Analysis of Bacterial Lipid-Linked Oligosaccharide Intermediates

Sep 22, 2009 - To date, most techniques to analyze lipid-linked oligosaccharides (LLOs) of these pathways involve the use of radiolabels and chromatog...
1 downloads 10 Views 964KB Size
Download PDF
Anal. Chem. 2009, 81, 8472–8478

Analysis of Bacterial Lipid-Linked Oligosaccharide Intermediates Using Porous Graphitic Carbon Liquid Chromatography-Electrospray Ionization Mass Spectrometry: Heterogeneity in the Polyisoprenyl Carrier Revealed Christopher W. Reid,† Jacek Stupak,† Christine M. Szymanski,‡ and Jianjun Li*,† National Research Council, Institute for Biological Sciences, 100 Sussex Drive, Ottawa, ON, Canada, K1A 0R6, and Department of Biological Sciences, Alberta Ingenuity Centre for Carbohydrate Science, University of Alberta, Edmonton, AB, Canada, T6G 2E9

N-Glycosylation of proteins is recognized as one of the most common post-translational modifications. It was believed that N-glycosylation occurred exclusively in eukaryotes until the recent discovery of the general protein glycosylation pathway (Pgl) in Campylobacter jejuni, which has similarities to the eukaryotic system and adds proteins en bloc from a lipid carrier to a protein acceptor. In addition to N-linked glycans, a number of pathogenic bacteria such as Pseudomonas aeruginosa and Neisseria species have been shown to O-glycosylate their proteins through polyisoprene-linked intermediates. To date, most techniques to analyze lipid-linked oligosaccharides (LLOs) of these pathways involve the use of radiolabels and chromatographic separation. With the increasing frequency of reports of bacterial protein glycosylation that proceed through lipid-mediated steps, there is a need for technologies capable of characterizing these newly described bacterial systems as well as eukaryotic pathways from biologically relevant samples in an accurate, rapid, and cost-effective manner. In this paper, a new glycomics strategy based on porous graphite carbon (PGC) liquid chromatography mass spectrometry (LC-MS) was devised and validated on the C. jejuni N-glycan pathway. Lipidlinked oligosaccharide intermediates of the Pgl pathway from crude lipid extracts were separated using online chromatography on a capillary PGC column with a chloroform gradient. By exploiting the retention properties of hydrophobic and polar analytes on PGC, baseline separation of LLOs with minor changes in oligosaccharide structure and polyisoprene chain length was obtained. This method is capable of analyzing low levels of LLOs (from approximately 106 bacterial cells) and distinguishing the LLOs that differ by as little as one monosaccharide or polyisoprene unit. Furthermore, we have demonstrated for the first time that oligosaccharides of the C. jejuni Pgl pathway are assembled on different * To whom correspondence should be addressed. Tel: (613) 990-0558. Fax: (613) 952-9092. E-mail: [email protected]. † National Research Council, Institute for Biological Sciences. ‡ University of Alberta.

8472

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

polyisoprenes, e. g. C45, C60, and apparent hydroxylated forms, in addition to those previously reported (i.e., C50 and C55). The hydroxylated forms of the LLOs are believed to be an intermediate in the degradation of accumulated LLOs for polyisoprene carrier recycling. Traditionally believed to occur solely in Eukarya, it is now evident that both Bacteria and Archaea are capable of protein N-glycosylation.1-4 These organisms assemble a diversity of glycan compositions and structures compared to those found in eukaryotic cells. There is a burgeoning interest in protein glycosylation in bacteria, due to the increasing frequency with which this post-translational modification is seen in pathogenic species.5 In the past decade, protein N-glycosylation has been identified in the intestinal pathogen Campylobacter jejuni,5 and O-linked protein glycosylation has been identified in pathogens such as Neisseria,6 Helicobacter,7 and Pseudomonas.8 The protein associated glycans are generally attached to serine or threonine (O-glycosylation) or to asparagine (N-glycosylation) residues. There are two different mechanisms for protein glycosylation that can be differentiated on the basis of the mode in which the glycans are transferred to proteins. The first mechanism involves the transfer of carbohydrates directly from nucleotide-activated sugars to acceptor proteins. This mechanism is typical of O-glycosylation in eukaryotes9 and flagellar glycosylation in several bacterial species.8-10 In the alternate mechanism, an oligosaccharide is (1) Abu-Qarn, M.; Eichler, J.; Sharon, N. Curr. Opin. Struct. Biol. 2008, 18, 544–550. (2) Eichler, J.; Adams, M. W. W. Microbiol. Mol. Biol. Rev. 2005, 69, 393– 425. (3) Messner, P. J. Bacteriol. 2004, 186, 2517–2519. (4) Weerapana, E.; Imperiali, B. Glycobiology 2006, 16, 91R–101. (5) Szymanski, C. M.; Yao, R.; Ewing, C. P.; Trust, T. J.; Guerry, P. Mol. Microbiol. 1999, 32, 1022–1030. (6) Banerjee, A.; Ghosh, S. K. Mol. Cell. Biochem. 2003, 253, 179–190. (7) Schirm, M.; Soo, E. C.; Aubry, A. J.; Austin, J.; Thibault, P.; Logan, S. M. Mol. Microbiol. 2003, 48, 1579–1592. (8) Castric, P.; Cassels, F. J.; Carlson, R. W. J. Biol. Chem. 2001, 276, 26479– 264585. (9) Spiro, R. G. Glycobiology 2002, 12, 43R–56. (10) Prendergast, M. M.; Moran, A. P. J. Endotoxin Res. 2000, 6, 341–359. 10.1021/ac9013622 CCC: $40.75  2009 American Chemical Society Published on Web 09/22/2009

Figure 1. Protein N- and O-glycosylation pathways in Bacteria and Archaea1 proceeding through polyisoprene-linked intermediates.

preassembled on a polyisoprene carrier before being transferred en bloc to protein acceptors by an oligosaccharyltransferase (Figure 1). This mechanism is utilized in N-glycosylation pathways in eukaryotes, bacteria,4 and archaea11 and the unique O-linked glycosylation pathways of P. aeruginosa and Neisseria species. The N-glycosylation system of C. jejuni is encoded by the pgl operon.5 The glycan in the C. jejuni pathway is a conserved heptasaccharide with the structure GalNAc-R1,4-GalNAc-R1,4-[Glcβ1,3]-GalNAc-R1,4-GalNAc-R1,4-GalNAc-R1,3-Bac-β1, where Bac is 2,4-diacetamido-2,4,6-trideoxyglucopyranose.12 This glycan is synthesized on the polyisoprenyl carrier undecaprenyl phosphate (Und-P) on the cytoplasmic face of the inner membrane through the action of a UDP-Bac transferase (PglC) and several glycosyltransferases (PglA, PglH, PglI, and PglJ; Figure 1 top panel, Table 1). The assembled oligosaccharide is then flipped to the periplasm by the action of the membrane-associated flippase (PglK). Once in the periplasm, the oligosaccharide is transferred to proteins (11) Chaban, B.; Voisin, S.; Kelly, J.; Logan, S. M.; Jarrell, K. F. Mol. Microbiol. 2006, 61, 259–268. (12) Young, N. M.; Brisson, J.-R.; Kelly, J.; Watson, D. C.; Tessier, L.; Lanthier, P. H.; Jarrell, H. C.; Cadotte, N.; St.Michael, F.; Aberg, E.; Szymanski, C. M. J. Biol. Chem. 2002, 277, 42530–42539.

Table 1. Components of the C. jejuni Pgl Pathway and Their Corresponding Phenotype upon Deletion protein

function

PglF

PglJ PglI

UDP-GlcNAc C4,6 dehydratase UDP-4-keto-6-deoxyGlcNAc aminotransferase acetyl transferase 2,4-diacetamido,2,4,6trideoxyglucose-1phosphate transferase GalNAc transferase GalNAc transferase/ polymerase GalNAc transferase Glc transferase

PglK PglB

flippase oligosaccharyltransferase

PglE PglD PglC PglA PglH

presence of glycoprotein in deletion mutant none none minor amounts not reported none none none similar amounts of glycoprotein without Glc none none

which possess the extended sequon Asp/Glu-X1-Asn-X2-Ser/Thr (where X1 and X2 are any amino acid except Pro)13,14 by the Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

8473

Scheme 1. Experimental Flow Chart for Analysis of LLOs by LC-MS/MS

Table 2. Effect of NH4OAc Concentration on LLO Signal Intensity Using LLO Standards Prepared from E. coli pACYCpglmut peak intensitya [NH4OAc] (mM)

m/z 1153b

m/z 1166c

5 25 50 100

15000 ± 2646 22333 ± 4509 37333 ± 6807 27333 ± 4619

10333 ± 1528 11000 ± 4000 19667 ± 5132 16333 ± 3215

a Averaged peak heights of samples run in triplicate, reported with their corresponding standard deviation from the mean. b Corresponds to the [M - 2H]2- ion of Und-PP-HexNAc6Hex.25 c Corresponds to the [M - 2H]2- ion of Und-PP-BacHexNAc5Hex.

oligosaccharyltransferase PglB. The similarities between the eukaryotic and bacterial N-glycan pathways include the assembly of an oligosaccharide on a polyisoprenyl carrier anchored in the membrane and transfer of the oligosaccharide en bloc to asparagine residues of proteins with specific sequons.13 Furthermore, the C. jejuni PglB shows modest homology to Stt3p, which is an essential subunit in the eukaryotic oligosaccharyltransferase (OST) complex.5,15 Given that the N-glycosylation pathway in C. jejuni is not required for bacterial viability and possesses many similarities with its eukaryotic counterpart, it provides an ideal system to study protein N-glycosylation. Since the discovery that the C. jejuni pgl pathway can be functionally transferred into E. coli,16 there has also been a growing interest in exploiting components of this pathway to metabolically engineer E. coli strains for the production of glycoconjugate vaccines.17-21 The detection of lipid-linked oligosaccharides (LLOs) has been achieved using a variety of techniques including metabolic (13) Kelly, J.; Jarrell, H.; Millar, L.; Tessier, L.; Fiori, L. M.; Lau, P. C.; Allan, B.; Szymanski, C. M. J. Bacteriol. 2006, 188, 2427–2434. (14) Kowarik, M.; Young, N. M.; Numao, S.; Schulz, B. L.; Hug, I.; Callewaert, N.; Mills, D. C.; Watson, D. C.; Hernandez, M.; Kelly, J. F.; Wacker, M.; Aebi, M. EMBO J. 2006, 25, 1957–1966. (15) Chavan, M.; Rekowicz, M.; Lennarz, W. J. Biol. Chem. 2003, 278, 51441– 51447. (16) Wacker, M.; Linton, D.; Hitchen, P. G.; Nita-Lazar, M.; Haslam, S. M.; North, S. J.; Panico, M.; Morris, H. R.; Dell, A.; Wren, B. W.; Aebi, M. Science 2002, 298, 1790–1793. (17) Alaimo, C.; Catrein, I.; Morf, L.; Marolda, C. L.; Callewaert, N.; Valvano, M. A.; Feldman, M. F.; Aebi, M. EMBO J. 2006, 25, 967–976. (18) Faridmoayer, A.; Fentabil, M. A.; Haurat, M. F.; Yi, W.; Woodward, R.; Wang, P. G.; Feldman, M. F. J. Biol. Chem. 2008, 283, 34596–34604. (19) Faridmoayer, A.; Fentabil, M. A.; Mills, D. C.; Klassen, J. S.; Feldman, M. F. J. Bacteriol. 2007, 189, 8088–8098. (20) Kowarik, M.; Numao, S.; Feldman, M. F.; Schulz, B. L.; Callewaert, N.; Kiermaier, E.; Catrein, I.; Aebi, M. Science 2006, 314, 1148–1150. (21) Wacker, M.; Feldman, M. F.; Callewaert, N.; Kowarik, M.; Clarke, B. R.; Pohl, N. L.; Hernandez, M.; Vines, E. D.; Valvano, M. A.; Whitfield, C.; Aebi, M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 7088–7093.

8474

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

radiolabeling,22 solid-state analysis of LLOs with conjugated lectins,23 OST assays with 125I-labeled acceptor peptides, or fluorophore assisted carbohydrate electrophoresis (FACE).24 Metabolic radiolabeling of LLOs is hampered by the variable rates of LLO turnover, isotope dilution effects, and difficulties associated with certain cells or tissues not being amenable to metabolic labeling. The solid-state assay employing horseradish peroxidase-conjugated Concanavalin A (ConA) lectin provides a selective, semiquantitative, and nonradioactive assay for LLOs (which cannot be applied to the enrichment of bacterial LLOs), while both the OST assay and FACE analysis of LLOs rely on detection of the LLO-associated glycans. With the exception of the solid-state assay, all current methods require chromatographic or electrophoretic separation and are not readily amenable to structural characterization. Previously, we demonstrated the utility of affinity-capture capillary electrophoresis-mass spectrometry (CE-MS) for glycomics and its application to the study of the C. jejuni N-glycosylation pathway.25 Affinity-capture CE-MS is a rapid nonradioactive method for the analysis of N-glycan associated LLOs that readily provided structural information on the LLOs of interest. However, due to the limitation of injection volume in capillary electrophoresis, the sensitivity limits of the previous methodology required prior knowledge of the LLO glycan (leading to further enrichment with lectins) and relatively large sample sizes. Thus, we attempted to develop a liquid chromatography mass spectrometry (LC-MS) based tool for a nonselective general approach to the analysis of LLOs that does not require radioisotopes, lectins, or chemical derivatization. Porous graphite carbon (PGC) is a conductive crystalline material made up of smooth graphene sheets. Its unique structure and physicochemical properties cause carbon based adsorbents to exhibit a retention mechanism that is different from conventional reversed-phase packing materials.26-28 The unique properties of carbon columns allow for the resolution of structurally similar compounds such as geometric isomers and diastereomers,28-31 as well as polar32-35 and nonpolar36-39 (22) Spiro, R. G.; Spiro, M. J.; Bhoyroo, V. D. J. Biol. Chem. 1976, 251, 6409– 6419. (23) Kelleher, D. J.; Karaoglu, D.; Gilmore, R. Glycobiology 2001, 11, 321–333. (24) Gao, N.; Lehrman, M. A. Glycobiology 2002, 12, 353–360. (25) Reid, C. W.; Stupak, J.; Chen, M. M.; Imperiali, B.; Li, J.; Szymanski, C. M. Anal. Chem. 2008, 80, 5468–5475. (26) Colin, H.; Guiochon, G.; Jandera, P. Chromatographia 1982, 15, 133–139. (27) Liang, C.; Dai, S.; Guiochon, G. Anal. Chem. 2003, 75, 4904–4912. (28) Pereira, L. J. Liq. Chromatogr. Relat. Technol. 2008, 31, 1687–1731. (29) Creaser, C. S.; Al-Haddad, A. Anal. Chem. 1989, 61, 1300–1302.

Figure 2. PGC-LC-MS analysis of LLOs. (A) Reconstructed ion chromatograms of LLOs isolated from the glycoengineering strain E. coli pACYCpglmut. LLOs are separated according to polyisoprene length (C50 or C55) as well as minor structural changes in the oligosaccharide (HexNAc versus Bac at reducing end). (B) Reconstructed ion chromatograms of LLOs from the C. jejuni pglK mutant showing chromatographic separation by the size of the carrier polyisoprene. Note: Bac is 2,4-diacetamido-2,4,6-trideoxy-D-glucose. Note: in panel A, the size of peaks 2 and 5 was increased for clarity.

compounds. The unique separation modes reported for PGC provide an intriguing stationary phase for the separation of polyisoprene-linked oligosaccharides with the possibility of differentiating them based on either the carbohydrate portion or the lipid carrier, depending on solvent conditions. METHODS Chemicals and Materials. All chemicals were of analytical reagent grade and used as received. Chloroform (CHCl3) and methanol (MeOH) were purchased from EMD Chemicals (Gibbstown, NJ). Distilled water was deionized on a Millipore Milli-Q water reagent system (Billerica, MA). (30) Gundersen, J. L. J. Chromatogr., A 2001, 914, 161–166. (31) Inamoto, Y.; Inamoto, S.; Hanai, T.; Tokuda, M.; Hatase, O.; Yoshii, K.; Sugiyama, N.; Kinoshita, T. J. Chromatogr., B 1998, 707, 111–120. (32) Elfakir, C.; Lafosse, M. J. Chromatogr., A 1997, 782, 191–198. (33) Pabst, M.; Bondili, J. S.; Stadlmann, J.; Mach, L.; Altmann, F. Anal. Chem. 2007, 79, 5051–5057. (34) Robinson, S.; Bergstrom, E.; Seymour, M.; Thomas-Oates, J. Anal. Chem. 2007, 79, 2437–2445. (35) Xing, J.; Apedo, A.; Tymiak, A.; Zhao, N. Rapid Commun. Mass Spectrom. 2004, 18, 1599–1606. (36) Delobel, A.; Roy, S.; Touboul, D.; Gaudin, K.; Germain, D. P.; Baillet, A.; Brion, F.; Prognon, P.; Chaminade, P.; Laprevote, O. J. Mass Spectrom. 2006, 41, 50–58. (37) Gaudin, K.; Chaminade, P.; Baillet, A. J. Chromatogr., A 2002, 973, 69–83. (38) Gaudin, K.; Hanai, T.; Chaminade, P.; Baillet, A. J. Chromatogr., A 2007, 1157, 56–64. (39) Kriz, J.; Adamcova, E.; Knox, J. H.; Hora, J. J. Chromatogr., A 1994, 663, 151–161.

Bacteria. Wild-type C. jejuni 11168 and the pglK flippase mutant were previously described.13,25 E. coli DH5R harboring pACYCpglmut (expressing the C. jejuni glycosylation locus with a point mutation in the active site of the oligosaccharyltransferase, PglB) was previously described.16,17 E. coli cells were grown on 2× YT agar plates overnight at 37 °C, while C. jejuni was grown on brain-heart infusion (BHI) agar at 37 °C under microaerobic conditions. Bacteria were enumerated using the technique described by Gross et al.40 LLO Extraction. Bacterial cells were scraped from a single agar plate and resuspended in glass test tubes containing methanol. The cell suspension was evaporated to dryness under a stream of N2 to drive out residual water. The dried cell pellet was extracted sequentially with 2:1 CHCl3/MeOH and 10:20:3 CHCl3/MeOH/H2O (2 × 1 mL); the cell debris was removed by centrifugation (500g, 5 min), and the supernatants were combined and retained. The pooled organic extract was evaporated to dryness under a stream of N2. Prior to LC-MS analysis, tightly adhering material was removed from the samples using TopTip Carbon 10 µL SPE tips (Glygen, Colombia, MD) under the conditions described below for LCMS analysis of LLOs. LC-MS Analysis of LLOs. The experimental flowchart for analysis of LLOs by LC-MS/MS is illustrated in Scheme 1. MS was performed on a 4000 Q-Trap (Applied Biosystems/MDS Sciex, Concord, ON, Canada) coupled to an UltiMate 3000 HPLC (Dionex, Sunnyvale, CA). Liquid chromatography was performed on a Hypercarb PGC column (0. 32 × 100 mm) (Thermo Scientific, San Jose, CA) at a flow rate of 6 µL min-1 with a gradient as follows: 100% A for 5 min, increasing to 70% B over 20 min, and held at 70% B for 20 min. Mobile phase A consisted of MeOH/CHCl3 (90:10) with 50 mM ammonium acetate. Mobile phase B consisted of MeOH/CHCl3 (10:90) with 50 mM ammonium acetate. For MS/MS experiments, a collision energy of -75 eV was used for experiments in the negativeion mode; for MS/MS experiments in the positive-ion mode, the orifice voltage was set at 400 V to allow for in-source CID cleavage of the pyrophosphate bond, thereby producing oligosaccharide fragment ions amenable to MS/MS analysis with a typical collision energy of +45 eV. RESULTS AND DISCUSSION Development of Polyisoprene Separation on PGC. The increasing number of reports of the use of PGC for the separation of lipids37,38,41 and its reported greater methylene selectivity compared to octadecyl silane (ODS) based stationary phases41 led to our investigation of PGC as a stationary phase for the separation of polyisoprene-linked oligosaccharides. Separation and ionization conditions were optimized using purified LLOs from E. coli pACYCpglmut, an E. coli strain into which the C. jejuni pgl gene cluster was functionally transferred with a point mutation in the active site of the oligosaccharyltransferase, PglB. Thus, E. coli pACYCpglmut can assemble the C. jejuni heptasaccharide but is incapable of transferring the oligosaccharides to proteins.25 The resultant accumulating LLOs were previously isolated from a large (40) Gross, M.; Marianovski, I.; Glaser, G. Mol. Microbiol. 2006, 59, 590–601. (41) West, C.; Cilpa, G.; Chaminade, P.; Lesellier, E. J. Chromatogr., A 2005, 1087, 77–85.

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

8475

Table 3. Polyisoprenylpyrophosphate Lipids Observed in the C. jejuni pglK Mutant and E. coli Glyco-Factory Harbouring the Plasmid pACYCpglmut As Determined by MS/MS Fragmentation of the Corresponding LLOsa observed LLO ion ([M - 2H]2-)

theoretical mass of LLO (Da)b

1097.6 1131.5 1139.6 1165.4 1173.6 1199.8

2196.1 2264.1 2280.1 2332.2 2348.2 2400.3

1118.8 1131.5 1160.6 1173.6 1169.1 1181.1 1187.1 1199.5

2239.1 2264.1 2323.2 2348.2 2339.2 2364.2 2375.2 2400.3

observed lipid ion ([M - H]-)

theoretical mass of lipid (Da)

composition of LLO

C. jejuni pglK 771.5/789.5 839.5/857.5 855.5/873.5 907.5/925.5 923.5/941.5 975.5/993.5

790.5 858.6 874.6 926.6 942.6 994.7

C45-PP-Bac(HexNAc)5Hex C50-PP-Bac(HexNAc)5Hex OH-C50-PP-Bac(HexNAc)5Hex C55-PP-Bac(HexNAc)5Hex OH-C55-PP-Bac(HexNAc)5Hex C60-PP-Bac(HexNAc)5Hex

E.coli pACYCpglmut 839.5/857.5 839.5/857.5 923.5/941.5 923.5/941.5 939.5/957.5 939.5/957.5 975.5/993.5 975.5/993.5

858.6 858.6 942.6 942.6 958.6 958.6 994.7 994.7

C50-PP-(HexNAc)6Hex C50-PP-Bac(HexNAc)5Hex OH-C55-PP-(HexNAc)6Hex OH-C55-PP-Bac(HexNAc)5Hex 2OH-C55-PP-(HexNAc)6Hex 2OH-C55-PP-Bac(HexNAc)5Hex C60-PP-(HexNAc)6Hex C60-PP-Bac(HexNAc)5Hex

a See Figure S1 in the Supporting Information for corresponding MS/MS spectra. b The theoretical masses were calculated using the monoisotope mass of each atom.

scale bacterial growth and purified by preparative diethylaminoethyl (DEAE) chromatography as described.25 Optimal LLO separation conditions were observed using a methanol/chloroform gradient. In this study, the effect of an organic modifier on the separation efficiency and detection sensitivity was investigated. Ammonium acetate (NH4OAc) is the most common organic modifier for eluents in LC-MS, and its concentration has an impact on separation efficiency and sensitivity. To this end, the effect of the NH4OAc concentration in the mobile phase on the separation of purified Und-PP-heptasaccharides and mass spectrometer signal intensity was investigated (Table 2). While increasing concentrations of the ion-pairing agent did not have a noticeable effect on peak shape or retention time, concentrations of NH4OAc above 50 mM resulted in signal suppression and an increased potential for arcing of the MS source. To achieve reproducible results while maintaining maximum sensitivity, a concentration of 50 mM NH4OAc was utilized in all subsequent separations. Generally, the method of choice for lipid analysis is reversedphase high performance liquid chromatography (RP-HPLC) with ODS or with silver nitrate-silica (AgNO3) stationary phases.42-44 While these methods are valuable tools for the analysis of triacylglycerols,43 dolichols,42 and phospholipids,44 they are either not suited to separating differences in carbon chain length45(e.g., AgNO3-RP-HPLC) or have been shown to be less than ideal for detection of LLOs directly from biological samples .42 The ideal stationary phase should be able to distinguish between differences in the polyisoprenyl carrier as well as distinguishing structural isomers and sugar substitutions in the oligosaccharide portion of the LLO. PGC has unique properties as a stationary phase in HPLC. For nonpolar analytes, such as lipids, (42) Garrett, T. A.; Guan, Z.; Raetz, C. R. H. Methods Enzymol. 2007, 432, 117– 143. (43) Momchilova, S.; Itabashi, Y.; Nikolova-Damyanova, B.; Kuksis, A. J. Sep. Sci. 2006, 29, 2578–2583. (44) Sestak, T. L.; Subbaiah, P. V.; Jaskowiak, N. T.; Bagdade, J. D. Anal. Biochem. 1990, 191, 156–159. (45) Muller, A.; Mickel, M.; Geyer, R.; Rngseis, R.; Eder, K.; Steinhart, H. J. Chromatogr., B 2006, 837, 147–152.

8476

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

PGC behaves as a strongly retentive alkyl-bonded silica gel, while the separation mechanism for polar analytes and structurally related compounds is complex and is influenced by electrostatic and steric interactions between the PGC surface and polar analyte.28,46 The strength of these interactions depends on both the molecular area of an analyte that comes into contact with the graphite surface and the nature and type of functional groups at the point of interaction.28 Although the separation mechanisms in PGC chromatography are diverse, under correct conditions PGC has the potential to separate highly similar LLOs which would be of tremendous advantage in glycoengineering. As demonstrated in Figure 2, LLOs from E. coli pACYCpglmut (Figure 2A) and the C. jejuni pglK flippase mutant (Figure 2B), two strains that accumulate LLOs due to their inability to glycosylate proteins, were well separated by PGC-LC-MS, in which LLOs were resolved on the basis of polyisoprene size and oligosaccharide structural differences. The detected ions and their corresponding compositions, deduced from the MS/MS experiments, are summarized in Table 3. Using this method, we have demonstrated for the first time that lipid-linked carbohydrate synthesis in bacteria can proceed on other polyisoprenes in addition to C50 or C55 (Figure 2 and Table 3). More precisely, the Pgl N-glycan in C. jejuni is synthesized on polyisoprenes ranging from C45 to C60. As shown in Table 3, apparent hydroxylated forms of the polyisoprene carrier were observed in both the C. jejuni pglK mutant and the E. coli glyco-factory (Figure S1 in the Supporting Information). These proposed hydroxylated forms could be a step in the degradation of accumulated LLOs that allow for recycling of the polyisoprene into other pathways (e.g., lipopolysaccharide, capsular polysaccharide, peptidoglycan). The results presented here highlight for the first time the heterogeneity of the polyisoprene carrier in bacterial carbohydrate biosynthesis. Sensitivity of PGC-LC-MS. In order to determine the minimal biomass required for LLO analysis, cells from varying size cultures, from 1 L of broth culture to a single 10 µL inoculating (46) Hanai, T. J. Chromatogr., A 2003, 989, 183–196.

Figure 4. Analysis of LLOs from wild-type C. jejuni 11168, prepared from a single Mueller-Hinton (MH) plate (approximately 1010 cells). (A) Total ion chromatogram showing the fragment ions for the native LLO ions at m/z 1131.0 [M - 2H]2-. (B) and (C) Negative-ion mode tandem MS of doubly charged parent ion at m/z 1131.0 extracted at 20.74 and 38.07 min, respectively, identifying the polyisoprenyl lipid (m/z 839.5, 857.5). The results confirm C50 polyisoprenyl as the lipid carrier for both species and that the separation observed is most likely due to cis/trans isomers of the polyisoprene. Figure 3. Single loop analysis of C. jejuni pglK flippase mutant. LLOs were extracted from a 10 µL inoculating loop of cells grown on solid media (106 cells) and analyzed by PGC-LC-MS. (A) Extracted-ion chromatogram of the pglK LLO at m/z 1165.4 [M - 2H]2- (circled). (B) Negative-ion mode full scan spectrum extracted from the chromatographic peak at 28. 05 min. (C) Negative-ion mode MS/MS spectrum of the LLO at m/z 1165.4, showing the size of the polyisoprenyl lipid (m/z 925. 5, 907. 5). (D) Positive-ion mode MS/ MS of ion at m/z 1406.5 [M + H]+, generated from in-source CID fragmentation, which corresponds to the released oligosaccharide. Note: Bac is 2,4-diacetamido-2,4,6-trideoxy-D-glucose.

loop of cells grown on solid media, were subjected to extraction and analysis by PGC-LC-MS. The results revealed that the most abundant LLOs could be detected in the C. jejuni pglK mutant and E. coli glyco-factory harboring the plasmid pACYCpglmut from as little as 106 bacterial cells, the equivalent of a 10 µL inoculating loop of plate grown cells (Figure 3). The small size of sample required for analysis places PGC-LC-MS as a practical alternative to FACE for LLO analysis. In addition, this method has the advantage of one simple extraction procedure and does not require extensive off-line purification or chemical derivatization prior to analysis. A single inoculating loop of cells is sufficient for MS/MS analysis of the LLOs, which shows that PGC-LC-MS is able to provide structural information in a rapid, facile, and sensitive manner (Figure 3C and D). Major improvements over affinity-capture CE-MS are the ability to isolate LLOs without prior knowledge of the oligosaccharide structure and increased sensitiv-

ity that even allows LLO detection from wild-type C. jejuni cells without genetic perturbation to intentionally cause accumulation of the metabolites (Figure 4). Analysis of wild-type cells revealed the presence of two peaks with retention times of 20. 74 and 38. 07 min which had identical masses corresponding to C50-PPBac(HexNAc)5Hex. This observation and confirmation of composition by MS/MS (Figure 4B and C) prompts us to suggest that cis/trans isomers of the polyisoprenes are present and are still amenable to chromatographic separation on PGC. This is consistent with the currently proposed mechanisms of retention on graphite stationary phases, since cis/trans isomerism would affect the general shape of the molecules, thereby altering their interaction with the flat graphite sheets.28 Interestingly, recent in vitro work by Chen et al. showed relaxed specificity of the N-linked glycosylation pathway enzymes in C. jejuni toward polyisoprene carriers of differing lengths, with the length of the polyisoprene carrier having little effect on enzyme activity. The authors also observed that mixed cis/trans double bond geometry in the polyisoprene was tolerated by three of the Pgl pathway enzymes tested, while an all-trans configuration was not.47 We have, thus, demonstrated that the PGC-LC-MS method is a novel and viable alternative to FACE, which is currently used for the characterization of intermediates of N-linked glycan biosynthesis and other glycosylation pathways that proceed (47) Chen, M. M.; Weerapana, E.; Ciepichal, E.; Stupak, J.; Reid, C. W.; Swiezewska, E.; Imperiali, B. Biochemistry 2007, 46, 14342–14348.

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

8477

through polyisoprenyl intermediates. The improved sensitivity and nonspecific enrichment of LLOs now opens the door to the analysis of other bacterial pathogens and their protein glycosylation machinery. In addition, PGC-LC-MS can be employed for the validation of bacterial glycoengineering platforms, through ensuring the fidelity of synthesis of the designer LLO intermediates. CONCLUSIONS Until recently, the analysis of LLOs of complex carbohydrate biosynthetic pathways has been a laborious process. Recently, we reported a method based on affinity-capture CE-MS for LLO analysis.26 This method was capable of detecting LLOs from the C. jejuni Pgl pathway without the use of radioactivity or chemical modification of the oligosaccharide. However, the method was limited by the requirement for prior knowledge of the glycan structures or lipid carrier under study and required relatively large sample sizes. For a nonradioactive LLO assay to be a viable alternative, it must be sensitive enough to be applicable to the small sample sizes typical of most biological experiments. Our investigation suggests that the PGC-LC-MS method not only provides the sensitivity required to compete with traditional methods of LLO analysis but also has the advantage of providing structural information regarding both the oligosaccharide and lipid portion. We have demonstrated that C. jejuni assembles LLOs on a heterogeneous pool of polyisoprenes, similar to the eukaryotic pathway. This new methodology can be applied to a myriad of questions in glycobiology. In microbial glycobiology, this method has applications in lipopolysaccharide and capsular polysaccharide studies for the characterization of lipid intermediates. For instance, in group II capsular polysaccharide biosynthesis, the identity of the lipid carrier during cytoplasmic biosynthesis of the repeat units is currently unknown. Additionally, the role of several proteins in group II capsular polysaccharide assembly (e.g., KpsS, KpsT)

8478

Analytical Chemistry, Vol. 81, No. 20, October 15, 2009

could be further dissected. With the increase in reports of bacterial O-linked protein glycosylation pathways that proceed through lipid-linked intermediates, this method provides a fast and effective way of characterizing these newly discovered pathways. In addition, the archaeal protein glycosylation pathways are still relatively uncharacterized compared to their bacterial and eukaryotic counterparts and may provide a source of unique enzymes for glycoengineering. Several questions still remain regarding protein N-linked glycosylation in archaea, most notably the initiating glycosyltransferase and the source of the donor sugars (dolichol-P-monosaccharides or NDP-sugars), a question PGCLC-MS could address. In terms of biotechnological applications, PGC-LC-MS can play a significant role in glycoengineering of vaccine candidates through the monitoring of assembly of designer oligosaccharides and the identification of bottlenecks in oligosaccharide production. Finally, PGC-LC-MS could find applications in eukaryotic glycobiology particularly in the study of congenital disorders of glycosylation. Currently, applications of this methodology to eukaryotic and archaeal glycobiology are being explored. ACKNOWLEDGMENT We would like to thank Drs. Anne Reid and Sue Twine for critical review of the manuscript. C.W.R. and J.S. contributed equally to this work. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review June 23, 2009. Accepted September 11, 2009. AC9013622