Silver Complexation and Tandem Mass Spectrometry for


Silver Complexation and Tandem Mass Spectrometry for...

0 downloads 93 Views 211KB Size

Anal. Chem. 2005, 77, 1761-1770

Silver Complexation and Tandem Mass Spectrometry for Differentiation of Isomeric Flavonoid Diglycosides Junmei Zhang and Jennifer S. Brodbelt*

Department of Chemistry and Biochemistry, The University of Texas, Austin, Texas 78712

For detection and differentiation of isomeric flavonoids, electrospray ionization mass spectrometry is used to generate silver complexes of the type (Ag + flavonoid)+. Collisionally activated dissociation (CAD) of the resulting 1:1 silver/flavonoid complexes allows isomer differentiation of flavonoids. Eighteen flavonoid diglycosides constituting seven isomeric series are distinguishable from each other based on the CAD patterns of their silver complexes. Characteristic dissociation pathways allow identification of the site of glycosylation, the type of disaccharide (rutinose versus neohesperidose), and the type of aglycon (flavonol versus flavone versus flavanone). This silver complexation method is more universal than previous metal complexation methods, as intense silver complexes are observed even for flavonoids that lack the typical metal chelation sites. To demonstrate the feasibility of using silver complexation and tandem mass spectrometry to characterize flavonoids in complex mixtures, flavonoids extracted from grapefruit juice are separated by high-performance liquid chromatography and analyzed via a postcolumn complexation ESI-MS/MS strategy. Diagnostic fragmentation pathways of the silver complexes of the individual eluting flavonoids allow successful identification of the six flavonoids in the extract. The impact of diet on human health has become an active area of research in recent years due to the growing recognition of the numerous chemopreventive properties of naturally occurring compounds, such as flavonoids,1-3 in fruits and vegetables. There are thousands of different flavonoids,1-3 with some possessing far greater biological activities than others, and thus, the mapping of structure/activity relationships is an important goal, as well as the development of versatile analytical methods for enhanced detection and structural characterization of trace quantities of flavonoids and biotransformation products in food sources, urine, plasma, and tissue. Tandem mass spectrometry has been particu* Corresponding author. Phone: (512) 471-0028. Fax: (512) 471-8696. E-mail: [email protected]. (1) Bohm, B. A. Introduction to Flavonoids; Harwood Academic Publishers: Singapore, 1998; Chapter 2. (2) Middleton, E., Jr.; Kandaswami, C. In The Flavonoids: Advances in Research since 1986; Harborne J. B., Ed.; Chapman & Hall: London, 1994; Chapter 15. (3) Kaur, C.; Kapoor, H. C. Int. J. Food Sci. Technol. 2001, 36, 703-725. 10.1021/ac048818g CCC: $30.25 Published on Web 02/04/2005

© 2005 American Chemical Society

larly effective for the structural characterization of flavonoids.4-25 In the past four years, our group has focused considerable attention on developing novel mass spectrometric ways to detect and differentiate isomeric flavonoids based on metal complexation, including the use of transition metals,20-24 with20-23 or without24 an auxiliary ligand, alkaline earth metals,24 and aluminum.25 Metal complexation alters the fragmentation pathways of flavonoids, which allows differentiation of isomeric flavonoids in a systematic fashion.20-25 Though success has been achieved, one limitation of the above metal complexation approaches is that flavonoids must have a 4-keto group and at least one neighboring hydroxyl group for formation of strong complexes with the metals. Therefore, an alternative method is needed for flavonoids that do not have such structural features. In this work, we report the use of silver(I) as a metal complexation reagent for electrospray ionization of flavonoids followed by tandem mass spectrometry. (4) Justesen U. J. Chromatog., A 2000, 902, 369-379. (5) Justesen, U. J. Mass Spectrom. 2001, 36, 169-178. (6) Fabre, N.; Rustan, I.; de Hoffmann, E.; Quetin-Leclercq, J. J. Am. Soc. Mass Spectrom. 2001, 12, 707-715. (7) Cuyckens, F.; Rozenberg, R.; de Hoffmann, E.; Claeys, M. J. Mass Spectrom. 2001, 36, 1203-1210. (8) Hughes, R. J.; Croley, T. R.; Metcalfe, C. D.; March R. E. Int. J. Mass Spectrom. 2001, 210/211, 371-385. (9) Hvattum, E.; Ekeberg, D. J. Mass Spectrom. 2003, 38, 43-49. (10) Ferreres, F.; Llorach, R.; Gil-lzquierdo, A. J. Mass Spectrom. 2004, 39, 312321. (11) Zhang, J.; Brodbelt, J. S. J. Mass Spectrom. 2003, 38, 555-572. (12) Zhang, J.; Satterfield, M. B.; Brodbelt, J. S.; Britz, S. J.; Clevidence, B.; Novotny, J. A. Anal. Chem. 2003, 75, 6401-6407. (13) Zhang, J.; Brodbelt, J. S. J. Am. Chem. Soc. 2004, 126, 5906-5919. (14) Cuyckens, F.; Claeys, M. J. Mass Spectrom 2004, 39, 1-16. (15) Ma, Y. L.; Li, Q.; Van den Heuvel, H.; Claeys, M. Rapid Commun. Mass Spectrom. 1997, 11, 1357-1364. (16) Ma, Y. L.; Vedernikova, I.; Van den Heuvel, H.; Claeys, M. J. Am. Soc. Mass Spectrom. 2000, 11, 136-144. (17) Ma, Y. L.; Cuyckens, F.; Van den Heuvel, H.; Claeys, M. Phytochem. Anal. 2001, 12, 159-165. (18) Cuyckens, F.; Shahat, A. A.; Pieters, L.; Claeys, M. J. Mass Spectrom. 2002, 37, 1272-1279. (19) Franski, R.; Matlawska, I.; Bylka, W.; Sikorska, M.; Fiedorow, P.; Stobiecki, M. J. Agric. Food Chem. 2002, 50, 976-982. (20) Satterfield, M.; Brodbelt, J. Anal. Chem. 2000, 72, 5898-5906. (21) Satterfield, M.; Brodbelt, J. S. J. Am. Soc. Mass Spectrom. 2001, 12, 537549. (22) Pikulski, M.; Brodbelt, J. S. J. Am. Soc. Mass Spectrom. 2003, 14, 14371453. (23) Zhang, J.; Brodbelt, J. S. J. Am. Soc. Mass Spectrom. 2005, 16, 139-151. (24) Davis, B. D.; Brodbelt, J. S. J. Am. Soc. Mass Spectrom. 2004, 15, 12871299. (25) Zhang, J.; Brodbelt, J. S. J. Mass Spectrom. In press..

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005 1761

Silver complexation has been used for effective ionization of several classes of compounds in both liquid secondary ion mass spectrometry and electrospray ionization applications to enhance the ionization efficiencies of the compounds, improve the selectivity and sensitivity of the mass spectrometric methods, or alter the fragmentation patterns of the molecular ions.26-51 Within the past decade, silver complexation has been applied to a variety of compounds that are nonpolar and not easily ionizable, such as methyl glycosides,26 fatty acid methyl esters,27 tocopherols and carotenoids,28 olefins, polyolefins and aromatic compounds,29 diacyl peroxides,30 barbiturates and chlorinated alkyl phenoxypropanoates,31 heavy aromatic petroleum fractions,32 monosaccharides,33 polyaromatic hydrocarbons,34 peroxidation products of cholesterols such as phospholipids and docosahexaenoate ester hydroperoxides,35-38 and rubber vulcanization products.39 In addition to enhanced isomer differentiation or improved selectivity and sensitivity for some of the previous work, silver complexation enables ionization of nonpolar compounds. Silver(I) can either be mixed directly with compounds of interest or be introduced as a postcolumn complexation reagent after the compounds are

separated by high-performance liquid chromatography (HPLC). Silver(I) also has been investigated as a complexation reagent for amino acids, peptides, and proteins by several groups in the past few years.40-51 Argentinated (silver-containing) peptides and amino acids have altered fragmentation pathways (compared to the protonated species) that can be useful for sequencing. In the current work, the formation of silver/flavonoid complexes by electrospray ionization and differentiation of isomers by tandem mass spectrometry is reported for flavonoid glycosides and aglycons. In general, abundant 1:1 silver(I)/flavonoid complexes of the type (Ag + flavonoid)+ are observed. Collisionally activated dissociation (CAD) is applied for the structural characterization of isomeric flavonoid diglycosides after silver complexation. Unique fragments of high abundances are used to differentiate the flavonoids in each isomeric series. The structural features that lead to specific fragmentation patterns are also discussed in detail. The feasibility of the method for LC/MS applications via a postcolumn complexation strategy is demonstrated for the separation and identification of six flavonoids (three groups of isomers) in a grapefruit juice extract.

(26) Berjeaud, J. M.; Couderc, F.; Prome, J. C. Org. Mass Spectrom. 1993, 28, 455-458. (27) Suma, K.; Raju, N. P.; Vairamani, M. Rapid Commun. Mass Spectrom. 1997, 11, 1939-1944. (28) Rentel, C.; Strohschein, S.; Albert, K.; Bayer, E. Anal. Chem. 1998, 70, 4394-4400. (29) Bayer, E.; Gfrorer, P.; Rentel, C. Angew. Chem., Int. Ed. 1999, 38, 992995. (30) Yin, H.; Hachey, D. L.; Porter, N. A. J. Am. Soc. Mass Spectrom. 2001, 12, 449-455. (31) von Brocke, A.; Wistuba, D.; Gfrorer, P.; Stahl, M.; Schurig, V.; Bayer, E. Electrophoresis 2002, 23, 2963-2972. (32) Roussis, S. G.; Proulx, R. Anal. Chem. 2002, 74, 1408-1414. (33) Boutreau, L.; Leon, E.; Salpin, J. Y.; Amekraz, B.; Moulin, C.; Tortajada, J. Eur. J. Mass Spectrom. 2003, 9, 377-390. (34) Ng, K. M.; Ma, N. L.; Tsang, C. W. Rapid Commun. Mass Spectrom. 2003, 17, 2082-2088. (35) Havrilla, C. M.; Hachey, D. L.; Porter, N. A. J. Am. Chem. Soc. 2000, 122, 8042-8055. (36) Milne, G. L.; Porter, N. A. Lipids 2001, 36, 1265-1275. (37) Seal, J. R.; Havrilla, C. M.; Porter, N. A. J. Am. Soc. Mass Spectrom. 2003, 14, 872-880. (38) Seal, J. R.; Porter, N. A. Anal. Bioanal. Chem. 2004, 378, 1007-1013. (39) Hayen, H.; Alvarez-Grima, M. M.; Debnath, S. C.; Noordermeer, J. W. M.; Karst, U. Anal. Chem. 2004, 76, 1063-1068. (40) Tang, X.; Ens, W.; Standing, K. G.; Westmore, J. B. Anal. Chem. 1988, 60, 1791-1799. (41) Talaty, E. R.; Perera, B. A.; Gallardo, A. L.; Barr, J. M.; Van Stipdonk, M. J. J. Phys. Chem. A 2001, 105, 8059-8068. (42) Anbalagan, V.; Perera, B. A.; Silva, A. M. T.; Gallardo, A. L.; Barber, M.; Barr, J. M.; Terkarli, S. M.; Van Stipdonk, M. J. J. Mass Spectrom. 2002, 37, 910-926. (43) Li, H.; Siu, K. W. M.; Guevremont, R.; Le Blanc, J. C. Y. J. Am. Soc. Mass Spectrom. 1997, 8, 781-792. (44) Lee, V. M. W.; Li, H.; Lau, T. C.; Guevremont, R.; Siu, K. W. M. J. Am. Soc. Mass Spectrom. 1998, 9, 760-766. (45) Lee, V. M. W.; Li, H.; Lau, T. C.; Siu, K. W. M. J. Am. Chem. Soc. 1998, 120, 7302-7309. (46) Chu, I. K.; Guo, X.; Lau, T. C.; Siu, K. W. M. Anal. Chem. 1999, 71, 23642372. (47) Chu, I. K.; Shoeib, T.; Guo, X.; Rodriquez, C. F.; Lau, T. C.; Hopkinson, A. C.; Siu, K. W. M. J. Am. Soc. Mass Spectrom. 2001, 12, 163-175. (48) Shoeib, T.; Hopkinson, A. C.; Siu, K. W. M. J. Phys. Chem. B 2001, 105, 12399-12409. (49) Shoeib, T.; Cunje, A.; Hopkinson, A. C.; Siu, K. W. M. J. Am. Soc. Mass Spectrom. 2002, 13, 408-416. (50) Chu, I. K.; Cox, D. M.; Guo, X.; Kireeva, I.; Lau, T. C.; McDermott, J. C.; Siu, K. W. M. Anal. Chem. 2002, 74, 2072-2082. (51) Cheguillaume, G.; Buchmann, W.; Desmazieres, B.; Tortajada, J. J. Mass Spectrom. 2004, 39, 368-377.

EXPERIMENTAL SECTION Chemical Reagents. (+)-Catechin hydrate, daidzein, hesperidin, naringin, neohesperidin, and rutin were purchased from Sigma (St. Louis, MO). Diosmin, isorhoifolin, narirutin, neodiosmin, and rhoifolin were purchased from Indofine (Somerville, NJ). Daidzin, datiscoside, didymin, eriocitrin, fortunellin, kaempferol7-O-neohesperidoside, kaempferol-3-O-rutinoside, linarin, neoeriocitrin, and poncirin were purchased from Extrasynthese (Genay, France). Biochanin A was purchased from Aldrich (Milwaukee, WI). Silver nitrate was purchased from Johnson Matthey (Ward Hill, MA). Formic acid was from EM Science (Gibbstown, NJ). All the above compounds were used without further purification. All the solvents were HPLC grade. The structures of all the flavonoid diglycosides are listed in Figure 1 and those of the aglycons and monosaccharide (daidzin) are shown in Figure 2. All the stock solutions of flavonoids (10-4-10-3 M) and silver nitrate (10-3-10-2 M), as well as the working solutions of flavonoids and silver/flavonoid complexes, were prepared in HPLC grade methanol. Direct Infusion. A Thermo Finnigan LCQ Duo quadrupole ion trap instrument equipped with an electrospray ionization (ESI) source (San Jose, CA) was used. The flow rate of the silver(I)/ flavonoid solutions was 5 µL/min. The ESI spray voltage was +5.0 kV. The heated capillary temperature was 210 °C. The flow rate of the sheath gas (nitrogen) was 20 arbitrary units, and auxiliary gas was not used. The injection time (i.e., ionization time) was 10-50 ms. The other instrumental parameters were tuned to optimize the relative abundance of a typical complex (Ag + rutin)+. An isolation window of 1.5 m/z was used for all the CAD experiments. The CAD energy (mass corrected, % of 5 V0-p) was varied such that only 5-15% of a parent ion survived the process. Each spectrum was an average of 10 scans, and each scan was an average of 10 microscans. The isomers in each isomeric series were run back-to-back on the same day. When the CAD patterns of deprotonated or sodium-cationized flavonoids were of interest, the flavonoids were introduced to the LCQ without mixing with silver nitrate. Very similar experimental conditions were used

1762

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

Figure 1. Flavonoids studied for isomer differentiation (molecular weight, class of flavonoid).

except that the ESI spray voltage was switched to -4.5 kV for deprotonation. LC/MS with Postcolumn Complexation. Flavonoids were extracted from a commercial grapefruit juice sample as described by Zhang and Brodbelt.52 The flavonoids in the extract were separated using a Waters Alliance 2690 system (Waters, Milford,

MA). A Waters Symmetry C18 column (2.1 × 50 mm, 3.5 µm) was used with a guard (2.1 × 10 mm, 3.5 µm). The mobile phase was composed of solvents A (0.33% formic acid), B (acetonitrile with 0.33% formic acid), and C (2-propanol with 0.33% formic acid). (52) Zhang, J.; Brodbelt, J. S. Analyst 2004, 129, 1227-1233.

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1763

Figure 2. Additional flavonoids screened for silver complexation (molecular weight, class of flavonoid).

A gradient method was used to separate the flavonoids. Solvent C was kept at 7% throughout the entire run. For the first 6 min, solvent B was increased from 10 to 28% and then brought back to 10% over the next 2 min. The initial conditions were held for 12 min prior to the next injection. The injection volume was 5 µL. The flow rate was 100 µL/min. The HPLC effluent was mixed with 1 × 10-3 M silver nitrate (introduced by a syringe pump at 20 µL/min) via a tee before entering the LCQ. For LC/MS analysis, the automated gain control option was engaged, and selected ion monitoring was used. The flow rate of the sheath gas was 50 arbitrary units and that of the auxiliary gas (nitrogen) was 20 arbitrary units. The heated capillary temperature was raised to 250 °C, and the other parameters were the same as for the direct infusion experiments. The maximum injection time was set at 2000 ms with 2-microscan averaging. The ions monitored were the silver complexes of narirutin and naringin (m/z 687), hesperidin and neohesperidin (m/z 717), and didymin and poncirin (m/z 701). For acquisition of the fragmentation patterns of the silver complexes in the LC/MS/MS experiments, the maximum injection time was 200 ms with 5 microscans. The isolation window was set at 1.5 m/z. The CAD voltage, expressed as percentage of 5 V0-p, was 33% (where 100% is 5 V0-p) for sufficient dissociation of all the complexes. RESULTS AND DISCUSSION Flavonoids usually give higher signals upon deprotonation in the negative mode than protonation or sodium cationization in the positive mode because of their acidic hydroxyl groups. However, deprotonated flavonoids tend to undergo very simple CAD dissociation pathways, dominated by the loss of the saccharide portion. For example, deprotonated datiscoside dissociates almost exclusively by the loss of the disaccharide group (m/z 285) (Figure 3A). The CAD patterns of sodium-cationized flavonoids are in general more complex, which include several fragmentation pathways as demonstrated by that of datiscoside (Figure 3B). Sodium-cationized datiscoside dissociates by the losses of the aglycon residue (m/z 331), the rhamnose moiety (m/z 471), and the disaccharide group (m/z 309) as well as dehydration (m/z 599). Despite the richer arrays of dissociation pathways of the sodium-cationized flavonoids, sodium cationization fails to serve as a general strategy for flavonoid characterization due to the very 1764 Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

Figure 3. CAD spectra of datiscoside after deprotonation, sodium and silver complexation. The parent ions are labeled with asterisks. Losses: R ) rhamnose residue (146), A ) aglycon moiety, and D ) disaccharide group (308).

Figure 4. Representative ESI full scan spectrum of a mixture of rutin (1 × 10-5 M) and silver nitrate (5 × 10-5 M) in methanol.

low (if any) abundances of the sodium complexes. Therefore, alternative methods are needed for more effective characterization of structurally similar flavonoids. Silver(I) forms strong complexes (their intensities are similar to or higher than the corresponding deprotonated flavonoids and much higher than the sodium-cationized flavonoids) of the type (Ag + L)+ with flavonoids, where L is the flavonoid. A typical ESI full-scan spectrum obtained from a mixture of rutin and silver nitrate is shown in Figure 4. The 1:1 silver/rutin complex dominates the mass spectrum, and this 1:1 stoichiometry is consistently favored for all the flavonoids with minimal competing products. In contrast to the simple fragmentation patterns of deprotonated flavonoids,13,20-25 the silver/flavonoid complexes provide richer arrays of dissociation pathways and more diagnostic product ions for flavonoid characterization. For example, the silver complex of datiscoside dissociates by the losses of the aglycon residue (m/z 415), the aglycon residue and one water molecule (m/z 397), the aglycon plus 64 Da (m/z 351), the aglycon plus the rhamnose moiety and one water molecule (m/z 251), and the rhamnose moiety (m/z 555) (Figure 3C). In the following sections, the conditions that influence formation of the silver complexes are investigated and optimized, and the ability to differentiate seven series of isomeric flavonoid

Table 1. Relative Intensity of the (Ag + Rutin)+ Complex in Methanol concn ratio (rutin/AgNO3)a complex intensityb

5:1 0.024

2:1 0.21

1:1 0.26

1:2 0.74

1:5 1.0

1:10 0.95

a The concentration of rutin was kept at 1 × 10-5 M. The solvent used was 100% methanol. b All the complex signals were compared to the one obtained when the concentration ratio was 1:5.

diglycosides by CAD is evaluated. As a demonstration of the feasibility of using this approach for characterizing flavonoids in complex mixtures, flavonoids extracted from grapefruit juice were separated by HPLC and then characterized by postcolumn silver complexation and tandem mass spectrometry. Investigation of Complexation Conditions. 1. Optimization of the Concentration of Silver Nitrate. Rutin was used as a model compound to investigate the influence of the silver nitrate concentration on the relative signal intensities of the complexes across a broad concentration range. The concentration of rutin was kept at 1 × 10-5 M while the concentration of silver nitrate was varied from 2 × 10-6 to 1 × 10-4 M. It can be seen from Table 1 that, as the concentration of silver nitrate increases, so does the complex intensity until the signal levels off when the concentration ratio reaches ∼1:5 (rutin/silver nitrate). Therefore, the 1:5 flavonoid/silver nitrate concentration ratio was used for all the other flavonoids of interest. 2. Structural Requirement of Flavonoids. One of the longterm objectives of our flavonoid work is to find a metal that promotes efficient ionization of all flavonoids, even those that do not have both a 4-keto group and at least one neighboring hydroxyl group as required for many other metal coordination strategies,20,24,25 to provide a more universal approach. To gain insight into the structural requirements of flavonoids for silver complexation by ESI-MS, a series of flavonoids were screened for their silver complexation abilities. Besides the flavonoid diglycosides listed in Figure 1, which all have a 4-keto group and a 5-OH group, several flavonoid aglycons and a monoglycoside (daidzin) that do not have the above structural features were also tested (Figure 2). (+)-Catechin is a flavonoid that does not have a 4-keto group, but the intensity of its silver complex is as high as that of biochanin A, which indicates that a 4-keto group is not essential for silver complexation. Daidzein and daidzin are two flavonoids that have the 4-keto group but lack a neighboring hydroxyl group (3- or 5-OH). The formation of abundant 1:1 complexes of these flavonoids with silver(I) suggests that the coexistence of a 4-keto group and at least one adjacent hydroxyl group is not required for silver complexation. Isomer Differentiation of Flavonoid Diglycosides. One unique feature of the silver(I) ion is that it has two isotopes (107 and 109 Da) with almost equal abundance. Therefore, the resulting ESI mass spectra for the 1:1 silver/flavonoid complexes have two adjacent peaks: (107Ag + L)+ and (109Ag + L)+ with similar intensities as shown in Figure 4. The presence of two isotopic peaks provides a convenient way to determine whether the fragment ions in the CAD mass spectra retain the silver ion or not based on comparison of m/z values for the fragment ions produced from each isotopic parent ion. Selective isolation and

Figure 5. CAD spectra of the (Ag + rutin)+ complex. The parent complex ions are labeled with asterisks. Losses: R ) rhamnose residue (146) and A ) aglycon moiety.

activation of each of the two parent ions leads to a series of fragment ions (Figure 5). If the m/z values are the same for a pair of such fragments, the fragments do not contain the silver ion. If the m/z values differ by 2 Da, then the pair of fragments retains the silver ion. Most of the fragment ions in this study involve retention of the silver ion. However, there are a few minor fragments that do not contain the silver ion. For example, the m/z 301 ions for the hesperidin and neohesperidin complexes, the m/z 287 ions for the eriocitrin and neoeriocitrin complexes, and the m/z 285 ions for the didymin and poncirin complexes all involve the loss of silver, the disaccharide group, and a hydrogen atom from the corresponding parent ions. Because these ions occur with such low intensities (less than 10% intensity) and are not highly diagnostic, they are not tabulated. For simplicity, only the CAD data obtained from the (107Ag + L)+ complexes are discussed below. It has been challenging to differentiate isomeric flavonoids because the structural differences among isomers are often subtle, involving only the inter-saccharide linkage or occasionally the glycosylation site or the type of aglycon as shown in Figure 1. According to their molecular weights, the flavonoid diglycosides in Figure 1 are divided into seven isomeric series (MW 610, 608, 596, 594, 592, 580, and 578). Two sets of representative CAD spectra for the MW 580 and 578 isomers are shown in Figures 6 and 7. The major CAD product ions of all the flavonoid complexes are summarized in Table 2. The silver complexes of structurally similar flavonoids have CAD fragmentation pathways distinct from each other, which are used for structural characterization and isomer differentiation as discussed below. Some of the major dissociation pathways are illustrated in Scheme 1 for the silver complex of rutin. The correlations between the structural features and CAD fragmentation patterns of the flavonoids are also discussed in detail. The 3-rutinosides have a richer array of fragments than the 7-rutinosides and 7-neohesperidosides, and thus, rutin serves as a good model for illustrating the diagnostic fragmentation pathways of the silver complexes. The five major fragmentation pathways include the losses of the rhamnose residue, the aglycon moiety, the aglycon moiety in conjunction with dehydration, the aglycon group plus the loss of 64 Da, and the aglycon group along Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1765

Figure 6. CAD spectra of the (107Ag + L)+ complexes of flavonoids with MW 580. The parent complex ions are labeled with asterisks. Losses: R ) rhamnose residue (146), A ) aglycon moiety, and D ) disaccharide group (308).

Figure 7. CAD spectra of the (107Ag + L)+ complexes of two flavonoids with MW 578. The parent complex ions are labeled with asterisks. Losses: R ) rhamnose residue (146), A ) aglycon moiety, and D ) disaccharide group (308).

with the rhamnose residue and water (confirmed by MSn) as shown in Scheme 1. The elimination of 64 Da mentioned above may occur via cleavage across one or both of the saccharides and involve fast consecutive losses (i.e., the loss of water and 46 Da (a C2H6O moiety or H2O and CO) from the saccharide residues, as confirmed by MSn experiments. The actual coordination sites of the silver ion are not known for either the parent or the resulting fragment ions. There may be more than one structure for a particular ion, but only one is shown in Scheme 1 for simplicity. For example, dehydration may occur on either of the two saccharide rings, and it is impossible to pinpoint where the water loss occurs with any certainty. 1. Isomer Differentiation. MW 610 Isomers. Hesperidin and neohesperidin are two 7-O-diglycosides of the same flavanone aglycon hesperetin. They differ only by one structural feature: the inter-saccharide linkage. Hesperidin has a 1-6 rhamnose-glucose linkage (rutinoside), while neohesperidin has a 1-2 rhamnoseglucose linkage (neohesperidoside). Despite this subtle structural difference, the silver complexes of these two flavonoids have 1766 Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

distinct CAD patterns (Table 2). For example, the dominant dissociation pathway of the hesperidin complex is the loss of the rhamnose residue (m/z 571) along with two minor fragmentation pathways (each with less than 15% relative abundance). There is a second major fragmentation pathway for the neohesperidin complex: the loss of the aglycon moiety (m/z 415). The third isomer in this series is rutin, a 3-O-rutinoside of quercetin, a flavonol. For the silver complex, the primary dissociation pathway is the elimination of the aglycon moiety (m/z 415) instead of the loss of the rhamnose residue that is the dominant fragmentation pathway of the complexes of the other two isomers (Figure 5 and Table 2). The rutin complex yields three additional key fragments: the loss of the aglycon moiety in conjunction with dehydration (m/z 397), the loss of the aglycon residue plus 64 Da (the sugar rearrangement and cleavage process) (m/z 351), and the loss of both the aglycon and the rhamnose moieties plus one water molecule (m/z 251). MW 608 Isomers. As two 7-O-diglycosides of diosmetin (a flavone), diosmin and neodiosmin differ from hesperidin and neohesperidin, respectively, only by the bond character between C2 and C3 of their aglycons. Diosmetin has a double bond, whereas hesperetin has a single bond. The silver complexes of diosmin and neodiosmin produce significantly different CAD spectra. The diosmin complex gives only one product ion via the loss of the rhamnose residue (m/z 569) upon CAD. The dominant product of the neodiosmin complex is also the elimination of the rhamnose moiety (m/z 569), but it has two additional fragments including the loss of the aglycon residue (m/z 415) and the loss of the rhamnose group plus one water molecule (m/z 551) (Table 2). MW 596 Isomers. The second series of flavanone isomers include eriocitrin and neoericitrin, two diglycosides of eriodictyol. Compared to hesperidin and neohesperidin, the substituent at the 4′-position of these two flavonoids is a hydroxyl instead of a methoxy group. Not surprisingly, the silver complexes of these two series of isomers have similar CAD patterns, which are governed by the inter-saccharide linkage rather than the aglycon structure (for the same type of aglycon). Loss of the rhamnose residue (m/z 557) is dominant for both of the isomeric complexes, in addition to the loss of the disaccharide moiety (m/z 395) (Table 2). The neoeriocitrin complex has the additional loss of the aglycon moiety (m/z 415), elimination of the aglycon and the rhamnose residues (m/z 269), and the dehydration product of the latter ion (m/z 251). MW 594 Isomers. This isomeric series includes five flavonoids: datiscoside (datiscetin-3-O-rutinoside), kaempferol-3-Orutinoside, kaempferol-7-O-neohesperidoside, didymin (isosakuranetin-7-O-rutinoside), and poncirin (isosakuranetin-7-Oneohesperidoside). The only difference between datiscetin and kaempferol is the position of the single hydroxyl group on the B ring (2′- vs 4′-). Isosakuranetin is a flavanone aglycon similar to hesperetin, while datiscetin and kaempferol are flavonols. The silver complexes of the two 3-O-rutinosides (datiscoside and kaempferol-3-O-rutinoside) share the same fragment ions, including losses of the aglycon residue (m/z 415), the rhamnose moiety (m/z 555), the aglycon residue plus one water molecule (m/z 397), the aglycon residue plus 64 Da (the sugar rearrangement and cleavage reaction) (m/z 351), the disaccharide group with addition

Table 2. Major CAD Product Ions of the (107Ag + L)+ Complexes of Flavonoids CAD product ionsb,c (m/z, (%)) - (R + H2O)

parenta

-R

hesperidin neohesperidin rutin diosmin neodiosmin eriocitrin neoeriocitrin datiscoside kaempferol-3-Orutinoside

717 (15) 717 (13) 717 (10) 715 (11) 715 (10) 703 (8) 703 (9) 701 (9) 701 (11)

571 (100) 571 (100) 571 (31) 569 (100) 569 (100) 557 (100) 557 (100) 555 (24) 555 (71)

551 (10)

kaempferol-7-Oneohesperidoside didymin poncirin linarin fortunellin narirutin naringin isorhoifolin rhoifolin

701 (13)

555 (100)

537 (23)

701 (12) 701 (8) 699 (11) 699 (11) 687 (10) 687 (10) 685 (8) 685 (11)

555 (100) 555 (100) 553 (100) 553 (100) 541 (100) 541 (100) 539 (100) 539 (100)

flavonoid

-A 415 (11) 415 (79) 415 (100)

- (A + H2O)

-D+ H2O

-D

- (A + 64)d

- (A + R)

- (A + R + H2O)

others

409 (14) 409 (22) 397 (29)

351 (25)

251 (12)

415 (45) 415 (62) 415 (100) 415 (100)

395 (32) 395 (38) 397 (29) 397 (42)

411 (10) 411 (15)

269 (12) 351 (26) 351 (44)

269 (13)

251 (10) 251 (13) 251 (25)

253 (10) 271 (11) 253 (20)

415 (45)

411 (11)

415 (14) 415 (66) 535 (14)

521 (16)

393 (17) 393 (40)

415 (38) 415 (12) 415 (75)

409 (11)

415 (42)

395 (16)

251 (10) 379 (26) 379 (28)

269 (11)

251 (12) 251 (11)

a The m/z of the parent complex and its relative intensity after CAD are given. b Only fragment ions with more than 10% relative intensities are listed. c The losses are abbreviated as follows: R ) rhamnose residue (146), A ) aglycon moiety, and D ) disaccharide group (308). d The elimination of 64 Da may involve consecutive losses, i.e., the loss of water and 46 Da (a C2H6O moiety or H2O and CO).

Scheme 1. CAD Fragmentation Pathways for the (Ag + Rutin)+ Complex Using Speculative Structures of the Fragment Ionsa

a Only the m/z values corresponding to one isotope of the silver ion (107) are listed for simplicity. The letters A and R represent the aglycon and rhamnose residues, respectively. The elimination of 64 Da may involve consecutive losses, i.e., the loss of water and 46 Da (a C2H6O moiety or H2O and CO). Please note that the dashed lines in the proposed structures indicate that the actual coordination sites of the silver ion are not known.

of one water molecule (m/z 411), and a few ions in the low-mass range (m/z 251, 253, 269, etc.) (Table 2). Despite the similarity between datiscoside and kaempferol-3-O-rutinoside, these two

flavonoids are distinguishable from each other by the significant differences in the relative intensities of product ions at m/z 555, 397, and 351, in addition to the low-mass fragment ions. The CAD Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

1767

pattern of the kaempferol-7-O-neohesperidoside complex is simpler and quite different from the above two flavonol complexes by having the loss of the rhamnose residue (m/z 555) (instead of the loss of the aglycon residue (m/z 415)) as the most dominant fragmentation pathway, and the other fragmentation pathways shared by the datiscoside and kaempferol-3-O-rutinoside complexes are much less pronounced for the kaempferol-7-O-neohesperidoside complex. In addition, the silver complex of kaempferol7-O-neohesperidoside has a unique fragment ion at m/z 537, stemming from the loss of the rhamnose residue plus one water molecule (Table 2). Similar to the complex of kaempferol-7-O-neohesperidoside, the CAD mass spectra of the complexes of the two flavanone diglycosides, didymin and poncirin (also with MW 594), are dominated by the loss of the rhamnose residue (m/z 555) (Table 2). These two flavanones can be distinguished from kaempferol7-O-neohesperidoside by the presence of an additional fragment ion at m/z 393 (from the loss of the disaccharide residue) and the absence of the m/z 537 ion (loss of the rhamnose group and one water molecule). The CAD pattern of the poncirin complex differs from the didymin complex by its much more pronounced fragments at m/z 415 (from the loss of the aglycon moiety) and 393. The dissociation patterns of the silver complexes allows successful differentiation of the five isomeric flavonoids, which demonstrates the advantage of silver complexation over deprotonation or transition metal complexation.21-23 MW 592 Isomers. The second flavone isomer series includes linarin and fortunellin, which are two acacetin 7-O-diglycosides. Similar to the complexes of the other flavone 7-rutinosides, the (linarin + Ag)+ complex dissociates exclusively via the loss of the rhamnose residue (m/z 553). The corresponding fortunellin complex yields four other fragment ions, one due to the loss of the aglycon moiety (m/z 415), one due to the loss of the rhamnose residue plus one water molecule (m/z 535), one due to the loss of the disaccharide moiety with rapid addition of a water molecule (m/z 409), and one involving elimination of both the aglycon and rhamnose residues plus loss of one water molecule (m/z 251) (Table 2). MW 580 Isomers. As the fourth flavanone pair, narirutin and naringin are two 7-O-diglycosides of naringenin. The CAD patterns of the silver complexes allow differentiation of these two flavonoids (Figure 6 and Table 2). Both complexes share the same dominant loss of the rhamnose residue (m/z 541) followed by the losses of the aglycon (m/z 415) and the disaccharide (m/z 379) residues, yet the product ion from the loss of the aglycon group is six times more intense for the naringin complex than for the narirutin complex. In addition, the naringin complex has more intense fragment ions at m/z 269 and 251 (stemming from the losses of the aglycon and rhamnose residues without and with a secondary dehydration reaction). MW 578 Isomers. Isorhoifolin and rhoifolin, two 7-O-diglycosides of apigenin, are the flavone versions of narirutin and naringin, respectively. The isorhoifolin complex only dissociates via the loss of the rhamnose moiety (m/z 539). In contrast, there are four additional pathways for the rhoifolin complex: loss of the aglycon moiety (m/z 415), loss of the rhamnose residue and one water molecule (m/z 521), loss of the disaccharide with addition of one water molecule (m/z 395), and elimination of both 1768

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

the aglycon and rhamnose residues in conjunction with dehydration (m/z 251) (Figure 7 and Table 2). 2. Correlations between Structural Features and CAD Fragmentation. The CAD fragmentation pathways cannot only be used to differentiate isomeric flavonoids, but they correlate with the specific classes of flavonoids, with the latter more important for characterizing newly discovered flavonoids. For example, the three flavones that are 7-O-rutinosides (diosmin, linarin, isorhoifolin) undergo only one dominant dissociation reaction: loss of the rhamnose residue. The four flavanones that are 7-O-rutinosides (hesperidin, eriocitrin, didymin, narirutin) undergo at least one or two additional fragmentation pathways besides the dominant loss of the rhamnose residue: loss of the disaccharide group and loss of the aglycon moiety. In contrast, the loss of the aglycon residue becomes dominant for the three 3-O-rutinosides (rutin, datiscoside, kaempferol-3-O-rutinoside) along with at least four other dissociation routes including loss of the rhamnose residue, loss of the aglycon moiety plus one water molecule, loss of the aglycon plus 64 Da (the sugar rearrangement and cleavage process), and elimination of the aglycon and rhamnose groups in conjunction with dehydration. All of the silver complexes containing flavonoid 7-O-neohesperidosides share the same dominant loss of the rhamnose residue with their 7-Orutinoside isomers, but the neohesperidoside complexes either have additional fragmentation pathways or the intensities of the fragment ions other than the loss of the rhamnose residue are much higher than the corresponding rutinoside complexes. For example, the loss of the aglycon residue is over four times more pronounced for a neohesperidoside complex than for the corresponding rutinoside counterpart. Therefore, the CAD fragmentation patterns of the silver complexes can be used to assist in classifying a flavonoid as either a rutinoside or a neohesperidoside, as either 7-O or 3-O glycosylated, and as either a flavanone or a flavone or a flavonol. LC/MS with Postcolumn Complexation. A good isomer differentiation method should not only work well for purified compounds but should also work well for compounds in complex mixtures. Several flavonoids, including isomeric ones, coexist in citrus products, i.e., naringin, narirutin, hesperidin, neohesperidin, didymin, and poncirin in some brands of grapefruit juice,53 and rutin, hesperidin, neohesperidin, naringin, narirutin, didymin, and poncirin in citrus leaves54 and citrus fruits.55 Therefore, identification of flavonoids including isomers in mixtures is an important analytical task. As an application of using silver complexation and tandem mass spectrometry for differentiating isomeric flavonoids, flavonoids extracted from grapefruit juice were analyzed based on a postcolumn complexation LC/MS strategy. Silver complexes for the eluting flavonoids were formed by introducing the silver nitrate reagent in excess via a tee to the HPLC effluent. The resulting silver complexes were then detected by ESI-MS with characterization by tandem mass spectrometry. A typical total ion chromatogram is shown in Figure 8A. To deconvolute the chromatogram and detect minor flavonoids in (53) Ross, S. A.; Ziska, D. S.; Zhao, K.; Elsohly, M. A. Fitoterapia 2000, 71, 154-161. (54) Kawaii, S.; Tomono, Y.; Katase, E.; Ogawa, K.; Yano, M.; Koizumi, M.; Ito, C.; Furukawa, H. J. Agric. Food Chem. 2000, 48, 3865-3871. (55) Kawaii, S.; Tomono, Y.; Katase, E.; Ogawa, K.; Yano, M. J. Agric. Food Chem. 1999, 47, 3565-3571.

Figure 8. Total ion chromatogram (TIC) and reconstructed selected ion chromatogram (SIC) of a grapefruit juice extract by LC/MS after postcolumn silver complexation.

the complex extract, selected ion chromatograms are reconstructed as shown in Figure 8B-D. It can be seen that two major isomeric flavonoids (retention times (RT) ) 8.4, 10.0 min) exist in the grapefruit juice studied, which have a molecular mass of 580 Da (i.e., 687-107 (Ag)). Besides the two major flavonoids, four minor flavonoids belonging to two isomeric series are also detectable in the extract: MW 594 (i.e., 701-107 (Ag)) with RT 15.7 and 17.0 min and MW 610 (i.e., 717-107 (Ag)) with RT 9.4 and 11.1 min. It can also be seen that good separation was achieved for each of the three isomeric series of flavonoids using the C18 column under the gradient conditions as described in the Experimental Section. To pinpoint the identity of each of the eluting flavonoids, CAD was used to probe the complexes. The corresponding CAD spectra for the six flavonoids with different retention times are shown in Figure 9. The two major flavonoids dissociate via the dominant loss of the rhamnose residue (m/z 541) and by the loss of the disaccharide (m/z 379) or the aglycon (m/z 415) residues (Figure 9A,B), which suggests that they are two flavanones. The fact that the loss of the aglycon group (m/z 415 fragment ion) is much more intense for the second eluting flavonoid (RT ) 10.0 min) than the first (RT ) 8.4 min) confirms that the second eluting flavonoid is naringin and the first is narirutin. The four minor flavonoids have similar CAD patterns to the above two flavanones, respectively, with the loss of the aglycon residue more pronounced for the flavonoids with longer retention times than their isomers (Figure 9C-F). Therefore, the four minor flavonoids are didymin (RT ) 15.7 min), poncirin (RT ) 17.0 min), hesperidin (RT ) 9.4 min), and neohesperidin (RT ) 11.1 min). The flavonoids identified are in accordance with the ones reported.53 This example demonstrates that postcolumn silver complexation can be used with tandem mass spectrometry to successfully identify flavonoids in mixtures.

Figure 9. LC/MS/MS analysis of a grapefruit juice extract after postcolumn silver complexation. 33% CAD energy was applied to each complex ion of interest (labeled with asterisks). Losses: R ) rhamnose residue (146), A ) aglycon moiety, and D ) disaccharide group (308).

CONCLUSIONS Silver complexation in an ESI-MS/MS strategy allows enhanced isomer differentiation of flavonoids. This metal complexation method is robust and more universal than previous metal complexation strategies that involved divalent transition metals with auxiliary ligands, trivalent aluminum, or alkaline earth metals. Unlike the previous metal complexation methods, the presence of a 4-keto group and at least one neighboring hydroxyl group (3-, 5-, or both) is not essential for the formation of strong silver complexes. The CAD patterns of the resulting 1:1 silver(I)/ flavonoid complexes are useful for structural characterization and isomer differentiation of seven series of isomeric flavonoid diglycosides. In addition to distinguishing isomeric flavonoids, correlations are identified between the CAD fragmentation pathways of the silver complexes and the structural features of the flavonoids. The 3-O-rutinoside flavonol complexes undergo the dominant loss of the aglycon moiety followed by the loss of the rhamnose residue and a few other fragmentation pathways. The complexes of the flavone 7-O-rutinosides exclusively dissociate by loss of the rhamnose residue, whereas the complexes of the flavanone 7-rutinosides dissociate by other routes as well as the loss of the rhamnose residue. For each neohesperidoside/ rutinoside isomeric pair, the silver complex containing the neohesperidoside typically has two or three additional diagnostic fragmentation pathways or its major fragment ions are substantially more intense compared to the rutinoside counterpart. Therefore, the use of silver complexation and tandem mass spectrometry allows consistent differentiation of flavonoids that differ only in the aglycon type (flavonol vs flavone vs flavanone) and those that differ only in the inter-saccharide linkage (rutinose vs neohesperidose). The glycosylation position also has a signifiAnalytical Chemistry, Vol. 77, No. 6, March 15, 2005

1769

cant impact on the CAD dissociation patterns of the silver complexes of flavonoids: the three flavonol-3-rutinosides have unique fragment ions that are not shared by their 7-rutinoside and 7-neohesperidoside analogues. The silver complexation method is also easily implemented into a versatile postcolumn complexation LC/MS/MS strategy for analysis of flavonoids in mixtures.

1770

Analytical Chemistry, Vol. 77, No. 6, March 15, 2005

ACKNOWLEDGMENT This work was supported by the Welch Foundation (F-1155) and the National Institutes of Health (R01-GM63512). Received for review August 10, 2004. Accepted December 22, 2004. AC048818G