Phosphopeptide Enrichment Using MALDI Plates Modified with High

Jun 26, 2008 - ... of Chemistry and Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, Michigan 48824. Anal. C...
0 downloads 0 Views 403KB Size
Download PDF
Anal. Chem. 2008, 80, 5727–5735

Phosphopeptide Enrichment Using MALDI Plates Modified with High-Capacity Polymer Brushes Jamie D. Dunn,† Elizabeth A. Igrisan,† Amanda M. Palumbo,† Gavin E. Reid,†,‡ and Merlin L. Bruening*,† Department of Chemistry and Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, Michigan 48824 Matrix-assisted laser desorption/ionization plates coated with poly(2-hydroxyethyl methacrylate) (PHEMA) brushes that are derivatized with Fe(III)-nitrilotriacetate (NTA) complexes allow selective, efficient phosphopeptide enrichment prior to analysis by mass spectrometry (MS). Fe(III)-NTA-PHEMA brushes (60 nm thick) have a phosphopeptide binding capacity of 0.6 µg/cm2 and exhibit phosphopeptide recoveries of over 70%, whereas much thinner polymer films containing Fe(III)-NTA afford a recovery of only 20%, and a monolayer of Fe(III)-NTA shows a recovery of just 10%. Recoveries are determined by comparing signals from enriched unlabeled phosphopeptides with those of their deuterium-labeled analogues that were added to the plate just prior to addition of matrix. Mass spectra of phosphopeptide-containing samples enriched using Fe(III)-NTA-PHEMA-modified plates also demonstrate higher recoveries or fewer interfering peaks than corresponding spectra obtained with enrichment using several commercially available Fe(III)containing films and resins or metal oxide materials. When analyzing tryptic digests of β-casein, the Fe(III)NTA-PHEMA brushes allow detection of as little as 15 fmol of phosphopeptide. Moreover, with both ovalbumin and β-casein digests, phosphopeptide signals dominate the mass spectra obtained using these modified plates. Protein phosphorylation is important in a host of cell regulation mechanisms, and changes in native phosphorylation states are involved in several diseases including cancer.1 Thus, identification of phosphorylation sites and quantitation of phosphorylation is of great interest for understanding biochemical pathways and disease states. Mass spectrometry (MS) has emerged as the premier tool for examining phosphorylation, but unfortunately a low abundance of phosphorylated peptides and suppressed ion signals for phosphopeptides in the presence of nonphosphorylated species make identification of phosphorylation sites difficult. Hence, a number of strategies have been developed for phosphopeptide enrichment prior to interrogation of phosphorylation by MS.2–17 Introduced in 1975 by Porath,18 immobilized metal affinity chromatography (IMAC), which relies on interactions between * Author to whom correspondence should be addressed. E-mail: bruening@ chemistry.msu.edu. Phone: 517-355-9715x237. Fax: 517-353-1793. † Department of Chemistry. ‡ Department of Biochemistry and Molecular Biology. (1) Johnson, S. A.; Hunter, T. Nat. Methods 2005, 2, 17–25. 10.1021/ac702472j CCC: $40.75  2008 American Chemical Society Published on Web 06/26/2008

metal-ligand complexes and specific functional groups such as phosphates, is the most common method used to enrich phosphopeptides. More recently, ZrO2 and TiO2 resins have also shown great promise for enrichment of phosphorylated species.3,6,10,19–22 At low pH, these metal oxides have a positively charged surface that selectively adsorbs phosphopeptides and exhibits outstanding enrichment behavior.10,23,24 As an example, Pinkse et al. showed that 90% of 125 fmol of a phosphorylated peptide could be recovered from its nonphosphorylated counterpart in a simple solution.3 Other studies showed that the effectiveness of metal oxide resins in phosphopeptide enrichment depends on loading (2) Lansdell, T. A.; Tepe, J. J. Tetrahedron Lett. 2004, 45, 91–93. (3) Pinkse, M. W. H.; Uitto, P. M.; Hilhorst, M. J.; Ooms, B.; Heck, A. J. R. Anal. Chem. 2004, 76, 3935–3943. (4) Xu, Y. D.; Bruening, M. L.; Watson, J. T. Anal. Chem. 2004, 76, 3106– 3111. (5) Chen, C. T.; Chen, Y. C. Anal. Chem. 2005, 77, 5912–5919. (6) Larsen, M. R.; Thingholm, T. E.; Jensen, O. N.; Roepstorff, P.; Jorgensen, T. J. D. Mol. Cell. Proteomics 2005, 4, 873–886. (7) Shen, J. W.; Ahmed, T.; Vogt, A.; Wang, J. Y.; Severin, J.; Smith, R.; Dorwin, S.; Johnson, R.; Harlan, J.; Holzman, T. Anal. Biochem. 2005, 345, 258– 269. (8) Belisle, C. M.; Chen, I. Y.; Mirshad, J. K.; Roth, U.; Steinert, K.; Walker, J. A., 54th ASMS Conference on Mass Spectrometry and Allied Topics, Seattle, WA, May 28-June 1, 2006. (9) Dunn, J. D.; Watson, J. T.; Bruening, M. L. Anal. Chem. 2006, 78, 1574– 1580. (10) Kweon, H. K.; Håkansson, K. Anal. Chem. 2006, 78, 1743–1749. (11) Lin, H. Y.; Chen, C. T.; Chen, Y. C. Anal. Chem. 2006, 78, 6873–6878. (12) Warthaka, M.; Karwowska-Desaulniers, P.; Pflum, M. K. H. ACS Chem. Biol. 2006, 1, 697–701. (13) Xu, S. Y.; Zhou, H. J.; Pan, C. S.; Fu, Y.; Zhang, Y.; Li, X.; Ye, M. L.; Zou, H. F. Rapid Commun. Mass Spectrom. 2006, 20, 1769–1775. (14) Xu, X. Q.; Deng, C. H.; Gao, M. X.; Yu, W. J.; Yang, P. Y.; Zhang, X. M. Adv. Mater. 2006, 18, 3289–3293. (15) Zhou, H. J.; Xu, S. Y.; Ye, M. L.; Feng, S.; Pan, C.; Jiang, X. G.; Li, X.; Han, G. H.; Fu, Y.; Zou, H. J. Proteome Res. 2006, 5, 2431–2437. (16) Chen, C. T.; Chen, W. Y.; Tsai, P. J.; Chien, K. Y.; Yu, J. S.; Chen, Y. C. J. Proteome Res. 2007, 6, 316–325. (17) Lo, C. Y.; Chen, W. Y.; Chen, C. T.; Chen, Y. C. J. Proteome Res. 2007, 6, 887–893. (18) Porath, J.; Carlsson, J.; Olsson, I.; Belfrage, G. Nature 1975, 258, 598– 599. (19) Thingholm, T. E.; Jorgensen, T. J. D.; Jensen, O. N.; Larsen, M. R. Nat. Protoc. 2006, 1, 1929–1935. (20) Jensen, S. S.; Larsen, M. R. Rapid Commun. Mass Spectrom. 2007, 21, 3635–3645. (21) Yu, L. R.; Zhu, Z. Y.; Chan, K. C.; Issaq, H. J.; Dimitrov, D. S.; Veenstra, T. D. J. Proteome Res. 2007, 6, 4150–4162. (22) Sugiyama, N.; Masuda, T.; Shinoda, K.; Nakamura, A.; Tomita, M.; Ishihama, Y. Mol. Cell. Proteomics 2007, 6, 1103–1109. (23) Blackwell, J. A.; Carr, P. W. J. Chromatogr. 1991, 549, 43–57. (24) Blackwell, J. A.; Carr, P. W. J. Chromatogr. 1991, 549, 59–75.

Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

5727

and washing solutions, and typical recoveries ranged from 50 to 80%.6,21,22 Immobilization of either IMAC materials or metal oxides on matrix-assisted laser desorption/ionization (MALDI) plates allows for on-probe purification that can lead to high-throughput analyses and reduce sample loss due to adsorption on container walls. In such analyses, a drop of sample containing nonphosphorylated and phosphorylated species is incubated on the MALDI plate, which is then rinsed to remove the unbound species prior to addition of matrix and subsequent analysis using MALDI-MS. Such purification methods are attractive for rapid analysis of microliter quantities of simple mixtures such as digests of immunoprecipitates. (More complicated samples such as whole cell lysates will require chromatographic methods for identification of large numbers of phosphopeptides.21) Previously examined surface modifications for on-plate enrichments include immobilization of TiO2 on glass, stainless steel, or polymeric substrates,11,25–27 formation of Fe(III) complexes with nitrilotriacetate (NTA) or iminodiacetate monolayers on Au or Si surfaces,7,13 synthesis of Zr(IV)-phosphonate monolayers on porous silicon,15 and functionalization of stainless steel with zirconium oxide.28 Modification of plates with polymer coating can provide high adsorption capacities,29–32 and two recent studies demonstrated the use of poly(acrylic acid) (PAA) films for onplate purification and concentration of phosphopeptides.4,9 Such films provide enhanced capacities relative to monolayers, and derivatization of PAA with Fe(III)-NTA complexes allows for efficient enrichment of phosphopeptides from protein digests.9 However, the binding capacity remains limited, and a low capacity can lead to low signals as well as selective binding of multiphosphorylated peptides.33–36 This manuscript describes our efforts to increase the binding capacity of modified MALDI plates by coating substrates with poly(2-hydroxyethyl methacrylate) (PHEMA) brushes derivatized with Fe(III)-NTA complexes. Such films can have thicknesses > 100 nm,37 and PHEMA derivatized with Cu(II)-NTA or Ni(II)-NTA complexes is capable of binding the equivalent of many mono(25) Ekstrom, S.; Wallman, L.; Helldin, G.; Nilsson, J.; Marko-Varga, G.; Laurell, T. J. Mass Spectrom. 2007, 42, 1445–1452. (26) Qiao, L.; Roussel, C.; Wan, J. J.; Yang, P. Y.; Girault, H. H.; Liu, B. H. J. Proteome Res. 2007, 6, 4763–4769. (27) Tan, F.; Zhang, Y. J.; Wang, J. L.; Wei, J. Y.; Qin, P. B.; Cai, Y.; Qian, X. H. Rapid Commun. Mass Spectrom. 2007, 21, 2407–2414. (28) Blacken, G. R.; Volny, M.; Vaisar, T.; Sadilek, M.; Turecek, F. Anal. Chem. 2007, 79, 5449–5456. (29) Brockman, A. H.; Orlando, R. Anal. Chem. 1995, 67, 4581–4585. (30) Brockman, A. H.; Orlando, R. Rapid Commun. Mass Spectrom. 1996, 10, 1688–1692. (31) Li, M.; Fernando, G.; van Waasbergen, L. G.; Cheng, X.; Ratner, B. D.; Kinsel, G. R. Anal. Chem. 2007, 79, 6840–6844. (32) Peterson, D. S.; Luo, Q. Z.; Hilder, E. F.; Svec, F.; Frechet, J. M. J. Rapid Commun. Mass Spectrom. 2004, 18, 1504–1512. (33) 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. (34) Ndassa, Y. M.; Orsi, C.; Marto, J. A.; Chen, S.; Ross, M. M. J. Proteome Res. 2006, 5, 2789–2799. (35) Nousiainen, M.; Sillje, H. H. W.; Sauer, G.; Nigg, E. A.; Korner, R. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 5391–5396. (36) Nuhse, T. S.; Stensballe, A.; Jensen, O. N.; Peck, S. C. Mol. Cell. Proteomics 2003, 2, 1234–1243. (37) Huang, W. X.; Kim, J. B.; Bruening, M. L.; Baker, G. L. Macromolecules 2002, 35, 1175–1179.

5728

Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

layers of protein.38–40 Use of an Fe(III)-NTA complex allows capture of phosphopeptides from proteolytic digests for direct analysis by MALDI-MS. This work compares phosphopeptide enrichment by Fe(III)-NTA-PHEMA-modified plates with enrichment using gold plates modified with either monolayers of Fe(III)NTA or relatively thin Fe(III)-PAA-NTA films. The high capacity of polymer brushes greatly enhances phosphopeptide recovery and signal. Enrichment with commercially available IMAC and metal oxide materials that can be applied to small volumes (1-2 µL) and femtomole levels of phosphopeptide is also examined. To make the comparison quantitative, signals from a recovered peptide are compared with signals from its partially deuterated analogue that was added to the sample spot after enrichment and prior to addition of matrix. The brush-modified plates show higher phosphopeptide signals than any of the enrichment techniques that were examined except adsorption on microtips in TiO2. Brushes and TiO2 show similar recoveries, but the spectra of materials recovered using TiO2 microtips contain a greater number of unwanted peaks, presumably due to leached resin. The Fe(III)-NTA-PHEMA brushes also show high selectivity and sensitivity in the enrichment of phosphoproteins from β-casein and ovalbumin digests. EXPERIMENTAL SECTION Materials and Solutions. Bovine β-casein, chicken egg ovalbumin, bovine serum albumin (BSA, MW ∼66 kDa), rabbit phosphorylase b (phos b, MW ∼62 kDa), and pig esterase (MW ∼97 kDa) were purchased from Sigma and digested using sequencing grade modified trypsin obtained from Promega. Other reagents employed in the digestion include Tris-HCl (Invitrogen), urea (J.T. Baker), 1,4-dithio-DL-threitol (BioChemika), iodoacetamide (Sigma), and ammonium bicarbonate (Columbus Chemical Industries). Human angiotensin II (DRVYIHPF) and phosphorylated angiotensin II (DRVpYIHPF) were acquired from Calbiochem. CH3CH2CO-LFTGHPEpSLEK (H5 peptide) and CD3CD2COLFTGHPEpSLEK (D5 peptide) were prepared using manual stepwise Fmoc-based solid-phase peptide synthesis on Wang resins as described in the Supporting Information. CH3CH2COLFpTGHPEpSLEK (2PH5 peptide) and CD3CD2CO-LFpTGHPEpSLEK (deuterated internal standard used for determining the recovery of 2PH5 peptide) were synthesized in a similar fashion to the H5 and D5 peptides. HPLC grade acetonitrile (EMD), HPLC grade methanol (Sigma), isopropyl alcohol (EMD), and ammonium hydroxide (∼28%, Columbus Chemical Industries) were used to clean the conventional stainless steel MALDI plate according to the “deep cleaning” procedure by Thermo Scientific. In this cleaning, the plate was rinsed with isopropanol and methanol, sonicated for 10 min in a solution containing 451 mL of acetonitrile, 451 mL of water, and 108 mL of NH4OH, and finally rinsed with methanol and water and dried. To create a hydrophobic surface on the plate, HPLC grade hexane (J.T. Baker) was rubbed over the plate, which was then rinsed with hexane, dried, and stored in a nitrogen (38) Dai, J. H.; Bao, Z. Y.; Sun, L.; Hong, S. U.; Baker, G. L.; Bruening, M. L. Langmuir 2006, 22, 4274–4281. (39) Jain, P.; Sun, L.; Dai, J.; Baker, G. L.; Bruening, M. L. Biomacromolecules 2007, 8, 3102–3107. (40) Sun, L.; Dai, J. H.; Baker, G. L.; Bruening, M. L. Chem. Mater. 2006, 18, 4033–4039.

glovebag when not in use. The matrix used in all MS experiments was 2,5-dihydroxybenzoic acid (2,5-DHB, Aldrich) in 1:1 HPLC grade acetonitrile (EMD)/1% aqueous phosphoric acid (Aldrich). For conventional MALDI analyses, 1 µL of the 10 mg/mL 2,5DHB solution was applied to the sample wells and mixed on-plate with 1 µL of the protein digest. Protein Digestion. Twenty 100 µg samples of each phosphoprotein (bovine β-casein and chicken ovalbumin) were dissolved separately in 20 µL of 6 M urea containing 50 mM TrisHCl (5 µg protein/µL). To treat all digests equally, even though β-casein does not contain disulfide linkages, 5 µL of 10 mM 1,4dithio-DL-threitol (DTT) was added to all the protein solutions prior to heating in a water bath at 70 °C for 1 h to cleave any disulfide bonds. After cooling, 160 µL of 50 mM ammonium bicarbonate was added to each protein solution, and a 10 µL aliquot of 100 mM iodoacetamide was added to the solutions, which were placed in the dark for 1 h. Subsequently, 10 µL of 0.5 µg/µL modified trypsin was added to each protein, and the solutions were incubated for ∼16 h at 37 °C. The digestions were quenched with addition of sufficient glacial acetic acid (∼11 µL) to achieve a 5% solution and stored in a -70 °C freezer until use. Typically, DTT is removed prior to application to IMAC materials since it is a strong reducing agent that can interfere with the affinity resin (i.e., it can reduce the metal ion),41,42 but it was not removed from the digest nor were any purification methods used prior to application to the polymer-modified plates. In preparing a nonphosphorylated protein digest mixture, a solution containing BSA, phos b, and esterase (100 µg of each protein) was digested in a similar fashion as described above except 30 µL of 100 mM iodoacetamide and 30 µL of 0.5 µg/µL modified trypsin were used. The digestions were quenched with addition of sufficient glacial acetic acid (∼12 µL) to achieve a 5% acetic acid solution, dried using a Speedvac, and stored in a -20 °C freezer until use. Dried protein digest mixture was dissolved in 5% acetic acid prior to its application to the Fe(III)-NTA-MUA-, Fe(III)-NTA-PAA-, and Fe(III)-NTA-PHEMA-modified MALDI plates. The digest solution applied to the modified plates contained ∼24 ng of each protein: BSA, phos b, and esterase (1 pmol total protein). The 1 pmol/µL protein digest mixture that was used in all of the enrichment analyses, except where noted, contained approximately 29 mM urea, 0.24 mM Tris-HCl, 0.012 mM DTT, 1.9 mM NH4HCO3, and 0.71 mM iodoacetamide. Quantitation of Phosphopeptide Binding Using Ellipsometry. Three gold wafers modified with Fe(III)-NTA-PHEMA films with thicknesses of 61, 55, and 66 nm were immersed in phosphoangiotensin II (pA) solutions in a 33 mm diameter Petri dish for 1 h. The pA solutions contained 0.002, 0.003, 0.004, 0.005, 0.01, 0.05, 0.1, or 0.2 mg pA/mL in 5% acetic acid. After incubation in one of the pA solutions, the wafers were rinsed thoroughly with 15 mL of 3:30:67 acetic acid/acetonitrile/water, followed by 6 mL of 250 mM acetic acid in 30% acetonitrile and then dried with a stream of nitrogen. Film thicknesses on these wafers were determined in triplicate using ellipsometry. The wafers were rinsed with ethanol and dried with nitrogen, and thicknesses were measured once again. The amount of bound pA was assumed to be directly proportional to the increase in ellipsometric thickness

after exposure to the pA solution. In all cases, the amount bound to the surface was less than 10% of the amount of peptide initially in solution. Between exposures to different pA solutions, the films were immersed in an ethylenediaminetetraacetic acid (EDTA, Aldrich) solution (50 mM, pH 7.3) for 30 min to remove Fe(III) and pA, rinsed with water, and immersed in 250 mM acetic acid in 30% acetonitrile for 15 min, followed by rinsing with ethanol and drying with nitrogen. The film was regenerated by exposure to 100 mM FeCl3, followed by rinsing with ethanol, drying with nitrogen, a 15-min immersion in 250 mM acetic acid in 30% acetonitrile, and drying with nitrogen. Thicknesses were measured using the ellipsometer. The films were rinsed with ethanol, and thicknesses were measured again. Thickness differences were calculated both after the 250 mM acetic acid in 30% acetonitrile rinse and after the ethanol rinse and averaged, and results after the two rinses were similar. A solution of 0.2 mg/mL nonphosphorylated angiotensin II in 5% acetic acid was used as a control and was applied to the three Fe(III)-NTA-PHEMA brushes as described for pA. Protocol for Using the Polymer-Modified MALDI Plates. For the analysis of protein digests, 1 µL of digest was spotted into the 2 mm diameter well of the Fe(III)-NTA-PHEMA-modified Au-coated Si wafer. The samples were then incubated on the polymer-modified plate for 1 h (or 10 min), rinsed with 15-20 mL of 3:30:67 glacial acetic acid/acetonitrile/water solution, and dried with a stream of N2 to remove any remaining rinse solution. (During incubation, 0.5 µL aliquots of 5% acetic acid were periodically added to the droplet to compensate for evaporation.) After incubation, rinsing, and drying, a 1 µL droplet of 1% phosphoric acid was added to each well immediately followed by 0.25 µL of 40 mg/mL 2,5-DHB solution in 1:1 acetonitrile/1% phosphoric acid. After crystallization of the matrix, the wafer was secured to the modified stainless steel MALDI sample plate using double-sided tape for subsequent MS analysis.43 Protocols for Determining Phosphopeptide Recovery. Stock solutions containing 100 pmol of either H5 or D5 peptides were prepared in 200 µL of deionized water, and all working solutions (125, 62, 31, 16, 8 fmol/µL) were made from the stock solutions using serial dilutions with deionized water. For on-plate enrichment techniques (Fe(III)-NTA-MUA, Fe(III)-NTA-PAA, and Fe(III)-NTA-PHEMA), recovery of the H5 peptide in the absence of nonphosphorylated protein digest was determined using the following protocol. First, 1 µL of 5% acetic acid was applied to the 2 mm wells on the plates prior to adding 1 µL of aqueous 125 fmol/µL H5 peptide. The samples were incubated for 1 h (or 10 min in a few cases), and 0.5 µL aliquots of 5% acetic acid were reapplied during the incubation as needed to prevent the sample from drying. The plates were then rinsed with 15-20 mL of 3:30: 67 acetic acid/acetonitrile/water solution and completely dried with nitrogen. A 1 µL aliquot of 125 fmol/µL D5 peptide was applied to the 2 mm well on the plate as an internal standard, followed by 1 µL of 1% phosphoric acid and 0.25 µL of 40 mg/mL 2,5-DHB in 1:1 acetonitrile/1% phosphoric acid. Analysis using MALDI-MS was then performed to compare signals due to the H5 and D5 peptides. Recovery of the H5 peptide in the presence of a digest of BSA, phos b, and esterase was determined using a

(41) AbDSerotec. http://www.ab-direct.com/about/faq-562.html, 2007. (42) Prochem. http://www.protein-chem.com/faqs-imac.html, 2007.

(43) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford, G.; Yost, R. A. Int. J. Mass Spectrom. 2007, 260, 166–176.

Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

5729

similar protocol with the exception that 1 µL of the digest mixture, which contained 1 pmol of total protein, was first spotted into the wells, instead of the 5% acetic acid, prior to adding 1 µL of 125 fmol/µL H5 peptide. In the case of ZipTipMC pipet tips and Qiagen IMAC MALDI plates, the phosphopeptide enrichment protocol provided by the manufacturer was followed. Details are provided in the Supporting Information. Procedures for utilizing metal oxide materials were adapted from the literature and are also given in the Supporting Information.21,22,44 Instrumentation and Data Analysis. The MALDI mass spectra shown here are representative of many similar spectra, as each experiment was repeated at least three times using different modified plates. All mass spectra were obtained using a MALDI linear ion trap mass spectrometer (Thermo vMALDI LTQ XL). Protein Prospector (http://prospector.ucsf.edu) was used to make preliminary assignments to intact precursor ion signals in the mass spectra, and assignments were confirmed using low energy collision-induced dissociation-tandem mass spectrometry (CID-MS/MS). The Protein Prospector Web site was also used to help confirm manual assignment of product ions observed in the tandem mass spectra. Typically, MS/MS, wideband MS/MS, and MS3 were implemented for the characterization and sequence analysis of phosphopeptides. The thicknesses of polymer films on modified plates were determined using a rotating analyzer spectroscopic ellipsometer (J. A. Woollam, M-44), assuming a film refractive index of 1.5. Reflectance FT-IR spectra of the modified plates were collected using a Nicolet Magna 560 spectrophotometer with a Pike grazing angle (80°) accessory. RESULTS AND DISCUSSION Synthesis and Characterization of Fe(III)-NTA-PHEMAModified Plates. Figure 1 shows the procedure for preparing the Fe(III)-NTA-PHEMA-modified MALDI plates using atom transfer radical polymerization. The polymerization and derivatization with NTA were described in detail previously.40 The fabrication steps are (1) adsorption of a monolayer of mercaptoundecanol on gold, (2) formation of an immobilized initiator, (3) growth of PHEMA brushes, (4) conversion of PHEMA to a film containing carboxylic acid groups, (5) creation of an active ester, (6) reaction with aminobutyl NTA to give an immobilized NTA derivative, and (7) formation of the Fe(III)-NTA complex. Reflectance FT-IR spectroscopy verified the derivatization process40 (see Figure S-1 of the Supporting Information). Quantitation of Phosphopeptide Binding to Fe(III)-NTAPHEMA-Modified Plates Using Ellipsometry. To determine the phosphopeptide binding capacity of Fe(III)-NTA-PHEMA brushes on gold-coated substrates, ellipsometry was used to measure the thicknesses of three Fe(III)-NTA-PHEMA films on Au before and after immersion in phosphoangiotensin (pA) solutions for 1 h. The increase in ellipsometric thickness (after rinsing) was assumed to be entirely due to bound pA, and Figure 2 shows the resulting pA adsorption isotherm for films with an average initial thickness of ∼60 nm. (A pA density of 1 mg/mL was employed to convert thickness increases to surface coverage, and the large error bars stem in part from the fact that even at (44) Simon, E.; Young, M. A.; Andrews, P. C. 55th ASMS Conference on Mass Spectrometry and Allied Topics, Indianapolis, IN, June 3-7, 2007.

5730

Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

Figure 1. Fabrication of an Fe(III)-NTA-PHEMA-modified gold substrate. Abbreviations are defined as follows: triethylamine (TEA), 2-bromoisobutyryl bromide (BIB), 2-hydroxyethyl methacrylate (HEMA), 2,2′bipyridine(bpy),4-dimethylaminopyridine(DMAP),N-(3-dimethylaminopropyl)N′-ethylcarbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS), NR,NR-bis(carboxymethyl)-L-lysine hydrate (aminobutyl NTA).

saturation, the increase in film thickness is only 6 nm on a 60 nm film.) The maximum amount of pA bound is about 0.6 µg/cm2, or 18 pmol in the area of a 2 mm diameter sample spot. Nonphosphorylated angiotensin II was also applied to the polymermodified plate to see whether its histidine and glutamic acid residues would cause nonspecific binding, and no binding of this peptide was detected even at a solution concentration of 0.2 mg/ mL. (Note that the ellipsometric method is not very sensitive and would not detect the low levels of nonspecific adsorption that may be observable by mass spectrometry.)

Figure 3. Structures of (a) Fe(III)-NTA-MUA- and (b) Fe(III)-NTAPAA-modified gold plates. Figure 2. Plot of the amount of pA bound to Fe(III)-NTA-PHEMA films on gold substrates as a function of pA concentration. The solid line represents the fit to the Langmuir adsorption isotherm, and the inset shows an expanded view of the data at low pA concentrations.

Figure 2 shows the fit of the binding data to the Langmuir adsorption isotherm, which is described by eq 1, where x (µg/ cm2) is the experimental amount of pA bound to the polymer film, xm (µg/cm2) is the maximum amount of pA that will bind to the polymer film, c (mg/mL) is the concentration of pA in solution, and K (mL/mg) is the Langmuir constant for the formation of the pA-Fe(III)-NTA complex. x)

xmc -1

K

+c

(1)

The fit yields a value of 0.6 µg/cm2 (0.5 nmol/cm2) for xm and a K of 250 mL/mg (2.8 × 108 mL/mol), but unfortunately, the uncertainty in the data and the fit are not sufficient to predict the amount of pA that would be bound to the film at the low concentrations used in mass spectrometry. However, the phosphopeptide recoveries of 70% described below for solutions initially containing 125 fmol of the H5 peptide in 2 µL of solution also give a K of about 250 mL/mg, assuming the binding capacity in a 2 mm diameter spot is 0.6 µg/cm2. Recovery of the H5 Peptide Using Monolayer-Modified and Polymer-Modified MALDI Plates. To examine whether relatively thick (∼500 Å) brushes have advantages over monolayer and thin polymer films, the enrichment capabilities of Fe(III)-NTAPHEMA-modified gold plates were compared with enrichment on plates modified with a monolayer of Fe(III)-NTA and a thin polymer film derivatized with Fe(III)-NTA. In these experiments, the phosphopeptide-binding moiety is kept constant, but the film architecture is varied. In brief, the monolayer of Fe(III)-NTA was immobilized onto a gold plate by forming a self-assembled monolayer (SAM) of mercaptoundecanoic acid (MUA) on the surface and coupling the amino-terminated NTA derivative to the MUA after activation of carboxylic acid groups using EDC/NHS. The monolayer was charged with Fe(III) by immersing the substrate into an aqueous ferric chloride solution, and these films will be referred to as Fe(III)-NTA-MUA. Thin polymer films were prepared in a similar way by functionalizing surface-grafted poly(acrylic acid) (PAA) with Fe(III)-NTA (Fe(III)-NTA-PAAmodified plates). These plates are essentially the same as those described in a prior publication, but the procedure was slightly

Figure 4. Protocol used to prepare samples for determination of the recovery of the H5 peptide from a protein digest mixture using Fe(III)-NTA-MUA-modified, Fe(III)-NTA-PAA-modified, or Fe(III)-NTAPHEMA-modified plates.

modified by using EDC/NHS instead of ethyl chloroformate/ 4-methylmorpholine to activate the carboxylic acid groups of PAA.9 Structures of the Fe(III)-NTA-MUA and Fe(III)-NTA-PAA films are given in Figure 3. The plates differ greatly in the ellipsometric thicknesses of the Fe(III)-NTA films, which were ∼10, ∼30, and ∼500 Å for Fe(III)-NTA-MUA, Fe(III)-NTA-PAA, and Fe(III)-NTAPHEMA, respectively. The different Fe(III)-NTA-modified plates were compared by examining their recovery of 125 fmol of the H5 peptide (CH3CH2COLFTGHPEpSLEK) from solutions containing a mixture of digested BSA, phos b, and esterase. The protocol for preparing samples for determining the recovery of the H5 peptide from a protein digest mixture is shown in Figure 4 and described in detail in the Experimental Section. By comparison of the relative signals due to H5 and D5 peptides to a calibration curve, which showed essentially equal sensitivities of MALDI-MS for the H5 and D5 peptides (Supporting Information, Figure S-2), the recovery of the H5 peptide was determined. The conventional MALDI analysis of 1 pmol of protein digest mixture spiked with 125 fmol of the H5 peptide and 125 fmol of the D5 peptide is shown in Figure 5a. The MALDI mass spectra in parts b-d of Figure 5 demonstrate the recovery of the H5 peptide from the protein digest mixture using Fe(III)-NTA-MUA-, Fe(III)-NTA-PAA-, and Fe(III)-NTA-PHEMA-modified plates, respectively. In the conventional spectrum, many signals of digested Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

5731

Figure 5. Mass spectra of a 1 pmol protein digest mixture containing 125 fmol of the H5 peptide. An amount of 125 fmol of D5 peptide was added as an internal standard. Spectra were obtained using (a) conventional MALDI-MS analysis, (b) an Fe(III)-NTA-MUA-modified plate, (c) an Fe(III)-NTA-PAA-modified plate, and (d) an Fe(III)-NTAPHEMA-modified plate.

peptides are present, but no signals were observed corresponding to the H5 and D5 peptides. Evidently, ionization of these peptides was suppressed by the excess of digest proteins (Supporting Information, Figure S-3). The Fe(III)-NTA monolayer-functionalized plate showed minimal signal from digest peptides but was only capable of recovering 9 ± 2% of the H5 peptide (Figure 5b). There was a slight improvement when the relatively thin (∼30 Å) Fe(III)-NTA-PAA film was utilized, as recovery increased to 23 ± 12%. As shown in the mass spectrum in Figure 5c, however, there were a number of peaks (m/z 547, 696, 902, 995, 1440, and 1852) due to peptides from the protein digest mixture. The binding of less nonphosphorylated peptide to the Fe(III)-NTA-MUAmodified plate compared to the Fe(III)-NTA-PAA-modified plate most likely occurs because the binding capacity of the monolayer is less than that of the Fe(III)-NTA-PAA film. As seen in Figure 5d, the Fe(III)-NTA-PHEMA film showed a remarkable 73 ± 12% recovery of the H5 peptide, even in the large excess of digest peptides. In fact, the recovery was the same in the presence and the absence of the protein digest mixture. The mass spectrum of the H5 peptide-spiked digest obtained using the Fe(III)-NTA-PHEMA film does, however, contain several ions corresponding to a number of nonphosphorylated peptides (m/z values of 665, 995, 1178, 1250, 1420, 1440, 1533, and 1902). Table S-1 (Supporting Information) shows that many of these peptides contain histidine residues, which could contribute to their binding to the modified plate. Additionally, there was an irregular distribution of the isotopic peaks for the D5 peptide due to a larger than expected signal at m/z 1399.7. This peak was determined to be due in part to the BSA peptide TVMENFVAFVDE by using CID MS/MS as shown in Figure S-4 (Supporting Information). For most of the analyses presented here, a 1 h sample incubation period was utilized. However, this limits the utility of MALDI-MS as a high-throughput technique. The rate of binding 5732

Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

of the H5 peptide to the Fe(III)-NTA complexes in the polymer brushes may be diffusion limited since no stirring or vortex mixing was used after the sample was applied to the Fe(III)-NTA-PHEMAmodified plate. (About every 5-10 min, acetic acid was replenished, which could allow for some on-plate sample mixing. Additionally, evaporation of the solvent molecules allows some mixing of the analyte to occur, which can be observed using a stereomicroscope.) When 2 µL of the digest mixture containing 125 fmol of the H5 peptide was applied to the 2 mm diameter well of the polymer-modified plate, it was found that a 10 min incubation time allowed for an H5 peptide recovery of 63 ± 15% based on MS data (or 78 ± 15%, based on MS/MS data, see Supporting Information Figure S-5 for an example of recovery determination using MS/MS). Thus, short incubation times should still allow efficient analyses. One limitation of the on-plate purification is that recoveries decrease when high concentrations of digest reagents are present. When the 1 pmol/µL digest mixture contained 86 mM urea, 0.72 mM Tris-HCl, 0.04 mM DTT, 5.8 mM NH4HCO3, 0.72 mM iodoacetamide, and 5% acetic acid (roughly a 3-fold increase in urea, Tris-HCl, DTT, and NH4CO3 concentrations relative to the results given above), the recovery of 125 fmol of H5 peptide decreased to 52 ± 9%. The recovery of 125 fmol of H5 peptide from solutions containing further 10-fold and 50-fold increases in both digest reagents and nonphosphorylated peptides (10 and 50 pmol total protein) was 9 ± 1% and 6 ± 1%, respectively. Control samples containing only digest reagents (the amounts in the 10 and 50 pmol mixtures) gave similar H5 peptide recoveries (11 ± 6% and 5 ± 2%, respectively) when using the Fe(III)-NTA-PHEMAmodified plates. This suggests that digest purification may be necessary before applying the sample to the Fe(III)-NTA-PHEMAmodified plates when the concentration of digest reagents is high (i.e., when urea exceeds 40 mM). Further studies need to be carried out for individual digest reagents to determine the limiting concentrations that are compatible with the Fe(III)-NTA-PHEMAmodified plates. Recovery of the H5 Peptide Using Commercially Available IMAC Materials. The performance of the Fe(III)-NTA-PHEMAmodified plates was also compared with enrichment using Millipore ZipTipMC pipet tips loaded with IMAC resin and Qiagen IMAC Mass Spec Focus Chips. These systems were chosen for comparison with Fe(III)-NTA-PHEMA-modified plates based on their applicability to low-volume samples and low amounts of a particular analyte. The ZipTipMC system is a 10 µL pipet tip that is filled with 0.6 µL of resin containing immobilized iminodiacetic acid, and enriched phosphopeptides can, in principle, be directly eluted onto the MALDI sample plate for subsequent MS analysis. The sample wells of the Qiagen IMAC Mass Spec Focus Chips contain an affinity zone with ligands that capture the phosphopeptides, and when the bound phosphopeptides are eluted from the affinity zone with matrix, they are concentrated or focused into a 0.6 mm diameter analysis zone. Although the affinity ligands on the chip are not described, in 2006, Qiagen presented a poster at the American Society for Mass Spectrometry conference on the use of chips modified with a NTA SAM for phosphopeptide purification and concentration.8 Use of the ZipTipMC, which contains Fe(III) complexes, resulted in only 12 ± 2% recovery of 125 fmol of the H5 peptide

from a 1 pmol protein digest mixture (Supporting Information, Figure S-6), and several ions corresponding to nonphosphorylated peptides (m/z 917, 1260, 1502, and 1749) were also present in the mass spectrum of such samples. When selective adsorption on the Qiagen IMAC chip was employed, the recovery of the H5 peptide from a protein digest mixture was 13 ± 3% (Supporting Information, Figure S-7). More recent results using the digest containing 3-fold higher reagent concentrations showed a recovery of 31 ± 9%. (Using a higher concentration of matrix (1 µL of 10 mg/mL or 20 mg/mL 2,5-DHB rather than 2 µL of 1 mg/mL) or adding phosphoric acid (0.5% or 1% overall) to the 10 mg/mL DHB matrix solution did not appear to significantly improve the H5 peptide recovery using the Qiagen chips.) Analyses with the Qiagen chip showed minimal signals due to nonphosphorylated peptides, most likely because of the relatively low binding capacity of the chip, which presumably contains a monolayer of Fe(III)NTA complexes. The binding capacity of the ZipTipMC is ∼290 pmol of the H5 peptide (400 ng as specified by the manufacturer), while the binding capacity of the Qiagen IMAC chip is roughly 5 pmol, assuming that the surface area of the affinity zone is 7 mm2 and a monolayer of peptide binds to this surface with a density of 1 peptide molecule per 2.5 nm2. Even though both the ZipTipMC and Qiagen IMAC chip have a large excess of binding capacity, this was not sufficient to attain high recoveries of 125 fmol of the H5 peptide from the digest mixture. Recovery of the H5 Peptide Using Commercial Metal Oxide Materials. Recoveries of the H5 peptide on Fe(III)-NTAPHEMA films were also compared with recoveries obtained using commercially available metal oxide-based enrichment methods capable of accommodating low sample volumes. Pipet tips containing 30 µg of TiO2 or ZrO2 embedded into the walls of the tip were purchased from Glygen. These systems have a binding capacity of 1-2 µg of peptide. When a sample of 1 pmol of protein digest mixture containing 125 fmol of the H5 peptide was treated with the tips using 1% TFA in the rinse solutions and 0.8% NH4OH in the elution solution, there was a recovery of 22 ± 8% and 68 ± 5% for the ZrO2-containing and TiO2-containing NuTips, respectively, as shown by the mass spectra (Supporting Information, Figures S-8 and S-9). More recent results using the digest containing 3-fold higher reagent concentrations showed a recovery of 46 ± 4% with the TiO2 tips. Efforts to increase recoveries using ammonium glutamate in washing solutions or β-hydroxypropanoic acid in the load and wash solutions, as suggested in recent literature,21,22 did not increase recoveries. When the ZrO2 NuTip was used, the mass spectrum showed only a few ions due to nonphosphorylated peptides. In contrast, signals presumably due to polymer, which most likely eluted from the tip with the TiO2, gave rise to prominent peaks separated by 338 Da (Supporting Information, Figure S-9a). The use of 0.5% piperidine as an eluent was also examined,10 but this caused more titanium dioxide to be leached out of the tips. The presence of the peaks separated by 338 Da makes identifying nonphosphorylated peptides from the protein digest mixture more complicated. Recovery of a Diphosphorylated Peptide Using Fe(III)NTA-PHEMA-Modified Plates. The specificity of the Fe(III)NTA-PHEMA-modified plates for diphosphorylated peptides over singly phosphorylated peptides was examined by analyzing the

Figure 6. Positive-ion MALDI mass spectra of 7 pmol of β-casein digest. Spectra were obtained using (a) conventional analysis and (b) a Fe(III)-NTA-PHEMA-modified plate with incubation and rinsing. In spectrum b, peaks labeled with triangles represent doubly charged phosphopeptides, peaks labeled with circles are due to fragmentation of phosphopeptides (1P2062, 4P2966, and 4P3122), and peaks labeled with stars stem from R-S1-casein and R-S2-casein phosphorylated peptides. Numbers in parentheses are peak intensities. Peak identification was facilitated by CID-MS/MS.

recovery of CH3CH2CO-LFpTGHPEpSLEK (2PH5), which is the diphosphorylated analogue of the H5 peptide. The recovery of 125 fmol of the 2PH5 peptide from a solution containing 1 pmol of total digested protein was 102 ± 9% when using the Fe(III)-NTAPHEMA-modified plates. This suggests that the Fe(III)-NTAPHEMA films have a higher affinity for multiply phosphorylated peptides than monophosphorylated peptides such as the H5 peptide, which had a recovery of 50-70%. Analysis of β-Casein Digests. To demonstrate the efficacy of Fe(III)-NTA-PHEMA-modified plates in the analysis of phosphoprotein digests, digests of β-casein, which is the protein most commonly used to validate phosphopeptide enrichment techniques, were initially examined. Tryptic digestion of this pentaphosphorylated protein typically results in the formation of three phosphorylated peptides (FQpSEEQQQTEDELQDK, m/z 2062, 1P2062; RELEELNVPGEIVEpSLpSpSpSEESITR, m/z 2966, 4P2966; and ELEELNVPGEIVEpSLpSpSpSEESITR, m/z 3122, 4P3122). Mono- and tetraphosphorylated peptides are represented by 1P and 4P, respectively. A volume of 1 µL of digest containing 7 pmol, 1 pmol, 500 fmol, 125 fmol, 31 fmol, or 15 fmol of β-casein was applied to a sample well (2 mm diameter), incubated for 1 h, rinsed to remove nonphosphorylated peptides and digest reagents, and dried with nitrogen. A volume of 1 µL of 1% phosphoric acid was immediately spotted onto each sample well, and 0.25 µL of 2,5DHB solution was added. All analyses using the Fe(III)-NTAPHEMA-modified plates were compared with conventional analysis by MALDI-MS. Figure 6 and Figure 7 show the mass spectra for 7 pmol and 125 fmol samples of digested β-casein, respectively. In the conventional MALDI mass spectrum of 7 pmol of β-casein digest (Figure 6a), phosphopeptide signals due to all three phosphopeptides were observed, but signals from the tetraphosphorylated species were weak. However, when the same digest was analyzed using the Fe(III)-NTA-PHEMA-modified plate (Figure 6b), strong signals from all three of the β-casein phosphorylated peptides were observed as well as 14 other peaks due to R-S1-casein and R-S2Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

5733

Figure 7. Positive-ion MALDI mass spectra of 125 fmol of β-casein digest. The spectra were obtained using (a) conventional analysis and (b) an Fe(III)-NTA-PHEMA-modified plate with incubation and rinsing. In spectrum b, the peak labeled with a triangle is due to a doubly charged phosphopeptide, the peak labeled with a circle is due to fragmentation of phosphopeptide (1P2062), and peaks labeled with stars stem from R-S1-casein and R-S2-casein phosphorylated peptide impurities. The number in parentheses is peak intensity.

casein phosphorylated peptides, which were present as impurities. (β-Casein is sold as a 90% pure mixture through Sigma-Aldrich.) After enrichment, there was an impressive 30-fold increase in the intensities of the phosphorylated peptides (1P2062, 4P2966, and 4P3122) relative to that observed in the conventional MS analysis. A β-casein phosphorylated peptide (NVPGEIVEpSLpSpSpSEESITR, m/z 2353, 1P2352) that originates from chymotryptic cleavage was also occasionally observed.6 Table S-2 (Supporting Information) lists the peaks in the positive-ion MALDI mass spectra that were attributed to β-casein, R-S1-casein, and R-S2-casein phosphopeptides. This table also lists peaks that were due to doubly charged phosphopeptide ions, [M + 2H]2+, and phosphopeptide ions that underwent fragmentation via loss of one or two phosphate groups, [M + H - H3PO4]+ or [M + H - 2H3PO4]+. No peptide sequences could be assigned to signals in Figure 6b at m/z 2599, 2603, 2616, and 3042, but these peaks have tentatively been assigned to phosphopeptides due to the apparent H3PO4 neutral loss (-98 Da) observed upon MS/MS activation. As the amount of β-casein in samples decreased, the number and intensity of signals in the mass spectra of these samples also decreased. Even at the 1 pmol level, neither tetraphosphorylated peptide of β-casein was observed in the conventional analysis of the diluted digest, but the polymer-modified plate was able to readily detect these peptides along with 11 phosphopeptides from R-S1- and R-S2-casein digests. However, when 7 pmol of β-casein was analyzed using the Fe(III)-NTA-PHEMA-modified plate, the peaks due to the tetraphosphorylated peptides were almost the same intensity (or more intense) as the peak due to 1P2062 as shown in Figure 6b. In contrast, when 1 pmol of digest was analyzed using the polymer-modified plate, the signals due to the tetraphosphorylated peptides were less than half the intensity of the peak due to 1P2062. This may suggest that at high amounts of digest, the tetraphosphorylated peptides displace the monophosphorylated peptides, while at low loadings this occurs to a lesser extent. (With 7 pmol of digest, the amount of phosphopeptides is near the 18 pmol capacity mentioned above.) 5734

Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

Figure 7 shows the mass spectra of a diluted digest containing 125 fmol of β-casein. Relative to conventional MALDI-MS (Figure 7a), the signal due to 1P2062 was improved by a factor of 20 when the Fe(III)-NTA-PHEMA-modified plate was used (Figure 7b). However, 4P2966 and 4P3122 did not give identifiable signals, but six phosphopeptides from the R-casein contaminants were still detected, one of which was due to the tetraphosphorylated peptide at m/z 3008. This tetraphosphorylated peptide contains only one more acidic residue than 4P2966 (ELEELNVPGEIVEpSLpSpSpSEESITR), so enrichment was not expected to be significantly better for the m/z 3008 tetraphosphorylated peptide than for 4P2966. At present, the high sensitivity for this species (4P3008) is not understood. To examine detection limits, the digest was further diluted and analyzed. When 31 fmol of β-casein was analyzed using conventional MALDI-MS, the mass spectrum showed a very weak signal due to 1P2062 and the signal-to-noise was only 3, whereas the signalto-noise was over 8 times greater when using the polymer modified plate. Only one R-casein phosphorylated peptide was observed on the modified plate at these low concentrations. Additionally, when 15 fmol of β-casein was analyzed using conventional analysis, the S/N was about 2, whereas the S/N observed using the polymer-modified plate was 6. From these results, the detection limit of conventional analysis by MALDIMS for β-casein (based on 1P2062) appears to be around 30 fmol whereas the detection limit when using the enrichment method is below 15 fmol.45 Analysis of Ovalbumin Digests. Various amounts (1 pmol, 500 fmol, 125 fmol, and 62 fmol) of tryptic digests of ovalbumin, which is a diphosphorylated protein whose digestion typically yields three phosphorylated peptides (EVVGpSAEAGVDAASVSEEFR, m/z 2088, 1P2088; LPGFGDpSIEAQCGTSVNVHSSLR, m/z 2511, 1P2511; and FDKLPGFGDpSIEAQCGTSVNVHSSLR, m/z 2901, 1P2901), were also analyzed. At 1 pmol levels, conventional MALDI mass spectra showed signals due to all three phosphorylated peptides, but the use of the polymer-modified plate greatly simplified the mass spectrum. The phosphopeptide peaks dominated the mass spectrum taken on the modified plate because most of the nonphosphorylated peptides were removed during the rinsing step. Moreover, the signals of the phosphorylated peptides increased by a factor of ∼40 when the modified plate was used. The use of the Fe(III)-NTA-PHEMA-modified plates allows for capture of phosphopeptides as well as desalting of the sample in a simple rinsing step. Figure 8 shows the spectra of a diluted digest containing 500 fmol of ovalbumin. In this case, the phosphopeptide signals in the conventional MALDI mass spectrum are weak compared to signals of nonphosphorylated peptides. In fact, one of the phosphopeptides, 1P2901, was not detected in the conventional analysis. However, when the same sample was applied to the polymermodified plates, all three phosphopeptides were detected, and signals from nonphosphorylated species were minimal. The spectra of 125 fmol of ovalbumin in diluted digest are shown in Figure S-10 (Supporting Information). No phosphopeptides were detected in the conventional MALDI mass spectrum, and when the digest was applied to the modified plate, only one (45) Dunn, J. D. Ph.D. Thesis, Michigan State University, East Lansing, MI, 2007.

(HIATNAVLFFGR), and 1247 (ADHPFLFCIK) remained while little to no signals due to the ovalbumin phosphopeptides were observed. This suggests that these plates should not be reused and that the nonphosphopeptide binding does not occur at the Fe(III) sites.

Figure 8. Positive-ion MALDI mass spectra of 500 fmol of ovalbumin digest. Spectra were obtained using (a) conventional analysis and (b) a Fe(III)-NTA-PHEMA-modified plate with incubation and rinsing. Numbers in parentheses are peak intensities.

phosphopeptide, 1P2088, was observed. Relative to this phosphopeptide peak at m/z 2088, signals due to nonphosphorylated peptides at m/z 1774, 1508, 1346, and 1264 were stronger than they were at higher digest concentrations. These peaks due to nonphosphopeptides were characterized using MS/MS and their assignments can be found in Table S-3 (Supporting Information). Ovalbumin contains seven histidine residues, and complete digestion of this protein yields five peptides containing at least one histidine residue and 20 peptides without histidine residues. One of the histidine-containing peptides, 1P2088, is phosphorylated, and this peptide gives the largest signal of the phosphorylated species. The other four histidine-containing peptides, m/z 1774, 1508, 1346, and 1264, were frequently detected when ovalbumin digests were analyzed using the Fe(III)-NTA-PHEMA-modified plate. Thus, the presence of the histidine residues in the peptides could be contributing to their binding to the polymer-modified plates. Similar results were found when the amount of ovalbumin digest was further decreased to 62 fmol (spectrum not shown). The conventional analysis detected 1P2088 minimally whereas the use of the polymer-modified plate again showed a stronger signal for this phosphopeptide than for the nonphosphorylated peptides, although signals due to histidine-containing and a few other peptides were present. A full assortment of mass spectra of both ovalbumin and β-casein digests enriched by the Fe(III)-NTAPHEMA-modified plates can be found elsewhere.45 To further examine nonspecific binding, films used for analyzing the 1 pmol ovalbumin digests were immersed in a 50 mM EDTA solution (pH 7.3) to remove Fe(III), sonicated for 1 h, rinsed with water, and dried. After addition of elution solution (1 µL of 1% phosphoric acid) and matrix (0.25 µL of 40 mg/mL 2,5-DHB containing 0.5% H3PO4) in the wells, the sample spots were again analyzed by MALDI-MS and MS/MS. Remnants due to nonphosphorylated peptides with m/z 1774 (ISQAVAAHAEINEAGR), 1346

CONCLUSIONS The binding capacity of ∼60 nm thick Fe(III)-NTA-PHEMA films on gold MALDI plates is 0.6 µg phosphoangiotensin/cm2 or about 18 pmol of phosphoangiotensin in a 2 mm diameter sample well. This high capacity is likely responsible for high recoveries of phosphopeptides by these modified plates. With the use of a a labeled peptide as an internal standard, the relatively thick (500 Å) Fe(III)-NTA-PHEMA films showed at least 3-fold higher recoveries of the H5 peptide than thinner (30 Å) Fe(III)PAA-NTA films or monolayer films. The Fe(III)-NTA-PHEMAmodified MALDI plates also allowed greater recoveries than the use of ZipTipMC pipet tips, ZrO2 NuTips, and Qiagen MALDI plates. Recovery from TiO2 NuTips was similar to that obtained with Fe(III)-NTA-PHEMA-modified plates, but interference from material leached out of the tips was frequently seen. Although most of the work on modified plates employed 60 min incubations, 10 min incubations with Fe(III)-NTA-PHEMA-modified plates were also sufficient for high recoveries. For β-casein digests, the limit of detection for 1P2062 was less than 15 fmol, and a number of phosphopeptides due to R-casein impurities were also detected at very low levels. Mass spectra of ovalbumin digests confirm that the use of the Fe(III)-NTA-PHEMA-modified plate greatly simplifies the MS analysis of digests of phosphoproteins. ACKNOWLEDGMENT We thank the National Science Foundation (Grant CHE0616795) and Michigan State University (Grant HBRI-639) for funding of this work. We would also like to thank Dr. Jetze J. Tepe and Samantha M. Frawley for their insightful discussions. SUPPORTING INFORMATION AVAILABLE Protocols describing the synthesis of the D5 and H5 peptides, the fabrication of the Fe(III)-NTA-PHEMA-modified plates, and the use of commercial materials for phosphopeptide enrichment; tables listing BSA, phos b, β-casein, R-S1-casein, R-S2-casein, and ovalbumin peptide sequences and m/z values for peaks observed in the mass spectra for the analyses presented here; reflectance FT-IR spectra of modified PHEMA brushes on a gold substrate; an H5/D5 peptide calibration curve; mass spectra of an ovalbumin digest and the H5 peptide recovered using commercial methods; MS/MS spectra of the H5 peptide recovered using Fe(III)-NTAPHEMA-modified plates. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review December 4, 2007. Accepted May 13, 2008. AC702472J

Analytical Chemistry, Vol. 80, No. 15, August 1, 2008

5735