Application of Directly Coupled HPLC−NMR−MS to the Identification

Anal. Chem. 2000, 72, 1793-1797

Application of Directly Coupled HPLC-NMR-MS to the Identification and Confirmation of Quercetin Glycosides and Phloretin Glycosides in Apple Peel A. Lommen,*,† M. Godejohann,‡ D. P. Venema,† P. C. H. Hollman,† and M. Spraul‡

State Institute for Quality Control of Agricultural Products (RIKILT), P.O. Box 230, NL-6700 AE Wageningen, The Netherlands, and Bruker Analytik GmbH, D-76287 Rheinstetten, Silberstreifen, Germany

Directly coupled HPLC-NMR-MS was used to identify and confirm the presence of quercetin O-glycosides and phloretin O-glycosides in an extract of apple peel. From the MS and MS/MS data, the molecular weights of the intact molecules as well as those of quercetin and phloretin and their sugar moieties were deduced. The NMR data provided information on the identity of the compounds as well as the r and β conformations and the position of the glycosides on quercetin and phloretin. The following O-glycosides of quercetin could be identified: quercetin-3-r-L-rhamnosyl-(1 f 6)-β-D-glucoside (rutin), quercetin-3-β-D-galactoside (hyperin), quercetin-3-β-Dglucoside (isoquercitrin), quercetin-3-β-D-xyloside (reynoutrin), quercetin-3-r-L-arabinofuranoside (avicularin), and quercetin-3-r-L-rhamnoside (quercitrin). Phloretin was present as phloretin-2′-β-D-glucoside (phloridzin) and the 2′-β-D-xylosyl-(1 f 6)-β-D-glucoside. Concentrations were between 0.2 and 5 mg/g of apple peel. Flavonoids, such as the flavonol quercetin and the dihydrochalcone phloretin (see Figure 2), are polyphenolic compounds that occur ubiquitously in foods of plant origin. They may have beneficial effects because of their antioxidant properties and their inhibitory role in various stages of tumor development in animal studies. It is estimated that the human intake of all flavonoids is a few hundreds of milligrams per day. The average intake of the flavonoid subclass of flavonols was inversely associated with subsequent coronary heart disease in most, but not all, prospective epidemiological studies; a protective effect on cancer is less likely.1 Flavonoids present in foods used to be considered nonabsorbable because they are bound to sugars as β-glycosides.2 However, it was found that the type of attached sugar is an important determinant of the bioavailability of these glycosides. Specific sugars may even enhance their bioavailability.3,4 Because of this predominant role of the type of glycoside in the bioavailability of * Correspondence to A. Lommen. E-mail:[email protected]. Fax: +31.317417717. † RIKILT. ‡ Bruker Analytik GmbH. (1) Hollman, P. C. H.; Katan M. B. Free Radical Res. 1999, 31, S75-S80. (2) Ku ¨ hnau, J. World Rev. Nutr. Diet. 1976, 24, 117-191. (3) Hollman, P. C. H.; van Trijp, J. M. P.; Buysman, M. N. C. P.; van der Gaag, M. S.; Mengelers, M. J. B.; de Vries, J. H. M.; Katan, M. B. FEBS Lett. 1997, 418, 152-156. 10.1021/ac9912303 CCC: $19.00 Published on Web 03/18/2000

© 2000 American Chemical Society

Figure 1. Schematic representation of HPLC-NMR-MS.

flavonoids, it is important to unambiguously identify these compounds. Apple peel is known to contain several quercetin and phloretin glycosides5,6 (see Figure 2A). Often these compounds have been identified with HPLC solely by comparing their retention times with those of reference compounds. However, only a limited number of reference flavonoid glycosides are available. Sometimes LC-MS/MS has been used for confirmation and identification.7 This technique alone, however, does not give conclusive results due to the relatively low gain of structural information given by mass spectrometry. It especially is not possiblesin contrast to NMRsto distinguish between possible different glycoside conformations, which are of importance in a biological and physiological sense. Several publications describe the direct coupling of liquid chromatographic (LC) methods with nuclear magnetic resonance (NMR) and mass spectroscopy (MS). Most applications, however, deal with the identification of metabolism products of drug substances in biofluids,8,9,10,11,12 and two examples describe the identification of natural products.13,14 Very recently, Hansen et al.14 (4) Hollman, P. C. H.; Buijsman, M. N. C. P.; van Gameren, Y.; Cnossen, P. J.; de Vries, J. H. M.; Katan, M. B. Free Radical Res. Commun. 1999, 31, 569573. (5) Suarez-Valles, B.; Santamaria-Victorero, J.; Mangas-Alonso, J. J.; BlancoGomis, D. J. Agric. Food Chem. 1995, 42, 2732-2736. (6) Andrade, P. B.; Carvalho, A. R. F.; Seabra, R. M.; Ferreira, M. A. J. Agric. Food Chem. 1998, 46, 968-972. (7) Justesen, U.; Knuthsen, P.; Leth, T. J. Chromatogr., A 1998, 799, 101-110. (8) Scarfe, G. B.; Wright, B.; Clayton, E.; Taylor, S.; Wilson, I. D.; Lindon, J. C.; Nicholson, J. K. Xenobiotica 1999, 29, 77-91. (9) Clayton, E.; Taylor, S.; Wright, B.; Wilson, I. D. Chromatographia 1998, 47, 264-268. (10) Scarfe, G. B.; Wright, B.; Clayton, E.; Taylor, S.; Wilson, I. D.; Lindon, J. C.; Nicholson, J. K. Xenobiotica 1998, 28, 373-388. (11) Scarfe, G. B.; Wilson, I. D.; Spraul, M.; Hofmann, M.; Braumann, U.; Lindon, J. C.; Nicholson, J. K. Anal. Commun. 1997, 34, 37-39. (12) Shockcor, J. P.; Unger, S. E.; Wilson, I. D.; Foxall, P. J. D.; Nicholson, J. K.; Lindon, J. C. Anal. Chem. 1996, 68, 4431-4435. (13) Wilson, D.; Morgan, E. D.; Lafont, R.; Shockcor, J. P.; Lindon, J. C.; Nicholson, J. K.; Wright, B. Chromatographia 1999, 49, 374-378.

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1793

in most cases. In the present study a crude extract of apple peel was injected directly onto a C18 analytical HPLC column. NMR and MS(/MS) spectra were acquired simultaneously. With this combined information we tried to identify or confirm the flavonoid glycosides present, in particular, the position and identity of the sugar moiety and the R or β linkage conformation.

Figure 2. (A) Chemical structures of quercetin (3,5,7,3′,4′-pentahydroxyflavone) (top) and phloretin (2′,4′,6′,4-tetrahydoxydihydrochalcone) (bottom); (B) DAD chromatogram at 370 nm showing quercetin glycosides in the extract; (C) DAD chromatogram at 280 nm indicating phloretin glycosides in the extract.

showed some of the possibilities of HPLC-NMR-MS in research toward structure elucidation of plant constituents such as certain flavonoids. In all these examples, HPLC is directly coupled to NMR and MS. Some practical aspects of this special coupling technique are given elsewhere.15 In the HPLC-NMR-MS setup, the NMR probe has been optimized to give increased sensitivity as compared with traditional NMR. Novel in our experimental setup is the use of MS/MS measurementssas well as UV-vissto trigger the transfer of an HPLC fraction either directly to the NMR or to a temporary sample storage unit. Although the use of MS/MS is, in principle, more selective and has more possibilities than UVvis for the selection of interesting fractions, in our case both MS/ MS and UV-vis will be shown to be selective with regard to the distinction between quercetin- and phloretin-derived glycosides. Also new in this type of research is the use of 1D total correlation spectroscopy (TOCSY) experiments for assignments and structure elucidation of plant constituents in crowded regions of the 1D spectra. Because of the technological innovations, direct identification and confirmation (also including the conformational information) of flavonoid glycosides in extracts, without laborious fractionation and purification of the extract, is shown to be possible (14) Hansen, S. H.; Gemal Jensen, A.; Cornett, C.; Bjornsdottir, I.; Taylor, S.; Wright, B.; Wilson, I. D. Anal. Chem. 1999, 71, 5235-5241. (15) Taylor, S.; Wright, B.; Clayton, E.; Wilson, I. D. Rapid Commun. Mass Spectrom. 1998, 12, 1732-1736.

1794

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

EXPERIMENTAL SECTION Instrumentation. The HPLC system consisted of a quaternary low-pressure gradient-mixing LC22 solvent delivery pump from Bruker Saxonia (Leipzig, Germany), a manual injector from Rheodyne, model 7725i (Cotati, USA), equipped with a 1-mL sample loop, and a diode array detector from J & M (Aalen, Germany). HPLC-NMR measurements were carried out using the BPSU-36 interface coupled to a DRX 500 NMR spectrometer equipped with a 1H/13C 4-mm inverse LC probe head (120 µL) from Bruker (Rheinstetten, Germany). The temporary sample storage unit (BPSU-36) was used for automated multiple NMR experiments overnight. For on-line MS detection, an ESQUIRELC ion trap mass spectrometer, equipped with an electrospray ion source, from Bruker Daltonics (Bremen, Germany) was used. Five percent of the eluent was splitted into the MS using a splitter from LC PACKINGS (Amsterdam, The Netherlands). Figure 1 shows a simple scheme of the directly coupled HPLC-NMRMS setup. Both UV-vis as well as MS/MS were used to trigger the transfer of HPLC fractions to the BPSU (for temporary storage) or directly to the NMR. Chemicals. Acetonitrile (NMR Chromasolv quality) was from Riedel-de-Haen (Seelze, Germany). Deuteriumoxide (99.9%) was obtained from Deutero GmbH (Kastellaun, Germany). Trifluoroacetic acid (99.5%) was purchased from Dr. Glaser AG Basel. Phloretin (2′4′6′4-tetrahydrodihydrochalcone) and phloridzin (2′-O-glucosyl-4′6′4-trihydrodihydrochalcon) were obtained from Extrasynthese (Genay, France). Hyperin, isoquercetrin, quercitrin, spiraeoside, guajeverin () quercetin-3-R-arabino-pyranoside)s wrongly labeled as avicularin (quercetin-3-R-arabino-furanoside)(see Results and Discussion)swere obtained from Roth (Karlsruhe, Germany). Rutin and quercetin were obtained from Sigma (St. Louis, MO). Xylose, arabinose, rhamnose, and glucose were obtained from Merck (Darmstadt, Germany). Apple Peel Extraction. Two and a half grams of apple peel was extracted with 12.5 mL of MeOD/D2O at 90 °C for 15 min. After cooling, the extract was centrifuged (10 min 3000g) and methanol was evaporated for further concentration. Measurements. HPLC: Chromatographic separation was carried out at 35 degrees Celsius using a 250 × 4 mm RP Select B column from Merck (Darmstadt, Germany) with a particle size of 5 µm at a flow rate of 1 mL/min. The injection volume was 100 µL. The mobile phase consisted of 80% D2O with 0.1% trifluoroacetic acid and 20% acetonitrile changed to 70% D2O within 40 min and held for another 10 min. The diode array detector was operated at 370 nm for detection of the quercetin glycosides and 280 nm for the detection of phloretin glycosides. 1H NMR spectra were obtained at 500 MHz (either at 30 or 37 degrees Celsius) using both the stop-flow and loop-sampling methods after applying a double solvent suppression on the acetonitrile and deuterium oxide signals. For structure elucidation, 1D TOCSY spectra were obtained using the pulse program selmlgp (Bruker, Rheinstetten, Germany) with mixing times

varying from 100 µs to 200 ms depending on the coupling constants of the anomeric protons. At least 128 scans were accumulated into 32 000 data points at a sweep width of 10 000 Hz. MS Detection. MS experiments were carried out in negative mode with a scan range from 100 to 700 m/z after adding 150 µL/min of a 20 mM aqueous NH4Ac solution to the splitted ratio of the eluent via a T-piece with a syringe pump. The mass spectrometer was operated in automatic MS/MS mode to detect the residual quercetin and phloretin fragments for triggering the NMR stop-flow measurements or transferral to the BPSU (temporary storage) prior to overnight NMR experiments. NMR Reference Measurements. 1H NMR reference spectra were obtained for several commercially available compounds at 400 MHz on a normal spectrometer. These compounds were dissolved in 75% D2O with 0.1% trifluoroacetic acid and 25% acetonitrile. RESULTS AND DISCUSSION Detection Limit of NMR in HPLC-NMR-MS. The detection limit of the least abundant quercetin glycoside, rutin, was 40 micrograms/mL. The time needed to get a reasonable signal-tonoise ratio of 4 micrograms of this compound (100 µL injected) was ∼1.5 h. Using a BPSU for overnight experiments and taking into account the time needed for the experimental setup and optimization, a successful run with multiple components as described here would take a minimum of 1 day. DAD Detection. Figure 2B shows the UV chromatograms acquired at 370 nm showing only quercetin glycosides. Phloretin glycosides, however, can be observed together with the quercetin glycosides at the detection wavelength of 280 nm (Figure 2C). Detection of Quercetin and Phloretin by MS/MS. Figure 3A shows the total ion current (TIC) chromatogram of apple peel obtained after automatic MS/MS acquisition. Figures 3B and 3C show the selected ion traces of the fragments 301 and 273 obtained in MS/MS mode. For quercetin and phloretin one expects the masses 301 and 273 for an MS operating in negative mode. From these experiments, we know at which position in the chromatogram the NMR experiments should be done. Also the information on the masses of the molecular ions is obtained (see Table 1). Detection of Quercetin and Phloretin Glycosides by NMR. Figure 4 shows representative examples of spectra, respectively, peaks Q1 (A), Q2 (B), Q3 (C), D1 (D), and D2 (E). Figure 4F shows the result of structure elucidation of the aliphatic region of D2 after acquisition of a 1D TOCSY using a spinlock time of 65 ms. Figure 4A shows impure quercetin-3-R-L-rhamnosyl(1 f 6)-β-D-glucoside (rutin). A macromolecular phenolic-type oligo/polysaccharide, such as would occur in the cell wall, seems also to be present at this retention time. Figure 4B shows a mixture of quercetin-3-β-D-galactoside (hyperin, major component indicated by +) and quercetin-3-β-D-glucoside (isoquercitrin, minor component indicated by () as well as traces of macromolecular impurities. Figures 4C, 3D, and 3E show, respectively, pure quercetin-3-β-D-xyloside (reynoutrin), phloretin-2′-β-D-xylosyl(1 f 6)-β-D-glucoside, and phloretin-2′-β-D-glucoside (phloridzin). Assignments. The NMR assignments of resonances as given in Table 1 are straightforward. This kind of assignment is relatively

Figure 3. C18 HPLC chromatogram of apple peel extract. (A) Detection by MS using total ion count (T. I. C.). (B) Detection by MS/ MS of mass 274 (D ) phloretin). (C) Detection by MS/MS of mass 302 (Q ) quercetin).

well documented for spectra acquired in DMSO in the literature.16 Resonance positions here deviate because of the difference in solvent. The quercetin in the glycosides is easily recognized by the distinct NMR resonance pattern in the aromatic region, the daughter ion mass of 301 (MS/MS in negative mode), and the UV-vis spectrum. Similarly, phloretin is recognized by its distinct NMR resonance pattern in the aromatic region and the triplet in the aliphatic region as well as the daughter ion mass of 273 (MS/ MS in negative mode). Normally in DMSO two triplets are observed for phloretin in the aliphatic region; this indicates free rotation of the aliphatic “bridge” between the aromatic rings. Here, due to a difference in solvent, as indicated in Figure 4F, free rotation is impeded (probably through hydrogen-bond formation); this follows from the fact that the protons next to the carbonyl group are not equivalent. The glycosidic nature of the flavonoids is also easily observed through NMR. Several resonances typical of sugar moieties can be observed. The linkage position of the flavonoid is derived from comparison with reference compounds. Typically a sugar moiety on the 3 hydroxyl group of quercetin would cause only very minor shifts on H5′, H6, H8 and small shifts on H2′ and H6′, whereas sugar moieties at other positions would have different patterns (16) Markham, K. R.; Geiger, H. The Flavonoids, advances in research since 1986, 1st ed.; Chapman & Hall: London, 1994; Chapter 10.

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

1795

Table 1. NMR and MS(-MS) Data on Peaks Q1-Q6 and D1-D2 (for notation see Figure 2) peak

H6-Da,b

H8-D

H2′-D

H5′-D

H6′-DDc

S H1-Dd

S CH3d

coupling

M. I.e

D. I.f

refg

Q1 (+)h Q2 (+)i Q2 (()j Q3 (+)k Q4l Q5m Q6n

6.32 (2)r 6.32 (2) 6.32 (2) 6.32 (2) 6.32 (2) 6.32 (2) 6.32 (2)

6.55 (2) 6.55 (2) 6.55 (2) 6.55 (2) 6.54 (2) 6.54 (2) 6.53 (2)

7.63 (2) 7.70 (2) 7.63 (2) 7.61(2) 7.51 (2) 7.48 (2) 7.36 (2)

6.96 (8) 6.97 (8) 6.97 (8) 6.97 (8) 6.98 (8) 6.99 (8) 7.01 (8)

7.58 (8;2) 7.57 (8;2) 7.57 (8;2) 7.57 (8;2) 7.48 (8;2) 7.44 (8;2) 7.33 (8;2)

4.95(8), 4.49 (2) 4.91(8) 5.00 (8) 4.99 (8) 5.35 (1) 5.52 (2) 5.22 (1)

1.01 (7)

3, β-pyranose 3, β-pyranose 3, β-pyranose 3, β-pyranose 3, R-furanose 3, β-? 3, β-pyranose

609 463 463 433 433 433 447

301 301 301 301 301 301 301

S S S ND L ND S

peak (+)o

D1 D2p

0.82

H3′-D

H5′-D

H2,6-D

H3,4-D

aliph-Tq

S H1-D

coupling

M. I.

D. I.

ref

6.23 (2) 6.20 (2)

6.06 (2) 6.05 (2)

7.08 (8) 7.08 (8)

6.74 (8) 6.74 (8)

2.74 (7) 2.74 (7)

5.13 (7), 4.31 (8) 5.12 (7)

2′,β -pyranose 2′,β -pyranose

567 435

273 273

S ND

a H is a Proton. b D is a doublet. c DD is a double doublet. d S is a sugar; CH is a desoxymethyl group. e M.I. is for molecular ion. f D.I. is for 3 daughter ion. g In the ref column, S is for the presence of standard, L is for literature, ND is for no data, aliph is for aliphatic protons. h Q1 is for i quercetin-3-R-L-rhamnosyl-(1 f 6)-β-D-glucoside (rutin). Q2 (+) is for quercetin-3-β-D-galactoside (hyperin). j Q2 (() is for quercetin-3-β-D-glucoside (isoquercitrin). k Q3 is for quercetin-3-β-D-xyloside (reynoutrin). lQ4 is for quercetin-3-R-L-arabinofuranoside (avicularin). m Q5 is for quercetin-3β-D-arabinopyranoside or 3-β-D-apiofuranoside. n Q6 is for quercetin-3-R-L-rhamnoside (quercitrin). o D1 is for phloretin-2′-β-D-xylosyl-(1 f 6)-β-Dglucoside. p D2 is for phloretin-2′-β-D-glucoside (phloridzin). q T is for triplet. r 6.32 (2) is for a resonance position at 6.32 ppm (coupling constant of 2 Hz).

Figure 4. Examples of 1H NMR spectra obtained from HPLC NMR. Peak codes Q1, Q2, Q3, D1, and D2 are taken from Figure 3. The most important peaks are indicated by + and (. (A) Peak Q1, quercetin-3-R-L-rhamnosyl-(1f 6)-β-D-glucoside (rutin). (B) Peak Q2, + ) quercetin3-β-D-galactoside (hyperin), ( ) quercetin-3-β-D-glucoside (isoquercitrin). (C) Peak Q3, quercetin-3-β-D-xyloside (reynoutrin). (D) Peak D1, phloretin-2′-β-D-xylosyl-(1f6)-β-D-glucoside. (E) Peak D2, region of the NMR spectrum of phloretin-2′-β-D-glucoside (phloridzin) containing the glucoside and aliphatic protons. (F) Same region and sample as in E: 1D TOCSY with a spinlock time of 65 ms obtained after selective irradiation of the triplet at 2.87 ppm.

and magnitudes of shifts. For phloretin, a similar hypothesis applies. Because of substitution of the 2′ hydroxyl position, resonance H3′ shifts more than H5′, while H2, H3, H5, and H6 remain unaffected. 1796 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

The sugar identity and conformation was obtained from information derived from several pieces of NMR and MS/MS data and by comparison to what is known in the literature. From the difference in mass between the molecular ion (M. I.) and the

daughter ion (D. I.), the mass of the sugar moiety can be calculated by adding the mass of water. From these MS results (see Table 1), it could be concluded that all glycosidic moieties consist of monosaccharides (Q3, Q4, Q5 ) pentose; Q2(+), Q2((), D2 ) hexose; Q6 ) desoxyhexose) except that of Q1 and D1, which are disaccharides; Q1 is consistent with a desoxyhexose and a hexose, and D1 is consistent with a hexose and pentose. The type of monosaccharide moieties known to date are limited to glucose, galactose, xylose, rhamnose, arabinose, mannose, allose, apiose, galacturonic acid, and glucoronic acid. Combining literature and the MS results here leadssin the case of monosaccharide moietiessto a smaller selection of possible sugars for each peak. However, even when including retention times, there is, in a number of cases, a lack of hard evidence with regard to the identity of compounds. For instance, peak Q2 contains two overlapping peaks of quercetin hexose molecules of identical mass. The evidence for this comes from the NMR spectrum (see Figure 4B) in which characteristic resonances of two quercetin glycosides are found. Furthermore, without NMR, difficulties also arise for Q3, Q4, and Q5, which are all quercetin pentose molecules; of these three peaks no commercially available reference standard exists (a commercially available avicularin - Q4 - reference was labeled wrongly!; see Experimental Section). The specific identification of the sugar identity and, together with this, its conformation is derived from comparison of, on one hand, multiplet patterns of NMR resonances obtained from the acquired normal and 1D TOCSY (not shown) spectra and, on the other hand, those of reference sugars and flavonoid glycosides. Using the information from NMR and MS spectra Q1, Q2 (double), Q6 and D2 were concluded to be identical to the appropriate references, respectively, quercetin-3-R-L-rhamnosyl(1 f 6)-β-D-glucoside (rutin, Q1), quercetin-3-β-D-galactoside (hyperin, Q2 +), quercetin-3-β-D-glucoside (isoquercitrin, Q2 (), quercetin-3-R-L-rhamnoside (quercitrin, Q6), and phloretin-2′-β-Dglucoside (phloridzin, D2). Also, Q4 could be identified as quercetin-3-R-L-arabinofuranoside (avicularin, Q4) by comparison of resonance multiplet patterns to literature data.15 For Q3, MS/ MS predicted the presence of a pentose. The multiplet patterns of Q3 sugar resonancessalthough shifted by the aromatic presence of and linkage to quercetinsare identical to that of the β-xylose. Therefore, Q3 can be identified as quercetin-3-β-Dxyloside (reynoutrin, Q3). In the case of Q5, according to MS a pentose should be present. In the case of the NMR, the difficulty lies in partial overlap with Q4 and low levels of the metabolite. Also, the anomeric proton resonance of Q5 exhibits a small coupling constant which makes this resonance an insensitive candidate for 1D TOCSY experiments. Of the known pentoses, rhamnose has the wrong mass and xylose has the wrong multiplet patterns (especially the anomeric proton). Both a 3-β-L-arabinopyranose and a 3-β-D-apiofuranoside are possible with respect to the anomeric proton position and coupling constant (∼2 Hz). The latter is an uncommon sugar with respect to flavonoid glycosides. The first has been reported for quercetin in the past although not in apple. A more precise identification is impeded. With regard to peak D1 the MS/MS spectra report the presence of a pentosehexose disaccharide. The NMR data show two anomeric protons,

which supports the presence of a disaccharide. 1D TOCSY experiments on the resonance at 5.13 ppm of the peak D2 (known to have a β-glucose) and 5.12 ppm of the peak of D1 show similar multiplet patterns as well as resonance positions for the first few resonances to which magnetization is transferred. From this it can be concluded that the sugar moiety is connected to the flavonoid in the same way. Therefore, the first sugar in D1 is β-glucose, which is connected to the 2′ hydroxyl group of phloretin. 1D TOCSY experiments on the anomeric resonance at 4.31 ppm show multiplet patterns similar to that of β xylose; the resonance positions do not coincide precisely as expected, because of the linkage to glucose. The linkage position of the xylose to glucose came from 1D TOCSY experiments on one of the H6 resonances of the glucose in D1 and D2 at, respectively, 4.00 and 3.84 ppm. Although multiplet patterns are similar, the positions of the H6, H6′, and H5 resonances differ approximately 0.15-0.1 ppm in chemical shift, while H1 to H4 remain nearly unchanged. From this it can be concluded that the linkage is 1 f 6-β. CONCLUSION Directly coupled HPLC-NMR-MS proves to be a powerful tool to identify and confirm flavonoid glycosides in a simple apple peel extract, without the necessity of laborious fractionation and purification. The use of MS(/MS) as detector for the transfer of a HPLC fraction to the NMR will besthrough its selectivity (as compared to UV-vis)sa useful future way of pinpointing compounds for NMR analysis. Furthermore, the 1D TOCSY NMR experiment proves to be a powerful tool in structure elucidation when used in the HPLC-NMR-MS setup. The analytical information obtained from only a few micrograms of compound present in a crude extract is, in most cases, sufficient to identify and/or confirm the structures of flavonoid glycosides. The presence of quercetin-3-R-L-rhamnosyl-(1 f 6)-β-D-glucoside (rutin), quercetin3-β-D-galactoside (hyperin), quercetin-3-β-D-glucoside (isoquercitrin), quercetin-3-β-D-xyloside (reynoutrin), quercetin-3-R-L-arabinofuranoside (avicularin), quercetin-3-R-L-rhamnoside (quercitrin), phloretin-2′-β-D-xylosyl-(1 f 6)-β-D-glucoside, and phloretin-2′-βD-glucoside (phloridzin) could be identified/confirmed unambiguously. The identity of one additional unknown quercetin could be narrowed down to quercetin-3-β-D-arabinopyranoside or 3-β-Dapiofuranoside. This technique is indispensable in the identification of flavonoids and their conjugates in plants, especially when no standards of these compounds are available. It is also a very valuable tool in determining the biologically and physiologically important information concerning the identity, conformation, and linkage position of sugars in flavonoid glycosides. ACKNOWLEDGMENT Part of this research was supported by the Wageningen NMR Centre by European Union contract number ERBFMGE-CT970085. Jacques Vervoort is thanked for critical reading. Received for review October 29, 1999. Accepted January 18, 2000. AC9912303

Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

1797