Anal. Chem. 2005, 77, 7774-7782
Identification of Unusual Bacterial Glycosylation by Tandem Mass Spectrometry Analyses of Intact Proteins Michael Schirm,†,‡ Ian C. Schoenhofen,§ Susan M. Logan,§ Karen C. Waldron,† and Pierre Thibault*,†,|
Department of Chemistry and Institute for Research in Immunology and Cancer, Universite´ de Montre´ al, Montre´ al, Que´ bec, Canada, and NRC Institute for Biological Sciences, Ottawa, Ontario, Canada
The characterization of protein glycosylation can be a complex and time-consuming procedure, especially for prokaryote O-linked glycoproteins, which often comprise unusual oligosaccharide structures with no known glycosylation motif. In this report, we describe a “top-down” approach that provides information on the extent of glycosylation, the molecular masses, and the structure of oligosaccharide residues on bacterial flagella, important structural proteins involved in the motility of pathogenic bacteria. Flagella from four bacterial pathogens, namely, Campylobacter jejuni, Helicobacter pylori, Aeromonas caviae, and Listeria monocytogenes, were analyzed by this top-down mass spectrometry approach. The approach needs minimal sample preparation and can be performed within a few minutes compared to the tedious and often time-consuming “bottom-up” approach involving proteolytic digestion and LC-MS-MS analyses of the suspected glycopeptides. Multiply protonated protein precursor ions subjected to low-energy collisional activation in a quadrupole time-of-flight instrument showed extensive and specific gas-phase deglycosylation resulting in the formation of abundant oxonium ions with very few fragment ions from peptidic bond cleavages. Structural information on individual carbohydrate residues is obtained using a second-generation product ion scan of oxonium ions formed by collisional activation of the intact protein ions in the source region. The four bacterial flagella examined differed not only by the extent of glycosylation but also by the nature of carbohydrate substituents. For example, the flagellin from the Grampositive bacterium, L. monocytogenes showed O-linked GlcNAc residues at up to 6 sites/protein monomer. In contrast, the three Gram-negative bacterial pathogens C. jejuni, H. pylori and A. caviae displayed up to 19 Ser/ Thr O-linked sites modified with residues structurally related to N-acetylpseudaminic acid (Pse5Ac7Ac) and in * Corresponding author. E-mail:
[email protected]. Phone: (514) 343-6910. Fax: (514) 343-6843. † Department of Chemistry, Universite ´ de Montre´al. ‡ Present address: Caprion Pharmaceuticals, Montre ´ al, Canada, H4S 2C8. § NRC Institute for Biological Sciences. | Institute for Research in Immunology and Cancer, Universite ´ de Montre´al.
7774 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
the case of Campylobacter include a novel N-acetylglutamine substituent on Pse5Am7Ac. Glycosylation is one of the most common posttranslational modifications in eukaryotic cells, and it is estimated that 50% of all proteins are glycosylated.1 Mass spectrometry has consistently played an important role in the analysis of eukaryote N-linked glycoproteins.2-12 Attached glycans can be released chemically or enzymatically prior to their labeling and mass spectrometry analyses.3-5 The identification of glycosylation sites is usually performed using a “bottom-up” approach involving proteolytic cleavage of the original glycoprotein and separation of the corresponding products by capillary electrophoresis-nanoelectrospray mass spectrometry (CE-MS)6,7 or nanoscale liquid chromatography coupled to mass spectrometry (nanoLC-MS).8-13 The analysis of suspected glycopeptides can be achieved via the detection of characteristic oxonium ions from the cleavage of glycosidic bonds obtained from in-source fragmentation or from mass-selected precursor ion in tandem mass spectrometry experiments (MS-MS).8,9 In particular, the occurrence of specific carbohydrate residues such as hexose (Man, Glc, Gal), Nacetylhexosamine (GlcNAc, GalNAc), or N-acetylneuraminic acid (Neu5Ac) can be monitored by the observation of characteristic oxonium ions at m/z 163, 204, and 292, respectively. More recently, a “top-down” mass spectrometry approach was described for the detection of posttranslational modifications (1) Apweiler, R.; Hermjakob, H.; Sharon, N. Biochim. Acta 1999, 1473, 4-8. (2) Dell, A.; Morris, H. R. Science 2001, 291, 2351-2356. (3) Rudd, P. M.; Guile, G. R.; Kuster, B.; Harvey, D. J.; Opdenakker, G.; Dwek, R. A. Nature 1997, 388, 205-207. (4) Rudd, P. M.; Colominas, C.; Royle, L.; Murphy, N.; Hart, E.; Merry, A. H.; Hebestreit, H. F.; Dwek, R. A. Proteomics 2001, 1, 285-294. (5) Harvey, D. J. Proteomics 2001, 1, 311-328. (6) Kelly, J. F.; Locke, S. J.; Ramaley, L.; Thibault, P. J. Chromatogr., A 1996, 720, 409-427. (7) Bateman, K. B.; White, R. L.; Yaguchi, M.; Thibault, P. J. Chromatogr., A 1998, 794, 327-344. (8) Carr, S. A.; Huddleston, M. J.; Bean, M. F. Protein Sci. 1993, 2, 183-196. (9) Huddleston, M. J.; Bean, M. F.; Carr, S. A. Anal. Chem. 1993, 65, 877884. (10) Burlingame, A. L. Curr. Opin. Biotechnol. 1996, 7, 4-10. (11) Schulz, B. L.; Packer, N. H.; Karlsson, N. G. Anal. Chem. 2002, 74, 60886097. (12) Kawasaki, N.; Itoh, S.; Ohta, M.; Hayakawa, T. Anal. Biochem. 2003, 316, 15-22. (13) Wuhrer, M.; Koeleman, C. A. M.; Hokke, C. H.; Deelder, A. M. Anal. Chem. 2004, 76, 833-838. 10.1021/ac051316y CCC: $30.25
© 2005 American Chemical Society Published on Web 10/26/2005
directly from intact proteins, using either Fourier transform mass spectrometry (FTMS) with electron capture dissociation (ECD)14,15 or ion/ion proton transfer on a ion trap instrument.16 ECD cleaves peptide backbones primarily at the CR-N bond, leaving intact the labile side-chain modifications such as phosphorylation or glycosylation, thus enabling the identification of the linkage site.17 The identification of these modifications is based on the observation of common mass discrepancies (e.g., 80 Da for phosphorylation or 162 Da for the addition of a hexose residue). It is noteworthy that earlier reports preceding those of FTMS including that of Nemeth-Cawley and Rouse,18 and Loo et al.19 also described the use of quadrupole time-of-flight (Q-TOF) and triple quadrupole instruments for sequencing intact proteins via low-energy collisioninduced dissociation. Successful protein identification relied on the detection of clearly resolved b- or y-type fragment ions, a situation that depends not only on instrumental resolution and sensitivity but also on the occurrence of consecutive peptide backbone cleavages. Although the glycan structures in eukaryotes are relatively well defined, the identification of prokaryote glycoproteins is significantly more difficult due to their unusual oligosaccharide structures and their low abundance in cell extracts. While it was originally thought that glycoprotein formation was restricted only to eukaryotes and Archaea, prokaryotic N- and O-linked glycosylation are only now becoming more widely accepted as outlined in recent reviews.20,21 The most comprehensively studied prokaryotic glycoproteins are the S-layer glycoproteins of Archeae.22 Several reports documented the identification and the characterization of a number of membrane-associated, surface-associated, exoenzymes and secreted glycoproteins from diverse bacterial species.23 In addition, the glycosylation of cell surface appendages such as pili and flagella has gained attention in view of their importance in motility and colonization of host cells.24,25 Most recently, we reported that flagella from Campylobacter jejuni and Helicobacter pylori were glycosylated with N-acetylpseudaminic acid analogues and flagella of Listeria monocytogenes were glycosylated with β-GlcNAc by the traditional bottom-up approach.26-29 (14) Fridriksson, K. E.; Beavil, A.; Holowka, D.; Gould, J. H.; Baird, B.; McLafferty, W. F. Biochemistry 2000, 39, 3369-3376. (15) Sze, K. S.; Ge, G.; Oh, H.; McLafferty, W. F. Proc. Natl. Acad. Sci. U.S.A. 2001, 99, 1774-1779. (16) Reid, E. G.; Stephenson, L. J., Jr.; McLuckey, A. S. Anal. Chem. 2002, 74, 577-583. (17) Kelleher, N. L.; Zubarev, R. A.; Bush, K.; Furie, B.; Furie, B. C.; McLafferty, F. W.; Walsh, C. T. Anal. Chem. 1999, 71, 4250-4253. (18) Nemeth-Cawley, J. F.; Rouse, J. C. J. Mass Spectrom. 2002, 37, 270-282. (19) Loo, J. A.; Edmonds, C. G.; Smith, R. D. Anal. Chem. 1991, 63, 24882499. (20) Messner, P. J. Bacteriol. 2004 186, 2517-2519. (21) Schmidt, M. A.; Riley, L. W.; Benz, I. Trends Microbiol. 2003, 11, 554-561. (22) Schaffer, C.; Messner, P. Biochemie 2001, 83, 591-599. (23) Upreti, K. R.; Kumar, M.; Shankar, M. Proteomics 2003, 3, 363-379. (24) Banerjee, A.; Gosh, S. K. Mol. Cell. Biochem. 2003, 253, 179-190. (25) Castric, P.; Cassels, F. J.; Carlson, R. W. J. Biol. Chem. 2001, 276, 2647926485. (26) Thibault, P.; Logan, S. M.; Kelly, J. F.; Brisson, J.-R.; Ewing, C. P.; Trust, T. J.; Guerry, P. J. Biol. Chem. 2001, 276, 34862-34870. (27) Logan, S. M.; Kelly, J. F.; Thibault, P.; Ewing, C. P.; Guerry, P. Mol. Microbiol. 2002, 46, 587-597. (28) Schirm, M.; Soo, E. C.; Aubry, A. J.; Austin, J.; Thibault, P.; Logan, S. M. Mol. Microbiol. 2003, 48, 1579-1592. (29) Schirm, M.; Kalmokoff, M.; Aubry, A.; Thibault, P.; Samdoz, M.; Logan, S. M. J. Bacteriol. 2004, 186, 6721-6727.
In an effort to facilitate the identification of unusual bacterial glycosylation by mass spectrometry, we investigated the application of a top-down approach using a low collision energy regime on a Q-TOF instrument. We evaluated this approach on glycosylated flagellin from H. pylori, C. jejuni, and L. monocytogenes and also report novel evidence for glycosylation on the flagellin from the bacterial pathogen, Aeromonas caviae. The selectivity and specificity of this approach is also presented for the identification of unusual glycosylation from crude protein extracts of H. pylori comprising less than 20% of bacterial flagellin. MATERIALS AND METHODS Bacterial Strains, Growth Conditions, and Purification of Flagellin. Helicobacter and Campylobacter bacteria are human pathogens, and adequate precautions should be exercised in biological sample handling. Cell cultures were prepared in a level two containment laboratory and all contaminated materials autoclaved prior to disposal. C. jejuni 81-176 and 81-176 pglC isogenic mutant were obtained from Dr. P. Guerry, NMRI, Bethesda, MD, and grown in Mueller Hinton broth under microaerophilic conditions at 37 °C for 24 h. Cells were harvested, and flagellin was purified as described by Power et al.30 The bacterial isolate H. pylori 1061 was obtained from Dr. P. Hoffman (Dalhousie University, Halifax, Nova Scotia, Canada), and H. pylori NCTC11637 (ATCC, Manassas, VA,USA) was grown in BHI broth containing 5% fetal bovine serum under microaerophilic conditions at 37 °C for 48 h. Flagella were purified as described by Kostrzynska et al.,31 and further purification was achieved by centrifugation using a Centricon YM-30 membrane filter (Millipore, Bedford, MA) to remove low molecular weight protein contaminants. The bacterial isolate L. monocytogenes 1174 was obtained from Listeria Reference Service collection in the Bureau of Microbial Hazards (Bureau of Microbial Hazards, Food Directorate, Health Products and Food Branch, Health Canada) and was grown at 21 °C for 72 h in BHI broth. The flagellin was purified as described previously.29 A. caviae UU51 was obtained from Dr. J. Dooley (School of Biological and Environmental Sciences, University of Ulster, Coleraine, Co. Londonderry, Northern Ireland) and was originally isolated from human feces. Cells were grown in BHI broth and grown statically overnight at 37 °C for production of polar flagella. Flagellin was purified according to the method of Rabaan et al.32 Flagellin protein yields varied for respective strains. Approximately 0.5-1.0 mg of purified bacterial flagellin was obtained from 500-mL culture volume grown overnight. Mass Spectrometry. For the intact mass spectrometry analyses, flagellin were dialyzed in H2O (0.2% formic acid) using Centricon YM-30 membrane filter (Millipore) with a molecular weight cut off of 30 000 to remove salts. Flagellin extracts (typically 200 µL) were concentrated to ∼100 µL of a 0.2 mg/mL solution. This solution was infused into a Waters Q-TOF Ultima mass spectrometer at a flow rate of 0.5 µL/min. External calibration of the TOF was performed by infusing a 150 fmol/µL solution of Glu-fibrinopeptide B (50% aqueous methanol, 0.2% formic acid) (30) Power, M. E.; Guerry, P.; McCubbin, W. D.; Kay, C. M.; Trust, T. J. J. Bacteriol. 1994, 176, 3303-3313. (31) Kostrzynska, M.; Betts, J. D.; Austin, J. W.; Trust, T. J. J. Bacteriol. 1991, 173, 937-946. (32) Rabaan, A. A.; Gryllos, I.; Tomas, J. M.; Shaw, J. G. Infect. Immun. 2001, 69, 4257-4267.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
7775
Table 1. Unusual Modifications and Extent of Glycosylation Found in Bacterial Flagellin Examined bacterium
strain
observed mass (Da)
expected mass from gene product (Da)
mass exces (%)
H. pylori
1061
55367
53153
4.2
A. caviae
UU51
33485-34010
7.2-8.9
L. monocytogenes
1174
31050-31659
31244 (FlaA) 31094 (FlaB) 30445
2.0-4.0
C. jejuni
81-176
65400-66400
59240
10.4-12.1
a
oligosaccharide residues and sites of modificationa Pse5Ac7Ac T180, T363, S207, S245, S353, S394, S402 Pse5Ac7Ac9Ac FlaA (6 sites), FlaB (7 sites) GlcNAc T141,T145, T173, T175, S181 Pse5Ac7Ac, Pse5Am7Ac, Pse5Ac7Ac8OAc, Pse5Am7Ac8GlnAc 19 O-linked Ser/Thr residues
Location of modification sites according to refs 26, 28, and 29.
at 0.6 µL/min. Mass accuracy was typically within (0.07 m/z unit across the acquisition range (m/z 50-2000) in both MS and MSMS modes. For internal calibration in MS-MS mode, fragment ions of known composition were used to bracket the ion of interest and provided mass assignment within 10 ppm of predicted measurements. Resolution was typically 9000 (50% valley definition) in V-mode detection. Molecular mass profiles were obtained through spectral deconvolution using MaxEnt (MassLynx software, Waters). Collision-induced dissociation (CID) experiments were performed using argon as collision gas with differential voltage values of 10-25 V between the collision cell and the incoming ions. Second-generation fragment ion spectra were obtained by increasing the rf lens1 voltage from 50 to 125 V, thereby forming fragment ions in the high-pressure region of the skimmer/cone region of the mass spectrometer. PCR Amplification and Sequencing of the Polar Flagellin Genes from A. caviae UU51. Polar flagellin genes were amplified from A.caviae strain UU51 by PCR using primers based on the 5′ and 3′ sequences from the flagellin genes of A. caviae Sch3 (accession number AF198617). The PCR-amplified products were gel purified and cloned into pUC19. An Applied Biosystems 373 DNA sequencer and Taq cycle sequencing kits with terminator chemistry were used to sequence the plasmid DNA and the corresponding flagellin nucleotide sequences deposited in Genbank (flaA accession AY839591, flaB AY839592). The predicted amino acid sequences of the flaA and flaB genes were used to determine the predicted molecular weight of the FlaA and FlaB proteins. RESULTS AND DISCUSSION Molecular Weight Determination of Flagellin Monomers. Preliminary experiments were first performed to profile the microheterogeneity of flagellin proteins from the different human pathogens and to optimize collisional activation conditions favoring the selective dissociation of the attached oligosaccharides. Bacterial flagellin extracts were subjected to dialysis with 0.2% aqueous formic acid on a 30-kDa cutoff membrane to remove low molecular weight contaminants present in the extracts. The dialyzed flagellin solutions were infused at a flow rate of 500 nL/min into the mass spectrometer and yielded characteristic multiply charged state envelopes for all four bacterial protein extracts examined (Figure 1). The nanoelectrospray mass spectra data were deconvoluted to obtain a molecular mass profile, which enabled comparison with expected masses from cDNA sequences (Table 1). The molecular masses observed and satellite peaks are typically associated with 7776 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
Figure 1. Molecular mass determination of intact bacterial flagellin from (a) L. monocytogens, (b) C. jejuni, (c) A. caviae, and (d) H. pylori. The molecular mass profile obtained from each corresponding spectrum is presented as an inset on the corresponding panel. In all cases, the observed molecular mass was higher than that predicted from the cDNA sequence, suggesting posttranslational modifications.
alkali adducts or lower abundance genetic variants that could be correlated with predicted masses deduced from cDNA sequences. In all four cases, the observed molecular masses were higher than those predicted from their corresponding cDNA sequences, thus suggesting posttranslational modifications. The L. monocytogenes flagellin revealed four peaks separated by 203 Da corresponding to mass excess of 605, 808, 1011, and 1214 Da from the expected protein molecular mass (Figure 1a). These mass shifts were consistent with the addition of single GlcNAc residues at up to six sites of O-linked glycosylation.29 Distinct mass excess of 2212 Da and of ∼6500 Da were observed on H. pylori (Figure 1d) and C. jejuni (Figure 1b) flagella in good agreement with measurements reported previously on the same bacterial strains.26,28 The deconvoluted spectrum of the flagellin from A. caviae (Figure 1c) indicated a flagellin protein with 2-3 kDa mass excess from the predicted FlaA or FlaB amino acid sequences (Table 1). However, the mass excess observed for the A. caviae flagellin could not be accounted to any of the previously reported modifications of other
Figure 2. Tandem mass spectrometry analyses of the multiply charged ion m/z 1049.7 from L. monocytogens flagellin acquired at (a) 5, (b) 10, (c) 15, and (d) 20 V collision energy using argon as target gas.
bacterial flagella. Further structural characterization was undertaken to probe the nature of these modifications using product ion mass spectra of intact flagellin monomers. Tandem Mass Spectrometry Analyses of Multiply Protonated Flagellin Precursor Ions. The labile nature of the glycosidic bond was exploited previously in CID experiments of glycoprotein digests to yield characteristic oxonium diagnostic ions that facilitated the location of the glycopeptides within complex proteolytic digests.6-13 MS-MS spectra of N- and O-linked glycopeptides are typically dominated by fragment ions associated with consecutive dissociation of glycosidic bonds with few peptide bond cleavages, a hallmark enabling the characterization of oligosaccharide chains of glycopeptides. These distinctive features prompted us to examine the analytical potential of tandem mass spectrometry for the direct identification of undigested glycoproteins. Collisional activation conditions favoring specific cleavage of carbohydrate residues from the intact glycoprotein were optimized using L. monocytogenes flagellin (Figure 2). Multiply protonated precursor ions subjected to low-energy CID in an rf-only quadrupole collision cell do not undergo significant dissociation at a collision voltage less than 10 V as evidenced in Figure 2a and b for the precursor ion m/z 1049.7 corresponding to [M + 30H]30+. However, a further increase in collision energy resulted in consecutive losses of GlcNAc residues from the precursor ions at 15 V (Figure 2c) whereas peptide backbone cleavages typically require a voltage of at least 20 V (Figure 2d). It is noteworthy that most of the abundant low-energy peptide fragment ions observed in Figure 2d are attributed to cleavage of the C-terminus amino acids giving rise to a series of y-type fragment ions. A collision voltage of 15 V was thus found satisfactory to yield specific glycosidic bond cleavages under the present collisional activation conditions. The specificity of the present approach for the identification of potential glycoprotein candidates was investigated with samples
Figure 3. Mass spectrometry analyses of a crude protein extract from H. pylori. (a) Conventional electrospray mass spectrum. Tandem mass spectra of (b) m/z 842, (c) 1017, and (d) 1154 from precursor ions highlighted in panel a. An intense oxonium ion corresponding to Pse5Ac7Ac is only observed for the glycosylated flagellin (d) whereas other H. pylori nonglycosylated proteins do not yield any carbohydrate specific fragment ions. Conditions as for Figure 2 except that collision energy was set to 15 V.
of higher complexity comprising both bacterial flagellin and nonglycosylated proteins of higher abundance. This is illustrated in Figure 3 for the analysis of H. pylori flagellin obtained from shearing of the bacterial cells and centrifugation without further dialysis. The nanoelectrospray mass spectrum of the corresponding sample showed a complex pattern of multiply charged ions associated with several protein components (Figure 3a). The SDS-PAGE analysis of this sample identified three major bands at approximately 20, 30, and 65 kDa, which upon in-gel digestion were identified as neutrophil activating protein, urease, and FlaA, respectively (data not shown). Although FlaA was clearly observed from the silver-stained SDS-PAGE (∼20% of the protein extract), its detection by nanoelectrospray mass spectrometry was compromised by the ionization of other proteins present in the same extract (Figure 3a). Product ion spectra were acquired on selected precursor ions for each of the identified proteins (Figures 3b-d). The MS-MS spectra of m/z 842 and 1017 showed low-abundance multiply charged fragment ions corresponding to cleavage of peptide bonds. In contrast, the MS-MS spectrum of m/z 1154 was characterized by an intense Pse5Ac7Ac oxonium ion at m/z 317 together with fragment ions at m/z 299 and 281 corresponding to consecutive losses of water molecules from the monosaccharide oxonium ion. The labile nature of glycosidic bonds thus favored the cleavage of O-linked monosaccharide residues from glycoproteins subjected to low collision activation, a unique feature that facilitates their identification from nonglycosylated proteins as shown in Figure 3b-d. Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
7777
Figure 4. MS-MS spectra of multiply charged precursor ions m/z 1155.7 from A. caviae and m/z 1196.5 from C. jejuni flagellin. Conditions as for Figure 2.
Additional evidence of this distinct feature is presented in Figure 4 for the MS-MS analyses of intact flagellin from A. caviae and C. jejuni strain 81-176. The MS-MS spectrum of m/z 1155.7 from A. caviae flagellin (Figure 4a) shows a predominant fragment ion at m/z 374 assigned to an unusual monosaccharide residue (see below). Fragment ions of lower m/z values were subsequently assigned to cleavages of functionalities associated with this unusual monosaccharide residue. The MS-MS spectrum of m/z 1196.5 from C. jejuni flagellin (Figure 4b) revealed the expected N-acetyl- and N-acetamidino-pseudaminic acid derivatives Pse5Ac7Ac and Pse5Am7Ac with their characteristic oxonium ions at m/z 317 and 316, respectively. Interestingly, an unexpected fragment ion at m/z 486 was also observed in Figure 4b, possibly suggesting the presence of an additional but yet unreported modification on C. jejuni flagellin. Further characterization of these unusual modifications was undertaken using second-generation tandem mass spectrometry of the intact proteins together with analyses of their corresponding tryptic digestion products. Structural Characterization of Unusual O-Linked Monosaccharide Residues by Tandem Mass Spectrometry Analyses of Second-Generation Product Ions. First-generation product ion mass spectra of intact multiply charged flagellin ions confirmed the presence of O-linked monosaccharide residues such as those of Pse5Ac7Ac and Pse5Am7Ac found in C. jejuni flagellin. However, more in-depth structural analyses were required to identify closely related pseudaminic acid analogues that showed increased complexity. In this context, in-source fragmentation was used on intact flagellin ions to promote the formation of oxonium fragment ions for subsequent tandem mass spectrometry analyses. This approach referred to as second-generation product ion was previously used by our group to aid the interpretation of glycopeptide tandem mass spectra and identify peptide backbone sequences7 or unusual carbohydrate moieties.26,28 Tandem mass spectrometry analyses of second-generation products of oxonium ions are shown in Figure 5 for all four flagellin proteins examined. In the case of L. monocytogenes (Figure 7778 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
Figure 5. Second-generation product ion spectra of oxonium ions obtained from in-source dissociation of multiply charged flagellin ions. (a) MS-MS spectrum of m/z 204 from L. monocytogenes, (b) MSMS spectrum of m/z 317 from H. pylori, (c) MS-MS spectrum of m/z 374 from A. caviae, and (d) MS-MS spectrum of m/z 486 from C. jejuni. Potential structures for the monosaccharide residues are shown as an inset on each corresponding mass spectrum. Conditions: Insource fragmentation was obtained by increasing the rf lens 1 from 50 to 125 V to produce oxonium fragment ions from the native flagellin in the orifice/skimmer region. Argon was used as target gas at collision energy of 30 V.
5a), dissociation of m/z 204 led to consecutive losses of water, together with cleavages of the side-chain CH2dO moiety from C5 and ketene (CH2dCdO) from the N-acetyl group consistent with that of a HexNAc residue. This spectrum was compared against a reference spectrum of GlcNAc from our laboratory, and both yielded superimposable fragmentation patterns (data not shown). Recent immunoreactivity experiments performed on Western blots of L. monocytogenes flagellin with specific monoclonal antibody confirmed this monosaccharide residue to be a β-O linked GlcNAc.29 Similarly, the MS-MS spectrum of the oxonium ion m/z 317 obtained from H. pylori flagellin (Figure 5b) was characterized by consecutive losses of neutral groups such as water, ketene, and formic acid, and fragmentation patterns observed here were identical to those previously observed for O-linked Pse5Ac7Ac tryptic glycopeptides of C. jejuni and H. pylori flagella.26,28 For convenience, the rationalization of the fragment ions observed for Pse5Ac7Ac is shown in Figure 6. Second-generation product ions from the intact flagellin multiply charged ions also facilitated the identification of novel and unexpected posttranslational modification as evidenced in Figure 5c for A. caviae flagellin. While evidence has been presented in the literature describing a group of genes (flmA, flmB, flmD, neuA, neuB) that appear to be involved in glycosylation of Aeromonas flagellin,33,34 the linkage type and structural nature of
Figure 6. Fragmentation pathways of N-acetylpseudaminic acid Pse5Ac7Ac, a monosaccharide residue found in both C. jejuni and H. pylori flagellin.
the glycan remains unknown. The molecular mass profile of A. caviae flagellin showed three discrete peaks at 33 485 ( 2, 33 708 ( 2, and 34 010 ( 3 Da (Figure 1c). MS-MS analyses of multiply charged ion precursors from all three protein peaks showed an abundant oxonium ion at m/z 374. Comparison of the observed molecular mass with that predicted from the cDNA sequence of FlaA (31 244 Da) and FlaB (31 094 Da), suggested that the peak at 33 485 Da corresponds to FlaA baring six modified residues (Mtheo., 33 483 Da) whereas the peak at 33 708 Da could be accounted for by FlaB with seven modified residues (Mtheo., 33 706 Da). The origin of the peak at 34 010 Da remains unknown but could be attributed to a genetic variant with substituted amino acids or a lateral flagellin protein unexpected to be expressed under the growth conditions used. Examination of fragmentation patterns of the oxonium ion m/z 374 produced from in-source dissociation of the intact A. caviae flagellin (Figure 5c) revealed structural features similar to those observed for Pse5Ac7Ac (Figure 5b). Indeed, backbone fragment ions at m/z 134, 162, 180, 221, 281, and 299 were observed for oxonium ions derived from O-linked monosaccharides of both A. caviae and H. pylori flagella. However, notable differences were also observed in the fragmentation of the oxonium ion at m/z 374 with the loss of water molecules and a 31 Da neutral moiety, presumably CH3NH2 from the precursor ion. Based on the mass of this residue and its similarity of fragmentation pattern to Pse5Ac7Ac (Figure 6), the O-linked monosaccharide of A. caviae flagellin could be accounted for by a Pse5Ac7Ac modified with an extra glycine residue via an ester bond as previously observed for H. influenzae core lipopolysaccharides.35 The position of this residue was tentatively assigned to C8 instead of C4 based on the previous observation of an O-acetyl substituent at C8.26 Although the formation of a tri-N-acetylated pseudaminic acid is possible based on the mass assignment, this proposal was deemed unlikely (33) Grillos, I.; Shaw, J. G.; Gavin, R.; Merino, S.; Tomas, J. M. Infect. Immun. 2001, 69, 65-74. (34) Power, M. P.; Jennings, P. M. FEMS Microbiol. Lett. 2003, 218, 211-222. (35) Li, J.; Bauer, S. H. J.; Mansson, M.; Moxon, E. R.; Richards, J. C.; Schweda, E. K. H. Glycobiology 2001, 11, 1009-1015.
in view of the biosynthetic requirements for an unusual amino sugar precursor. Further NMR structural characterization is currently underway to unambiguously identify the nature of this O-linked monosaccharide, and results from these analyses will be reported separately. Examination of the MS-MS spectrum of the intact C. jejuni flagellin (Figure 4b) revealed the expected oxonium ions for Pse5Am7Ac and Pse5Ac7Ac at m/z 316 and 317 but also provided additional evidence for another unusual modified residue giving rise to a fragment ion at m/z 486 not observed previously in bottom-up analyses of the same C. jejuni flagellin. Secondgeneration product ion spectrum of m/z 486 (Figure 5d) revealed that this novel O-linked glycan comprises a Pse5Am7Ac (m/z 316) together with an unexpected residue showing a fragment ion at m/z 171. Accurate mass determination on fragment ions at m/z 486 (observed, m/z 486.219 ( 0.005) and 171 (observed, m/z 171.077 ( 0.002) identified a single empirical formula C7H11N2O3 (calculated, m/z 171.0770) within the precision of the experimental mass measurements. The most probable structural assignment for this moiety is an N-acetylglutamine linked to the Pse5Am7Ac residue via an ester bond. This proposal was confirmed separately on a commercial sample of the N-acetylglutamine (GlnAc) where the fragment ion m/z 171 arising from the loss of a water molecule from the protonated molecule was selected for MS-MS analysis. The corresponding MS-MS spectrum yielded fragmentation patterns comparable to those observed for m/z 171 produced through second-generation product ion of the multiply charged flagellin ions (data not shown). In an effort to locate the site of modification of this unusual O-linked disaccharide, further structural analyses were conducted using the more traditional bottom-up approach. Results from these analyses are presented in Figure 7 for the nanoLC-MS-MS of tryptic digests from C. jejuni flagellin. Detailed examination of the released tryptic peptides revealed a single glycopeptide candidate with a molecular mass of 5608.5 ( 0.1 Da (Figure 7a). The MS-MS analysis of the corresponding pentuply charged ions at m/z 1123.1 confirmed the presence of the suspected Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
7779
Figure 7. Identification of glycosylation site from tryptic digest of C. jejuni flagellin using nanoLC-MS-MS analysis. (a) Extracted mass spectrum from nanoLC-MS analysis showing an heterogeneous population of glycopeptides. Inset shows the [M + 5H]5+ ion from the modified tryptic peptide 390-422 with a monoisotopic mass of 5608.5 Da. (b) MS-MS spectrum of m/z 1123.1 showing characteristic oxonium ions at m/z 317, 359, and 486 corresponding to Pse5Ac7Ac, Pse5Ac7Ac8OAc, and Ac-N-Gln- Pse5Am7Ac, respectively. Conditions as for Figure 5.
Pse5Am7Ac8GlnAc together with Pse5Ac7Ac and Pse5Ac7Ac8OAc residues (Figure 7b). Based on these data and those previously reported for C. jejuni 81-176 tryptic digest,26 the sequence of this glycopeptide was assigned to tryptic peptide 390-422 with an oxidized Met and containing Pse5Ac7Ac3, Pse5Am7Ac8OAc1, and Pse5Am7Ac8GlnAc2 residues. It is noteworthy that a more abundant glycopeptide with the same sequence but comprising Pse5Ac7Ac3, Pse5Am7Ac8OAc1, and two unusual residues showing oxonium ions at m/z 409 was also found in the same digest consistent with previous investigations.26 This finding could potentially indicate a degradation of the Pse5Am7Ac8GlnAc residue giving rise to a loss of two water molecules and the acetamidino group under the current tryptic digestion conditions used. Application of Top-Down Mass Spectrometry To Identify Glycosylation Status and To Infer Structure-Function Relationship in Isogenic Mutant Strains. The present top-down approach was used in two examples to assess its effectiveness as a tool for the probing the protein glycosylation status. In the first example, top-down analysis was used to determine the effect of selected C. jejuni 81-176 isogenic mutants on glycosylation status of the flagellin protein. A Cj1316 mutant of 81-176 obtained from insertional inactivation of the corresponding gene was previously shown by bottom-up analysis to abrogate the synthesis of the acetamidino variant, Pse5Am7Ac, and the resulting flagellin protein was glycosylated only with Pse5Ac7Ac residues.26 In the current study, the top-down tandem mass spectrometry approach also confirmed the combined loss of Pse5Am7Ac and Pse5Am7Ac8GlnAc consistent with our previous report (data not shown). 7780 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
Flagellar glycosylation status in C. jejuni 81-176 was shown previously to be affected by the pgl genetic locus as evidenced by altered immunoreactivity patterns.36 In particular, the immunoreactivity by Western blotting of flagellin from a pglC mutant examined with O:23 and O:36 typing sera was greatly reduced, suggesting an alteration of the flagellin protein in this strain. Unlike the Cj1316 mutation, where the IEF pattern of flagellin was altered upon loss of Pse5Am7Ac, the IEF pattern of the pglC mutant flagellin was comparable to that of the wild-type strain (P. Guerry, personal communication). Examination of the flagellin from a C. jejuni 81-176 pglC mutant by nanoelectrospray mass spectrometry (Figure 8a) showed a narrow molecular envelope in contrast to that observed previously for the wild-type strain (Figure 1b). Tandem mass spectrometry analysis performed on m/z 1212.5 ([M + 54H]54+ ion) revealed the expected oxonium ions m/z 316 and 317 corresponding to Pse5Am7Ac and Pse5Ac7Ac (Figure 8b). However, in comparison to C. jejuni 81-176 parent flagellin (Figure 4b), no evidence of the oxonium ion at m/z 486 was apparent, suggesting a loss of the Pse5Am7Ac8GlnAc residue on the flagellin protein. Based on these observations, the flagellin from the mutant strain comprised only Pse5Am7Ac and Pse5Ac7Ac residues at 19 Ser/Thr O-linked glycosylation sites. While the pgl gene locus had been shown in earlier work to affect flagellin immunoreactivity, this top-down analysis identifies a specific change in modification pattern as a consequence of inactivation of pglC. We are currently investigating the role of pglC in the biosynthesis of Pse5Am7Ac8GlnAc. (36) Szymanski, C.; Yao, R.; Ewing, C. P.; Trust, T. J.; Guerry, P. Mol. Microbiol. 1999, 32, 1022-1030.
Figure 8. Probing glycosylation status in flagellin from H. pylori NCTC11637 and in mutant strain plgC from C. jejuni 81-176 using top-down mass spectrometry. (a) Nanoelectrospray mass spectrum of flagellin from plgC from C. jejuni 81-176. Inset shows the reconstructed molecular mass profile. (b) Tandem mass spectrum of m/z 1212.5 (arrow in a) showing oxonium ions at m/z 316 and 317 with no detectable fragment ions at m/z 486 (Pse5Am7Ac8GlnAc). (c) Nanoelectrospray mass spectrum of flagellin from H. pylori NCTC11637 with its corresponding molecular mass profile. (d) Tandem mass spectrum of m/z 1046.2 (arrow in c) showing the presence of Pse5Ac7Ac residue as indicated by its corresponding oxonium ion at m/z 317. Conditions as for Figure 5.
In a second experiment, the top-down method was used to rapidly establish the glycosylation status of flagellin from a second strain of H. pylori NCTC11637. A recent report in the literature suggested that, unlike H. pylori strain 1061, the flagellin from H. pylori strain NCTC11637 was not glycosylated.37 We purified flagellin from this strain and subjected it to both intact mass analysis and top-down analysis to establish if indeed the flagellin was nonglycosylated. The results of this analysis are presented in Figure 8c and d. The deconvoluted molecular mass profile of the flagellin protein (Figure 8c) showed an intact mass of 55 395 ( 4 Da corresponding to an excess mass of 2108Da from the predicted sequence (53 287 Da, accession number AY319294). The tandem mass spectrometry analysis of the multiply charged ion m/z 1046.2 ([M + 51H]51+ ion) is presented in Figure 8d and is characterized by an intense Pse5Ac7Ac oxonium ion at m/z 317 together with a fragment ion at m/z 299 corresponding to the loss of water from the monosaccharide oxonium ion. The difference in mass of 2108 Da from the predicted sequence corresponds to seven sites of glycosylation with Pse5Ac7Ac. No other novel fragment ions corresponding to unique carbohydrate moieties were observed, and this analysis clearly demonstrates that the flagellin of H. pylori NCTC11637 flaA protein is indeed glycosylated with Pse5Ac7Ac in a fashion identical to that of H. pylori 1061. (37) Merkx-Jacques, A.; Obhi, R. K.; Bethune, G.; Creuzenet, C. J. Bacteriol. 2004, 186, 2253-2265.
It is noteworthy that phase variation is unlikely to explain the differences obtained between this study and that reported previously by Merkx-Jacques for H. pylori 1061 flagellin.37 Indeed, previous investigations indicated that insertional inactivation of flagellar glycosylation genes affects flagellar assembly and leads to loss of motility.28 Therefore, it is unlikely that a motile phenotype would be observed if glycosylation is downregulated through phase variation. More importantly, the earlier observation suggesting that H. pylori flagelin was not glycosylated was based on an indirect crude enzymatic treatment of bacterial cells to remove the glycan components.37 The positive control Campylobacter flagellin is unsheathed (i.e., lacks an extraneous membrane covering) and so appears to be accessible to the enzymatic treatment. However, Helicobacter flagellin is a sheathed flagellin, and as such, the extracellular covering makes the filament much more impervious to the enzymatic deglycosylation reaction resulting in unchanged migration patterns on SDS-PAGE. The direct examination of flagellin protein thus provides more detailed structural information, which facilitates the interpretation of glycosylation patterns and correlation of molecular profiles with that predicted from cDNA sequence of H. pylori 1081 flagellin. CONCLUSION This investigation presents a new approach to analyze samples and probe unusual posttranslational modifications present in bacterial glycoproteins. In contrast to the time-consuming protein Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
7781
digestion and nanoLC-MS-MS analyses, the top-down mass spectrometry approach provides a simple and rapid detection method for the identification of characteristic signature ions from suspected glycoproteins. In a first step, measurements on the intact protein molecular mass provide an estimate of the extent of glycosylation present when prior sequence information is available. In a second step, the nature of the glycosylation can be obtained from CID of one of the intact multiply charged protein ions, favoring extensive gas-phase deglycosylation and formation of abundant carbohydrate oxonium ions with minimal protein backbone cleavages. Confirmation of the potential structure of the carbohydrate residues can be achieved in a third step by acquiring second-generation MS-MS spectra of oxonium ions formed upon collisional activation in the orifice/skimmer region of the electrospray interface. The present approach was demonstrated for the analysis O-linked glycans but could also be applicable to N-linked glycoproteins due to labile glycosidic bonds resulting in the formation of oxonium fragment ions of truncated glycan chains.7 In addition, the present analytical method provides a rapid and relatively straightforward approach to routinely screen protein samples to identify unusual carbohydrate residues and to facilitate the profiling of glycosylation changes for genetically manipulated strains involved in the biosynthesis pathways of novel glycoproteins. In contrast to the laborious and costly bottom-up approach where particular modifications may be missed, the approach described here potentially provides a higher throughput for the examination of protein glycosylation and the identification of
7782
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
unusual carbohydrate features. On the other hand, this approach complements and facilitates the traditional bottom-up method by identifying characteristic carbohydrate features that can be probed using precursor ion scanning of glycoprotein proteolytic digests. These results also show that while glycosylation of flagellin protein is quite common in bacteria, there is significant diversity in terms of glycan residues found on respective flagella. Flagellin from the Gram-negative bacteria H. pylori, C. jejuni, and A. caviae were glycosylated with structurally related monosaccharides of N-acetylpseudaminic acid. In contrast, flagellin from the Grampositive bacterium L. monocytogenes was modified with a single GlcNAc at up to six sites. The flagellin of C. jejuni was also shown to be glycosylated with a novel glycan of mass 486 Da comprising an N-acetylglutamine. This novel posttranslational modification had not been identified by the traditional bottom-up approach possibly as a consequence of its instability. ACKNOWLEDGMENT The authors are grateful to Caprion Pharmaceuticals for access to mass spectrometry instrumentation. We thank Dr. Pat Guerry (NMRI, Bethesda, MD) for C. jejuni pglC flagellin and A. Aubry for technical assistance. Financial support to P.T. was provided through an NSERC discovery grant and to S.M.L. and I.S. by NRC’s Genomics and Health Initiative. Received for review July 25, 2005. Accepted September 20, 2005. AC051316Y
