Identification of Phosphorylation Sites in Proteins Separated by

We report a fast, sensitive, and robust procedure for the identification of precise phosphorylation sites in proteins separated by polyacrylamide gel ...
12 downloads 0 Views 171KB Size
Download PDF
Anal. Chem. 1998, 70, 2050-2059

Identification of Phosphorylation Sites in Proteins Separated by Polyacrylamide Gel Electrophoresis Xiaolong Zhang,† Christopher J. Herring,† Patrick R. Romano,§ Joanna Szczepanowska,‡ Hanna Brzeska,‡ Alan G. Hinnebusch,§ and Jun Qin*,†

Laboratory of Biophysical Chemistry and Laboratory of Cell Biology, NHLBI, and Laboratory of Eukaryotic Gene Regulation, NICHD, National Institutes of Health, Bethesda, Maryland 20892

We report a fast, sensitive, and robust procedure for the identification of precise phosphorylation sites in proteins separated by polyacrylamide gel electrophoresis by a combination of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI/TOF) and online capillary liquid chromatography electrospray tandem ion trap mass spectrometry (LC/ESI/MS/MS). With this procedure, a single phosphorylation site was identified on as little as 20 ng (500 fmol) of the baculovirusexpressed catalytic domain of myosin I heavy-chain kinase separated by gel electrophoresis. The phosphoprotein is digested in the gel with trypsin, and the resulting peptides are extracted with >60% yield and analyzed by MALDI/ TOF before and after digestion with a phosphatase to identify the phosphopeptides. The phosphopeptides are then separated and fragmented in an on-line LC/ESI ion trap mass spectrometer to identify the precise phosphorylation sites. This procedure eliminates any off-line HPLC separation and minimizes sample handling. The use of MALDI/TOF and LCQ, two types of mass spectrometers that are widely available to the biological community, will make this procedure readily accessible to biologists. We applied this technique to identify two autophosphorylation sites and to assign at least another 12 phosphorylation sites to two tryptic peptides in a series of experiments using a gel slice containing only 200 ng (3 pmol) of human double-stranded RNA-activated protein kinase expressed in a mutant strain of the yeast Saccharomyces cerevisiae.

sary to determine the sites that are phosphorylated in vivo. In the conventional biochemical approach, the protein is labeled with [32P]phosphate, purified, and enzymatically cleaved to peptides that are separated by HPLC, and the phosphopeptides are then subjected to Edman sequencing to identify the phosphorylation sites by monitoring the release of the radioactivity during the Edman cycles.4-6 This approach is laborious and can be unreliable, primarily because of the instability of phosphoamino acids under the Edman degradation conditions.5 Alternatively, the phosphorylated peptides can be separated by two-dimensional thinlayer chromatography and identified by autoradiography, the suspected phosphorylation sites mutated to unphosphorylatable residues, and the experiment repeated to determine if the phosphorylated peptide disappears.7,8 Although this procedure is sensitive, it requires considerable time and sufficient prior knowledge of the system under study to make educated guesses of the most likely sites of phosphorylation. When multiple phosphorylation sites are, or might be, present, the complexity of the peptide map and the multiplicity of mutations that are required make this approach very difficult, if not impossible. As regulatory proteins are often expressed only at low levels, a sensitive and rapid method for determining the phosphorylation sites in partially purified proteins would be very useful. In principle, a combination of mass spectrometry and polyacrylamide gel electrophoresis would seem to be the method of choice. Indeed, methodologies based on mass spectrometry have been developed to identify phosphopeptides and precise phosphorylation sites.9-17 These methods are inherently sensitive, consuming

* To whom correspondence and reprint requests should be addressed. Laboratory of Biophysical Chemistry, Building 10, Room 7N307, NHLBI, National Institutes of Health, 9000 Rockville Pike, Bethesda, MD 20892. Tel.: (301) 4028704. Fax: (301) 402-3404. E-mail: [email protected]. † Laboratory of Biophysical Chemistry. ‡ Laboratory of Cell Biology. § Laboratory of Eukaryotic Gene Regulation. (1) Cohen, P. Trends. Biochem. Sci. 1992, 17, 408-413. (2) Hunter, T.; Karin, M. Cell 1992, 70, 375-387. (3) Posada, J.; Cooper, J. A. Mol. Biol. Cell 1992, 3, 583-592.

(4) Boyle, W. J.; van der Geer, P.; Hunter, T. Methods Enzymol. 1991, 201, 110-149. (5) Aebersold, R.; Watts, J. D.; Morrison, H. D.; Bures, E. J. Anal. Biochem. 1991, 199, 51-60. (6) Wettenhall, R. E.; Aebersold, R. H.; Hood, L. E. Methods Enzymol. 1991, 201, 186-199. (7) Van der Geer, P.; Hunter, T. Electrophoresis 1994, 15, 544-554. (8) Picciotto, M. R.; Cohn, J. A.; Bertuzzi, G.; Greengard, P.; Nairn, A. C. J. Biol. Chem. 1992, 267, 12742-12752. (9) Biemann, K.; Scoble, H. A. Science (Washington) 1987, 237, 992-998. (10) Gibson, B. W.; Cohen, P. Methods Enzymol. 1990, 193, 480-501. (11) Huddleston, M. J.; Annan, R. S.; Bean, M. F.; Carr, S. A. J. Am. Soc. Mass Spectrom. 1993, 4, 710-717. (12) Ding, J.; Burkhart, W.; Kassel, D. B. Rapid Commun. Mass Spectrom. 1994, 8, 94-98. (13) Hunter, A. P.; Games, D. E. Rapid Commun. Mass Spectrom, 1994, 8, 559570. (14) Annan, R. S.; Carr, S. A. Anal. Chem. 1996, 68, 3413-3421. (15) Carr, S. A.; Huddleston, M. J.; Annan, R. S. Anal. Biochem. 1996, 239, 180192.

2050 Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

S0003-2700(97)01207-9 CCC: $15.00

Protein phosphorylation, principally on Ser, Thr, and Tyr residues, is one of the most common cellular regulatory mechanisms.1-3 Phosphorylation can modulate enzyme activity, alter affinity for other proteins, and transmit signals through kinase cascades that often are branched and interactive. To understand the molecular basis of these regulatory mechanisms, it is neces-

© 1998 American Chemical Society Published on Web 04/11/1998

as little as a few femtomoles of a phosphopeptide, although the amount of starting materials that are digested is much higher due to the difficulty of sample handling.15 These methodologies work well if the proteins are highly purified and soluble so that they can be digested in solution to retrieve phosphopeptides in a form suitable for MS analysis. However, technical difficulties have prevented the widespread use of such methodologies for the identification of phosphorylation sites in SDS-PAGE-separated proteins, and they often require a few micrograms of phosphoproteins as the starting material (the amount of phosphoprotein in a gel slice) in practice.18-23 Significant advances have been made recently for high-sensitivity sequencing by electrospray mass spectrometry of small quantities of proteins separated by gel electrophoresis,24-26 where only a few peptides need to be identified and sequenced to identify or clone a gene.27,28 This approach is not useful, however, for identification of phosphorylation sites when it is necessary to recover and identify peptides covering most of the protein sequence, determine which are phosphopeptides, and sequence them to identify unambiguously the phosphorylation sites. We previously utilized matrix-assisted laser desorption/ionization ion trap mass spectrometry of proteolytic digests of about 200 ng (5 pmol) of SDS-PAGE purified protein29 to identify the phosphorylated peptide of the baculovirus-expressed catalytic domain of myosin I heavy-chain kinase (MIHCK)30 and on-line HPLC electrospray ion trap mass spectrometry of another 200 ng of the soluble protein digested in solution to identify the single phosphorylated serine.29 Based on this work, we subsequently improved the procedure substantially. Now we report a fast and sensitive procedure for the identification of phosphorylation sites of gel-separated proteins, using a combination of MALDI/TOF and LC/ESI ion trap mass spectrometry. In this improved procedure, phosphopeptides in the peptide mixture extracted from the in-gel digestion are identified by the observation of an 80 Da (or multiples of an 80-Da) shift in mass from the MALDI/TOF spectrum after treatment with a phosphatase. This approach was first reported by Liao et al., using a HPLC fraction containing the phosphopeptide with a linear MALDI/TOF mass spectrometer.16 (16) Liao, P.; Leykam, J.; Andrews, P C.; Gage, D. A.; Allison, J. Anal. Biochem. 1994, 219, 9-20. (17) Qin, J.; Chait, B. T. Anal. Chem. 1997, 69, 4002-4009. (18) Watts, J. D.; Affolter, M.; Krebs, D. L.; Wange, R. L.; Samelson, L. E.; Aebersold, R. J. Biol. Chem. 1994, 269, 29520-29529. (19) Erickson, A. K.; Payne, D. M.; Martino, P. A.; Rossomando, A. J.; Shabanowitz, J.; Weber, M. J.; Hunt, D. F.; Sturgill, T. W. J. Biol. Chem. 1990, 265, 19728-19735. (20) Resing, K. A.; Johnson, R. S.; Walsh, K. A. Biochemistry 1995, 34, 94779487. (21) Townsend, R. R.; Lipniunas, P. H.; Tulk, B. M.; Verkman, A. S. Protein Sci. 1996, 5, 1865-1873. (22) Betts, J. C.; Blackstock, W. P.; Ward, M. A.; Anderton, B. H. J. Biol. Chem. 1997, 272, 12922-12927. (23) Jonscher, K. R.; Yates, J. R., III J. Biol. Chem. 1997, 272, 1735-1741. (24) Wilm, M.; Shevchenko, A.; Houthaeve, T.; Breit, S.; Schweigerer, L.; Fotsis, T.; Mann, M. Nature 1996, 379, 466-469. (25) Wilm, M.; Mann, M. Anal. Chem. 1996, 68, 1-8. (26) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68, 850-858. (27) Mann, M.; Wilm, M. Anal. Chem. 1994, 66, 4390-4399. (28) Lingner, J.; Hughes, T. R.; Shevchenko, A.; Mann, M.; Lundblad, V.; Cech, T. R. Science 1997, 276, 561-567. (29) Szczepanowska, J.; Zhang, X.; Herring, C.; Qin, J.; Korn, E. D.; Brzeska, H. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 8503-8508. (30) Brzeska, H.; Szczepanowska, J.; Hoey, J.; Korn, E. D. J. Biol. Chem. 1996, 271, 27056-27062.

Recent advances in MALDI/TOF mass spectrometry, such as the application of delayed extraction, which improves mass resolution and mass accuracy significantly,31 allow us to measure the total protein digest without HPLC separation. We documented the utility of the new procedure using 20 ng (0.5 pmol) of the catalytic domain of MIHCK and then, in a series of experiments using 200 ng (3 pmol) of gel electrophoretically purified protein, used the new procedure to identify precisely two new autophosphorylation sites and to assign another 12 autophosphorylation sites to two tryptic peptides in human double-stranded RNA-activated protein kinase (PKR) expressed in a mutant strain of Saccharomyces cerevisiae.32 EXPERIMENTAL SECTION Proteins. The catalytic domain of MIHCK was expressed as a N-terminal poly(His) fusion protein and purified as described previously.30 The protein concentration was quantified with amino acid analysis, and known amounts of protein were separated on a 4-15% gradient SDS-PAGE gel and stained with Cu stain (BioRad, Hercules, CA). A plasmid containing PKR cDNA, with the additional coding sequence for the amino acids MRGSHHHHHH inserted at the 5′ end, was introduced into a mutant strain of yeast, and the expressed autophosphorylated PKR was purified as described.32 Trypsin was purchased from Boehringer Mannheim (Indianapolis, IN), and calf intestinal phosphatase (CIP) was from New England Biolabs (Beverly, MA). In-Gel Digestion of Gel-Separated Proteins. Our previously published procedure for digestion of proteins in a gel slice33 was modified as follows. After conventional SDS-PAGE, a Cu-stained gel piece containing 20 ng or more of protein was destained twice for 3 min each with the Bio-Rad destain solution and soaked for 4 h in 1:1 (v/v) methanol:10% acetic acid solution (pH 2) at room temperature and then in water for 30 min. The cleaned, wet gel was ground to a fine powder in 10 µL of 50 mM NH4HCO3, 200 ng of trypsin in 10 µL of 50 mM NH4HCO3 was added, and the digestion was carried out at 37 °C for 90 min; lower amounts of trypsin reduce the digestion efficiency significantly. To extract the peptides, 60 µL of acetonitrile was added, the tube vortexed for 2 min, and the solution carefully removed with a gel-loading pipet tip. The above step was repeated with an additional 10 µL of acetonitrile. The extracted peptides were pooled and dried, and the dried peptides were dissolved in 10 µL of 50% acetonitrile in water for mass spectrometry. This procedure recovered >60% of the radioactivity of a 35S-labeled protein (the catalytic domain of Zap70) as soluble, extracted peptides (data not shown). We did not find it difficult to recover Cys-containing peptides without reduction and alkylation from a systematic study of optimizing our in-gel digestion procedure with the 35S-labeled catalytic domain of Zap70, which contains six Cys residues. We also found that prolonged sample handling at the level of ∼1 pmol of protein loaded in the gel tends to introduce contamination, such as keratin. Therefore, the step of reduction and alkylation was omitted. Tryptic peptides extracted from the in-gel digestion are predomi(31) Whittal, R. M.; Li, L. Anal. Chem. 1995, 67, 1950-1954. (32) Romano, P. R.; Garcia-Barrio, M. T.; Zhang, X.; Wang, Q.; Taylor, D. R.; Zhang, F.; Herring, C.; Mathews, M. B.; Qin, J.; Hinnebusch, A. G. Mol. Cell Biol., in press. (33) Qin, J.; Fenyo, D.; Zhao, Y.; Hall, W. W.; Chao, D. M.; Wilson, C. J.; Young, R. A.; Chait, B. T. Anal. Chem. 1997, 69, 3995-4001.

Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

2051

Figure 1. Identification of the phosphopeptides in the in-gel tryptic digest of 20 ng (0.5 pmol) of the catalytic domain of MIHCK by MALDI/TOF spectrometry. (a) Before CIP treatment. Arrows indicate peptides that are modified. (b) After CIP treatment. Arrows indicate new peaks arising from dephosphorylation of the phosphopeptides. The inset shows detailed spectra of the region of phosphopeptides before (top) and after (bottom) CIP treatment. T, trypsin autolysis peaks; *, methionine-oxidized peptides of mass 16 Da higher.

nantly complete digestions and partial digestions with one miss (one internal R/K is not cut by trypsin). Partial digestions with up to four misses can be observed too, (data not shown). We believe that this in-gel digestion procedure is a generic procedure. Phosphatase Treatment of Peptides. One-tenth (1 µL) of the dissolved peptide solution was mixed with 2 µL of 50 mM NH4HCO3 and 1 µL of 0.5 unit/µL of CIP in 50 mM NH4HCO3 (diluted from 10 unit/µL in storage buffer as supplied by the manufacturer) and incubated for 30 min at 37 °C (the CIP digestion buffer supplied by the manufacturer was not used). The reaction mixture was dried on a SpeedVac concentrator and redissolved in 2 µL of 50% acetonitrile in water for mass spectrometry. MALDI/TOF Mass Spectrometry. A 1.2-m MALDI/TOF mass spectrometer with delayed extraction (Voyager-DE, Persep2052 Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

tive Biosystems, Framingham, MA) was used. Data were collected with an external 2-GHz digital oscilloscope (Tektronix, Houston, TX). The working matrix solution was a 2-fold dilution of a saturated solution of 2,5-dihydroxybenzoic acid in acetonitrile: water (1:1). Aliquots of 0.5 µL of the peptide mixture and 0.5 µL of the working matrix solution were mixed on the sample plate and dried in air prior to MS analysis. Two spectra were taken for each sample, one under conditions optimized for high mass resolution (to measure masses with high accuracy) and one under conditions optimized for sensitivity (to observe the weaker peaks). Masses were calibrated internally with an added calibration standard (bradykinin fragment 1-5) and a peptide derived from trypsin autolysis. Masses of 2903.7 were not measured. To our knowledge, this is the first time that two phosphorylation sites have been identified and at least another 12 phosphorylation sites have been assigned to two tryptic peptides using just 200 ng (3 pmol) of a protein separated by gel electrophoresis. (41) Qin, J.; Chait, B. T. J. Am. Chem. Soc. 1995, 117, 5411-5412.

Figure 5. MALDI/TOF mass spectra of in-gel trypsin digestion of 200 ng (3pmol) of human PKR, showing only the regions containing the phosphopeptides. In each panel, the upper trace was obtained before and the lower trace after the peptides were incubated with CIP. The spikes within each peak represent isotopic isoforms and illustrate the resolution of the method. In panels c and d, the underlined symbols designate phosphorylated derivatives of peptide 80-107, and the normal symbols designate phosphorylated derivatives of peptide 81-107. Similarly, in panel d, phosphorylated peptides derived from peptide 315-356+β are underlined, and those derived from peptide 315-352+β are not underlined. β represents β-mercaptoethenal modification of the Cys residue. The methionine-oxidized peptides are labeled with asterisks.

DISCUSSION Procedure. The key features of this procedure are the optimized in-gel digestion, treatment with a phosphatase, and the combined use of MALDI/TOF and LC/ESI ion trap mass spectrometers. With our optimized in-gel digestion procedure, we could extract >60% of the tryptic peptides with minimal contamination by peptides from trypsin autolysis and other sources (such as keratin) that could arise from prolonged sample handling. The use of a phosphatase to treat the entire mixture of extracted peptides in conjunction with MALDI/TOF measurements provides a simple, fast, sensitive, and robust method to identify phosphopeptide candidates for the subsequent tandem MS experiments to find the precise phosphorylation site. This method has been

successfully used by another group to treat a HPLC fraction containing a phosphopeptide.16 Digestion of the peptide mixture with a phosphatase allows the identification of phosphopeptides that could otherwise be easily overlooked due to mass degeneracy. The use of MALDI mass spectrometry for the identification of phosphopeptides is advantageous. MALDI is highly sensitive and tolerant of salt and biological buffers, thus eliminating the need for a desalting step and accompanying loss of sample, while still having a sufficiently high signal-to-noise ratio to allow the identification of low-intensity ions. MALDI mass spectrometry requires only very small amounts of sample, and there is no danger of losing the sample during the mass measurement. The use of a short, linear MALDI/TOF mass spectrometer (1.2 m) Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

2057

Figure 6. Sequence coverage map for trypsin digestion of PKR. The boxed sequence is the sequence tag. Observed sequences in the MALDI/TOF mass spectrum are underlined. T20+β, T49+β, and T49-50+β are the corresponding Cys-containing tryptic peptides that are modified by β-mercaptoethenol.

with delayed extraction provides single-stage MS of the highest possible sensitivitysi.e., providing the potential for a high sequence coverage. In contrast, electrospray mass spectrometry often requires a desalting step for in-gel-digested samples that can lead to preferential loss of highly hydrophilic and hydrophobic peptides, and it also has a high chemical background, which often tends to mask the weaker peaks. We found that weaker peaks observed in the MALDI/TOF spectrum were often missing in ESI/LC/MS using the LCQ mass spectrometer. However, electrospray combined with the ion trap mass spectrometer has a much higher sensitivity in the MS/MS mode than in the MS mode. Peptide ions that cannot be observed in the LC/MS run can be detected with good S/N ratio in a LC/MS/MS run if the m/z value is known. Therefore, using MALDI/TOF as the firststage MS to identify the phosphopeptides, followed by MS/MS in an ESI/ion trap, provides a very sensitive means for the identification of phosphorylation sites. On-line LC/MS/MS separates peptides with the same m/z values, thus allowing use of a wide mass isolation window ((4 Da) for high sensitivity (see Figure 7a), concentrates the peptide, and allows phosphopeptides that elute in the HPLC void volume and very hydrophobic phosphopeptides that elute only in a highly organic mobile phase to be monitored. Our use of MALDI/TOF to identify phosphopeptides that otherwise cannot be detected in ESI/LC/MS is analogous to the elegant approach of parent-ion scan to identify phosphopeptides developed by Carr’s group.15 The two approaches are complementary yet are different from each other in one significant way. Our combined MALDI/TOF and LC/MS/MS strategy eliminates any off-line HPLC fractionation, thus minimizing sample handling and its associated sample losses, while the parent-ion scan technique, using nanosray, requires HPLC fraction and fraction collections. Parent-ion scan to identify phosphopeptides, of course, is more direct and less ambiguous than MALDI/TOF with a phosphatase treatment if collecting of HPLC fractions does not 2058

Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

introduce sample losses to a degree that the phosphopeptides are below the detection limit. Potential pitfalls do exist in our approach. Phosphopeptides that cannot be ionized by MALDI cannot be identified. Peptides with masses 200 fmol of protein that is fully phosphorylated to identify a single phosphorylation site. If the phosphorylation stoichiometry is 10%, for example, 2 pmol of total protein load in the gel is generally required; if there are multiple sites, more samples are needed as a function of the numbers of phosphorylation sites. We believe the reason that less phosphoprotein (200 fmol) is required for low phosphorylation stoichiometry is that the unphosphorylated protein protects the phosphorylated protein from excessive loss during sample handing. We did not find major problems with low phosphorylation stoichiometry (if the phosphorylation stoichiometry is >10%), as long as there was >200 fmol of fully phosphorylated protein. CONCLUSIONS We have devised a highly sensitive, rapid, and robust procedure for the determination of phosphorylation sites with small quantities of gel-separated proteins. This procedure eliminates any off-line HPLC separation and minimizes sample handling. The use of MALDI/TOF and LCQ, two types of mass spectrometers that are widely available to the biological community, will make this procedure readily accessible to biologists. We have used this procedure to identify two autophosphorylation sites and to assign at least another 12 phosphorylation sites to two tryptic peptides in PKR using a gel slice containing only 200 ng (3 pmol) of protein. The ability to identify phosphorylation sites from 200 ng of gel-

Figure 7. LC/ESI/MS/MS of phosphopeptide ions in the tryptic digest of PKR. The identified phosphorylated residue is shown in bold. (a) Top panel: total fragment ion chromatogram of the doubly charged tryptic phosphopeptide (m/z 873.4) that was identified as singly charged peptide, m/z 1745.86 in Figure 4a. The other peaks in the region of 7.5-9.5 min are other peptides with the same m/z values that elute later in the LC separation. Bottom panel: the CID spectrum of the doubly charged phosphopeptide, which is the average of 10 CID spectra that compose the peak in Figure 4a at 6.26 min. The vertical scale is expanded by a factor of 10, except for the region around m/z 824.6, corresponding to the doubly charged phosphopeptide minus the elements of H3PO4. (b) Top panel: total fragment ion chromatogram of the doubly charged phosphopeptide (m/z 1029.1) that was identified as the singly charged peptide, m/z 2057.11 in Figure 4b. Botton panel: the CID spectrum of the doubly charged phosphopeptide, which is the average of 10 CID spectra that comprise the peak in Figure 7b at 8.04 min. The vertical scale is expanded by a factor of 5, except for the region around m/z 980.5. Ions are identified as in Figure 2.

purified protein that are not 32P-labeled opens new avenues for biological research. This amount of protein can be purified from immunoprecipitation with reasonable effort. We have already applied this procedure to the identification of in vivo phosphorylation sites in the yeast eIF2R kinase GCN2, the mouse protooncogene c-cbl, the mouse colony stimulation factor receptor CSF1R, the mouse nuclear factor of activated T-cells NFAT, and human Stat5A and 5B. Based on our experience, with a better protocol to increase sequence coverage by mass spectrometry, we anticipate that mass spectrometry will become the method of choice for identification of phosphorylation sites.

ACKNOWLEDGMENT We thank Dr. Weiguo Zhang for supplying the 35S-labeled catalytic domain of Zap70 and Mrs. Angela Murphy for amino acid analysis of the catalytic domain of MIHCK. We thank Drs. Klaus Biemann and Edward Korn for suggestions during the preparation of the manuscript. Received for review November 3, 1997. Accepted March 5, 1998. AC971207M Analytical Chemistry, Vol. 70, No. 10, May 15, 1998

2059