Isolation and Characterization of Glycosylphosphatidylinositol

This method allowed analysis of GPI peptides derived from low picomole .... DHB (20 μg/μL) in 70% ACN/0.1% TFA were mixed directly on the MALDI-MS t...
3 downloads 0 Views 160KB Size
Download PDF
Anal. Chem. 2006, 78, 3335-3341

Isolation and Characterization of Glycosylphosphatidylinositol-Anchored Peptides by Hydrophilic Interaction Chromatography and MALDI Tandem Mass Spectrometry Miren J. Omaetxebarria,† Per Ha 1 gglund,†,⊥ Felix Elortza,†,| Nigel M. Hooper,‡ Jesus M. Arizmendi,§ and ,† Ole N. Jensen*

Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark, School of Biochemistry and Microbiology, Leeds Institute of Genetics, Health and Therapeutics, University of Leeds, Leeds LS2 9JT, United Kingdom, Department of Biochemistry and Molecular Biology, University of The Basque Country, PO Box 644, 48080 Bilbao, Spain, and Cooperative Research Centre on Biosciences (CIC-BioGUNE), Technology Park of Bizkaia, 801 A Building, 48160 Derio, Spain

Glycosylphosphatidylinositol-anchored proteins (GPIAPs) are posttranslationally processed proteins that become tethered to the extracellular leaflet of the plasma membrane via a C-terminal glycan-like moiety. Since the first GPI-AP was described in the 1970s, more than 500 GPI-APs have been reported in a range of species, including plants, microbes, and mammals. GPI-APs are probably involved in cell signaling, cell recognition, and cell remodeling processes, and they may potentially serve as cell surface antigens or vaccine targets in pathogenic microorganisms or transformed mammalian cells. Due to the structural complexity and physicochemical properties of GPI-APs, their identification and structural characterization is a demanding analytical task. Here, we report a simple, fast and sensitive method for isolation and structural analysis of GPI-anchors using a combination of hydrophilic interaction liquid chromatography and matrixassisted laser desorption/ionization (MALDI) quadrupole time-of-flight tandem mass spectrometry. This method allowed analysis of GPI peptides derived from low picomole levels of the porcine kidney membrane dipeptidase. Furthermore, it allowed unambiguous assignment of the omega site via amino acid sequencing of the modified peptides. GPI-anchor-specific diagnostic ions were observed by MALDI-MS/MS at m/z 162, 286, 422, and 447, corresponding to glucosamine, mannose ethanolamine phosphate, glucosamine inositol phosphate, and mannose ethanolamine phosphate glucosamine, respectively. Thus, the methodology described herein may enable sensitive and specific detection of GPI-anchored * Corresponding author. Tel.: +45 6550 2368. Fax: +45 6550 2467. E-mail: [email protected]. URL: www.protein.sdu.dk. † University of Southern Denmark. ‡ University of Leeds. § University of The Basque Country. | CIC-BioGUNE. ⊥ Current address: Biochemistry & Nutrition Group, Biocentrum DTU 224124, Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark. 10.1021/ac0517949 CCC: $33.50 Published on Web 04/07/2006

© 2006 American Chemical Society

peptides in large-scale proteomic studies of plasma membrane proteins. Glycosylphosphatidylinositol (GPI)-anchored proteins (GPIAPs) is a diverse class of proteins that are attached to the cell surfaces by a GPI moiety, a complex posttranslational, lipidcontaining modification that is covalently linked to the C-terminus of proteins. A variety of biological functions have been ascribed to GPI-APs. They can act as coat proteins, receptors, ectoenzymes, differentiation antigens, and adaptors, and it has also been suggested that they are involved in cell adhesion, cell-cell interactions, intracellular sorting, cell wall remodeling, and transmembrane signaling processes.1-6 Posttranslational processing of GPI-anchored proteins encompasses two steps.7 After synthesis on the ribosome, the pro-protein, containing an N-terminal signal peptide, is translocated to the endoplasmic reticulum (ER). Following proteolytic processing of the C-terminal domain of the pro-protein, the GPI moiety is attached to this new COOH-terminus (omega site) of the polypeptide by a transamidation reaction.8 Subsequently, the mature protein is translocated from the ER lumen via the secretory route and immobilized on the extracellular face of the plasma membrane. This type of cell surface targeting and anchoring of GPIAPs is an alternative to the hydrophobic transmembrane polypeptide domains of integral membrane proteins. GPI-anchored proteins were first described in 1976 when Ikezawa et al. discovered that alkaline phosphatase could be released by phosphatidylinositol-specific phospholipase C (PI(1) Hooper, N. M. Mol. Membr. Biol. 1999, 16, 145-156. (2) Ferguson, M. A. J. Cell Sci. 1999, 112, 2799-2809. (3) Horejsi, V.; Drbal, K.; Cebecauer, M.; Cerny, J.; Brdicka, T.; Angelisova, P.; Stockinger, H. Immunol. Today 1999, 20, 356-631. (4) Muniz, M.; Riezman, H. EMBO J. 2000, 19, 10-15. (5) Angst, B. D.; Marcozzi, C.; Magee, A. I. J. Cell Sci. 2001, 114, 629-641. (6) Ghiran, I.; Klickstein, L. B.; Nicholson-Weller, A. J. Biol. Chem. 2003, 278, 21024-21031. (7) Gerber, L. D.; Kodukula, K.; Udenfriend, S. J. Biol. Chem. 1992, 267, 1216812173. (8) Udenfriend, S.; Kodukula, K. Methods Enzymol. 1995, 250, 571-582.

Analytical Chemistry, Vol. 78, No. 10, May 15, 2006 3335

PLC).9 PI-PLC hydrolyzes the diphosphoester bond between the lipidic and carbohydrate moieties of the GPI-anchor, thereby releasing the protein and carbohydrate moiety from the glycerolipid membrane anchor. In general, the biological function of this enzyme-catalyzed release is unclear. In Trypanosoma brucei, the enzymatic release of GPI-APs is suggested to increase the antigenic variation to evade the immune response of the host organism.10 PI-PLC treatment is now widely used for release and biochemical characterization of GPI-APs from cells,11,12 and this specific method has been used in proteomic studies aimed at identification of GPI-APs from human and plant cells. Recently, the utility of phospholipase D (PLD) for the release of GPI-APs was demonstrated.13 Using the concept of modification-specific proteomics,14 we have previously identified human (Hela cell) lipid raft proteins and Arabidopsis thaliana proteins as GPI-APs by a “shave-and-conquer” strategy that combined plasma membrane preparation, Triton X-114 phase-partitioning, PI-PLC or PLD treatment, LC-MS/MS, and computational sequence analysis.13,15 Borner et al. used 2D-gel electrophoresis, PI-PLC treatment and mass spectrometry to identify A. thaliana proteins as GPI-APs.16 Computational sequence analysis enables prediction of GPIanchored proteins via detection of omega sites (ω-sites) and absence of transmembrane domains.17-19 As a result of genomic sequence analysis, an estimated 2% of open reading frames in model organisms were predicted to encode GPI-APs.19 GPI-anchors described to date contain a core structure, which is well-conserved among species (Figure 1). The presence of a nonacetylated glucosamine (GlcN) residue is a unique feature of GPI-anchors. The GlcN residue in the reducing end of the core tetrasaccharide is glycosidically linked to the 6-hydroxyl group of the phosphatidylinositol residue. The mannose in the nonreducing end of the tetrasaccharide is linked via a phosphodiester bond to ethanolamine (EtN), which in turn is linked to the R-carboxyl group of the protein C-terminal residue via an amide bond. GPI-anchors may differ widely in the side chains attached to this core as well as in the lipid moiety of the anchor.20 The highest degree of diversity between GPI-anchors from different species is due to glycan substitutions on the conserved mannose residues. Furthermore, in some anchors, hydroxyl groups of the inositol residue are acylated, and this variation may be responsible for the resistance of some proteins to PI-PLC digestion. This resistance can be overcome by using PLD instead of PI-PLC or (9) Ikezawa, H.; Yamanegi, M.; Taguchi, R.; Miyashita, T.; Ohyabu, T. Biochim. Biophys. Acta 1975, 450, 154-164. (10) Low, M. In PNH and the GPI-linked Proteins; Young, N. S., Moss, J., Ed.; Academic Press: San Diego, 2000, pp 239-268. (11) Bordier, J. Biol. Chem. 1981, 256, 1604-1607. (12) Hooper, N. M.; Low, M. G.; Turner, A. J. Biochem. J. 1987, 244, 465-469. (13) Elortza, F.; Mohammed, S.; Bunkenborg, J.; Foster, L. J.; Nu ¨ hse, T. S.; Brodbeck, U.; Scott, C. P.; Jensen, O. N. J. Proteome Res. 2006, (web release date: 18 Feb 2006). (14) Jensen, O. N. Curr. Opin. Chem. Biol. 2004, 8, 33-41. (15) Elortza, F.; Nuhse, T. S.; Foster, L. J.; Stensballe, A.; Peck, S. C.; Jensen, O. N. Mol. Cell. Proteomics 2003, 2, 1261-1270. (16) Borner, G. H.; Lilley, K. S.; Stevens, T. J.; Dupree, P. Plant Physiol. 2003, 132, 568-577. (17) Eisenhaber, B.; Bork, P.; Eisenhaber, F. J. Mol. Biol. 1999, 292, 741-758. http://mendel.imp.onivie.ac.at/sat/gpi/gpi_server.html. (18) Kronegg, J.; Buloz, D. retrieved from http//129.194.185.165/dgpi/1999. (19) Fankhauser, N.; Ma¨ser, P. Bioinformatics 2005, 21, 1846-1852. http:// gpi.unibe.ch. (20) McConville, M. J.; Collidge, T. A.; Ferguson, M. A.; Schneider, P. J. Biol. Chem. 1993, 268, 15595-15604.

3336

Analytical Chemistry, Vol. 78, No. 10, May 15, 2006

Figure 1. General core structure for GPI-anchored proteins. X stands for HexNAc and HexHexNAc, and Y, for ethanolamine phosphate possible additional side-chains.29

by first deacylating the inositol with alkaline hydroxylamine prior to PI-PLC treatment.21 The core structure of a GPI-anchor was first determined for the T. brucei variant surface glycoprotein.22 The route for the complete structural characterization was based on enzymatic and chemical microsequencing reactions and a series of chromatographic techniques. Several GPI core glycans have been determined by nitrous acid deamination or Triton X-114 partitioning and subsequently analyzed using selective exo- and endoglycosidase treatments in combination with NMR, FAB-MS, or GC/MS. These include the GPI-anchor of rat Thy-1,23 human acetylcholinesterase,24 Leishmania major promastigote surface protease,25 yeast glycoprotein,26 Dictyostelium discoideum prespore-specific antigen,27 human urine CD59,28 and porcine and human membrane dipeptidase (MDP).29 In more recent studies, electrospray ionization mass spectrometry (ESI-MS) combined with collision induced (21) Toutant, J. P.; Roberts, W. L.; Muray, N. R.; Rosenberry, T. L. Eur. J. Biochem. 1989, 180, 503-508. (22) Ferguson, M. A.; Homans, S. W.; Dwek, R. A.; Rademacher, T. W. Science 1988, 239, 753-759. (23) Homans, S. W.; Ferguson, M. A.; Dwek, R. A.; Rademacher, T. W.; Anand, R.; Williams, A. F. Nature 1988, 333, 269-272. (24) Roberts, W. L.; Santikarn, S.; Reinhold, V. N.; Rosenberry, T. L. J. Biol. Chem. 1988, 263, 18776-18784. (25) Schneider, P.; Ferguson, M. A.; McConville, M. J.; Mehlert, A.; Homans, S. W.; Bordier, C. J. Biol. Chem. 1990, 265, 16955-16964. (26) Fankhauser, C.; Homans, S. W.; Thomas-Oates, J. E.; McConville, M. J.; Desponds, C.; Conzelmann, A.; Ferguson, M. A. J. Biol. Chem. 1993, 268, 26365-26374. (27) Haynes, P. A.; Gooley, A. A.; Ferguson, M. A.; Redmond, J. W.; Williams, K. L. Eur. J. Biochem. 1993, 216, 729-737. (28) Nakano, Y.; Noda, K.; Endo, T.; Kobata, A.; Tomita, M. Arch. Biochem. Biophys. 1994, 311, 117-126.

dissociation was used to confirm the carbohydrate moiety and fragmentation pattern of GPI-anchor peptides of human acetylcholinesterase30 and to deduce the peptide sequence of the GPIanchor peptide of bovine erythrocyte acetylcholinesterase.31 ESIMS/MS and MALDI-TOF-MS with post-source decay were applied to characterize GPI-anchored peptides from 5′-nucleotidase.32 Mass spectrometry-based strategies are now extensively used for determination of posttranslational modifications of proteins, such as phosphorylation and glycosylation;14,33 however, some posttranslationally modified peptides are still difficult to detect in MS, presumably due to their physicochemical properties (size, charge, hydrophobicity, lability) and matrix effects during sample preparation and mass analysis (ionization bias). In our laboratory, miniaturized enrichment and solid-phase extraction techniques, such as immobilized metal affinity chromatography, titanium dioxide, graphite powder, and hydrophilic interaction chromatography (HILIC), proved very useful for the recovery of posttranslationally modified peptides, including phosphopeptides and glycopeptides, for mass spectrometric analysis.34-41 We investigated whether these miniaturized sample preparation techniques could be useful for the preparation and analysis of peptide derived from GPI-APs and found that HILIC42,43 in combination with MALDI-QTOF tandem mass spectrometry provided a sensitive method for enrichment, detection, and characterization of peptides that contain GPI-anchors. The method was applied to PI-PLC-treated and -purified porcine kidney membrane dipeptidase and allowed for assignment of the ω-site and characterization of the anchor structure. In addition, diagnostic marker ions specific for GPI-anchored C-terminal peptides were detected, which may eventually be of use in large-scale experiments aimed at characterization of posttranslationally modified membrane proteins. MATERIAL AND METHODS Purification of Porcine Renal Membrane Dipeptidase. MDP from porcine kidney cortex was purified by affinity chromatography on cilastatin-Sepharose following solubilization with (29) Brewis, I. A.; Ferguson, M. A.; Mehlert, A.; Turner, A. J.; Hooper, N. M. J. Biol. Chem. 1995, 270, 22946-22956. (30) Deeg, M. A.; Humphrey, D. R.; Yang, S. H.; Ferguson, T. R.; Reinhold, V. N.; Rosenberry, T. L. J. Biol. Chem. 1992, 267, 18573-18580. (31) Haas, R.; Jackson, B. C.; Reinhold, B.; Foster, J. D.; Rosenberry, T. L. Biochem. J. 1996, 314, 817-825. (32) Taguchi, R.; Hamakawa, N.; Maekawa, N.; Ikezawa, H. J. Biochem. 1999, 126, 421-429. (33) Mann, M.; Jensen, O. N. Nat. Biotechnol. 2003, 21, 255-261. (34) Stensballe, A.; Andersen, S.; Jensen, O. N. Proteomics 2001, 1, 207-222. (35) Stensballe, A.; Jensen, O. N. Rapid Commun. Mass Spectrom. 2004, 18, 1721-1730. (36) Nuhse, T. S.; Stensballe, A.; Jensen, O. N.; Peck, S. C. Mol. Cell. Proteomics 2003, 2, 1234-1243. (37) Gruhler, A.; Olsen, J. V.; Mohammed, S.; Mortensen, P.; Faergeman, N. J.; Mann, M.; Jensen, O. N. Mol. Cell. Proteomics 2005, 4, 310-327. (38) Larsen, M. R.; Cordwell, S. J.; Roepstorff, P. Proteomics 2002, 9, 12771287. (39) Larsen, M. R.; Hojrup, P.; Roepstorff, P. Mol. Cell. Proteomics 2004, 4, 107119. (40) Larsen, M. R.; Thingholm, T. E.; Jensen, O. N.; Roepstorff, P.; Jorgensen, T. J. Mol. Cell. Proteomics 2005, 4, 873-876. (41) Ha¨gglund, P.; Bunkenborg, J.; Elortza, F.; Jensen, O. N.; Roepstorff, P. J. Proteome Res. 2004, 3, 556-566. (42) Alpert, A. J. J. Chromatogr. 1990, 499, 177-96. (43) Alpert, A. J.; Shukla, M.; Shukla, A. K.; Zieske, L. R.; Yuen, S. W.; Ferguson, M. A.; Mehlert, A.; Pauly, M.; Orlando, R. J. Chromatogr. 1994, 676, 191122.

PI-PLC.44,45 Protein concentrations were determined using the bicinchoninic acid method of Smith et al.46 SDS-PAGE and in-Gel Digestion of Protein. SDS-PAGE was performed using the system of Laemmli.47 Polyacrylamide gels (10%) were used, and protein was visualized by Coomassie blue staining. Proteins were in-gel-digested using trypsin.48 Excised protein bands were cut in pieces and washed with 30 µL of acetonitrile (ACN). The ACN was removed, and 50 µL of 10 mM DTT in 100 mM NH4HCO3, pH 7.8, was added, and incubation was carried out for 45 min at 56 °C. After removal of the supernatant, 30 µL of freshly prepared 55 mM iodoacetamide in 100 mM NH4HCO3, pH 7.8, was added, and the mixture was incubated in the dark at room temperature for 30 min. The supernatant was discarded, and the gel pieces were washed with 30 µL of ACN for 10 min. The supernatant was removed, and the gel pieces were dried by vacuum centrifugation. Sequence grade trypsin (10 ng/µL, Promega) in 50 mM NH4HCO3, pH 7.8, was added to the dried gel pieces, which were incubated for 1 h on ice. After the incubation, the supernatant was discarded, and an additional 10 µL of 50 mM NH4HCO3, pH 7.8, was added, and the digestion was incubated overnight (up to 18 h) at 37 °C. The overnight supernatant was then transferred to a new tube, and 20 µL of 50 mM NH4HCO3, pH 7.8, was added to the gel pieces for further peptide extraction for 10 min. One volume equivalent of ACN was added, and the incubation was continued for another 10 min. This supernatant was recovered and pooled with the overnight supernatant. The peptide extraction was repeated one more time and pooled with the peptide extracts, and the peptide samples were then dried in a vacuum centrifuge and stored at -20 °C. Hydrophilic Interaction Chromatography (HILIC). ZICHILIC chromatography resin (10-µM particles) was a gift from Sequant (Umeå, Sweden). HILIC microcolumns were prepared and used as previously described.41 The HILIC column was packed in a partially constricted GeLoader tip (Eppendorf), to a column height of ∼5 mm. The column was equilibrated in 80% ACN/0.5% formic acid (FA). Dry peptide samples were redissolved in 5 µL of 80% ACN/0.5% FA and loaded onto the HILIC microcolumn using air pressure. The HILIC column was washed three times with 20 µL 80% ACN/0.5% FA, and then the bound peptides were eluted with 2 µL of 95% H2O/5% FA and collected in a microcentrifuge tube. Endoglycosidase Digestion. N-linked glycans were released by overnight incubation with 1 mU PNGase F (Flavobacterium meningosepticum, Roche) in 100 mM NH4HCO3 buffer (pH 7.8) at 37 °C. In all cases, 1 U was defined as the amount of enzymatic activity needed to digest 1 µmol of substrate in 1 min, as described by the manufacturer. MALDI-TOF-MS and MALDI-QTOF-MS/MS. MALDI-TOFMS was performed using a Voyager Elite DE STR instrument (Applied Biosystems, Framingham, MA). MALDI mass spectra (44) Littlewood, G. M.; Hooper, N. M.; Turner, A. J. Biochem. J. 1989, 257, 361-367. (45) Hooper, N. M.; Turner, A. J. Biochem. J. 1989, 257, 361-367. (46) Smith, P. K.; Krohn, R.; Hermanson, G. T.; Mallia, A. K.; Gartner, F. H.; Provenzano, M. D.; Fujimoto, E. K.; Goeke, N. M.; Olson, B. J.; Klenk, D. C. Anal. Biochem. 1985, 150, 76-85. (47) Laemmli, U. K. Nature 1970, 227, 680-685. (48) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68, 850-858.

Analytical Chemistry, Vol. 78, No. 10, May 15, 2006

3337

Figure 2. MALDI-TOF mass spectra of peptides from MDP tryptic digest. (A) Peptides from the tryptic digest of MDP. (B) MDP tryptic digest, HILIC void volume. (C) HILIC-enriched fraction analyzed in MALDI reflector TOF-MS mode. (D) HILIC-enriched fraction analyzed in MALDI linear TOF-MS mode. In all cases, 0.5 µL of sample and 0.5 µL of DHB (20 µg/µL) in 70% ACN/0.1% TFA were mixed directly on the MALDI-MS target.

Figure 3. MALDI-TOF mass spectra of peptides from MDP tryptic digest after HILIC enrichment: (A) 5 pmol of peptide mix, (B) 2.5 pmol of peptide mix, (C) 1 pmol of peptide mix. In all cases, 0.5 µL of HILIC eluate and 0.5 µL DHB (20 µg/µL) in 70% ACN/0.1%TFA were mixed directly on the MALDI-MS target.

were acquired in the positive ion linear and positive ion reflector modes. Ions were generated by irradiation of analyte-matrix deposits by a nitrogen laser (337 nm) and analyzed using an accelerating voltage of 20 kV. Analyte-matrix samples were prepared by the dried-droplet method using 20 µg/µL of 2,5 dihydroxybenzoic acid (DHB) in 70% ACN, 5% FA as matrix. Mass spectra were externally calibrated in the m/z range of 800-4000 using peptides generated by tryptic digestion of bovine β-lactoglobulin. MALDI tandem mass spectra were recorded on a MALDIQTOF tandem mass spectrometer (Ultima HT, Waters//Micromass, Manchester, U.K.) equipped with a nitrogen laser (337 nm). CID-MS/MS experiments were performed by using argon as the collision gas, and a collision energy in the range of 50-80 eV was applied to fragment singly protonated peptide ions. Poly(ethylene glycol) was used for external calibration (m/z 4004000). 3338 Analytical Chemistry, Vol. 78, No. 10, May 15, 2006

Figure 4. MALDI-TOF-MS spectra of the PNGase F-treated HILICenriched fraction. (A) HILIC eluate (control); (B) PNGase F-treated HILIC eluate.

RESULTS AND DISCUSSION Enrichment of GPI-Anchored Peptides by Hydrophilic Interaction Chromatography. In the present study, we explored the use of HILIC for capturing GPI-anchored peptides after tryptic digestion of porcine renal dipeptidase. In HILIC, hydrophilic, polar analytes dissolved in an apolar mobile phase are retained through hydrophilic interactions with apolar stationary phase and eluted in the order of increasing hydrophilicity. HILIC is, therefore, suitable for analysis of polar compounds, such as glycopeptides, that are not retained efficiently by reversed-phase columns. We investigated whether HILIC was useful for the preparation of GPIpeptides for analysis by MALDI-QTOF-MS/MS, since we find this MS method very efficient for analysis of glycopeptide and glycan structures and their heterogeneity. MALDI-QTOF-MS/MS enables detailed structural studies of large (>2 kDa) glycan structures that often escape detection in LC/ESI-MS/MS. The well-characterized porcine membrane dipeptidase protein MDP was chosen as the model protein to establish suitable conditions for sample preparation and mass spectrometry.29 MDP was purified from porcine kidney cortex preparations by using PI-PLC treatment and affinity chromatography on cilastatinSepharose, followed by SDS-PAGE and trypsin digestion.44,45 The resulting peptide mixture was analyzed by MALDI-MS (Figure 2A). A total of 15 peptides were assigned to the MDP amino acid sequence, corresponding to 48% sequence coverage. Next, the MDP tryptic peptide mixture was loaded onto a custom-made HILIC microcolumn, and the flowthrough and the retentate, respectively, were analyzed by MALDI-TOF-MS. The void volume generated a MALDI mass spectrum (Figure 2B) that appeared very similar to the crude peptide mixture (Figure 2A) and contained no signals from glycosylated peptides. In contrast, the MALDI-TOF mass spectrum of the peptides that were eluted from the HILIC column exhibited three major and several minor ion signals (Figure 2C) that were observed neither in the crude sample nor in the HILIC flow-through fraction. The peptide ion signal at m/z 1839.6 matched the calculated mass of the core GPI-anchor (Figure 1) attached to the tryptic COOHterminal peptide TNYGYS from the MDP protein. The two other signals at m/z 2042.7 and 2204.79 matched to the core GPI-anchor with mass increases of 203 and 365 Da that were tentatively

Figure 5. (A) MALDI-QTOF-MS/MS spectra of the core GPI-peptide (m/z 1839.6). Fragment ions and corresponding carbohydrate units are indicated. Diagnostic ions at m/z 162 (GlcN), 286 (ManEtNP), 422(GlcNInsP), and 447 (ManEtNPGlcN) are evident. (B) Expanded view of the low m/z range where b and y fragment ions corresponding to the COOH-terminal peptides were detected to allow for the assignment of the ω-site.

assigned as additional N-acetylhexosamine (HexNAc) (+203 Da) and the combination of one hexose and one N-acetylhexosamine (Hex + HexNAc) (+365 Da) moieties. We performed the mass analysis of the HILIC eluate in reflector TOF (Figure 2C) and linear TOF (Figure 2D) modes to exclude the possibility that the candidate GPI-peptide signals were produced by metastable decay of glycopeptide ions during reflector TOF analysis. As anticipated, the signals of all the three variants of the GPI anchor signals (core, core + HexNAc, core + Hex + HexNAc) were also present in the spectrum recorded in linear mode (Figure 2D). This demonstrates that the GPI-peptides are rather stable, since they generate no significant metastable decay products in reflector TOF mode. To study the sensitivity of the method to capture GPI-anchored peptides, different amounts of MDP tryptic digests were loaded onto the HILIC microcolumns, and the eluates were analyzed by MALDI-TOF. As shown in Figure 3, detection of the GPI-anchored peptides was still possible when loading only 1 pmol of MDP tryptic digest onto the HILIC column. Several high m/z ions were observed in the HILIC eluted fraction (Figure 2D), which were identified as N-glycosylated peptides of MDP. Two potential N-glycosylation sites at residues Asn57 and Asn279 were predicted for MDP, and Asn57 was previously suggested to be glycosylated.49 Release of the glycans by PNGase F treatment of the HILIC enriched fraction of the MDP (49) Rached, E.; Hooper, N. M.; James, P.; Semenza, G.; Turner, A. J.; Mantei, N. Biochem. J. 1990, 271, 755-760.

tryptic digest allowed identification of the MDP glycosylated peptide 278ANLSQVADHLDHIK291 at m/z 1561.79. This mass value corresponds to the molecular weight of the unmodified peptide with the addition of 0.98 Da due to the conversion of the asparagine residue to an aspartate residue as a consequence of the PNGaseF treatment (Figure 4). The peak at m/z 1689.89 corresponds to the peptide 278ANLSQVADHLDHIKK,292 corresponding to the previously assigned peptide carrying an additional lysine residue (Figure 4). After PNGase F treatment of the peptides in the HILIC eluate, no signals from the candidate GPIanchored peptides were detected. The PNGase F enzyme was not expected to cleave the trimannose-GlcN-InsP structure in the GPI-anchor, so we speculate that the GPI-peptide is suppressed by the presence of the two abundant peptide species at m/z 1561.79 and 1689.89, since deglycosylated peptides are expected to ionize much more efficiently than the glycosylated counterparts. GPI-Peptide Sequencing by MALDI-MS/MS. To determine the structure of the HILIC-enriched GPI-peptides, we employed MALDI-QTOF-MS/MS for the analysis of the ions at m/z 1839.6, 2042.6, and 2204.7, respectively. Figure 5A displays a MALDIMS/MS spectrum of the precursor ion at m/z 1839.6. The fragment ions observed at m/z 1579.6, 1418.5, 1133.5, and 971.4 were assigned as fragments of the core-GPI-peptide precursor ion, allowing confirmation of the complete structure of the GPIanchored peptide of MDP. The fragment ion at m/z 729.3 was assigned as the tryptic COOH-terminal peptide containing an Analytical Chemistry, Vol. 78, No. 10, May 15, 2006

3339

Figure 6. MALDI-QTOF tandem mass spectra of the GPI-anchored peptides containing side-chain subtituents: (A) Precursor ion at m/z 2042.6 corresponding to the core GPI-anchor carrying an additional HexNAc unit. (B) Precursor ion at m/z 2204.7 corresponding to the core GPI-anchored peptide carrying an extra HexHexNAc/HexNAcHex. Fragment ions and corresponding carbohydrate units are indicated. Diagnostic ions at m/z 162 (GlcN), 286 (ManEtNP), 422 (GlcNInsP), and 447 (ManEtNPGlcN) are evident.

ethanolamine residue, generated upon gas-phase cleavage of the phosphodiester bond in the GPI-anchor. The mass of the MDP tryptic COOH-terminal peptide TNYGYS was calculated to be 703 Da. The TNYGYS-EtN ion fragment appeared at m/z 729 with low intensity. The most intense ion in the MS/MS spectrum at m/z 711.3 corresponds to the loss of water from TNYGYS-EtN, which is most likely eliminated from serine 384 or threonine 379, since these residues are prone to lose water in low-energy collision processes. Similar fragmentation patterns were previously described in studies of human and bovine acetylcholinesterase GPIanchor.30,31 The analogous fragment ions composed of the COOHterminal peptide and the EtN moiety were also detected by Taguchi and co-workers while studying the structure of the GPIanchored protein 5′-nucleotidase by ESI-MS/MS.32 This observation may prove useful in future mass spectrometry studies of GPIAPs because the mass of the COOH-terminal peptide with the EtN moiety can be predicted and extracted from large peptide MS datasets to confirm the assignment of GPI-APs in proteomic experiments. In the low m/z range of the tandem mass spectrum of the m/z 1839.6 precursor ion, b-type fragment ions from the tryptic COOHterminal anchored peptide TNYGYS were detected (Figure 5B). Several y ions were also detected, including y4 at m/z 496, y3 at m/z 333, and y2 at m/z 276, all of which contained the EtN moiety, and they all exhibited water loss (Figure 5B). Since no water 3340

Analytical Chemistry, Vol. 78, No. 10, May 15, 2006

losses were observed in the b ions that were detected, it seems likely that the water loss originated from serine 384 and not from threonine 379. The peptide sequence information is in agreement with a previous study showing that serine 384 is the ω-site of MDP.29 It is also in agreement with the general trend that ω-sites are often defined by small amino acid residues. Serine (48%), glycine, asparagine, aspartic acid, and cysteine are the most common residues found at ω-sites in metazoa, whereas serine (44%), aspartic acid, asparagine, alanine, and glycine are the most common residues found at ω-sites in protozoa. Several GPI-anchor-specific ion signals were detected in the MS/MS spectrum. The ion at m/z 162.1 corresponds to GlcN, a carbohydrate unit that is unique for core GPI-anchors. This fragment ion is not generated by N-linked or O-linked glycans, and therefore, it is a candidate diagnostic signal for GPI-anchors. The ion signal at m/z 286.1 corresponds to mannose ethanolamine phosphate (ManEtNP), whereas the signal at m/z 422.1 corresponds to glucosamine inositol phosphae (GlcNInsP), and the signal at m/z 447.2 corresponds to mannose ethanolamine phosphate glucosamine (ManEtNPGlcN) (Figure 5A). The presence of this set of ions in MALDI-MS/MS is a strong indication for the presence of a GPI-anchor in a protein or peptide, as also reported previously for ESI-MS/MS.32 These ion signals could potentially be used in MS/MS precursor ion scanning experiments to detect with high specificity and sensitivity the presence of GPI-

anchored peptides in complex samples. Next, MALDI-MS/MS analysis of the ions at m/z 2042.6 and m/z 2204.7 was performed (Figure 6). The mass difference of 203 Da between the GPI-peptide observed at m/z 1839.6 and the ion at m/z 2042.6 suggested the presence of an additional HexNAc moiety. This was further supported by MS/MS fragmentation, which generated a signal at m/z 204, indicating formation of a HexNAc oxonium ion (Figure 6A). The HexNAc side chain of the GPI-peptide ion at m/z 2042.6 was easily lost in the fragmentation process, and consequently, an intense fragment at m/z 1839.6 was detected. Further analysis of the tandem mass spectrum allowed us to localize the attachment point of the HexNAc side chain to the R-mannose residue adjacent to the GlcN. The ion at m/z 1782.6 corresponded to the mass of the precursor ion that lost InsP, and the ion at m/z 1621.5 corresponded to the precursor ion that eliminated GlcNInsP. The rest of the GPI-peptide fragment ions did not contain the HexNAc side chain, suggesting that all were fragment ions derived from the core-GPI-peptide at m/z 1839.6, that is, after elimination of the HexNAc side chain from the precursor ion at m/z 2042.6 (Figure 6A). The mass difference of 365 Da between the ion at m/z 2204.7 and the GPI-peptide at m/z 1839.7 indicated the addition of HexHexNAc, HexNAcHex, or separate HexNAc and Hex moieties. Detection of an ion signal at m/z 366 in the MS/MS fragment ion spectrum of the precursor ion at m/z 2204.7, suggested the presence of a HexHexNAc or HexNAcHex oxonium ion, indicating that Hex is bonded to HexNAc in a single side chain. The presence of a Galβ1-3GalNAcβ1-4 side chain bound to the R-mannose adjacent to the GlcN was reported by Brewis et al.29 The m/z 366 ion signal was not present in the MSMS spectra of the other two GPI-peptide precursor ions at m/z 1839.6 and 2042.6. The hexoseN-acetylhexoxamine oxonium ion at m/z 366 is often used to identify glycosylated peptides by MS/MS product ion and precursor ion scans.50,51 A signal at m/z 204 was also detected, which is most likely generated by further decomposition of the HexHexNAc/HexNAcHex oxonium ion. The position of the HexHexNAc/ HexNAcHex side chain could be readily localized to the third mannose adjacent to the GlcN by considering the ion signals at m/z 1945.8, corresponding to the loss of InsP, and the ion at m/z 1783.6, corresponding to the loss of GlcNInsP from the precursor ion at m/z 2204.7 (core + 365). The remaining abundant fragment ions were assigned to the fragmentation of the core-GPI-peptide at m/z 1839.7 derived from the precursor ion at m/z 2204.7 after loss of the HexHexNAc/HexNAcHex side chain. (50) Carr, S.; Huddleston, M.; Bean, M. Protein Sci. 1993, 2, 183-196. (51) Jebanathirajah, J.; Steen, H.; Roepstorff, P. J. Am. Soc. Mass Spectrom. 2003, 14, 777-784. (52) Zubarev, R. A. Curr. Opin. Biotechnol. 2004, 15, 12-16. (53) Syka, J. E. P.; Coon, J. J.; Schroeder, M. J.; Shabanowitz, J.; Hunt, D. F. Proc. Natl. Acad. Sci U.S.A. 2004, 101, 9528-9533.

In summary, these MALDI-MS and MS/MS experiments established the MDP COOH-terminal peptide sequence and ω-site of the HILIC-purified GPI-peptides. In addition, the structure of the GPI core glycan structure was analyzed, including determination of HexNAc and either a HexHexNAc or a HexNAcHex moiety that gives rise to molecular heterogeneity. These data are in agreement with previous analyses of the MDP GPI-anchor structure with the exception that no signal corresponding to an additional ethanolamine phosphate (EtNP) side chain on the second mannose was detected in the present study.29 CONCLUSION Using the GPI-anchored protein MDP, we have shown that HILIC enrichment and MALDI tandem mass spectrometry is a sensitive and simple method for purification and analysis of GPIanchored peptides. The method is compatible with standard biochemical protein separation methods, such as SDS-PAGE, and allows analysis of low-picomole levels of GPI-anchored proteins. MALDI mass spectra of the HILIC-enriched peptide fractions revealed microheterogeneity of the GPI-anchor, and MS/MS analysis facilitated direct assignment of the ω-site and determination of the core structure of GPI-anchors by amino acid sequencing of the C-terminal, modified peptide. GPI-specific diagnostic ions detected in tandem mass spectra could potentially be used in future large-scale proteomic experiments to track GPIpeptides in complex mixtures. The complex appearance of the MS/MS spectra of GPI-peptides suggests that multistage MS/ MS analysis in combination with nonergodic ion dissociation techniques, such as electron capture dissociation52 or electrontransfer dissociation,53 could be beneficial to gain more detailed insights into the structure of these species. These analytical technologies and their application in proteomics will generate a more solid foundation for the characterization of GPI-anchored proteins and for the understanding of the biology of this diverse class of plasma membrane proteins. ACKNOWLEDGMENT This study was supported by a project grant from the Danish Natural Sciences Research Council (O.N.J). M.J.O. was supported by ETORTEK Grant IE019 from the Industry Department of the Basque Government. P.H. was supported by a long-term postdoctoral fellowship from the Federation of European Biochemical Societies. O.N.J. is a Lundbeck Foundation Professor and the recipient of a Young Investigator Award from the Danish Natural Sciences Research Council. Received for review October 7, 2005. Accepted March 8, 2006. AC0517949

Analytical Chemistry, Vol. 78, No. 10, May 15, 2006

3341