Highly Specific Capture and Direct MALDI MS Analysis of

Dec 7, 2009 - D-60438 Frankfurt, Germany, and Qiagen GmbH, Qiagenstr. 1, D-40724 Hilden, Germany. The dynamic range and low stoichiometry of protein...
0 downloads 0 Views 412KB Size
Anal. Chem. 2010, 82, 219–228

Highly Specific Capture and Direct MALDI MS Analysis of Phosphopeptides by Zirconium Phosphonate on Self-Assembled Monolayers Tri Hoang,‡,† Udo Roth,‡,§ Karen Kowalewski,§ Christopher Belisle,§ Kerstin Steinert,§ and Michael Karas*,† Institute of Pharmaceutical Chemistry, Goethe-University Frankfurt/Main, Max-von-Laue-Str. 9, D-60438 Frankfurt, Germany, and Qiagen GmbH, Qiagenstr. 1, D-40724 Hilden, Germany The dynamic range and low stoichiometry of protein phosphorylation frequently demands the enrichment of phosphorylated peptides from protein digests prior to mass spectrometry. Several techniques have been reported in literature for phosphopeptide enrichment, including metal oxides such as TiO2 and ion metal affinity chromatography (IMAC). While the metal oxides have been used with reasonable success, IMAC has suffered from reduced selectivity and poor reproducibility. In this report, we present the first demonstration of the use of immobilized zirconium on a phosphonateterminated self-assembled monolayer (SAM) for specific phosphopeptide capture and direct analysis by MALDI MS. By using the herein described functionalized-surface-based technology, efficient enrichment of phosphopeptides in different standard test systems such as r- or β-casein digests or synthetic phosphopeptides spiked in nonphosphorylated protein digest has been demonstrated. The limit of detection for a β-casein phosphopeptide was assessed to be at the low femtomole level. Compared to other state-of the-art technologies, like use of TiO2 and Fe-IMAC, the presented technique demonstrated a superior performance with respect to specificity and bias with respect to singly or multiply phosphorylated peptides. Additionally, this platform was also successfully applied for ESI sample preparation, providing detailed sequence information of the investigated phosphopeptide. This technology was also proven to be applicable for real life samples such as phosphorylation site analysis of recombinant human MAPK1 and HSP B1 isolated from a 2D-gel spot by phosphopeptide enrichment and direct MALDI MS/MS. Protein phosphorylation is one of the most important posttranslational modifications (PTM) in eukaryotic cells and is a primary regulatory pathway involving both kinases and phosphatases.1,2 The mechanism of protein phosphorylation and * Corresponding author. E-mail: [email protected]. Fax: +49-69-798-29918. † Goethe-University Frankfurt/Main. ‡ These authors equally contributed to the work. § Qiagen GmbH. (1) Hubbard, M. J.; Cohen, P. Trends Biochem. Sci. 1993, 18, 172–177. 10.1021/ac9017583  2010 American Chemical Society Published on Web 12/07/2009

dephosphorylation controls a wide range of cellular events, including cell growth, division, and differentiation. In order to gain a detailed understanding of these pathways, it is imperative to determine the phosphorylation state of proteins with high precision at a given metabolic stage. Detection and quantitation of phosphorylated proteins is complicated by the substoichiometric occurrence and the dynamic range presented versus unmodified proteins. Particularly, in MSdriven phosphoproteomics experiments analyzing whole cell lysates or tissue samples, phosphopeptides are strongly outnumbered by their nonphosphorylated counterparts. Thus, it is barely possible to obtain a global view of protein phosphorylation in a biological system, not to mention quantitative aspects, by classical LC-MS approaches without enrichment of phosphopeptides.3,4 For that reason, tremendous effort was spent to develop specific enrichment technologies suitable for coupling to MS. Among the existing enrichment methods for phosphopeptides, immobilized metal affinity chromatography (IMAC) is one of the mostly employed methods, using various metals such as Fe3+and Ga3+.5 However, nonspecific binding of nonphosphorylated acidic peptides to IMAC results in low specificity and sensitivity. Although it was reported that chemical modification of the acid groups to methyl esters has minimized this problem,6 the success of this approach was hampered, due to undesired side reaction and incomplete modification of the peptides.7 Recently, oxides of titanium, zirconium, and aluminum have been used for the specific enrichment of phosphopeptides from complex peptide mixtures.7-10 The major improvement of titanium dioxide over IMAC is the increased tolerance of higher concentration of strong acids in the binding and washing steps, thus minimizing (2) Cohen, P. Nat. Cell Biol. 2002, 4, E127-E130. (3) Annan, R. S.; Huddleston, M. J.; Verma, R.; Deshaies, R. J.; Carr, S. A. Anal. Chem. 2001, 73, 393–404. (4) Tsay, Y. G.; Wang, Y. H.; Chiu, C. M.; Shen, B. J.; Lee, S. C. Anal. Biochem. 2000, 287, 55–64. (5) Posewitz, M. C.; Tempst, P. Anal. Chem. 1999, 71, 2883–2892. (6) Ficarro, S. B.; McCleland, M. L.; Stukenberg, P. T.; Burke, D. J.; Ross, M. M.; Shabanowitz, J.; Hunt, D. F.; White, F. M. Nat. Biotechnol. 2002, 20, 301–305. (7) Thingholm, T. E.; Jensen, O. N.; Larsen, M. R. Proteomics 2009, 9, 1451– 1468. (8) Wolschin, F.; Wienkoop, S.; Weckwerth, W. Proteomics 2005, 5, 4389– 4397. (9) Kweon, H. K.; Hakansson, K. Anal. Chem. 2006, 78, 1743–1749. (10) Zhou, H.; Tian, R.; Ye, M.; Xu, S.; Feng, S.; Pan, C.; Jiang, X.; Li, X.; Zou, H. Electrophoresis 2007, 28, 2201–2215.

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

219

the nonspecific binding of acidic nonphosphorylated peptides.11 Another study concluded that both titanium dioxide and IMAC were found to be reproducible techniques, but since they do not enrich the same phosphopeptides from the same sample, the technologies provide complementary information as to the whole picture of the phosphoproteome.12 The use of zirconium alkyl-phosphonate linker structures (in the following abbreviated “ZrPO3”) for capturing phosphatecontaining molecules was originally reported in the context of DNA microarrays,13,14 on which oligonucleotide probes were immobilized by ZrPO3 linkers attached to the array surface. Just recently, ZrPO3 was also described for phosphopeptide enrichment by several groups.15-19 As compared to metal oxides and IMAC, ZrPO3 has been shown to bind more phosphopeptides with better selectivity from different protein digests. Several formats have been used, including magnetic nanoparticles, nanoplatelets, polymer beads, and monolithic capillary columns.15-19 To our knowledge, we report herein the first example of immobilized zirconium on a phosphonate self-assembled monolayer (SAM) for successful enrichment of phosphopeptides and direct analysis by MALDI-MS. The background of the SAM technology is reviewed in detail by Zhou and Walker.20 We used this approach to establish a functionalized surface for phosphopeptide affinity enrichment in combination with a special surface patterning, resulting in virtual sample wells for the application and concentration of higher sample volumes. By applying this functionalized MALDI plate, several phosphoprotein digests or peptide mixtures were analyzed with respect to the successful enrichment of phosphopeptides present in the sample. EXPERIMENTAL METHODS Chemicals. All chemicals used were at least of analytical grade or specified for mass spectrometry. Phosphorylated and unphosphorylated human angiotensin II were purchased from Invitrogen (Karlsruhe, Germany). All standard proteins and ZrOCl2 were obtained from Sigma-Aldrich (Taufkirchen, Germany). 2,5Dihydroxybenzoic acid (DHB) was purchased from Bruker Daltonics (Bremen, Germany); trifluoroacetic acid (TFA) was obtained from Thermo Scientific (Hagen, Germany). Phosphoric acid and acetonitrile (gradient grade) was from Merck (Darmstadt, Germany). Deionized water used throughout all experiments was prepared by a Milli-Q water system (Millipore, Milford, MA). (11) Thingholm, T. E.; Larsen, M. R. Methods Mol. Biol. 2009, 527, 57–66. (12) Bodenmiller, B.; Mueller, L. N.; Mueller, M.; Domon, B.; Abersold, R. Nat. Methods 2007, 4, 231–237. (13) Nonglaton, G.; Benitez, I. O.; Guisle, I.; Pipelier, M.; Le´ger, J.; Dubreuil, D.; Tellier, C.; Talham, D. R.; Bujoli, B. J. Am. Chem. Soc. 2004, 126, 1497– 1502. (14) Bujoli, B.; Lane, S. M.; Nonglaton, G.; Pipelier, M.; Le´ger, J.; Talham, D. R.; Tellier, C. Chemistry 2005, 11, 1980–1988. (15) Zhao, L.; Wu, R.; Han, G.; Zhou, H.; Ren, L.; Tian, R.; Zou, H. J. Am. Soc. Mass Spectrom. 2008, 19, 1176–1186. (16) Wei, J.; Zhang, Y.; Wang, J.; Tan, F.; Liu, J.; Cai, Y.; Qian, X. Rapid Commun. Mass Spectrom. 2008, 22, 1069–1080. (17) Feng, S.; Ye, M.; Zhou, H.; Jiang, X.; Gong, B. Mol. Cell. Proteomics. 2007, 6, 1656–1665. (18) Xu, S.; Whitin, J. C.; Yu, T. T.; Zhou, H.; Sun, D.; Sue, H. J.; Zou, H.; Cohen, H. J.; Zare, R. N. Anal. Chem. 2008, 80, 5542–5549. (19) Dong, J.; Zhou, H.; Wu, R.; Ye, M.; Zou, H. J. Sep. Sci. 2007, 30, 2917– 2923. (20) Zhou, C.; Walker, A. Langmuir 2006, 22, 11420–11425.

220

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

Manufacturing of the Functionalized MALDI Sample Support. Substrate. MALDI sample supports were made by coating the stainless steel substrate with 25 nm of Ti/W (9:1) for adhesion followed by 100 nm of Au using a CPA 9900 RF sputtering system (base pressure 10-7 Torr). SAM and UV-Patterning. The two-zone patterned MALDI surfaces were made by UV-photopatterning of the substrate. Self-assembled monolayers of HS(C11)PO3H2 and HS(C11)OCH2CH2(CF2)5CF3 were formed on the Au-sputtered substrates. For creation of the functionalized sample spot patterns, the substrates were first thiolized with background thiol (HS(C11)OCH2CH2(CF2)5CF3 (for details on thiol synthesis, see the Supporting Information) by exposing the surface to a 0.1 mM solution of the thiol in ethanol for 1 h. Unbound thiol was removed by two washing steps with absolute ethanol, yielding a hydrophobic plain SAM surface. After drying the surface, a mask providing a 5 × 5 spot array with holes of 2 mm diameter was placed on the MALDI target. The area within the sample spots was then exposed 25 times to strong UV light (Qiagen UV systems with fusion UV lamps, DDU Enterprises) with a passing velocity of 4 ft/min in order to remove the hydrophobic thiols at the sample spot positions. After another wash with ethanol, the substrates were incubated in a 0.1 mM solution of phosphonate thiol [HS(C11)PO3H2] in ethanol and incubated overnight to yield a SAM of phosphonate thiol at the sample spot positions. After washing as described above, the modified MALDI target was dried and ready to be used. Purification of Phosphoproteins and Preparation for 2DPAGE. Phosphoproteins were isolated from HELA cells (DMSZ, No. ACC 57, Braunschweig, Germany) with the help of a PhosphoProtein Purification Kit (Qiagen GmbH, Germany). Eluate fractions containing putative phosphoproteins were pooled and precipitated using the Na-deoxycholate/TCA method. The resulting protein pellet was then resolubilized in IEF-compatible sample buffer.21 2D-PAGE and Gel Staining. Two-dimensional separation of proteins was essentially carried out according to Clemen et al.21 In brief, proteins were separated by IEF with a total of approximately 34 kV h using 18 cm IPG strips (pH 3-10, nonlinear gradient), whereas SDS-PAGE [12.5% (w/v) acrylamide] as the second dimension was carried out according to Laemmli22 on a Hoefer SE 600 chamber. Upon termination of electrophoresis, the gel was stained with Pro-Q Diamond (Invitrogen, Carlsbad, CA) to visualize putative phosphoproteins following the manufacturer’s instructions and scanned on an Ettan DIGE Imager (GE Healthcare, Uppsala, Sweden). In-Solution Protein Digestion. R- And β-casein were digested according to Zhou et al.23 Briefly, 1 mg of R- and β casein was directly dissolved in 1 mL of digestion buffer (50 mM NH4HCO3) and digested overnight at 37 °C with trypsin (sequencing grade, Promega, Madison, WI) in 1:50 ratio and diluted with loading buffer to the respective concentrations applied in the experiments. BSA was dissolved in 1 mL of 25 mM NH4HCO3 and (21) Clemen, C. S.; Fischer, D.; Roth, U.; Simon, S.; Vicart, P.; Kato, K.; Kaminska, A. M.; Vorgerd, M.; Goldfarb, L. G.; Eymard, B.; Romero, N. B.; Goudeau, B.; Eggermann, T.; Zerres, K.; Noegel, A. A.; Schro¨der, R. FEBS Lett. 2005, 579, 3777–3782. (22) Laemmli, U. K. Nature 1970, 22, 680–685. (23) Zhou, H.; Xu, S.; Ye, M.; Feng, S.; Pan, C.; Jiang, X.; Li, X.; Han, G.; Fu, Y.; Zou, H. J Proteome Res. 2006, 5, 2431–2437.

reduced using 10 µL of 10 mM DTT for 1 h at 56 °C. Alkylation was performed using 10 µL of 55 mM iodoacetamide. Digestion with trypsin applying an E/S ratio of 1:50 was carried out at 37 °C for 10 h and stopped by addition of 0.1% (v/v) TFA. Aliquots used were diluted to the respective target concentration using 30% (v/v) ACN/300 mM acetic acid. Two micrograms of recombinant MAPK1 (ERK2) expressed in Escherichia coli was solubilized in 8 M urea, 10 mM DTT in 50 mM Tris-HCl (pH 8.0). After incubation for 30 min at 30 °C, the same volume of 110 mM iodocatamide in 50 mM Tris-HCl (pH 8.0) was added, and alkylation of the protein was carried out for 20 min at room temperature in the dark. The alkylated protein was further diluted with 50 mM Tris-HCl (pH 8.0) to a final urea concentration of 800 mM, and trypsin was added in a 1:50 ratio. After overnight digestion at room temperature, the resulting peptides were purified using PepClean C-18 spin columns (Thermo Scientific, Hagen) according to the manufacturer’s instructions and dried in a Speed Vac. For processing on the ZrPO3 sample carrier, the peptides were resolubilized in 10 µL of 30% ACN in 300 mM acetic acid. In-Gel Digestion of 2D-Gel Spot Proteins. Gel spots of interest were picked from the gel after overstaining with Coomassie and basically processed according to Roth et al.24 In brief, gel plugs were washed with deionized water and two subsequent washing steps with 50% acetonitrile (ACN) and 100% ACN. After removal of the ACN, the pieces were covered with 12.5 ng/µL trypsin in 50 mM NH4HCO3 and incubated on ice for 1 h. Subsequently, the remaining trypsin solution was removed and 8 µL of digestion buffer (25 mM NH4HCO3) was added. After overnight digestion at 37 °C, 2 µL of 1% (v/v) trifluoroacetic acid (TFA) was added and the digest incubated for 30 min at room temperature. Of the acidified digest, 1 µL was applied on Mass Spec Turbo Chips (Qiagen GmbH, Hilden, Germany) prespotted with CHCA for protein identification, whereas the remaining sample was applied to the ZrPO3-modified MALDI sample support. Sample Processing on the ZrPO3-Modified MALDI Sample Carrier. Prior to sample loading, the plate surface was cleaned by sequential incubation of the sample spots with 70% ACN and 300 mM acetic acid for 2 min, respectively. After removal of the washing solution, each sample spot was charged with Zr4+ by incubation of 10 µL of an aqueous 100 mM ZrOCl2 solution for 15 min. Unbound Zr4+ ions were then removed from the sample spots by pipetting 10 µL of 300 mM acetic acid solution up and down for approximately 2 min. This procedure was repeated twice, before 5 µL of sample was applied. After an incubation of 20 min, the remaining liquid was removed and the sample spots were washed again three times with 10 µL of sample loading buffer containing 30% ACN in 300 mM acetic acid. After the final washing step, 2 µL of matrix solution (1 mg/mL DHB in 0.1% phosphoric acid)25 was spotted and the preparation dried down. Phosphopeptide Purification Applying Fe-IMAC. Twenty microliters of Fe-IMAC resin (PHOS-Select Iron Affinity Gel, Sigma Aldrich, Taufkirchen, Germany) was resuspended with 60 (24) Roth, U.; Mu ¨ ller, S.; Hanisch, F. G. In Methods in Molecular Biology Vol. 346, Dictyostelium discoideum Protocols; Eichinger, L., Rivero F., Eds.; Humana Press: Totowa, NJ, 2006; pp 95-109. (25) Kjellstro ¨m, S.; Jensen, O. N. Anal. Chem. 2004, 76, 5109–5117.

µL of deionized water and centrifuged to separate the beads from the supernatant, which was discarded. Subsequently, the beads were equilibrated with 20 µL of acetic acid (100 mM), and 10 µL of sample was added. Incubation was performed for 60 min at room temperature. Then the resin was filled in a gel loader tip (Eppendorf, Hamburg, Germany) and washed twice with 30% ACN in 100 mM acetic acid according to Gobom et al.26 Bound peptides were eluted with 3 µL of DHB (10 mg/mL) matrix in 50% ACN/ 1% phosphoric acid directly onto a stainless steel MALDI target. Phosphopeptide Purification Using TiO2 Pipet Tips. Gel loader tips packed with TiO2 particles (Glygen, Columbia, MD) were equilibrated by two washing steps with 20 µL of 100 mM acetic acid. After binding of phosphopeptides from 10 µL sample solution, the columns were again washed twice with 30% ACN in 100 mM acetic acid and peptides were eluted directly on a stainless steel MALDI target with 3 µL of DHB matrix as described under Fe-IMAC. MALDI-MS and MS/MS Spectra Acquisition and Processing. MS spectra were recorded either on a Voyager STR MALDI-TOF instrument or 4800 MALDI TOF/TOF analyzer (both ABI, Framingham, MA) in reflectron positive ion mode. Each spectrum represents the average of 1000 shots for the 4800 TOF/ TOF or 500 shots for the Voyager instrument, respectively. To identify peptides or assess peptide phosphorylation, MS/MS experiments were carried out on the same preparation using the 4800 MALDI TOF/TOF analyzer. Precursor ions were separated by timed ion selection (300 resolution window, fwhm), and fragmentation was triggered either by PSD or CID on ions decelerated to an energy of 2 kV. Here, each spectrum exhibits an average of 1500 shots (50 shots per subspectrum). Spectra were annotated using the Data Explorer software 4.9 and manually inspected for neutral loss of phosphoric acid (dominant fragment ion with a mass shift of -98 Da in relation to the precursor mass). For protein (2D-gel spots) and peptide identification, peak lists were extracted employing the “Peaks to Mascot” tool embedded in the 4000 Series Explorer software of the 4800 TOF/TOF instrument and searched against the human IPI database using the Mascot Server (v.2.2) search engine. ESI-MS/MS of β-Casein 2061 m/z Phosphopeptide. A 500 fmol sample of β-casein digest was purified on the ZrPO3 MALDI target, and bound peptides were eluted using 2 µL of 150 mM (NH4)OH. Subsequently, 2 µL of 2% formic acid/50% MeOH was added. The sample was filled into a laboratory-made, goldcoated glass capillary, and MS as well as MS/MS spectra were acquired on a LCQ Classic (Thermo-Finnigan, San Jose) in positive ion mode applying static nanospray conditions. The spray voltage was set at 2.1 kV, whereas the capillary voltage and the tube lens were both set to 42 V. The temperature of the capillary was kept at 200 °C. Spectra were averaged over 30 scans, each scan consisting of three microscans. For MS2 fragmentation, precursor ions were selected using an isolation width of 5 Th. The relative collision energy was set between 16% and 22%. RESULTS AND DISCUSSION Focusing Effect. Due to the special design of the SAM-coated MALDI target, functionalized sample spots, which provide virtual (26) Gobom, J.; Nordhoff, E.; Mirgorodskaya, E.; Ekman, R.; Roepstorff, P. J Mass Spectrom. 1999, 34, 105–116.

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

221

Figure 1. Sensitivity assay. MS spectra after processing of 10 fmol (A) or 1 fmol (B) of β-casein digest (with R-casein contaminations) on the ZrPO3-functionalized MALDI target. Phosphopeptides are annotated: R ) phosphopeptide derived from R-casein, β ) phosphopeptide derived from β-casein. For illustration purposes, spectra expansions are shown as insets in part A.

wells with capacities of up to 20 µL surrounded by a hydrophobic area, are created (see the structured sample support outline in the Supporting Information, Figure 1). This feature allows the application of sample amounts exceeding normal sample volumes applied for classical dried droplet preparations by far. Furthermore, this design results in a concentration of the sample on the sample spot with a diameter of approximately 2 mm and in a gain of sensitivity beneficial for the in depth analysis of the purified peptides. This focusing effect is maintained throughout the whole sample processing, including matrix application (see the Supporting Information, Figure 2) and distinguishes this kind of sample preparation from an approach published by Zhou et al.23 They fabricated a ZrPO3-modified porous silicon sample carrier but did not provide any further patterning to obtain a sample-welllike structure on the plates. It is also dissimilar to the approach described by Blacken et al.,27 who also reported the enrichment of phosphopeptides in different samples by application of modified MALDI plate and direct MS. However, in this case, the modification was carried out by creation of an active ZrO2 layer employing reactive landing technology. Characterization of the Zirconium-Phoshonate Surface. Aiming at the implementation of the phosphonate surface in proteomics applications, this platform was intensively characterized. The herein described method for the synthesis of the phosphonate surface using the SAM technology was highly reproducible and reliable. This could be demonstrated with a series of experiments investigating issues that are essential for a phosphopeptide purification technique. The first topic was the binding capacity of the surface. For this, different amounts of a synthetic monophosphopeptide (FQpSEEQQTEDELQDK) from β-casein were applied on a sample well, and after an incubation period, the supernatant was withdrawn and checked for the occurrence of the unbound peptide. For complementary results, (27) Blacken, G. R.; Volny´, M.; Vaisar, T.; Sadı´lek, M.; Turecek, F. Anal. Chem. 2007, 79, 5449–5456.

222

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

the amount of peptide bound to the spot was also monitored after elution of the analyte (for details, see the Experimental Section of the Supporting Information). With this approach, the binding capacity was determined to be about 650 fmol of the β-casein phosphopeptide (see the Supporting Information, Figure 3). In order to assess the reproducibility of the target manufacturing process, a series of five independent experiments (five ZrPO3modified targets) was carried out, and the signal intensity of the monophosphopeptide and the tetraphosphopeptide from 300 fmol of a β-casein digest applied to five wells of each of the ZrPO3 target was monitored. In these experiments, the signals for the monitored phosphopeptides showed reproducible signal intensity profiles with error bars of less than 15% (see the Supporting Information, Figure 4). These minor deviations are more likely to be attributed to the MALDI process, i.e., due to the complex desorption/ionization process and incorporation of analyte in the MALDI matrix, than to the purification process using the introduced enrichment device. Thus, a high reproducibility in preparation of the MALDI target surface with the Phosphonate-SAM film allowing for reliable and reproducible results for the purification of phosphopeptides could be demonstrated. In proteomics applications, sensitivity is a key performance criterion, as frequently low amounts of analytes are present. Therefore, techniques for the enrichment of phosphopeptides ideally have to enable the detection of amounts in the lower femtomole range. Figure 1 shows representative MS spectra for the purification of β-casein digests on the level of only a few femtomoles of analyte. From a 20 fmol β-casein digest, a number of phosphorylated peptides can be detected with high S/N ratios (Figure 1A). Even at this level, not only phosphopeptides of the β-casein were revealed but also phosphopeptides of R-casein, which is present as a 10% impurity in the β-casein sample as declared by the supplier, were detectable. Due to the focusing capability of the surface, the amount of only 1 fmol of β-casein

Table 1. Influence of Different Metal Cations on the Enrichment of Phosphopeptides on Phoshonate Surfacesa MH+

amino acid sequence

protein

Fe3+

Ga3+

Zr4+

1466.61 1539.60 1594.71 1660.80 1927.64 1951.95 2061.83 2432.05 2747.00 2556.09 3008.03 3122.27

TVDMpSTEVTK EQLpSTpSEENSKK TVDMEpSTEVFTKK VPQLEIVPNpSAEER DIGpSEpSTEDQAMEDIK YKVPQLEIVPNpSAEER FQpSEEQQQTEDELQDK IEKFQpSEEQQQTEDELQDK KNTMEHVpSpSpSEEpSIISQETYK FQpSEEQQQTEDELQDKIHPF NANEEYSIGpSpSpSEEpSAEVATEEVK RELEELNVPGEIVEpSLpSpSpSEESITR

R-S1-casein R-S1-casein R-S1-casein R-S2-casein R-S2-casein R-S2-casein β-casein β-casein R-S2-casein β-casein R-S2-casein β-casein

× × × × × × + × × × × +

× × × × + × + × + × + +

+ + + + × + + + × + × +

a PO3-MALDI targets were charged with different metal ions, and a mixture of 700 fmol of β-casein and R-casein digest was processed. Peaks observed under the respective conditions are marked by “+”, and peaks remaining undetected are marked by “×”.

applied was sufficient for the enrichment of the monophosphopeptide and the tetraphosphopeptide as well as one R-casein phosphopeptide (Figure 1B). Considering the low amount of sample, the S/N ratios observed are satisfactory. In general, the sensitivity is significantly higher than with different off-line purification tools like TiO2 or Fe-IMAC microcolumns or magnetic bead approaches. For example, Larsen et al.28 are typically working with analyte amounts of 500 fmol. This effect is most likely caused by the avoidance of additional sample transfer steps and hence the corresponding loss of analytes. When phosphonate surfaces are used for the purification of DNA as well as phosphopeptides, usually Zr4+ is added to the device prior to sample application. To our knowledge, there are no conclusive investigations comparing the performance of this metal cation compared to others. To address this issue, the phosphonate-modified target was loaded with Fe3+, Ga3+, and Zr4+, respectively (Table 1 and the Supporting Information, Figure 5) and probed with 700 fmol of a β-casein digest. The results for the three applied metal ions revealed that Fe3+ only enabled the detection of two phosphopeptides (Table 1), whereas with Ga3+ four phosphopeptides were visible in the mass spectra. In contrast to these two metal cations, after application of Zr4+, nine phosphopeptides could be determined. Interestingly, three of the bound phosphopeptides with Ga3+ as metal ion are complementary to the analytes detected with Zr4+, indicating different specificities for capturing of phosphopeptides. Comparison of IMAC, TiO2, and ZrPO3. IMAC and TiO2 are techniques that are frequently used for the purification of phosphopeptides. In order to characterize the performance of the presented ZrPO3-modified target in comparison to these standard methods,7 the ability to selectively enrich phosphopeptides was tested. In a first experiment, an R-casein digest (100 fmol) was purified using the three enrichment platforms (Figure 2). For this model protein, IMAC and titanium dioxide showed comparable results (Figure 2, Table 2) with respect to the number and S/N ratios of phosphopeptides detected. In contrast to these standard methods, the ZrPO3-modified target allowed the detection of seven additional phosphopeptides. Altogether 13 phosphorylated peptides were observed using (28) Larsen, M. R.; Thingholm, T. E.; Jensen, O. N.; Roepstorff, P.; Jørgensen, T. J. D. Mol. Cell. Proteomics 2005, 4, 873–886.

this device with decent S/N ratios (Figure 2, Table 2). The lower S/N ratios can be attributed to the larger number of phosphopeptides being bound to the surface. This result shows that the modified target has less discriminating properties than the other two techniques. After the model protein R-casein, a more complex system was investigated to compare the specificity and robustness of the purification techniques with respect to unphosphorylated background. A synthetic phosphopeptide (angiotensin II) was spiked into a BSA digest in 1:10 ratio (Figure 3A). Although capable of enriching phosphopeptides with a high phosphorylation ratio, as shown with the R-casein digest (Figure 2), IMAC is not suitable when samples with a higher unphosphorylated background are present (Figure 3B). At a phosphorylation ratio of 1:10, IMAC does not seem to selectively bind phosphopeptides, since the number of bound nonphosphorylated peptides represented by BSA peptides is very high. With respect to lower unspecific binding, TiO2 is more suitable than IMAC, showing significantly reduced numbers of signals attributed to unphosphorylated peptides (Figure 3C). Sample application to the ZrPO3 modified target, however, revealed that unspecific binding at the level of a 10fold excess of nonphosphorylated peptides was even lower than when using TiO2 (Figure 3D). In the mass spectra, no signals of nonspecifically binding peptides above S/N of 3:1 were registered, which demonstrates the superiority of the technique compared to the two standard methods IMAC and TiO2. This huge increase of specificity can most likely be explained by the strong exposition of the analytes to the washing solvents, due to the plain surface built by the monolayer. Furthermore, compared to IMAC and TiO2, ZrPO3 seems not to discriminate between different phosphopeptides contained in one sample. This effect can clearly be observed on the exemplary purification of R-casein phosphopeptides (Figure 2). Whereas the two standard techniques IMAC and TiO2 showed a bias to a certain subset of phosphopeptides, the results using the ZrPO3modified MALDI target were highly complementary (Table 2). Interestingly, the additionally detected phosphorylated peptides were in the higher mass range from 2500 to 3200 Da. Analysis of Protein Phosphorylation in Recombinant Human MAP Kinase1. In order to investigate the functionality of the ZrPO3-modified target with different sample types than classical standard proteins, the functionalized MALDI target Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

223

Figure 2. Comparison between different phosphopeptide purification technologies. A 100 fmol sample of R-casein digest was processed by Fe-IMAC (A), TiO2 (B), or on ZrPO3 surfaces (C), and eluates were analyzed by MALDI-MS. For illustration purpose,s spectra expansions are shown as insets in part C. Table 2. Phosphopeptides Detected after Purification by IMAC, TiO2, or on ZrPO3 Surfaces and Analysis by MALDI MSa MH+

Amino acid sequence

IMAC

TiO2

Phosphonate

1466.6 1539.6 1594.7 1660.8 1927.7 1943.2 1951.9 2678.0 2721.0 2935.2 3008.0 3088.0 3132.2

TVDMEpSTEVFTK EQLpSTpSEENSKK TVDMEpSTEVFTKK VPQLEIVPNpSAEER DIGpSEpSTEDQAMEDIK TKLpTEEEKNRLNFLK YKVPQLEIVPNpSAEER VNELpSKDIGpSEpSTEDQAMEDIK QMEAEpSIpSpSpSEEIVPNpSVEQK EKVNELpSKDIGpSEpSTEDQAM EDIK NANEEEYSIGpSpSpSEESAEVATEEVK NANEEEYSIGpSpSpSEEpSAEVATEEVK KNTMEHVpSpSpSEEpSIISQETYKQEK

267 91 443 730 928 752 -

272 29 421 781 835 1126 -

51 43 21 228 962 237 334 33 40 20 44 12 13

a Samples of 100 fmol of R-casein digest were processed and MS spectra were acquired as described under Experimantal Methods. The table shows averaged signal-to-noise ratios of five preparations. Phosphorylated amino acids are marked by “p”.

was probed with a digest of human MAPK1, which was produced in-house by recombinant expression in E. coli. Since activation of this kinase is dependent on its phosphorylation state,29 it was of interest to identify possible (auto)phosphorylation sites, even in the absence of upstream MAPK phosphorylating protein kinases in the bacterial host cells. Payne et al.30 found a characteristic “TEY” motif in the protein, on which both threonine and tyrosine residues were phosphorylated upon stimulation. Hence, our focus was on the detection (29) Boulton, T. G.; Cobb, M. H. Cell Regul. 1991, 2, 357–371.

224

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

of amino acid phosphorylation within this motif. Upon analyzing the MS spectrum of the digested protein, a sequence coverage of 34% was observed and enabled the detection of a dominant peak at 2143 m/z, which could be matched to the mass of the unphosphorylated peptide covering the TEY motif (Figure 4A). Subsequent MS/MS experiments confirmed the sequence VADPDHDHTGFLTEYVATR comprising the phosphorylation motif. Thus, these data suggest that the majority of the protein present was not phosphorylated at this site. However, a small peak at 2223 m/z pointing to the singly phosphorylated counterpart could also

Figure 3. Specificity of IMAC, TiO2,and ZrPO3 surfaces. A 100 fmol sample of phosphorylated angiotensin II was spiked in 1 pmol of BSA digest and purified using the investigated methods: (A) control, (B) IMAC, (C) TiO2, and (D) ZrPO3 surfaces. The phosphopeptide is marked with “P”.

be observed. Since the abundance of this peak was not sufficient to obtain reasonable fragmentation data, the digest was further processed on the ZrPO3-modified MALDI target and a MS spectrum of the eluate was recorded. Here, the signal of the putative singly phosphorylated peptide was strongly enhanced, whereas the majority of the nonphosphorylated peptides were removed by the washing steps (Figure 4B). In the next step, the peptide peak was selected for MS/MS and a CID fragment ion spectrum was recorded for identification (Figure 4C). Although DHB is in comparison to CHCA not the most suitable MALDI matrix for peptide fragmentation,31 a spectrum showing a number of clearly resolved peaks with good S/N ratios could be observed. A Mascot search of the peak list extracted from the fragment ion spectrum yielded one hit with a score of 43 for the phosphorylated peptide sequence, proving the presence of the phosphorylated MAPK1 isoform in the protein preparation. The indepth analysis of the MS/MS spectrum signals revealed the (30) Payne, D. M.; Rossomando, A. J.; Martino, P.; Erickson, A. K.; Her, J. H.; Shabanowitz, J.; Hunt, D. F.; Weber, M. J.; Sturgill, T. W. EMBO J. 1991, 10, 885–92. (31) Schulz, E.; Karas, M.; Rosu, F.; Gabelica, V. J. Am. Soc. Mass Spectrom. 2006, 17, 1005–1013.

neutral loss of H3PO4 and HPO3 from the precursor and also provided sequence information by matching five peaks to a short y-ion series (no loss of phosphoric acid) as well as internal fragments and immonium ions supporting the correct identification of the peptide (see annotations in Figure 4C). Furthermore, this fragment ion spectrum points most likely to tyrosine as phosphorylation site, which is known to be more stable than serine/threonine phosphorylation in MS/MS experiments.32 This was also supported by the observation that the putative neutral loss peak is by far not as dominant as in MS/MS spectra of serine/ threonine phosphorylated precursor ions (e.g., Figure 5C). For a more detailed interpretation of the fragment spectrum, please see the Supporting Information. In summary, the presence of the phosphorylated form of this kinase was demonstrated after successful enrichment of the phosphorylated peptide, which was not possible by MALDI MS analysis of the crude digest alone. Thus, the practicability of the modified target approach insofar uncharacterized samples could be proven. (32) Steen, H.; Ku ¨ ster, B.; Fernandez, M.; Pandey, A.; Mann, M. Anal. Chem. 2001, 73, 1440–1448.

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

225

Figure 4. Analysis of protein phosphorylation of recombinant human MAPK1 protein using the ZrPO3-modified MALDI target. A 2 µg sample of the recombinant protein was digested in-solution and applied to the sample carrier. (A) MS spectrum without further washing steps; (B) after processing, phosphopeptides are marked with “P”; (C) MS/MS spectrum of the 2223 m/z precursor enables validation as phosphopeptide. The peptide phosphorylation site was determined to be located at the tyrosine residue with high probability.

Identification of Phosphopeptides in 2D-Gel Spot Digest. Prior to the analysis of 2D-gel spots as the most demanding sample type in this set of experiments, the general compatibility of typical in-gel digestion protocols with phosphopeptide purification on the ZrPO3-modified sample carrier was tested successfully using ovalbumin separated by SDS-PAGE (see Results & Discussion in the Supporting Information and Figure 6 therein). 2D-PAGE is used in proteomic workflows to separate proteins due to its high resolution power. At the same time, the gel can be used for micropreparative purposes, since spots of interest can be cut out and digested for protein identification by MS. However, protein amounts are very limited, particularly in spots only visible after staining with highly sensitive fluorescent dyes, thus rendering phosphoprotein validation challenging.33 In the selected approach, intact phosphoproteins of Hela cells were first enriched by affinity chromatography and separated by 2D-PAGE, and finally, phosphoproteins were detected by staining with a specific fluorescent dye34 (Supporting Information, Figure 7). Several protein spots showing positive signals were picked, identified, and validated as phosphoproteins by detection of phosphopeptides after processing on the functionalized MALDI target. Among those, we found proteins such as lamin-A/C, elongation factor 2, heterogeneous (33) Eymann, C.; Dreisbach, A.; Albrecht, D.; Bernhardt, J.; Becher, D.; Gentner, S.; Tam, T.; Bu ¨ ttner, K.; Buurman, G.; Scharf, C.; Venz, S.; Vo ¨lker, U.; Hecker, M. Proteomics 2004, 4, 2849–2876. (34) Agrawal, G. K.; Thelen, J. J. Proteomics 2005, 5, 4684–4688.

226

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

nuclear ribonucleoprotein L, PDHA1, Vil2, tubulin β, nucleophosmin, and HSP B1. Here, the analysis of HSP B1 will be presented in more detail to exemplify the workflow: After digestion, the resulting supernatants were acidified and split for protein identification (peptide mass fingerprint and MS/MS) and on-target phosphopeptide enrichment. Figure 5A shows a MS spectrum with a distinct number of peaks of the crude protein digest. This information in combination with further MS/MS experiments allowed the unambiguous identification of HSP B1 (total Mascot score 374, sequence coverage 58%). After processing 5 µL of the remaining digest on the modified MALDI sample carrier, the peak pattern of the eluate changed dramatically in comparison to the unprocessed sample (Figure 5B). In this spectrum, a peak at 1662 m/z became the most prominent peak, which was a good phosphopeptide candidate because of the presence of a poorly resolved peak at 1579 m/z as a result of PSD. In the next step, all peaks occurring in the spectrum were selected for MS/MS experiments to investigate the phosphorylation state of the peptides. Exemplarily, Figure 5C shows the fragment ion spectrum of the 1662 m/z precursor: the strongest signals could be matched to masses resulting from the neutral loss of H3PO4 and HPO3 and the combination of both events [M - H3PO4 - HPO3]. With the help of this spectrum and the in-silico digest, it was possible to clearly assign two phosphorylation sites to the peptide ALSRQLSSGVSEIR, which was supported by the

Figure 5. Identification of HSP B1 and validation as phosphoprotein after separation of HeLa cell phosphoproteins on 2D-PAGE. A 2D-gel spot was picked, digested according to the Experimental Methods and identified as HSP B1. (A) MS spectrum of crude digest after preparation on Mass Spec Turbo Chips. (B) Enrichment of phosphopeptides after processing the digest on the modified MALDI target. Phosphopeptides are marked with “P”, and the corresponding PSD fragment is marked with “*”. (C) MS/MS spectrum of 1662 m/z precursor demonstrating the presence of two phosphorylation sites.

literature.35 In contrast to the MS/MS spectrum of hMAPK1 discussed above, the number of other fragment ion peaks except those resulting from neutral loss was very limited, but could be assigned to y4, y6, y7 of the suggested peptide sequence. Thus, the phosphorylation sites can be matched to the first and the second serine in the sequence. It should be noted, that the occurrence of strong neutral loss peaks in combination with much less pronounced b- and y-ion signals is well-known for phosphopeptides analyzed by MALDI-TOF MS/ MS and discussed in the literature.36,37 Probing less abundant peptide peaks present in the MS spectrum, two other phosphopeptides were verified, which could be assigned to the same site, but exhibiting two (2243 m/z, ALSRQLSSGVSEIRHTADR) and three (2585 m/z, ALSRQLSSGVSEIRHTADRWR) trypsin miscleavage sites, respectively. Due to these results, detection and validation of phosphopeptides using MALDI-TOF MS(/MS) is feasible, but precise mapping of phosphorylation sites is maybe hampered by the often occurring lack of indicative fragment ion peaks. This issue is partially solved by electron transfer dissociation (ETD) of phospho(35) Landry, J.; Lambert, H.; Zhou, M.; Lavoie, J. N.; Hickey, E.; Weber, L. A.; Anderson, C. W. J. Biol. Chem. 1992, 267, 794–803. (36) Schmidt, A.; Csaszar, E.; Ammerer, A.; Mechtler, K. Proteomics 2008, 8, 4577–4592. (37) Bennet, K. L.; Stensballe, A.; Podtelejnikov, A. V.; Moniatte, M.; Jensen, O. N. J. Mass Spectrom. 2002, 37, 179–190.

peptides, which however requires multiply charged ions.38 Hence, ESI-driven MS creating multiply charged peptide ions is the method of choice for these kinds of experiments. Thus, in the next step it was investigated whether the functionalized surface can also be used for ESI-MS/MS sample preparation. Usage of the ZrPO3-Modified for Phosphopeptide Sample Preparation Upfront to ESI-MS. To adapt the protocol to ESIMS-compatible sample preparation, only a modification of the elution step had to be introduced: Since phosphate is not compatible with electrospray ionization, the analytes were eluted under alkaline conditions using ammonium hydroxide solution. Here 150 mM (NH4)OH turned out to be ideal, as lower concentrations did not quantitatively remove the analyte from the surface, whereas more basic solutions did not increase the extent of eluted phosphopeptides. Using this approach, a monoand tetraphosphopeptide of β-casein were observed. Due to the higher S/N of the monophosphopeptide, this peptide peak was chosen for MS/MS and fragmented under CID conditions. The fragmentation pattern reveals an almost complete amino acid sequence and enables the unambiguous determination of the (38) McAlister, G. C.; Berggren, W. T.; Griep-Raming, J.; Horning, S.; Makarov, A.; Phanstiel, D.; Stafford, G.; Swaney, D. L.; Syka, J. E.; Zabrouskov, V.; Coon, J. J. J Proteome Res. 2008, 7, 3127–3136.

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

227

Figure 6. ESI-MS/MS spectrum of 1031.7 m/z precursor after purification on ZrPO3 surfaces. A 500 fmol sample of a β-casein digest was processed on a ZrPO3-modified MALDI target and eluted with (NH4)OH, and spectra were acquired on a LCQ ion trap applying static nanospray. The putative phosphopeptide FQpSEEQQQTEDELQDK (2061 m/z) was selected for fragmentation, and MS/MS scans were acquired. Fragment ion peaks are annotated, allowing unambiguous location of the phosphorylation site.

exact phosphorylation site (Figure 6). Although this approach is, due to the remaining manual transfer step of the analyte, not fully streamlined, the general practicability for electrospray phosphopeptide purification could be demonstrated. Moreover, this combination offers the possibility to apply also other fragmentation strategies (MSn, MSA, ETD) than TOF-TOF MS. However, to further develop the usability for ESI, also online coupling to a chip-based infusion system and/or downstream nano HPLC, or even the transfer of this particular SAM approach from a MALDI plate to a bead-sized format would be a desirable. CONCLUSIONS A novel technological platform for direct purification and analysis of phosphopeptides on a prestructured MALDI plate from different sample types has been presented. The results from the reproducibility experiments demonstrated that the synthesis and subsequent attachment of the ZrPO3-terminated SAM onto the MALDI target is highly reproducible and allows the highly sensitive, selective, and robust assessment of protein phosphorylation. Its purification efficiency was shown to be superior to standard techniques such as IMAC and TiO2. It significantly facilitates the analysis of phosphoprotein digests and gives complementary results compared to the established methods. Beyond the analysis of pure standard proteins, the applicability to real-life samples possessing limited analyte amounts and substoichometric presence of phosphopeptides was demon-

228

Analytical Chemistry, Vol. 82, No. 1, January 1, 2010

strated successfully. Overall, these features qualify the zirconium-phosphonate surface for implementation in phosphoproteomics applications. Although it is primarily used in direct MALDI MS, this tool has the potential to become a substantial element for phosphopeptide purification in various proteomic workflows like LC-MALDI setups or automated purification of 2D-spot samples. In terms of suitable MS instrumentation downstream of this “chip-like” phosphopeptide purification tool, also ESI-MS was demonstrated to be an attractive application. In a large-scale phosphoproteomics investigation, the comparison of MALDI and complementary ESI results using this technique should be of high interest to detect a maximum of phosphorylation sites. ACKNOWLEDGMENT We are grateful to John Walker, Qiagen Inc., for manufacturing the ZrPO3-modified MALDI targets and Jan Petzel, Qiagen GmbH, for critically reading the manuscript. SUPPORTING INFORMATION AVAILABLE Additional information as outlined in the text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review August 5, 2009. Accepted November 18, 2009. AC9017583