Analysis of phospholipase A
2
glycosylation
patterns from venom of individual bees by
capillary electrophoresis/electrospray ionization
mass spectrometry using an ion trap mass
spectrometer
Chien-Chen Lai and Guor-Rong Her*
Department of Chemistry, National Taiwan University, Taipei, Taiwan, R.O.C.
A method based on tryptic digestion, ultrafiltration and capillary electrophoresis/mass spectrometry (CE/MS) has been developed for the analysis of the glycosylation pattern in the phospholipase A2(PLA) of individual honeybees. Without reducing the disulfide bonds, PLA was digested with trypsin and filtered with a 3 kDa molecular weight (MW) cut-off membrane. With this procedure, the glycopeptides could be isolated from the nonglycosylated peptides. After tryptic digestion and ultrafiltration, the disulfide bonds were reduced before analysis by CE. To reduce the adsorption, CE separation was performed on successive multiple ionic-polymer (SMIL) polybrene (PB) coated capillary columns. The SMIL-PB columns allowed partial separation of the glycopeptides and eight glycopeptides were identified by on-line coupling of CE with electrospray ionization (ESI) mass spectrometry. The analysis of phospholipase A2from the venom of individual bees indicated that the variation and relative abundances of different glycopeptides were similar between the younger and the older bees. Copyright# 2000 John Wiley & Sons, Ltd.
Received 20 July 2000; Revised 4 September 2000; Accepted 5 September 2000
Bee venom contains a complex mixture of biologically active agents, such as peptides, enzymes, biogenic amines, etc.1,2The main components of honeybee (Apis mellifera) venom are the enzymes phospholipase A2 (PLA) and hyaluronidase, and the low-molecular-weight proteins melittin and apamine. Bee venom exhibits bacteriostatic and bactericidal properties,3,4and a slight direct hemolytic effect.5
PLA, the major protein constituent of honeybee venom (12–14% of the dry weight of bee venom),1is an allergenic, basic glycoprotein6with a molecular mass of 15.7 kDa. The protein consists of 134 amino acids with a single glycosylation site at Asn13, and is cross-linked by five disulfide bridges.7–9Numerous biochemical and pharmaco-logical studies have been carried out with PLA.10–14
During the past three decades, the sugar moieties of glycoproteins have been found to be involved in important interactions with the immunological specificity of antigens, and to participate in a variety of cellular functions. There are several reports which indicate a contribution of covalently linked carbohydrate to the allergenicity of a glycoprotein PLA allergen.11–13,15,16 Since different glycoforms of a given glycoprotein may exhibit different biological proper-ties,17,18 characterization and monitoring of glycosylation patterns are of growing importance.
Many methods have been developed for the characteriza-tion of honeybee PLA. Convencharacteriza-tional preparative and analytical methods, such as ion-exchange chromatography,
gel filtration and sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), are relatively time consum-ing and therefore, are often replaced by high-performance liquid chromatography (HPLC).19–21 Characterization of the oligosaccharide structures on PLA has been reported and fourteen different structures of heterogeneous N-glycans from honeybee venom PLA have been elucidated by HPLC in combination with composition analysis, methylation analysis and NMR spectroscopy.22
In comparison with HPLC, CE offers extraordinary resolving power and speed of analysis. Proteins and peptides, having chromophores and electric charges under most pH conditions, are appropriate analytes for analysis by CE/UV and CE/MS. On major problem in the application of CE to proteins and peptides is their inherent tendency to adsorption on capillary walls owing to the electrostatic attractive forces between the silanol groups on the wall surface and the positively charged protein molecules.23,24 Among the approaches used for wall modification, capil-laries modified by polybrene (PB) using successive multiple ionic-polymer (SMIL) coating25 have proved to be stable and efficient for the separation of proteins.
CE has been applied previously to the analysis of honeybee venom.26,27These reports mainly focused on the analysis of the major component melittin, and the separation of melittin and PLA in bee venom was the goal of these studies. There was no report on the separation and analysis of the different glycoforms of PLA.
Because of its low detection limit, high specificity and, more importantly, ability to provide abundant structural information, ESI mass spectrometry (MS) and related techniques have been considered one of the ideal devices *Correspondence to: G.-R. Her, Department of Chemistry, National
Taiwan University, Taiwan, R.O.C. E-mail: grher@mail.ch.ntu.edu.tw
# 2000 John Wiley & Sons, Ltd.
cetyltrimethylammonium bromide (CTAB),b-cyclodextran (b-CD) and formic acid were purchased from Sigma (St. Louis, MO, USA). HPLC grade methanol was obtained from LAB-SCAN Analytical Science (Labscan Ltd. Dublin, Ireland). Ammonium acetate was purchased from J. T. Baker (Phillipsburg, NJ, USA). Deionized (18 M) water (Milli-Q water system, Millipore Inc., Bedford, MA, USA) was used in the preparation of the samples and buffer solution. Prior to use, the buffer solution was filtered through a 0.45mm membrane filter (Gelman Sciences, Michigan, USA).
Collection of bee venom sample
Honeybees (Apis mellifera) were collected near a hive entrance with an insect net. The sting was removed directly from each individual bee without anesthetizing. Each sting was placed immediately in buffer (50 mM NH4OAc, pH 7.2, 200mL) and was removed after vortexing for 10 min. Ultrafiltration
Two ultrafiltration filters were used for desalting, buffer exchange, and purification of PLA or its digests. The Ultrafree-0.5 filter with a 5 kDa MW cut-off membrane (Millipore Corporation, Bedford, MA) was used to purify and desalt bee venom before enzyme digestion. The Microcon-0.3 microconcentrator with a MW cut-off of 3 kDa (Millipore Corporation, Bedford, MA) was used to purify and concentrate the digests, and for buffer exchange before reducing the disulfide bonds.
Preparation of PLA and digests
For the nanospray analysis of intact PLA, PLA was dissolved in a water/methanol/formic acid (49:49:2 v/v/v) solution at a concentration of 1 mg/mL. For the digestion experiments, PLA was digested by trypsin with a substrate to enzyme ratio of 30:1 in 50 mM NH4OAc buffer solution at pH 7.2 for 3 h at 37°C. For the single bee venom analysis, the matrix plus the low MW proteins (melittin and apamine) in the bee venom solution were removed with an Ultrafree-0.5 filter, before digestion.
Concentration and buffer exchange (50 mM NH4OAc, pH 7.2 to 8.0) were achieved with a Microcon-0.3 filter. The reduction of the digests was carried out by adding DTT to the digests at a 100:1 molar ratio (DTT/PLA) and incubating the mixture at 37°C for 1 h. All ultrafiltration procedures
were processed with a desktop centrifuge (HERMLE Z 231 M, BHG, F. R., Germany).
CE/UV system
The CE system was constructed in-house and has been
Man 1-3 Man 1-4GlcNAc 1-4GlcNAc
4A Man 1-6 Fuc 1-6 n n Man 1-4GlcNAc 1-4GlcNAc = Fuc 1-3 4B Man 1-6 Fuc 1-6 n n =Man 1-4GlcNAc 1-4GlcNAc Man 1-3 5A Man 1-6 n Man 1-4GlcNAc 1-4GlcNAc = = Man 1-3 Fuc 1-3 5B Man 1-6 = n
Man 1-3 Man 1-4GlcNAc 1-4GlcNAc = Man 1-3 6 Man 1-6 Fuc 1-6 n n Man 1-4GlcNAc 1-4GlcNAc = = Man 1-3 Fuc 1-3 7 Man 1-6 n Man 1-4GlcNAc 1-4GlcNAc = GalNAc 1-4GlcNAc 1-2Man 1-3 = Fuc 1-3 8 Man 1-6 Fuc 1-6 n n Man 1-4GlcNAc 1-4GlcNAc = GalNAc 1-4GlcNAc 1-2Man 1-3 = Fuc 1-3 9 Man 1-6 n Man 1-4GlcNAc 1-4GlcNAc = =
GalNAc 1-4GlcNAc 1-2Man 1-3 Fuc 1-3 = Fuc 1-3 10 Man 1-6 Fuc 1-6 n n Man 1-4GlcNAc 1-4GlcNAc = =
GalNAc 1-4GlcNAc 1-2Man 1-3 Fuc 1-3 =
described elsewhere.28 CE columns were fused-silica capillaries (Polymicro Technologies, Phoenix, AZ, USA), 50mm i.d., 375 mm o.d. and 1 m length (0.9 m to detector). A small area of the polyimide coating was burned off to form a window for UV detection. On-column detection was performed with a UV detector (UV-C Rainin, Emeryville, CA, USA) operated at 205 nm. The SMIL coating procedure was based on a literature method.25 The capillary column was rinsed with 1 M NaOH for 15 min, 0.1 M NaOH for 15 min, and then deionized water for 15 min. After this preconditioning, the capillary was rinsed with 10% PB solution for 15 min to form the first cationic layer and 3% dextran sulfate for 15 min to form the second. Finally, 10% PB solution was applied to rinse over the anionic layer for 15 min. The capillary to which PB was attached as a third layer was named the SMIL-PB capillary. The capillary column was equilibrated with running buffer for 20 min before each CE experiment.
Electrospray and mass spectrometry
Mass spectrometry was performed using a Finnigan LCQ quadrupole ion trap mass spectrometer (Finnigan Corp., San Jose, CA, USA). Two different electrospray sources were used: (1) the pneumatically assisted electrospray ionization source for CE/MS with a sheath gas of nitrogen, and (2) a commercial nanospray ESI source using pulled glass capillaries (Protana, Odense, Denmark), which were posi-tioned with the tip at a distance of<1mm from the entrance hole of the heated transfer capillary with the help of a stereomicroscope. The capillaries were gold-coated for electrical contact. The glass capillaries were filled by gel loader tips with 1–3mL of analyte solution. Nebulization gas was not necessary in this spray mode.
The mass spectrometer was operated in positive ion mode by applying a voltage of 4.5 kV to the ESI needle and 800 V to the nanospray needle. The temperature of the heated capillary in the ESI source was set at 200°C. To avoid space
charge effects, the number of ions stored in the trap was regulated by the automatic gain control, which was set at 4 107 ions for full scan mode, 2 107for selected ion monitoring (SIM) mode, and 1 107for ZoomScan mode. Helium was used as the damping gas at a pressure of 10ÿ3 Torr. In SIM analysis, the maximum ion collection time was 0.06 s for each step and 3 scans were added for each spectrum.
RESULTS AND DISCUSSION Nanospray of intact PLA
PLA exists as glycosylated and unglycosylated variants. The glycosylated PLA consists of fourteen glycoforms. Table 1 lists the structures22 of N-glycans from honeybee venom PLA. There are four isobaric pairs (2 and 3A, 3B and 3C, 4B and 5A, 8 and 9) in the PLA glycoforms. For proteins or glycoproteins, nanospray-MS produces a series of multiply charged molecular ions with different charge states. The mass resolution is sometimes insufficient to reveal the full pattern of ions generated by ESI because the molecular weight (15.7 kDa) of the protein is significantly larger than the differences between masses of carbohydrate moieties. For example, the mass difference between the isobaric sugar chain (2, 3A) and (3B, 3C) was16 u. Since the PLA glycoforms carried multiple charges (10–13), a mass resolution of more than 10 000 was needed to differentiate these two isobaric pairs. To study the potential for direct analysis of PLA by ESI, intact PLA was analyzed by nanospray. The results indicated that the mass resolution of our instrument was insufficient to differentiate (2, 3A) and (3B, 3C) glycoforms (data not shown). A similar result was also observed for (4B, 5A) and 5B glycoforms. Moreover, some minor glycoforms were not detected. These nanospray data suggested that the ion trap mass spectrometer alone was not capable of supplying sufficient information for the analysis of PLA glycoforms.
Figure 1. CE/UV electrophorogram of intact PLA standard. Electrophoresis was carried out with a 1.1 m 50 mm i.d. (1 m effective length) SMIL coated capillary in the presence of 10 mM b-CD (pH 6.7) in 50 mM ammonium acetate. The potential drop wasÿ15 kV and the detection wavelength was 205 nm.
CE/UV analysis of intact PLA
In order to detect different glycoforms, CE separation was performed prior to analysis by MS. In analysis of proteins by CE, strong analyte-capillary wall interaction is the major cause for the loss of efficiency and poor reproducibility of migration times. In order to reduce the adsorption of PLA, at first the electrophoresis was carried out using dynamic coating with the cationic surfactant CTAB (0.5 mM) as an
because the PLA was detected in positive ion mode, and the CD interference in positive ion mode ESI was found to be more serious than in negative ion mode.
Digestion of PLA
Due to the difficulty in the analysis of intact PLA glycoforms by CE/MS, PLA was digested to peptides before the analysis. Table 2 lists all the theoretical peptides of PLA hydrolyzed by trypsin. In general, tryptic digestions of glycoproteins are usually carried out after reducing the disulfide bonds. A different tactic was used in this study; PLA was digested without reducing the disulfide bonds. Without reducing the disulfide bonds the PLA glycopep-tides should be significantly larger than other pepglycopep-tides, and
T16 125–132 1125.56 VYQWFDLR T17 133–133 146.11 K
T18 134–134 181.07 Y
Scheme 1. Possible structures of glycopeptides resulting from incomplete digestion. Figure 2. Positive ion nanospray mass spectrum of a tryptic digest of PLA standard.
Figure 3. CE/UV electrophorogram of the tryptic digest of PLA standard which had been purified with an ultrafiltration filter and reduced with DTT. Electrophoresis was carried out with a 1.1 m 50 mm i.d. (0.9 m effective length) SMIL coated capillary in the presence of 0.5% formic acid. The potential drop wasÿ20 kV and the detection wavelength was 205 nm.
Figure 4. CE/MS electrophorograms of the purified and reduced glycopeptides of PLA standard in SIM mode. Sheath liquid composition was methanol/water 80:20 containing 0.5% formic acid. The potential drop used for CE separation wasÿ15 kV and the ESI voltage was 4.5 kV.
therefore the glycopeptides should be easily isolated from other peptides by ultrafiltration. The glycan moiety is located at Asn13. The glycopeptide T1 is linked with peptide T4 which is also linked with peptide T8 by disulfide bonds. Without reduction of the disulfide bonds, the tryptic glycopeptide ought to be the peptide T1-T4-T8. The predicted molecular weights are about 4–5 kDa for the glycopeptides and 0.2–2.2 kDa for other tryptic peptides (Table 2). This difference made it possible to purify and concentrate glycopeptides using a 3000 Da MW cut-off ultrafiltration filter.
To check the results of digestion, the digests were analyzed by nanospray prior to ultrafiltration and analysis by CE (Fig. 2). The m/z values corresponding to the glycopeptides were found to be significantly larger than the expected m/z values. Incomplete digestion was believed to be the reason for the mass shifts. The masses of the ions suggested that the Arg32-Thr33 or Lys66-Phe67 peptide bonds in PLA were not cleaved by trypsin (Scheme 1). The glycopeptides were observed as two groups of ions. For example, the ions at m/z 3468 and 2313 were assigned as the
doubly and triply protonated glycopeptide with the glycans 3B and 3C. The molecular weights of the glycopeptides were about 7 kDa, and this made it convenient to isolate them from other peptides by ultrafiltration.
Figure 5. CE/MS electrophorograms of the purified and reduced glycopeptides of PLA from the venom of a single bee. Other conditions were the same as in Fig. 4.
Table 3. Relative abundance and CV (n = 5) of PLA glycopeptides of older (30 days after eclosion) and younger (10 days) honeybees
Glycopeptides
Relative abundance in PLA
glycopeptides (%) CV(%)
Older Younger Older Younger
T1-1 13.1 12.8 4.3 4.2 T1-(2, 3A) 7.2 7.1 3.3 4.2 T1-(3B, 3C) 57.3 58.6 2.7 2.4 T1-(4B, 5A) 7.7 7.5 5.4 5.6 T1-5B 7.5 7.4 5.9 5.6 T1-6 2.9 3.0 5.8 5.9 T1-7 2.2 2.1 5.9 6.8 T1-(8, 9) 1.6 1.5 6.9 6.7
Several modified ions of the type [M 2H 16]2and [M 3H 16]3 were observed with much higher abun-dances (Fig. 2) than those of the unmodified peptides. For example, m/z 2318 and 2324 were the modified molecular ions [M 3H 16]3and [M 3H 32]3for the glyco-peptides with the glycans 3B and 3C. These modifica-tions are attributed to the selective oxidation of methionyl residues in the glycopeptide (Met36 and Met41 residues in T5). Oxidation of peptides during electrospray ioniza-tion has been reported for peptides containing methionyl residues.31To eliminate the complexity resulting from oxi-dation, the disulfide bonds were reduced before analysis by CE but after ultrafiltration, so that the T5 peptide was no longer connected to the glycopeptide T1.
Analysis of PLA digests by CE/UV and CE/MS
The glycopeptides were analyzed using a PB-SMIL capillary column and the best result obtained is shown in Fig. 3. After this CE/UV analysis, the glycopeptides were further analyzed by CE/MS and numerous glycopeptides could be identified even though they were only partially separated (Fig. 4). Glycopeptides were observed as [M 3H]3 ions in the acidic buffer. Apart from the two low abundance glycopeptides, T1-4A and T1-10, eight glyco-peptides including four isobaric pairs [T1-(2, 3A), T1-(3B, 3C), T1-(4B, 5A) and T1-(8, 9)] could be identified (Fig. 4).
Analysis of PLA glycopeptides in venom of individual single bees
As mentioned earlier, there are many components in bee venom. To reduce the interference of other components, PLA in bee venom was separated using a 5 kDa MW cut-off filter. Digestion and ultrafiltration without reducing the disulfide bonds, followed by DTT reduction before analysis, by CE were performed as on the standard PLA sample.
The CE/MS electropherogram obtained from the analysis of glycopeptides from single bee venom is similar to Fig. 4 except that there were more interferences in the traces of T 1-6, T 1-7, and T 1-8,9 (Fig. 5). This technique was used to investigate the PLA glycopeptides of individual honeybees. The relative abundances and coefficient of variance (CV) of each glycopeptide for older (30 days after eclosion) and younger (10 days after eclosion) honeybees are listed in Table 3. The relative abundances of each glycopeptide were similar for younger and older honeybees. The CV values were below 7%.
CONCLUSIONS
Baseline separation of all PLA glycoforms was difficult to achieve in CE/UV analysis. Under optimized conditions, up to nine peaks were observed. The analysis of different glycoforms by CE/MS was even more difficult because of the incompatibility of nonvolatile additives such as b-cyclodextran in ESI analysis. Partial separation was achieved if the PLA was digested to glycopeptides before analysis by CE and CE/MS. Under optimized CE/MS
conditions, eight glycopeptides could be identified in SIM mode. Several steps were critical to the isolation of glycopeptides before analysis by CE and CE/MS. These steps include isolation of the PLA with ultrafiltration, tryptic digestion and ultrafiltration without reducing the disulfide bridges, and DTT reduction before analysis by CE and CE/MS. This method was found to be useful to estimate the relative concentration of each glycoform in the venom of single honeybees. In order to quantify all the different glycoforms, including the isobaric pairs, baseline or near baseline resolution is needed. Work is in progress to investigate the possibility of better separation if the glycans are removed from PLA before analysis by CE and CE/MS.
REFERENCES
1. Habermann E. Science 1972; 177: 314.
2. Banks BEC, Dempsey CE, Pearce FL, Vernon CA, Wholley TE. Anal. Biochem. 1981; 116: 48.
3. Ortel S, Markwardt F. Pharmazie 1955; 38: 743.
4. Benton AW, Morse RA, Kosikowski FV. Nature 1963; 168: 295. 5. Neumann W, Habermann E. Arch. Exp. Pathol. Pharmakol. 1954;
222: 367.
6. Marz L, Kuhne C, Michl H. Toxicon. 1983; 21: 893.
7. Kuchler K, Gmachl M, Sippl MJ, Kreil G. Eur. J. Biochem. 1989; 184: 249.
8. Scott DL, Sigler PB. Adv. Protein Chem. 1994; 45: 53.
9. Nordvall SL, Uhlin T, Einarsson R, Johansson SGO, Ohman S. Clin. Allergy 1984; 14: 341.
10. Sobotka AK, Franklin RM, Adkinson NF, Valentine M, Baer H, Lichtenstein LM. J. Allergy Clin. Immunol. 1976; 57: 29. 11. Weber A, Schroder H, Thalberg K, Marz L. Allergy 1987; 42: 464. 12. Prenner C, Mach L, Glossl J, Marz L. Biochem. J. 1992; 284: 377. 13. Muller UR, Dudler T, Schneider T, Crameri R, Fischer H, Skrbic D, Maibach R, Blaser K, Suter M. J. Allergy Clin. Immunol. 1995; 96: 395.
14. Akdis CA, Akdis M, Blesken T, Wymann D, Alkan SS, Muller U, Blaser K. J. Clin. Invest. 1996; 98: 1676.
15. Tretter V, Altmann F, Kubelka V, Marz L, Becker WM. Int. Arch. Allergy Appl. Immunol. 1993; 102: 259.
16. Okano M, Nishizaki K, Satoskar AR, Yoshino T, Masuda Y, Harn DA Jr. Allergy 1999; 54: 811.
17. Cumming DA. Glycobiology 1991; 1: 115. 18. Varki A. Glycobiology 1993; 3: 97.
19. Azhitskii Giu, Azhitskii DG. Ukr Biokhim Zh. 1993; 65: 7. 20. Ameratunga RV, Hawkins R, Prestidge R, Marbrook J. Pathology
1995; 27: 157.
21. Raeder K, Wildfeuer A, Wintersberger F, Bossinger P, Muecke W. J. Chromatogr. 1987; 408: 341.
22. Kubelka V, Altmann F, Staudacher E, Tretter V, Marz L, Hard K, Kamerling JP, Vliegenthart JFG. Eur. J. Biochem. 1993; 213: 1193.
23. Li SFY. Capillary Electrophoresis: Principles, Practice and Applications, Elsevier: Amsterdam, 1993; 158–200.
24. Greve KF, Nashabeh W, Karger BL. J. Chromatogr. 1994; 680: 15. 25. Katayama H, Ishihama Y, Asakawa N. Anal. Chem. 1998; 70:
5272.
26. Zhu SQiT, Zhao T, Fang X, Sun Y. J. Chromatogr. B 1994; 658: 397.
27. Pacakova V, Stulik K, Hau PT, Jelinek I, Vins I, Sykora D. J. Chromatogr. A 1995; 700: 187.
28. Tsai CY, Her GR. J. Chromatogr. A 1996; 743: 315. 29. Ju DD, Lai CC, Her GR. J. Chromatogr. A 1997; 779: 195. 30. Li DT, Sheen JF, Her GR. J. Am. Soc. Mass Spectrom. 2000; 11:
292.
31. Morand K, Talbo G, Mann M. Rapid Commun. Mass Spectrom. 1993; 7: 738.
