Multistage and Tandem Mass Spectrometry of Glycosylated

Anal. Chem. 2007, 79, 8214-8221

Multistage and Tandem Mass Spectrometry of Glycosylated Triterpenoid Saponins Isolated from Bacopa monnieri: Comparison of the Information Content Provided by Different Techniques Martin Zehl,† Ernst Pittenauer,† Leopold Jirovetz,‡ Pamita Bandhari,§ Bikram Singh,§ Vijay K. Kaul,§ Andreas Rizzi,⊥ and Guenter Allmaier*,†

Institute of Chemical Technologies and Analytics, Vienna University of Technology, Getreidemarkt 9/164, A-1060 Vienna, Austria, Department of Clinical Pharmacy and Diagnostics, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria, Department of Natural Plant Products, Institute of Himalayan Bioresource Technology, P.O. Box No. 6, Palampur 176 061, India, and Institute of Analytical and Food Chemistry, University of Vienna, Waehringer Street 38, A-1090 Vienna, Austria

Whereas all state-of-the-art techniques in mass spectrometry (MS) have been extensively applied to oligosaccharides derived from glycoproteins, less effort has been devoted to the analysis of smaller glycoconjugates. In the present study, the application of a variety of MS techniques for the analysis of two dammarane-type triterpenoid saponins isolated from B. monnieri is reported. The structural information provided by ESI-ion trap (IT)-, APMALDI-IT-, and MALDI-IT/reflectron time-of-flight (RTOF)MS, all utilizing low-energy collision-induced dissociation (CID), and MALDI-TOF/RTOF-MS, facilitating postsource decay and high-energy CID analysis, was compared. The applied desorption/ionization technique does not influence the fragmentation of identical precursor ions in lowenergy CID. All three fragmentation techniques clearly yield the sequence and branching of the glycan moiety as well as the molecular mass of the intact aglycon. Crossring cleavage of the branching sugar, which gives some information about the sugar linkages, was mainly observed in low-energy CID. High-energy CID, on the other hand, yielded some additional diagnostic fragment ions from the aglycon moiety. Internal cleavage ions are formed by alternative mechanisms in high-energy CID and are assumed to be diagnostic for some linkages. However, none of the applied MS techniques facilitates the identification of those saponins that differ only by their aglycon moiety (i.e., jujubogenin or pseudojujubogenin). Bacopa monnieri (L.) WETTST. (Family, Scrophulariaceae), also referred to as Bacopa monniera or Herpestis monniera and water hyssop, is a small, creeping plant. It is found in wet, damp, and marshy areas of the Indian subcontinent (India and Pakistan) and in southern United States (from Texas to Florida), and it is a popular aquarium plant.1,2 More important, B. monnieri has been * Corresponding author. Phone: +43-1-58801-15160. Fax: +43-1-58801-15199. E-mail: [email protected]. † Vienna University of Technology. ‡ Department of Clinical Pharmacy and Diagnostics, University of Vienna. § Institute of Himalayan Bioresource Technology. ⊥ Institute of Analytical Chemistry and Food Chemistry, University of Vienna. (1) Godfrey, R. K.; Wooten, J. W. Aquatic and Wetland Plants of Southeastern United States: Dicotyledons; University of Georgia Press: Athens, GA, 1981.

8214 Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

used in India’s Ayurvedic system of medicine for almost 3000 years (vernacular name Brahmi). Traditionally, it is recommended as a brain tonic to enhance memory, learning, and concentration, but also for the use in a variety other physiological conditions.2,3 Several biological effects of the plant and plant extracts were investigated in recent pharmacological and clinical studies, thus confirming its efficacy in the Indian system of medicine.2-4 The main focus was on the neuropharmacological activities of B. monnieri, particularly its nootropic action. Cognition-facilitating, antidementic, and anticholinesterase effects were observed in rats and mice, which indicate that B. monnieri contains phytochemicals that could be used to alleviate debilitating disorders such as Alzheimer’s disease. The nootropic activity was furthermore supported by several clinical studies, which showed memory and learning enhancing effects upon chronic administration to adults and children.2-9 B. monnieri was also reported to have an anxiolytic effect as well as antidepressant and anticonvulsive (antiepileptic) activities. Besides the neuropharmacological properties, B. monnieri was shown to possess antioxidant, antiinflammatory, antiulcerogenic, anti-Helicobacter pylori, anthelmintic, adaptogenic, anticancer, cardiotonic, bronchodilatatory, and mast cell stabilizing activities.2,3 The pharmacological actions of B. monnieri, particularly the neuropharmacological activities, have been mainly attributed to the presence of dammarane-type triterpenoid saponins.2-4 Up to the now, 20 such structures have been characterized by spectroscopic methods. They have been named rather unsystematically as bacosides A1-A3, bacopasaponins A-H, and bacopasides I-V, X, N1, and N2 (with two different structures having been defined (2) Russo, A.; Borrelli, F. Phytomedicine 2005, 12, 305-317. (3) Altern. Med. Rev. 2004, 9, 79-85. (4) Deepak, M.; Amit, A. Phytomedicine 2004, 11, 264-268. (5) Singh, R. H.; Singh, L. Res. Ayur. Siddha 1980, 1, 133-148. (6) Sharma, R.; Chaturvedi, C.; Tewari, P. V. J. Res. Educ. Indian Med. 1987, 6, 1-12. (7) Dave, U. P.; Chauvan, V.; Dalvi, J. Indian J. Pediatr. 1993, 60, 423-428. (8) Stough, C.; Lloyd, J.; Clarke, J.; Downey, L. A.; Hutchison, C. W.; Rodgers, T.; Nathan, P. J. Psychopharmacology 2001, 156, 481-484. (9) Roodenrys, S.; Booth, D.; Bulzomi, S.; Phipps, A.; Micallef, C.; Smoker, J. Neuropsychopharmacology 2002, 27, 279-281. 10.1021/ac070008s CCC: $37.00

© 2007 American Chemical Society Published on Web 10/06/2007

as bacopaside III).10-20 All structures possess either jujubogenin or pseudojujubogenin as an aglycon moiety, and a mono-, di-, or branched trisaccharide moiety (containing arabinose, glucose, or both) that is attached to C-3 of the aglycon. In addition, bacopasaponins A, E, and F have an arabinopyranose attached to C-20 of the aglycon,12,15 whereas bacopaside I and one of the two published structures for bacopaside III carry a sulfate group at the 6-hydroxy group of a glucose.16,17 Despite the progress in isolation and characterization of the triterpenoid saponins, only a few analytical procedures have been described for the determination in B. monnieri extracts and preparations. Renukappa et al. reported the use of HPLC-MS for the qualitative analysis of saponins in a B. monnieri extract.21 They have also applied HPLC-atmospheric pressure chemical ionizationMS, but without giving a detailed interpretation and discussion of the structural information derived by this technique. Later, four studies showed the qualitative and quantitative analysis of B. monnieri saponins by HPLC using UV or electrospray ionization (ESI)-MS detection.20,22-24 However, neither the UV spectrum nor the molecular mass is specific enough to differentiate the structurally closely related saponins. Thus, these methods require welldefined reference standards. Consequently, most of the pharmacological and clinical studies mentioned above were not performed with well-defined compounds or compound mixtures, but with alcoholic extracts of B. monnieri that were only characterized by their concentration of “bacoside A and B”.4 Meanwhile, several products derived from B. monnieri are commercially available in the international nutraceutical market, which are also only characterized by their content of “bacoside A” and “bacoside B”.24 However, recent studies have revealed that “bacoside A” and “bacoside B”, which were initially reported to be the two major saponins in B. monnieri extracts,25-27 are actually mixtures of different saponins that show quantitative variations of their constituents.4,20,23,24 Hence, there is an urgent need for analytical methods that facilitate the identification of known and (10) Jain, P.; Kulshreshtha, D. K. Phytochemistry 1993, 33, 449-451. (11) Rastogi, S.; Pal, R.; KulshreshthaDk Phytochemistry 1994, 36, 133-137. (12) Garai, S.; Mahato, S. B.; Ohtani, K.; Yamasaki, K. Phytochemistry 1996, 42, 815-820. (13) Garai, S.; Mahato, S. B.; Ohtani, K.; Yamasaki, K. Phytochemistry 1996, 43, 447-449. (14) Rastogi, S.; Kulshreshtha, D. K. Indian J. Chem. B 1999, 38, 353-356. (15) Mahato, S. B.; Garai, S.; Chakravarty, A. K. Phytochemistry 2000, 53, 711714. (16) Chakravarty, A. K.; Sarkar, T.; Masuda, K.; Shiojima, K.; Nakane, T.; Kawahara, N. Phytochemistry 2001, 58, 553-556. (17) Hou, C. C.; Lin, S. J.; Cheng, J. T.; Hsu, F. L. J. Nat. Prod. 2002, 65, 17591763. (18) Chakravarty, A. K.; Garai, S.; Masuda, K.; Nakane, T.; Kawahara, N. Chem. Pharm. Bull. 2003, 51, 215-217. (19) Mandal, S.; Mukhopadhyay, S. Indian J. Chem. B 2004, 43, 1802-1804. (20) Sivaramakrishna, C.; Rao, C. V.; Trimurtulu, G.; Vanisree, M.; Subbaraju, G. V. Phytochemistry 2005, 66, 2719-2728. (21) Renukappa, T.; Roos, G.; Klaiber, I.; Vogler, B.; Kraus, W. J. Chromatogr., A 1999, 847, 109-116. (22) Ganzera, M.; Gampenrieder, J.; Pawar, R. S.; Khan, I. A.; Stuppner, H. Anal. Chim. Acta 2004, 516, 149-154. (23) Deepak, M.; Sangli, G. K.; Arun, P. C.; Amit, A. Phytochem. Anal. 2005, 16, 24-29. (24) Murthy, P. B. S.; Raju, V. R.; Ramakrisana, T.; Chakravarthy, M. S.; Kumar, K. V.; Kannababu, S.; Subbaraju, G. V. Chem. Pharm. Bull. 2006, 54, 907911. (25) Chatterjee, N.; Rastogi, R. P.; Dhar, M. L. Indian J. Chem. 1963, 1, 212215. (26) Chatterjee, N.; Rastogi, R. P.; Dhar, M. L. Indian J. Chem. 1965, 3, 24-29. (27) Basu, N.; Rastogi, R. P.; Dhar, M. L. Indian J. Chem. 1967, 5, 84-86.

characterization of new members of this class of compounds without the need for reference standards. In the present study, we report the application of different mass spectrometric techniques for the analysis of two triterpenoid saponins isolated from B. monnieri. ESI-IT-, atmospheric pressurematrix-assisted laser desorption/ionization (AP-MALDI)-IT-, and MALDI-IT/reflectron time-of-flight (RTOF)-MS, all utilizing lowenergy collision-induced dissociation (CID) to obtain structural information, and MALDI-TOF/RTOF-MS, facilitating postsource decay (PSD) and high-energy CID analysis, were used. The main focus is on the comparison of the structural information provided by these techniques in order to verify their suitability for the identification and characterization of the saponins. EXPERIMENTAL SECTION Isolation and Purification of the Saponins. The plant material was collected from the herbal garden of the Herbarium Research Institute at Joginder Nagar (H.P.), India, during May to June 2004. A voucher specimen is deposited in the herbarium section of the institute. Air-dried, powdered material (1.2 kg) of the whole plant of B. monnieri was defatted with petroleum ether (bp 60-80 °C) in a Soxhlet apparatus. The defatted plant material was dried and further extracted in a percolator with methanol at room temperature (cold extraction). The methanol extract was concentrated and partitioned between water and 1-butanol. The 1-butanol layer was washed with water (2-3 times) and then distilled under vacuum. The residue (38.7 mg) was dissolved in a minimum volume of methanol, adsorbed on silica gel, and dried. The adsorbed extract was loaded on a chromatography column packed with silica gel 60-120 mesh (Merck, Darmstadt, Germany) and eluted successively with CHCl3, CHCl3/CH3OH 9:1 (v/ v), and CHCl3/CH3OH 4:1 (v/v). The CHCl3/CH3OH 4:1 (v/v) fraction was repeatedly crystallized with methanol to yielded 27 mg of purified sample. The melting point of the compound could not be recorded due to its decomposition in the melting point apparatus. Homogeneity of this fraction was checked by thin-layer chromatography (TLC) on silica gel-G plates (Merck) using three different solvent systems: (a) EtOAc/CH3OH/H2O 60:14:10 (v/ v/v) , (b) CHCl3/EtOAc/CH3OH/H2O 75:10:14:1 (v/v/v/v), and (c) CHCl3/pyridine/H2O 80:19:1 (v/v/v). The detection was done by applying the Liebermann-Burchard spray reagent.28 All the above-mentioned solvent systems gave a single TLC band. For the mass spectrometric analysis, the sample was dissolved to ∼50 mg L-1 either in CH3OH/H2O 1:1 (v/v) or in CH3OH/200 µM aqueous NaCl 1:1 (v/v). ESI-IT-MS. The positive ion mode ESI-MSn (n ) 2-3) measurements were performed on an Esquire 3000plus 3D-ion trap mass spectrometer equipped with an orthogonal ESI ion source (Bruker Daltonics, Bremen, Germany). The following experimental conditions were applied: spray high voltage, 4.0-4.1 kV; capillary exit, tube lens, skimmer, and quadrupole/octopole voltages were set according to the precursor ion mass (“smart parameter settings”); dry gas, nitrogen at 4 L min-1 and 250 °C; nebulizer gas, nitrogen with 10 psi; isolation width, 1 or 4 Th in MS2 and 4 Th in MS3; activation width, 10 Th. The sample was delivered by a syringe pump with a flow rate of 180 µL h-1. Data (28) Huang, T. C.; Chen, C. P.; Wefler, V.; Raftery, A. Anal. Chem. 1961, 33, 1405-1407.

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8215

were acquired with a scan rate of 5500 m/z s-1 using ion charge control (maximum accumulation time, 200 ms). LC-ESI-IT-MS. The measurements were carried out on a LaChrom Elite HPLC system (VWR International Merck-Hitachi, Darmstadt, Germany) on-line coupled to the Esquire 3000plus instrument via an orthogonal ESI ion source. Chromatographic separation of the saponins was performed with a Hypersil ODS column (150 × 2.1 mm, 5-µm particle size, 120-Å pore size, C18) from Thermo Electron (Waltham, MA) and an ACE C4 column (250 × 4.6 mm, 5-µm particle size, 300-Å pore size, C4) from Advanced Chromatography Technologies (Aberdeen, UK). The column temperature was 50 °C for the Hypersil ODS and room temperature for the ACE C4. The solvents used were 10 mM ammonium acetate in doubly distilled water (solvent A) and 10 mM ammonium acetate in methanol (solvent B). Good separation was achieved when the gradient was performed from 60 to 68% solvent B for 30 min at a flow rate of 200 µL min-1 (Hypersil ODS) or from 65 to 75% solvent B for 30 min at a flow rate of 500 µL min-1 (ACE C4). The ESI source was operated in positive ion mode under the following experimental conditions: spray high voltage, 4.0 kV; capillary exit, tube lens, skimmer, and octopole voltages were tuned for maximum transmission of [M + Na]+ ions; dry gas, nitrogen at 8 L min-1 and 350 °C; nebulizer gas, nitrogen with 25 psi. Data were acquired with a scan rate of 5500 m/z s-1 using ion charge control (maximum accumulation time, 300 ms). AP-MALDI-IT-MS. The positive ion mode AP-MALDI-MSn (n ) 2-3) measurements were performed on a HCTplus 3D-ion trap mass spectrometer (Bruker Daltonics) equipped with a secondgeneration AP-MALDI-pulsed dynamic focusing source and a nitrogen laser (λ ) 337 nm) operating at 10 Hz repetition rate (Agilent Technologies, Palo Alto, CA made by Mass Tech, Columbia, MD). The following experimental conditions were applied: HV capillary, 2.5 kV; capillary exit, tube lens, skimmer, and octopole voltages were set according to the precursor ion mass (smart parameter settings); dry gas, nitrogen at 6 L min-1 and 350 °C; isolation width 2 Th in MS2 and 4 Th in MS3; activation width 4 Th. Data were acquired with a scan rate of 5500 m/z s-1 using a fixed accumulation time of 100 ms. AP-MALDI sample preparation was performed on a titanium-covered target (type called “target plates for AP-MALDI LC/MS”, Agilent Technologies) using a commercial R-cyano-4-hydroxy-cinnamic acid (CHCA) matrix solution containing 6.2 g L-1 in CH3OH/CH3CN/H2O 36: 56:8 (v/v/v) (Agilent Technologies). MALDI-IT/RTOF-MS. The positive ion mode vacuum MALDIlow-energy CID-MSn (n ) 2-3) measurements were performed on an Axima-QIT (Shimadzu Biotech Kratos Analytical, Manchester, UK). This first-generation hybrid instrument consists of a 3D quadrupole ion trap coupled to a double-stage gridless reflectron TOF analyzer.29 The Axima-QIT is equipped with a pulsed nitrogen laser (λ ) 337 nm, 4 ns pulse width) and an integrated 1-GHz recorder. A total acceleration voltage of 4 kV was applied. The ion trap was operated at a helium back pressure of roughly 5 × 10-5 mbar to allow collisional cooling of the trapped ions. In addition, a short pulse of argon was injected into the IT immediately before the ion introduction to enhance the ion-cooling

efficiency and immediately before the radio frequency-induced precursor ion excitation to enhance the fragmentation efficiency during CID in MSn mode. CID spectra were acquired by averaging 5041 unselected single laser shots using a precursor ion selection width of 1/250 of the precursor ion mass in MS2 and 1/70 of the precursor ion mass in MS3. MALDI sample preparation was performed on stainless steel target slides (Shimadzu Biotech Kratos Analytical) using a solution of 20 g L-1 2,4,6-trihydroxyacetophenone monohydrate (Fluka, Buchs, Switzerland) in methanol. MALDI-TOF/RTOF-MS. The vacuum MALDI-seamless PSD and - high-energy CID measurements were performed on a TOF/ RTOF instrument (Axima-TOF,2 Shimadzu Biotech Kratos Analytical) fitted with a high-resolution dual Bradbury-Nielsen ion gate and a differentially pumped high-energy gas collision cell located in the first drift region shortly before the curved field reflectron.30 This instrument is equipped with a pulsed nitrogen laser (λ ) 337 nm, 4 ns pulse width) and an integrated 2-GHz recorder. It was operated in the positive ion mode at a source potential of 20 kV, yielding an effective collision energy of ELAB ) 20 keV. For high-energy CID experiments, helium was introduced into the collision cell generating a pressure of 5 × 10-6 mbar. PSD and CID spectra were acquired by averaging 4900 unselected single laser shots using a precursor ion selection width of 12 Th. No deceleration of the precursor ion or reacceleration of the fragment ions was performed in this device. MALDI sample preparation was performed on stainless steel target slides (Shimadzu Biotech Kratos Analytical) using the above-mentioned commercial CHCA matrix solution.

(29) Ding, L.; Kawatoh, E.; Tanaka, K.; Smith, A. J.; Kumashiro, S. Proc. SPIEIn. Soc. Opt. Eng. 1999, 3777, 144-155.

(30) Belgacem, O.; Bowdler, A.; Brookhouse, I.; Brancia, F. L.; Raptakis, E. Rapid Commun. Mass Spectrom. 2006, 20, 1653-1660.

8216

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

RESULTS AND DISCUSSION LC-ESI-IT-MS. Although the fraction isolated from B. monnieri has given a single band upon TLC with three different solvent systems, reversed-phase LC-MS was performed to verify the homogeneity of the sample. Chromatographic separation of the saponins on a C18 column (with the conditions given in the Experimental Section) yielded two well-resolved, Gaussian-shaped peaks (corroborated by reversed-phase columns of different quality). The first component, eluting at a retention time of 7.8 min, gave a [M + Na]+ ion at m/z 951.5 and a [M + K]+ ion at m/z 967.5. Two previously characterized isobaric saponins from B. monnieri would fit to the corresponding monoisotopic molecular mass of 928.5 Da, namely, bacopaside II (3-O-[R-L-arabinofuranosyl(1f2)-{β-D-glucopyranosyl-(1f3)}-β-D-glucopyranosyl] pseudojujubogenin)16 and bacoside A3 (3-O-[R-L-arabinofuranosyl-(1f2){β-D-glucopyranosyl-(1f3)}-β-D-glucopyranosyl] jujubogenin).11 The second component eluted after 10.0 min and yielded a [M + Na]+ ion at m/z 921.5 and a [M + K]+ ion at m/z 937.5. The corresponding monoisotopic molecular mass of 898.5 Da matches with three isobaric saponin structures, namely, bacopasaponin C (3-O-[R-L-arabinofuranosyl-(1f2)-{β-D-glucopyranosyl-(1f3)}-R-Larabinopyranosyl] pseudojujubogenin),12 bacopaside X (3-O-[R-Larabinofuranosyl-(1f2)-{β-D-glucopyranosyl-(1f3)}-R-L-arabinopyranosyl] jujubogenin),20,23 and bacoside A2 (3-O-[R-L-arabinofuranosyl(1f6)-{R-L-arabinopyranosyl-(1f5)}-R-D-glucofuranosyl] pseudojujubogenin).14 No additional compounds were detected upon

varying the gradient or upon using a C4 column instead of the C18 column. Differences in MS1 Generated by Three Desorption/ Ionization Techniques Attached to Different Types of Analyzers. Positive ion mode ESI-IT- (Figure 1A), vacuum MALDI-IT/ RTOF-MS and vacuum MALDI-TOF/RTOF-MS (Figure 1B) all yielded exclusively alkali adduct ions of the saponins. Saponin monomers were observed as singly charged species, although nonspecific oligomers were observed with more charge (e.g., [2M + Me]+ and [2M + Me]2+ (Me ) Na or K)). In-source or postsource fragmentation was not observed on either the MALDITOF/RTOF instrument or the ESI-IT instrument, unless intentionally forced by harsh conditions. However, nonprovoked fragment ions of the [M + Na]+ ions of the saponins were observed with low relative intensities (