Comparative quantification and identification of phosphoproteins using stable isotope labeling and liquid chromatography/mass spectrometry

Transcription

1 RAPID COMMUNICATIONS IN MASS SPECTROMETRY Comparative quantification and identification of phosphoproteins using stable isotope labeling and liquid chromatography/mass spectrometry Wolfram Weckwerth*, Lothar Willmitzer and Oliver Fiehn Max-Planck-Institute of Molecular Plant Physiology, Potsdam, Germany SPONSOR REFEREE: Dr. David R. Goodlett, The Institute for Systems Biology, Seattle, WA, USA A new liquid chromatography/mass spectrometry (LC/MS) method is described for relative quantification of phosphoproteins to simultaneously compare the phosphorylation status of proteins under two different conditions. Quantification was achieved by b-elimination of phosphate from phospho-ser/thr followed by Micheal addition of ethanethiol and/or ethane-d 5 -thiol selectively at the vinyl moiety of dehydroalanine and dehydroamino-2-butyric acid. The method was evaluated using the model phosphoprotein a S1 -casein, for which three phosphopeptides were found after tryptic digestion. Reproducibility of the relative quantification of seven independent replicates was found to be 11% SD. The dynamic range covered two orders of magnitude, and quantification was linear for mixtures of 0 to 100% a S1 -casein and dephosphoa S1 -casein (R 2 = 0.986). Additionally, the method allowed protein identification and determination of the phosphorylation sites via MS/MS fragmentation. Received 20 July 2000; Accepted 24 July 2000 In the post genomic era, information will be important that cannot be derived from genome sequences alone. Today, protein identification of fully sequenced organisms by mass spectrometry in conjunction with database search algorithms is routine. 1 However, regulation of enzymatic activities by post-translational s cannot be predicted by gene sequences. Further, it has been recognised that protein identification alone is not sufficient to gain insight into regulation at the cellular level and that accurate profiling of expression levels of proteins is also needed. Recently, quantifying relative protein expression levels has been achieved by Aebersold et al. using a combination of chemical stable isotope labeling in vitro and affinity purification (i.e. isotope coded affinity tag or ICAT) prior to analysis by liquid chromatography/mass spectrometry (LC/MS). 2 A different approach was chosen by Chait et al. who labeled proteins in vivo using growth media isotopically enriched for 15 N-leucine prior to matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) analysis. 3 Chemical labeling by in vitro techniques such as the ICAT method, however, may prove somewhat more universal since growth of cells on isotopically enriched media is limited to cells that can be cultured whereas in vitro labeling methods are applicable to proteins from, for instance, solid tumors or plants. In both approaches, the two cell states under investigation were combined in order to overcome analytical errors caused by differences in matrix/peptide co-crystallization in MALDI-TOF analyses or, resp., run-to-run variations of electrospray ionization (ESI) efficiencies in LC/MS analyses. Current ICAT methods use chemical attachment of biotin groups to *Correspondence to: W. Weckwerth, Max-Planck-Institute of Molecular Plant Physiology, Potsdam, Germany. weckwerth@mpimp-golm.mpg.de cysteine peptide residues. Therefore, peptides bearing post-translational s will be lost during biotin/ avidin affinity purification. Among these, reversible protein phosphorylation at the hydroxyl group of serine, threonine or tyrosine is regarded as one of the major regulatory mechanisms for controlling the intracellular protein functionality, particularly in plant research. For instance, protein phosphorylation has emerged as an important mechanism controlling activities and localisation of key enzymes in sucrose metabolism 4 in higher plants as well as sugar sensing, 5 defense 6 and stress mechanisms via signal transduction and altered gene expression. However, a detailed picture in plants is still missing, and an analytical approach to quantify phosphoprotein ratios in differentially regulated cell states is of particular importance. For more accurate and faster phosphoprotein analysis, including identification and quantification, new approaches have to be developed and to be evaluated. Based on stable isotope labeling, we describe here a new method for quantitative analysis of phosphoproteins that also allows MS/MS identification of phosphorylation sites. 7,8 EXPERIMENTAL a S1 -Casein and dephosphorylated a S1 -casein were purchased from SAF (Deisenhofen, Germany) and were not further purified prior to analysis. Trypsin was obtained at sequencing grade from Roche Diagnostics (Mannheim, Germany). Ethane-d 5 -thiol was obtained from Campro Scientific (Emmerich, Germany). HPLC gradient grade acetonitrile was obtained from Merck (Darmstadt, Germany). Protein digestion was performed overnight at trypsin/protein ratios of 1:100 in 50 mm ammonium bicarbonate at ph 8.5 and 37 C. For chemical replacement of phosphopeptides through alkylthiols, 9 10 ml of the

2 1678 QUANTIFICATION OF PHOSPHOPROTEINS USING STABLE ISOTOPES AND LC/MS tryptic digest were added to 50 ml of a 2:2:1:0.65 mixture of H 2 O/DMSO/EtOH/5 M NaOH. After addition of 60 ml of alkylthiol, the reaction mixture was heated for 3 h at 50 C. After cooling to room temperature, the reaction was quenched by the addition of 60 ml of 20% acetic acid and 10 ml of acetonitrile. This mixture was subjected to LC/ES- MS analysis without further purification. The LC/MS analysis was performed on a Waters Alliance 2690 separation module (Waters, Milford, MA, USA) coupled to a QTOF hybrid mass spectrometer from Micromass (Manchester, UK) equipped with an orthogonal electrospray source (Z-spray) operated in the electrospray positive ion mode (ESI ). Chromatographic separation was achieved using a 100 1mm, 3mm, Luna C18(2) column (Phenomenex, Aschaffenburg, Germany). HPLC gradient elution was performed by water/0.1% formic acid (A) and acetonitrile/0.1% formic acid (B) starting at 5% B, 5 min isocratic, 80% B at 15 min, 5 min isocratic with a flow rate of 100 ml/min. 10 ml of the crude mixture were injected onto the column without prior purification. [Glu 1 ]-fibrinopeptide B from SAF (Deisenhofen, Germany) was used for mass calibration checks, and optimal parameter tuning was performed using flow injection of standard solutions. Conditions optimised for sensitivity were: ESI capillary voltage 2900 V, cone voltage 30 V, source temperature 100 C, desolvation temperature 150 C. For preliminary flow injection and LC/MS analyses, Q1 was set to bandwidth bypass mode with all ions transmitted into the TOF analyser which was scanned over m/z 300 to 2000 with 1s integration time. All TOF measurements were performed at high resolution settings (5000 FWHM at mass 785), and data were recorded in continuum mode. For MS/MS fragmentation, the quadrupole was set up to pass precursor ions of selected masses to the hexapole collision cell (using argon as collision gas), and product ion spectra were acquired with the TOF analyser. Using data dependent MS/MS switching, the three most abundant peptides were selected from each survey scan. For each peptide, a collision energy profile was used in five steps from 30 to 50 V for 1 s each. The resolution of the quadrupole mass filter was set such that the peak width was 1.2 mass units at half height. Quantification via integration was performed from smoothed, extracted ion chromatograms covering the limited mass ranges of the monoisotopic peaks in order to obtain best signal-to-noise ratios. RESULTS AND DISCUSSION Figure 1 shows a principle scheme for the relative quantification of two independent protein samples. For phosphopeptides, Michael addition of ethanethiol to the double bond of dehydroalanine after alkaline b-elimination of the phosphate moiety is a selective reaction prone to be extended to relative quantification. Cysteine residues require partial oxidation by peroxo organic acids since free thiol groups may also undergo alkaline b-elimination. Several Michael reagents were tested such as methylamine or mercaptoethanol, but these reagents did not show a sufficient selectivity or reactivity (data not shown) which is in agreement with the early work of Meyer et al. 10 To prove the validity of this new approach, a S1 -casein was used that has also served as model phosphoprotein in previous studies. First, optimum conditions were evaluated by testing the reaction s temperature and time dependence. As a compromise between reaction completeness and peptide Figure 1. Analytical scheme for relative quantification of two different phosphoprotein samples. breakdown, the temperature optimum was found at 50 C (Fig. 2). At this temperature, 3 h heating gave best results when testing reaction times between 30 min and 12 h (data not shown). Secondly, several parameters were tested that are important for evaluating whether an analytical method is suited for routine comparative quantifications. Among these, within-run analytical precision (i.e. reproducibility) is at least as important as sensitivity and dynamic range. Reproducibility was tested dividing an aqueous solution of 200 pmol a S1 -casein into two equal aliquots. After phosphate b-elimination and addition of ethanethiol, and, resp. ethane-d 5 -thiol, both aliquots were combined and subjected to analysis by LC/MS with an expected 1:1 ratio of labeled and non-labeled modified phosphocasein. After analysis of seven independent replicates, an average ratio of was found. This standard deviation is in good agreement with the accuracy obtained by the ICAT method. 2 Dynamic range was tested from 35 to 2500 pmol modified phosphocasein, injected onto the column (Fig. 3). Better sensitivities would have been achieved by down scaling the LC system from 1 mm to <100 mm i.d. columns as has been shown by Goodlett et al., 11 which would have Figure 2. Comparison between quantitative b-elimination and peptide degradation.

3 QUANTIFICATION OF PHOSPHOPROTEINS USING STABLE ISOTOPES AND LC/MS 1679 Figure 3. Dynamic range of quantification of a sl -casein, modified phosphopeptide at m/z been necessary for low abundant biological phosphoproteins but was regarded not to be necessary for proof of concept using model phosphoproteins. Analysing these samples, three different phosphopeptides were found both in the modified and in the non-modified forms (Table 1). For the phosphopeptide YKVPQLEIVPNS p AEER, no dephosphorylated form was detectable in commercial a S1 -casein; however, in the dephosphorylated a S1 -casein, 37% of this peptide was still found to be phosphorylated which was used for calculations of expected dephospho/phosphopeptide ratios. In order to simulate the quantification of different phosphorylation states between two hypothetical in vivo situations I and II, we set out for an experimental setup where the total protein content was held constant, but the amount of the phosphorylated forms varied. To achieve this, a set of phosphocasein/dephosphocasein mixtures were divided into two aliquots and subjected to tryptic digestion. One aliquot was modified with ethanethiol (in vivo situation I), and the other aliquot was modified with ethane-d 5 -thiol (in vivo situation II). Both aliquots were combined in different ratios, and analysed by LC/MS. Plotting the expected to the observed phosphopeptide ratios of the two situations, the resulting calibration curve was found to be linear from 0 to 100% dephospho/phosphocasein with a linearity coefficient of R 2 = (Fig. 4). This finding demonstrates the suitability of stable isotope labeling for comparative quantification of phosphorylated proteins. To demonstrate that the method is also applicable for determination of multiphosphorylated peptides, the relative quantification and identification of the diphosphorylated peptide DIGS p ES p TEDQAMEDIK within the tryptic digest of phosphocasein is shown in Fig. 5. The expected peptide Figure 4. Ratio of the d 5 -EtSH-modified and EtSH-modified phosphopeptide m/z in mixtures of different amounts of phospho-casein. Total content of casein was held constant, but ratios of phospho/dephosphoprotein were gradually increased (for details, see text). masses and the mass difference of 5 u for the labeled and non-labeled form of the diprotonated molecules at m/z and after were observed as can be seen in lanes 1 6 of Fig. 5. It also gives a good example how the presence of unexpected phosphopeptides is revealed when investigating complex peptide digests. Besides the relative quantification, the high resolving power of the QTOF mass spectrometer guided phosphopeptide identification by calculating charge states and mass differences between labeled and non-labeled modified phosphopeptides. [M 3H] 3 charge states may be recognised up to m/z 1500 which is not achievable by triple quadrupole or ion trap instruments. Identification of phosphorylation sites was achieved by MS/MS fragmentation of modified peptides within the QTOF collision cell and subsequent automated peptide sequencing using the instrument s MassLynx software (Fig. 6). The proposed method has the potential to be extended to simultaneous quantification of the phosphorylation states of proteins in relation to their parental non-phosphorylated enzymes for biological situations, in which the total content of parental protein does not stay constant (such as in mutant and wild-type plants). For low abundant phosphoproteins, preconcentration and purification steps such as ICAT techniques have to be included where a biotinylated tag is present allowing specific chemical selection and purification for phosphopeptides. However, for ICAT-like preconcentration strategies normalisation of phosphopeptide amounts have to be done on the basis of mg tissue or overall Table 1. Detected phosphopeptides in the tryptic digest of a sl -casein Tryptic fragment Peptide sequences No. of phosphorylation sites Calculated Observed Calculated after with EtSH Observed after with EtSH Calculated after with d 5 -EtSH Observed after with d 5 -EtSH T15 16 (aa ) YKVPQLEIVPNS P AEER T16 (aa ) VPQLEIVPNS P AEER T8 (aa 58 73) DIGS P ES P TEDQAMEDIK

4 1680 QUANTIFICATION OF PHOSPHOPROTEINS USING STABLE ISOTOPES AND LC/MS Figure 5. Combination of 30 scans of a 50 pmol LC/MS run showing elution of the EtSH- and d 5 -EtSH-modified phosphopeptide mixtures with different ratios. At m/z and , resolution and relative quantification of the doubly charged modified diphosphopeptide DIGS p ES p TEDQAMADIK is demonstrated (lane 1 100% deuterated/0% undeuterated mod. peptide, lane 2 80% deuterated/20% undeuterated mod. peptide, lane 3 60% deuterated/40% undeuterated mod. peptide, lane 4 40% deuterated/60% undeuterated mod. peptide, lane 5 20% deuterated/80% undeuterated mod. peptide, lane 6 0% deuterated/100% undeuterated mod. peptide). Figure 6. MS/MS fragmentation of doubly charged ion signal at m/z Automated sequencing parameters were set as 0.75% for intensity threshhold and 0.15 for the mass tolerance. The modified phosphoserine residue is indicated as mod. S p. protein content, since parental non-phosphorylated proteins will get lost during purification. CONCLUSIONS Chemical attachment of stable isotopes to functional groups of proteins is a promising tool for the comparative quantification of two different samples, for instance a wild-type and a mutant plant. Using a S1 -casein as a model phosphoprotein, reproducibility, sensitivity and dynamic range were proven to be satisfactory when simulating two different phosphorylation states of abundant proteins. For quantification of low abundant phosphoproteins, further purification and preconcentration steps such as IMAC, 12 ICAT or isoelectric focusing would be needed as well as down scaling of the LC/MS method to capillary liquid chromatography. Further method improvements will also exploit QTOF mass resolution for developing algorithms that automatically find phosphopeptides in crude extracts by calculating charge states and mass differences between labeled and non-labeled modified phosphopeptides. 13 REFERENCES 1. Burlingame AL, Boyd RK, Gaskell SJ. Anal. Chem. 1998; 70: 647R 2. Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH, Aebersold R. Nat. Biotech. 1999; 17: Oda Y, Huang K, Cross FR, Cowburn D, Chait BT. Proc. Natl. Acad. Sci. USA 1999; 96: Winter H, Huber SC. Crit. Rev. Plant Science 2000; 19: Smeekens S. Curr. Opin Plant Biol. 1998; 1: Ehness R, Ecker M, Godt DE, Roitsch T. Plant Cell 1997; 9: Neubauer G, Mann M. Anal. Chem. 1999; 71: Hunter AP, Games DE. Rapid Commun. Mass Spectrom. 1994; 8: Meyer HE, Hoffmann-Posorske E, Korte H, Heilmeyer LMG. FEBS 1986; 204: 61

5 QUANTIFICATION OF PHOSPHOPROTEINS USING STABLE ISOTOPES AND LC/MS Meyer HE, Hoffmann-Posorske E, Heilmeyer LMG. Methods Enzymol. 1991; 205: Goodlett DR, Bruce JE, Anderson GA, Rist B, Pasa-Tolic L, Fiehn O, Smith RD, Aebersold R. Anal. Chem. 2000; 72: Gallis B, Corthals GL, Goodlett DR, Ueba H, Kim F, Presnell SR, Figeys D, Harrison DG, Berk BC, Aebersold R, Corson MA. J. Biol. Chem. 1999; 274: Drexler DM, Tiller PR, Wilbert SM, Bramble FQ, Schwartz JC. Rapid Commun. Mass Spectrom. 1998; 12: 1501