragment. Fig. 6B displays the highest charge state for each glycosylation form at the Asn337 site, and Fig. 6C presents an expanded scale of the 5+ state for Man8 from which quantitation was made. The percentage of Man7, Man8, and Man9 as a function of stimulation time is shown in Fig. 6D. There was no change in the relative amounts of each form at the different time points. Our previous work involved a commercial source of EGFR that was purified from A 431 cells cultured under unknown (to us) growth conditions. Although the specific sites that were fully or partially glycosylated agreed with what was found in the present study, significant differences in the percentage of the forms were observed. For example, at Asn337, the percentage of Man9 was 5-fold higher, and Man6 was 5-fold lower in the present sample relative to the commercial sample. Thus, measurement of the relative amounts of the glycoforms at specific sites is important for comprehensive characterization.


Figure 6
View larger version (34K):
[in this window]
[in a new window]
  		FIG. 6. Temporal LC-MS analysis of the high mannose glycoforms of the Lys-C fragment 337–372 of EGFR. A, base peak chromatogram of the Lys-C digest of immunoprecipitated mixture at t = 0 prior to stimulation by EGF. The specific glycopeptide fragment 337–372, with different numbers of mannoses at the Asn337 site eluted between 34 and 35 min, are indicated. B, FTICR mass spectrum of the specific glycopeptide fragment 337–372 with different numbers of mannoses at the Asn337 site. Only the most intense charge state of the glycoforms (5+) is shown. The structure of each glycoform was deduced by ERPA (see supplemental material). In the glycan structures, the solid circle (•) represents mannose, and the solid square ({blacksquare}) represents N-acetylglucosamine. C, expanded scale spectrum of the 5+ charge state of Man8 glycoform. The three most intense isotopes were selected for XIC for quantitation. D, the %ES of the three major glycoforms (Man8, Man7, and Man9) as a function of EGF stimulation is shown.

 
A second example of relative glycosylation of glycopeptides is shown in Fig. 7 for site Asn420, which was again found to be fully glycosylated. The structure of this site is a complex type with biantennary branching with multiple fucosylation and sialic acids. Fig. 7A presents the base ion chromatogram, and Fig. 7B details the highest intensity of individual forms summed over the retention time interval of 33.6–34.5 min. Each of the complex structures are determined from the MS2 and MS3 spectra of individual precursor ions (illustrated in Supplemental Figs. B and C). Fig. 7C shows an expanded scale of the 3+ charge of glycoform W. Finally the percentage of forms W, X, Y, and Z at Asn420 as a function of time of EGF stimulation is presented in Fig. 7D. As with Asn337, essentially no change from 0 to 4 h was found for any of the complex-type forms. The major glycoforms at other sites with their relative percentages are detailed in the supplementary material (Supplemental Table A).


Figure 7
View larger version (38K):
[in this window]
[in a new window]
  		FIG. 7. Temporal LC-MS analysis of the complex-type glycoforms of the Lys-C fragment 408–430 of EGFR. A, base peak chromatogram of the Lys-C digest of immunoprecipitated mixture at t = 0 prior to stimulation by EGF. The specific glycopeptide fragment 408–430, with different complex-type glycoforms at the Asn420 site eluted between 33.6 and 34.5 min, are indicated. B, FTICR mass spectrum of the specific glycopeptide fragment 408–430 with different complex-type glycoforms at the Asn420 site. Only the most intense charge state of the glycoforms (3+) are shown. The structure of each glycoform (W, X, Y, or Z) was deduced by ERPA (see supplemental material). In the glycan structures, the solid circle (•) represents mannose, the open circle ({circ}) represents galactose, the solid square ({blacksquare}) represents N-acetylglucosamine, the solid triangle ({blacktriangleup}) represents fucose, and the solid diamond ({diamondsuit}) represents sialic acid. C, expanded scale spectrum of the 3+ charge state of the complex-type glycoform W. The three most intense isotopes were selected for XIC for quantitation. D, the %ES of the four major glycoforms (W, X, Y, and Z) as a function of EGF stimulation is shown.

 
Interaction Partners of EGFR—
EGF binding to the extracellular domain of EGFR activates the intrinsic kinase domain resulting in phosphorylation on specific residues within the intracellular tail and the start of specific signaling pathways. The anti-EGFR antibody used in this study does not interfere with the EGF binding domain or its interior EGFR phosphorylation sites. Thus, in this experiment, immunoprecipitation can become the starting point for the discovery of binding partners docking to exposed phosphorylation sites. Importantly the relative change (after normalization) of the binding partners as a function of time of EGF stimulation can be measured. Those proteins that changed significantly over the time of stimulation represent likely binding partners of EGFR.

A double filter was used to minimize identification of proteins that were nonspecifically pulled down in the precipitation. The first filter was the standard negative control experiment consisting of immunoextraction of the A 431 cell line by a non-EGFR antibody (see "Experimental Procedures"). Proteins appearing in both the negative control and EGFR pulldown were eliminated. Then the additional filter was applied such that only proteins that changed by at least a factor of 3 over the time of stimulation were included in the list of likely binding partners. A 3-fold change was selected to minimize variations due to washing of the precipitate from run to run and also to exclude non-changing binding partners during EGF stimulation. Normalization over the time points of the experiment was achieved by dividing the XIC peak area of a specific peptide of the interaction partner by the average XIC peak area of three representative Lys-C peptides of EGFR that were not post-translationally modified. This normalization approach has the further advantage of correcting for differences due to LC sample loading or to immunoprecipitation procedures.

Fig. 8 illustrates the approach for the known EGFR interaction partner GRB2 (41). Fig. 8A, Western blotting, shows a general increase, relative to EGFR, in the amount of GRB2 pulled down in the immunoprecipitation as a function of stimulation time. Fig. 8B indicates a similar trend from the ERPA LC-MS analysis in which the relative change in GRB2 level from 0 to 4 h was over 6-fold. It is interesting to note that the temporal changes for GBR2 follow those of phosphorylation of Tyr(P)1068 (Fig. 4), the known position of association (42).


Figure 8
View larger version (24K):
[in this window]
[in a new window]
  		FIG. 8. Comparison of the temporal changes of GRB2, a known EGFR interaction partner, by Western blotting (A) and mass spectrometry (B). As in Fig. 4, aliquots of the immunoextracted EGFR at different time points of EGF stimulation were blotted with the antibody specific for GRB2, and aliquots of the same immunoextracted EGFR were analyzed by LC-MS analysis (ERPA).

 
Tables III and IV present a list of 19 likely docking proteins and their normalized relative changes as a function of EGF stimulation based on the filter criteria described above. Tables III and IV include a number of proteins known to bind to EGFR, e.g. CAV1, GRB2, GAPDH, and stratifin (43–45). Tables III and IV are divided into those proteins that increase in relative amount due to stimulation (Table III) and those proteins that decrease (Table IV) relative to the zero time point. Separately Western blotting was conducted on several of these proteins with available antibodies, and as seen in Fig. 9, the trends in protein levels with EGF stimulation qualitatively agreed with the results in Tables III and IV. As shown in Tables III and IV, some proteins displayed a general overall increase in amount of binding with time, e.g. calmodulin and caveolin-1, whereas others have maxima at 30 s or 2 min during stimulation, e.g. calgranulin B and clathrin light chain A. On the other hand, for other proteins, the amount pulled down appeared to decrease gradually, e.g. stratifin, or immediately upon stimulation, e.g. CD98 antigen and CD166 antigen precursor. The down-regulation of individual proteins could delay or reduce specific signaling pathways that control cell function just as up-regulation can activate pathways.


View this table:
[in this window]
[in a new window]
  		TABLE III Identification of EGFR interaction partners by mass spectrometry: proteins that increase upon EGF stimulation

Only proteins with more than a 3-fold change over the time of EGF stimulation are listed. Note that the listed proteins were identified with more than 10% sequence coverage containing either at least two unique peptides or one large Lys-C peptide at every time point.

 

View this table:
[in this window]
[in a new window]
  		TABLE IV Identification of EGFR interaction partners by mass spectrometry: proteins that decrease upon EGF stimulation

Only proteins with more than a 3-fold change over the time of EGF stimulation are listed. Note that the listed proteins were identified with more than 10% sequence coverage containing either at least two unique peptides or one large Lys-C peptide at every time point.

 

Figure 9
View larger version (43K):
[in this window]
[in a new window]
  		FIG. 9. Comparison of the temporal changes of the EGFR interaction partners by Western blotting. As in Fig. 8, aliquots of the immunoextracted EGFR at different time points of EGF stimulation were blotted with specific antibodies of potential EGFR interaction partners (listed in the figure, lanes B–F). In addition, prior to the immunoextraction, the A 431 lysate sample, without EGF stimulation, was also blotted with the antibodies of potential EGFR interaction partners (lane A). Lamin A/C, a known protein that does not bind to EGFR, served as a negative control. The protein was detected in the A 431 lysate sample (lane A) but not in the aliquots of the immunoextracted EGFR samples (lanes B–F). IP, immunoprecipitate; LCA, light chain A.

 

   DISCUSSION
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
REFERENCES
 
The functional machinery of protein signal transduction is embodied in the post-translational modifications and binding partner complexes that assemble and disassemble over time following a stimulus. ERPA is an effective approach for the comprehensive characterization of fluctuating binding partners and post-translational modifications over time. An important aspect of ERPA is its quantitative nature in which the high sequence coverage of the approach allows either a non-post-translationally modified peptide counterpart or other peptides of EGFR to be used as internal standards without the need for labeling. As shown under "Results," we validated the quantitative assessments of protein phosphorylation and protein binding partner stoichiometry by comparing the results with the semiquantitative Western blotting approach.

As a demonstration of the depth of coverage, we report procedures for label-free quantitation of PTMs, both phosphorylation and glycosylation, following stimulation of the EGFR by EGF. The estimated phosphorylation in its cytoplasmic tail or glycosylation stoichiometries in its extracellular domain fluctuating over time represent a unique picture of EGFR upon EGF stimulation. In addition, the quantitative changes that occur over time for PTMs can potentially be matched to the quantitative fluctuation of binding partners, thereby providing clues as to which binding partners relate to specific PTMs. Finally the estimated stoichiometries reveal significant differences in the relative levels of phosphorylation, and these differences likely also play a role in the signaling processes.

Although past studies of tyrosine kinase receptors, including EGFR, have evaluated sites of phosphorylation following stimulation, these studies have typically involved some form of enrichment such as IMAC or phosphotyrosine antibody pulldown. Enrichment was required because phosphorylation at individual sites is low. Although effective, especially in global searches for phosphorylation sites on proteins in cells (17), the enrichment approach may be less useful for quantitation of changes over time because enrichment can lead to bias (15). Moreover enrichment requires normalization to correct for (a) initial sample amount differences, (b) differential losses during sample preparation, and (c) differences in sample injection. Temporal studies of EGFR by proteomic approaches have previously involved either isotopic growth medium, e.g. SILAC (29), or chemical labeling, e.g. iTRAQ (isobaric tags for relative and absolute quantitation) (46), but without estimation of the stoichiometry because of the difficulty to obtain the non-phosphorylated or non-glycosylated counterparts in the same run. Our goal was to develop quantitative methods using mass spectrometry that did not include enrichment steps or the use of SILAC or a labeling procedure to provide a general approach that could ultimately be used in tissue.

The developed methodology will allow in-depth exploration of key protein molecules that may presently or ultimately be drug targets. The approach also opens up possibilities to explore such molecules in tissue or blood under various physiological states, e.g. disease versus normal. Such capabilities could prove highly effective in the discovery phase of a study. Although antibodies have been demonstrated to be a powerful approach to quantitation in later stages (47), specific site modifications must first be known. Moreover not all isoforms will produce highly specific antibodies; this is especially the case for the carbohydrate forms.

The results in Figs. 4 and 5 reveal quite distinct kinetic behavior of individual phosphorylation sites. Six tyrosine phosphorylation sites increased with the extent of EGF stimulation. In contrast, a seventh site, Tyr(P)992, decreased upon stimulation (shown in both MS and Western blotting results). A similar downward trend by Western blotting has also been observed by others (48). The temporal trends in Tyr(P)1068, Tyr(P)1086, and Tyr(P)1148 generally agree with those found using labeling approaches (17, 48, 49), leading to further confidence in our label-free results.

We demonstrated temporal changes of phosphorylation for specific serine and threonine sites of EGFR. Although there is a growing compendium of antibodies recognizing specific Ser/Thr site residues, phospho-Ser and -Thr antibodies have historically performed poorly when used for blotting or pulldown purposes. Although phosphorylation did not vary greatly with stimulation time for Ser(P)1057, there are examples where increases, or decreases, in phosphorylation were observed, e.g. Ser(P)1002, Ser(P)1046, and Ser(P)967. Although the biological significance of these specific phosphorylation is unclear, it has been established previously that serine hreonine phosphorylation of the EGFR can play a role in the attenuation of kinase activity (50).

The relationship of extracellular glycosylation to phosphorylation kinetics using the ERPA approach is an avenue that can now be explored and correlated with other functional information. Although, as expected, no temporal changes in glycosylation were observed in this study, variations in growth conditions or the use of different cell lines would be expected to yield a variety of glycosylation patterns that could affect phosphorylation and thus signaling. In addition, it is well known that glycosylation is altered in disease, e.g. cancer (51, 52), and this ability to characterize in detail the glycoforms will prove useful in future studies.

We additionally were able to determine binding partners and estimate the kinetics of binding of these proteins to EGFR with a tight criteria, i.e. only proteins that changed by at least a factor of 3 over the time of stimulation. We acknowledge that some binding partners may not reach that extent of change upon stimulation. To avoid the potential nonspecific interaction and also the measurement variations due to washing of the precipitate from run to run, only 19 of 49 possible interaction partners (data not shown) were included at this stage, and many of these 19 proteins had a trend similar to that found by Western blotting (Fig. 9). We showed that some proteins are up-regulated, whereas others are down-regulated. In some other cases, there appears to be release of specific proteins upon initial stimulation. It may well be that this release is either important for receptor homo- or heterodimerization to occur or that removal of protein exposes sites on the EGFR tail for phosphorylation to take place.

As noted earlier, one of our ultimate goals is to provide comprehensive characterization of such molecules in tissue. Key to this will be the capability to characterize and quantitate at low levels of sample. Although LC-MS is a sensitive method, we can lower detection levels at least 10-fold relative to the several hundred fmol in this study by using LC columns narrower than 50 µm (as discussed in a separate study)2 and by targeting individual precursors for SIM. With these and other advances, the ability for comprehensive characterization of EGFR in disease tissue will be available. Using reverse protein tissue arrays to determine pathways (25), the relationship of the structure of EGFR, e.g. glycosylation, to signaling pathways in disease will be possible to determine. Furthermore the high sequence coverage will allow analysis of specific mutant forms known to influence phosphorylation patterns (6).


   ACKNOWLEDGMENTS
 
We acknowledge Dr. Victor Andreev for development of an algorithm to automate the selection of chromatographic peaks.


  FOOTNOTES
 
Received, March 28, 2006, and in revised form, June 23, 2006.

Published, MCP Papers in Press, June 23, 2006, DOI 10.1074/mcp.M600105-MCP200

1 The abbreviations used are: EGFR, epidermal growth factor receptor; ERPA, Extended Range Proteomic Analysis; EGF, epidermal growth factor; PTM, post-translation modification; DMEM, Dulbecco’s modified Eagle’s medium; FBS, fetal bovine serum; DMP, dimethyl pimelimidate dihydrochloride; Bis-Tris, 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol; RIPA, radioimmunoprecipitation assay; SIM, selected ion monitoring; SRM, selected reaction monitoring; XIC, extracted ion chromatogram; ES, estimated stoichiometry; Man, mannose; SILAC, stable isotope labeling by amino acids in cell culture; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; LTQ-FT, linear ion trap integrated with Fourier transform. Back

2 J. Zhang, J. K. Kim, S. L. Wu, and B. L. Karger, manuscript in preparation. Back

* This work was supported by National Institutes of Health Grant GM 15847. This research was also supported by the Intramural Research Program of the National Institutes of Health, NCI Center for Cancer Research. This paper is Contribution Number 880 from the Barnett Institute. 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. Back

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

§ Both authors contributed equally to this work. Back

** To whom correspondence should be addressed. E-mail: b.karger{at}neu.edu


   REFERENCES
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Blume-Jensen, P., and Hunter, T. (2001) Oncogenic kinase signalling. Nature 411, 355 –365[CrossRef][Medline]

  2. Yarden, Y., and Sliwkowski, M. X. (2001) Untangling the ErbB signalling network. Nat. Rev. Mol. Cell. Biol. 2, 127 –137[CrossRef][Medline]

  3. Pawson, T. (2004) Specificity in signal transduction: from phosphotyrosine-SH2 domain interactions to complex cellular systems. Cell 116, 191 –203[CrossRef][Medline]

  4. Pawson, T., and Nash, P. (2003) Assembly of cell regulatory systems through protein interaction domains. Science 300, 445 –452[Abstract/Free Full Text]

  5. Cohen, P. (2002) Protein kinases—the major drug targets of the twenty-first century? Nat. Rev. Drug Discov. 1, 309 –315[CrossRef][Medline]

  6. Noble, M. E., Endicott, J. A., and Johnson, L. N. (2004) Protein kinase inhibitors: insights into drug design from structure. Science 303, 1800 –1805[Abstract/Free Full Text]

  7. Pao, W., Miller, V., Zakowski, M., Doherty, J., Politi, K., Sarkaria, I., Singh, B., Heelan, R., Rusch, V., Fulton, L., Mardis, E., Kupfer, D., Wilson, R., Kris, M., and Varmus, H. (2004) EGF receptor gene mutations are common in lung cancers from "never smokers" and are associated with sensitivity of tumors to gefitinib and erlotinib. Proc. Natl. Acad. Sci. U. S. A. 101, 13306 –13311[Abstract/Free Full Text]

  8. Paez, J. G., Janne, P. A., Lee, J. C., Tracy, S., Greulich, H., Gabriel, S., Herman, P., Kaye, F. J., Lindeman, N., Boggon, T. J., Naoki, K., Sasaki, H., Fujii, Y., Eck, M. J., Sellers, W. R., Johnson, B. E., and Meyerson, M. (2004) EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science 304, 1497 –1500[Abstract/Free Full Text]

  9. Lynch, T. J., Bell, D. W., Sordella, R., Gurubhagavatula, S., Okimoto, R. A., Branniga