SPE-NMR Hyphenation. A Tool for the Analysis of


LC-DAD-MS/SPE-NMR Hyphenation. A Tool for the Analysis of...

0 downloads 184 Views 219KB Size

Anal. Chem. 2005, 77, 878-885

LC-DAD-MS/SPE-NMR Hyphenation. A Tool for the Analysis of Pharmaceutically Used Plant Extracts: Identification of Isobaric Iridoid Glycoside Regioisomers from Harpagophytum procumbens Christoph Seger,*,† Markus Godejohann,‡ Li-Hong Tseng,‡ Manfred Spraul,‡ Anny Girtler,† Sonja Sturm,† and Hermann Stuppner†

Institute of Pharmacy, Department of Pharmacognosy, Leopold Franzens University Innsbruck, Innrain 52, A-6020 Innsbruck, Austria, and Bruker-Biospin GmbH, D-76287 Rheinstetten, Germany

LC-DAD-MS monitored fractionation of a Harpagophytum procumbens DC. (Pedaliaceae) root extract was combined with a hyphenated LC-DAD-MS/SPE-NMR technique, thus providing the spectral data needed for structure elucidation. This approach allowed the characterization of isobaric iridoid glycoside regioisomers present only as minor constituents. The analytes were identified as the (E/Z) pairs of 6′-O-(p-coumaroyl)harpagide (6′PCHG) and 8-O-(p-coumaroyl)-harpagide (8-PCHG). The fact that 8-(Z)-PCHG constitutes a new natural product underlines the analytical power of this combined approach. Furthermore, derivatives 6′-(Z)- and 6′-(E)-PCHG are new constituents for H. procumbens. Medicinal plants and plant-derived phytopharmaceuticals account for ∼25% of prescribed medicines and ∼50% of the market share in over-the-counter products in industrial countries.1-3 Safety and efficacy of these products are a central issue for legal authorities, manufacturing companies, and the general public. Addressing these topics usually relies on validated analytical methods, which allows for the identification and quantification of relevant secondary metabolites. A prerequisite of establishing such analytical tools is, besides the availability of reference material, a detailed knowledge of the secondary metabolite pattern under investigation. The analysis of these highly complex matrixes is still one of the major challenges in analytical chemistry. It has been one of the most prominent driving forces for chromatographic and spectroscopic method development within the past several decades. Currently, besides miniaturization of equipment, which is cumulating in the development of microseparation techniques (µ-HPLC, CE, CEC), the development of hyphenated techniques has gained an unforeseen acceleration within the last several years. The combination of powerful separation techniques, * Corresponding author. E-mail: [email protected]. Tel: +43 512 5075344. Fax: +43 512 5072939. † University of Innsbruck. ‡ Bruker-Biospin GmbH. (1) Hostettmann, K.; Wolfender, J. L.; Terreaux, C. Pharm. Biol. 2001, 39 (Suppl.), 18-32. (2) Cordell, G. A. Phytochem. Rev. 2002, 1, 261-273. (3) Hostettmann, K.; Marston, A. Phytochem. Rev. 2002, 1, 275-285.

878 Analytical Chemistry, Vol. 77, No. 3, February 1, 2005

(HPLC, GC, CE, CEC) with sophisticated structure characterizing detection devices such as the mass spectrometer (MS) or nuclear magnetic resonance spectrometer (NMR), has remarkably increased the analytical capacities. In the field of NMR hyphenations,4-10 which has become widely accepted in the bio-analytical community, one of the latest technological improvements was the recently introduced LC-SPE-NMR hyphenation.11-13 It allows the temporal decoupling of analytical separation and spectroscopical characterization of the analytes, as previously realized with LC NMR off-line peak parking (e.g., loop collection) systems. This method has several advantages compared to these other approaches. Analytes eluting from the separation system are trapped on solid-phase extraction (SPE) cartridges under appropriate solvent conditions and are transferred to the NMR spectrometer for measurements at a later time. The LC-SPE-NMR setup allows changing the sample solvent to a well defined deuterated NMR solvent in contrast to partially deuterated HPLC solvent mixtures encountered when working with loop collection systems. Therefore, this approach allows the comparison of NMR data obtained from analytes with even pronounced polarity differences directly with the NMR data already compiled from the literature over the past 30 years. In addition, if aprotic deuterated NMR solvents are chosen, signals stemming from acidic protons, which often have high diagnostic value in the structure elucidation process (e.g., (4) Albert, K.; Dachtler, M.; Glaser, T.; Haendel, H.; Lacker, T.; Schlotterbeck, G.; Strohschein, S.; Tseng, L. H.; Braumann, U. J. High Resolut. Chromatogr. 1999, 22, 135-143. (5) Bringmann, G.; Wohlfarth, M.; Rischer, H.; Heubes, M.; Saeb, W.; Diem, S.; Herderich, M.; Schlauer, J. Anal. Chem. 2001, 73, 2571-2577. (6) Wolfender, J. L.; Ndjoko, K.; Hostettmann, K. Phytochem. Anal. 2001, 12, 2-22. (7) Albert, K. On-Line LC-NMR and Related Techniques; Wiley & Sons: Chichester, 2002. (8) Rapp, E.; Jakob, A.; Schefer, A. B.; Bayer, E.; Albert, K. Anal. Bioanal. Chem. 2003, 376, 1053-1061. (9) Spraul, M.; Freund, A. S.; Nast, R. E.; Withers, R. S.; Maas, W. E.; Corcoran, O. Anal. Chem. 2003, 75, 1546-1551. (10) Wolfender, J. L.; Ndjoko, K.; Hostettmann, K. J. Chromatogr., A 2003, 1000, 437-455. (11) Nyberg, N. T.; Baumann, H.; Kenne, L. Magn. Reson. Chem. 2001, 39, 236240. (12) Exarchou, V.; Godejohann, M.; Van Beek, T. A.; Gerothanassis, I. P.; Vervoort, J. Anal. Chem. 2003, 75, 6288-6294. (13) Nyberg, N. T.; Baumann, H.; Kenne, L. Anal. Chem. 2003, 75, 268-274. 10.1021/ac048772r CCC: $30.25

© 2005 American Chemical Society Published on Web 12/23/2004

the hydrogen bond stabilized hydroxy proton at C-5 in flavonoids), can be detected.12 Our current investigation on the composition of the methanolic extract obtained from the secondary root tubers of Harpagophytum procumbens DC (Pedaliaceae) is driven by the goal to establish new analytical methods for the quality control of this often adulterated or insufficiently characterized herbal medicine.14-16 This plant, also known as “devils claw”, is currently one of the most promising phytomedicines used for the treatment of rheumatism, polyarthritis, and osteoarthritis. Its medicinal values are due to its analgesic, antiinflammatory, and antiphlogistic properties.17-22 These activities can be associated with the presence of iridoid glycosides and phenylpropanoid glycoside derivatives of broad structural variability. Since isolation and NMR-based characterization of a large number of bioactive H. procumbens metabolites has already been achieved,23 rapid identification of already known major and minor constituents is mandatory to optimize work flow in terms of labor and financial resources. This approach of accelerated analyte identification has become known as “rapid dereplication” if applied to analyte mixtures of bioactive extracts.24 It is based on online gathering of UV, MS, and NMR data from hyphenated systems and has become a central issue in the field of phytochemical and biomedical analysis.1,5,6,10,25,26 Often analytes can be elucidated using only microgram quantities present in an HPLC injection volume. This makes the timeconsuming and expensive isolation of milligram quantities of pure compounds for the sole need of structure elucidation unnecessary. However, currently this method is still limited by the relative insensitivity of NMR and is most often used only for major constituent identification. In this paper, we will demonstrate, how this approach can be extended to minor constituents by combining HPLC-DAD-MS monitored extract breakdown using simple chromatographic procedures with HPLC-DAD-MS/SPE-NMR hyphenation-based analyte identification. EXPERIMENTAL SECTION Instrumentation: LC-DAD-MS (Innsbruck), LC-DAD-MS/ SPE-NMR, and LC-DAD-NMR (Bruker BioSpin, Rheinstetten) measurements were carried out using Agilent 1100 liquid chromatographs (Waldbronn, Germany), consisting of HPLC quaternary pumps, autosamplers, column ovens, and diode array (14) Eich, J.; Schmidt, M.; Betti, G. Pharm. Pharmacol. Lett. 1998, 8, 75-78. (15) Sporer, F.; Chrubasik, S. Z.. Phytother. 1999, 20, 335-336. (16) Cygan, F. C.; Hiller, K. In Teedrogen und Phytopharmaka; Wichtl, M., Ed.; WVG: Stuttgart, 2002; pp 271-273. (17) Erdo ¨s, A.; Fontaine, R.; Friehe, H.; Durand, R.; Po ¨ppingaus, T. Planta Med. 1978, 34, 97-108. (18) Lanhers, M. C.; Fleurentin, J.; Mortier, F.; Vinche, A.; Younos, C. Planta Med. 1992, 58, 117-123. (19) Baghdikian, B.; Lanhers, M. C.; Fleurentin, J.; Olliver, E.; Maillard, C.; Balandard, G.; Mortier, F. Planta Med. 1997, 63, 171-176. (20) Chrubasik, S.; Junck, H.; Breitschwerdt, H.; Conradtt, C.; Zappe, H. Eur. J. Anaesthesiol. 1999, 16, 118-129. (21) Chrubasik, S.; Model, A.; Black, A.; Pollak, S. Rheumatology 2003, 42, 141148. (22) Andersen, M. L.; Santos, E. H. R.; Seabra, M. L. V.; da Silva, A. A. B.; Tufik, S. J. Ethnopharmacol. 2004, 91, 325-330. (23) Holz, W. In Hagers Handbuch der pharmazeutischen Praxis; Ha¨nser, R., Keller, K., Rimpler, H., Schneider, G., Eds.; Springer: Berlin, 1999; Vol. 5, pp 384390. (24) Cordell, G. A.; Shin, Y. G. Pure Appl. Chem. 1999, 71, 1089-1094. (25) Kraus, W.; Ngoc, L. H.; Conrad, J.; Klaiber, I.; Reeb, S.; Vogler, B. Phytochem. Rev. 2002, 1, 409-411. (26) Corcoran, O.; Spraul, M. Drug Discovery Today 2003, 8, 624-631.

detectors. MS measurements were performed on ESQUIRE-3000 ion trap mass spectrometers (Bruker Daltonik, Bremen, Germany) with electrospray ion sources. For the LC-DAD-MS/SPE-NMR hyphenation, 5% of the eluent was split into the MS using a BNMI (Bruker BioSpin Rheinstetten NMMS-Interface). The Bruker/ Spark Prospekt 2 solid-phase extraction unit (Bruker BioSpin, Rheinstetten, Germany & Spark, Emmen, Holland) was used to automatically trap the chromatographic peaks on HySphere Resin GP or HySphere C18 cartridges (10 × 2 mm, Spark, Emmen, Holland) after postcolumn addition of water using a Knauer K120 HPLC pump (Berlin, Germany). The trapped peaks were first dried with nitrogen and eluted with deuterated acetonitrile (CD3CN) or deuterated methanol (CD3OD) into an AVANCE 600-MHz NMR spectrometer equipped with a LC SEI 13C, 1H probe head with an active volume of 30 µL from Bruker BioSpin. The LCDAD-NMR hyphenation consisted of a Bruker LC22 quaternary gradient HPLC pump, a Bruker diode array detector, and a peak sampling unit (BPSU-36) that allowed the storage of 36 peaks for subsequent NMR measurement on a AVANCE DRX600 spectrometer (Bruker Biospin). The spectrometer was equipped with an identical probe head. Chemicals. Acetic acid (p.a. grade), acetonitrile (GC grade quality), dichloromethane, ethyl acetate, methanol, 1-propanol (all p.a. grade), silica gel (KG60, 40-63 µm), TLC plates (silica gel KG60-F254), and water (LiChrosolv quality) were obtained from Merck (Darmstadt, Germany). Formic acid was purchased from Fluka (Buchs, Switzerland) and Sephadex LH-20 was obtained from Sigma-Aldrich (Sigma-Aldrich, Vienna, Austria). Helium 5.0 was supplied by Messer-Griesheim (Krefeld, Germany). Nitrogen (99.995%) for the mass spectrometry was produced by a nitrogen generator (Peak Scientific Instruments Ltd., Fountain Crescent, U.K.). CD3CN (99.8%) was obtained from Cambridge Isotope Laboratories (Andover, MA), and CD3OD was from Deutero GmbH (Kastellaun, Germany). All solvents for preparative separations were distilled prior to use, and water for the counter current chromatography (HSCCC) was produced by reverse osmosis followed by distillation. Sample Preparation. Dried H. procumbens secondary root tubers (400 g) of European Pharmacopoeia quality (Radix harpagophyti, purchased at a local reseller) were ground and extracted exhaustively with methanol. A voucher specimen (KL2848/98) was deposited at the Department of Pharmacognosy, Institute of Pharmacy, University of Innsbruck. The LC-DAD-MS/SPE-NMR sample was obtained by crude fractionation of an aliquot (30 g) of this extract (150 g) over a silica gel column (dichloromethane/ methanol step gradient) followed by HSCCC (ethyl acetate/1propanol/water, lower phase ) mobile phase) of a subfraction of 1.4 g. Final enrichment was facilitated by methanolic SephadexLH20 column chromatography of an 0.36 g HSCCC fraction. Sample preparation for the LC NMR hyphenation experiments were carried out by dissolving an aliquot (7 mg) of the final fraction (50 mg) in 0.5 mL of methanol. The purification process was monitored by TLC, LC-DAD, and LC-DAD-MS. TLC was performed on silica gel plates, with dichloromethane/methanol/ water (80:20:2) as mobile phase and vanillin/sulfuric acid as the detection reagent. LC-DAD-MS. Separations were performed on a Zorbax SBC18 column (150 mm × 4.6 mm) with a particle size of 3.5 µm Analytical Chemistry, Vol. 77, No. 3, February 1, 2005

879

(Agilent Technologies) and a solvent gradient of 0.1% acetic acid in water (A) and 0.1% acetic acid in acetonitrile (B). The gradient consisted of program: 0 min, 2% B; 4 min, 10% B; 8 min, 18% B; 14 min, 60% B; and 20 min 98% B at a flow rate of 1 mL/min. The system was operated at 50 °C. The injection volume was 10 µL. Chromatograms were recorded at 205, 280, and 325 nm. DAD spectra between 190 and 600 nm were stored for all peaks exceeding a threshold of 0.1 mAU. Between runs, the column was equilibrated with 98% A for 10 min. The LC flow was split 1:5, and the temperature of the ESI interface heated capillary was 300 °C. The nebulizer gas (N2) pressure was set to 2.75 bar (40 psi), and a dry gas (N2) flow of 10 L/min was used. The spray voltage was set to -4.5 kV, and MS data were acquired in the negative mode over a scan range between 50 and 1000 Da. LC-UV-MS/SPE-NMR. Chromatography. The chromatographic separation was carried out on a Supersphere C18 endcapped column (250 mm × 2 mm) with a particle size of 5 µm, from Trentec (Gerlingen, Germany). The flow rate was 0.2 mL/ min, and the injection volume was 10 µL. Gradient elution was performed using water (solvent A) and acetonitrile (solvent B), both containing 0.1% (v/v) formic acid, with the following linear gradient combination: 0 min, 2% B; 8 min, 18% B; 18 min, 45% B; and 25 min, 98% B. After the chromatographic separation, water with 0.1% (v/v) formic acid was added to the eluent of the column with a Knauer K-120 HPLC pump (Knauer, Berlin, Germany) at a flow rate of 0.8 mL/min (makeup pump), to provide proper retention of the peaks under study on the sorbing material of the cartridges. In a second step, the cartridges were dried with nitrogen gas to remove all residual solvents and the trapped analyte was subsequently transferred to the NMR probehead with 255 µL of pure deuterated organic solvent. NMR. 1D 1H NMR spectra of eluted cartridges was recorded using multiple solvent suppression with time-shared double presaturation of the residual water and organic solvent signal proton frequencies. Shaped low-power rf pulse and CW decoupling on the F2 channel for the decoupling of the 13C satellites were utilized. All spectra was recorded at 300 K and referenced to residual solvent peaks (δH ) 1.93 ppm for CD2HCN and δH ) 3.30 ppm for CD2HOD).The data were collected into 32K computer data points with a spectral width of 10 000 Hz. Ninety degree pulses were used with an acquisition time of 1.6 s, and the spectra were acquired by accumulation of 128 scans. Prior to Fourier transformation, an exponential multiplication was applied to the FID that corresponded to a line broadening of 1 Hz. MS. MS data were acquired in a scan range between 50 and 1000 Da with an electrospray interface under positive ionization conditions. The end plate of the ion source was set to 4.5 kV with respect to the grounded needle. The nebulizer was operated at 60 psi with a nitrogen dry gas stream of 11 L/min at 300 °C. The capillary exit voltage was set to 80 V. LC-DAD-NMR. Chromatography was preformed on a Zorbax SB-C18 column (4.6 mm × 150 mm) with a particle size of 5 µm (Agilent Technologies). A solvent gradient of 0.1% phosphoric acid in D2O (A) and 0.1% phosphoric acid in acetonitrile (B) was applied with a gradient composed of the following: 0 min, 2% B; 8 min, 18% B; 18 min, 55% B; and 25 min 98% B at a flow rate of 1 mL/ min. The column temperature was 35 °C. The injection volume was 10 µL, and the chromatograms were taken at 205 and 220 880 Analytical Chemistry, Vol. 77, No. 3, February 1, 2005

Figure 1. LC-DAD-MS chromatogram of H. procumbens methanol extract in comparison to the LC-DAD-MS/SPE-NMR sample. The analytes under investigation are denoted 1-4. (A) LC-DAD chromatogram (280 nm) of the methanol extract. (B) LC-MS negative ESI ion trace ([M - H]- ) 509) of the methanol extract showing the isobaric (M ) 510) analytes. (C) LC-DAD chromatogram (280 nm) of the enriched sample used for the LC NMR experiments.

nm. The analyte peaks were transferred to the spectrometer via the peak sampling unit. The NMR data were acquired under identical conditions as described in the SPE-NMR setup above. RESULTS AND DISCUSSION The methanolic extract of H. procumbens root tubers was fractionated by combining three distinctively different separations systems (silica gel column chromatography, liquid-liquid partition chromatography, and size exclusion dominated column chromatography). Besides the well-known major Harpagophytum metabolites such as harpagide and harpagoside, several small fractions of minor constituents with only minute differences in physicochemical properties such as polarity and molecular size were obtained. ESI-MS spectra acquired with a LC-DAD-MS hyphenation system revealed that some of these mixtures contained groups of two to four isobaric metabolites. Thus, the possibility of further analyte separation via repeated size exclusion column chromatography was excluded (Figure 1). The common approach to tackle this problem would result in the reisolation of the analyte mixture. Semipreparative RP-18 HPLC would be utilized to obtain quantities (at least 5 mg/analyte) sufficient enough to facilitate NMR-based structure elucidation. To avoid this problematic process, LC-DAD-MS/SPE-NMR hyphenation was applied. As a test case sample, an aliquot of a fraction containing four nearly coeluting isobaric derivatives with a molecular mass of m/z ) 510 was chosen (Figure 2). With a single trapping step (∼50 µL of mobile phase, depending on the peak adjusted trapping time), sufficient amounts of the analytes were transferable to the spectrometer and 1D 1H NMR spectra were obtained within 10 min (Figure 3). To optimize the process of SPE peak trapping and transferring, two stationary phases (silica-based RP-18 and poly(divenylbenzene) polymer resin based GP) and two different transfer solvents (CD3CN and CD3OD) were tested. Additionally,

Figure 2. LC-DAD-MS chromatogram of the investigated H. procumbens Sephadex LH-20 fraction. Peaks denoted 1-4 are the isobaric (M ) 510) iridoid regioisomers. (A) LC-MS negative ESI ion trace ([M - H]- ) 509) with mass spectra of the respective peak as inset. (B) LC-DAD chromatogram at 280 nm with UV spectra of the respective peaks as insets.

a LC-DAD-NMR experiment using a loop collection device for analyte transfer was performed for comparison. The signal-to-noise (S/N) ratios of the obtained spectra was assessed, and remarkable differences between the three investigated stationary phase/eluent combinations were observed (Table 1). The loop collection experiment (experiment 1), using equivalent transfer volumes for all analytes, allowed the analyte ratio from the respective S/N values of 7:3:25:65 (1:2:3:4) to be estimated. Therefore, with a 140 µg compound mixture on column, the amount of the least populated analyte (2) was ∼5 µg (∼10 nmol). Similar ratios with superior overall S/N values were achieved by combining the RP18 SPE stationary phase with CD3CN as eluent (experiment 2). Changing to GP resin as trapping phase and using CD3CN as transfer solvent (experiment 3) resulted in losses of the S/N sum. Particularly, analyte 3 was insufficiently transferred to the spectrometer, whereas 1 experienced a slight S/N increase. In contrast, when combining the polymer resin phase with CD3OD as eluent (experiment 4), a remarkable drop in the S/N ratios of

analytes 1 and 4 were observed, whereas the S/N of 3 was increased. The overall achievable S/N was identical to experiment 2 and lower than that for the loop collection system. It can be concluded that the polymer-based resin showed an overall reduced retention capability for the analyte class under investigation (lowered overall S/N). Furthermore, differences in the elution power of the transfer solvent (e.g., lowered S/N of 3 in experiment 3 and 1 and 4 in experiment 4) had to be observed. The increased S/N of 1 in experiment 3 indicates that in most cases, including the RP-18 phase, analyte retention was not completely optimized. Signal-to-noise ratios obtained from SPE hyphenation-based NMR spectra should not be used as a measure for relative amounts of analyte present in mixtures. The overall irregular pattern of achievable S/N ratios does prove the necessity to optimize transfer conditions (SPE phase and NMR solvent chosen) on a case-tocase basis. Monitoring analyte trapping and transferring by recording 1H NMR spectra during repeated transfers supports this optimization process experimentally. The simplification of the Analytical Chemistry, Vol. 77, No. 3, February 1, 2005

881

Figure 3. LC-DAD-derived chromatogram (205 nm) of the LC-DAD-MS/SPE-NMR experiment. The inset NMR spectra (600 MHz, solvent CD3CN) were obtained by single trapping of the respective isobaric peaks 1-4 on RP-18 SPE cartridges. Table 1. Comparison of S/N Ratios Obtained for Different Sample Transfer Conditions Using an Off-Line LC-DAD-NMR System (Experiment 1) and the LC-DAD-MS/SPE-NMR Hyphenation (Experimenta 2-4)a expt

sampling conditions

NMR solvent

1

2

3

4

sum 1-4

1 2 3 4

loop collection HySphere C18 HySphere Resin GP HySphere Resin GP

D2O/CH3CN gradient CD3CN CD3CN CD3OD

18.9 (7) 37.4 (10) 42.4 (23) 2.2 (1)

6.7 (3) 12.6 (3) 14.0 (8) 9.2 (6)

69.2 (25) 81.0 (21) 9.7 (5) 93.1 (55)

177.8 (65) 260.1 (66) 119.6 (64) 63.9 (38)

272.6 391.1 185.7 168.4

a Ratios were calculated from the NMR spectra (9.0-5.5 ppm) using the spectrometer manufacturers’ software and relative S/N ratios (% of the S/N sum of all four analytes in one experiment) are given in parentheses.

general analytical problem by extract breakdown allows the use of shortened HPLC methods and enables the optimization of the LC-DAD-MS/SPE-NMR hyphenation within a few hours. The isobaric metabolites 1-4 (M ) 510, deduced from negative ESI LC-MS data) showed LC-DAD-derived UV spectra with striking pair wise (1/4, 2/3) similarities (Figure 2B). Since the UV maximums was above 300 nm for all analytes (1, 312 nm; 2, 308 nm; 3, 302 nm; 4, 312 nm), the presence of a conjugated aromatic system with a size ranging between benzoic acid and flavonoids was assumed. Thus, derivatives bearing a 2-phenylpropenoic acid subunit (cinnamic acid derivatives) are likely candidates for 1-4. This is in good agreement with the secondary metabolite profile of H. procumbens. This assumption was supported by the obtained proton spectra (Figure 4) that showed signals of a para-substituted aromatic ring system and an isolated double bond. Both signal pairs did show pronounced polarization 882

Analytical Chemistry, Vol. 77, No. 3, February 1, 2005

effects (∆δ > 0.65 ppm for the phenyl unit and ∆δ > 1.06 ppm for the double bond; for details on shift values and coupling constants, see Table 2), making a para substitution by an hydroxy function and the presence of an carboxyl group in the vicinity of the double bond likely. The coupling constant of the double bond doublet showed pairwise grouping that paralleled the UV spectra. Analyte pair 1/4 does bear an (E) double bond (J ) 16.2 and 15.8 Hz, respectively), whereas 2 and 3 are (Z) isomers (J ) 12.8 Hz). Hence, the aromatic subunit present was assigned as (E/ Z)-[2-(4′-hydroxyphenyl)propenoic acid] ) (E/Z)-p-coumaric acid. This (E/Z) isomerization is accompanied by remarked shift differences, the double bond signals of the (Z) isomers 2 and 3 are lowered by at least ∆δ ) 0.50 ppm compared to the (E) derivatives 1 and 4. Two more signals groups in the downfield region of the NMR spectra also showed shift differences: (i) the AB part of an ABX system was downfield shifted (∆δ > 0.60 ppm)

Figure 4. Downfield region of the 1H NMR spectra of analyte 1-4 with compete signal assignments and chemical structures.

in the case of analytes 1 and 2 (1, δ ) 4.42 ppm and δ ) 4.35 ppm; 2, δ ) 4.38 ppm and δ ) 4.30 ppm) and (ii) the only singlet signal in this region at δ ) 5.61 ppm (1) experienced an upfield shift (∆δ > 0.60 ppm). However, since this grouping is 1/2 and 3/4 in contrast to the 1/4 and 2/3 pairing as found for the (E/ Z) isomers, additional regiochemical differences in the analytes have to be taken into consideration. The remaining signals in the downfield region of the spectrasa pair of polarized AB doublets (∆δ > 1.50 ppm, J ∼ 6.3 Hz, one coupling partner showed an additional coupling of J ∼ 1.5 Hz) and one doublet (J ) 8.1 Hz) without an obvious coupling partnersdid not show shift differences in all cases. The shifted AB signals in combination with the stationary doublet at δ ) 4.55 ppm (1) do indicate the presence of a glycoside moiety, with the AB system representing the -CH2OH methylene (H2-6′) and the doublet being the anomeric proton H-1′. The remaining unassigned signals, including a singlet methyl group at δ ) 1.13 ppm (1) with a pronounced shift difference (∆δ > 0.3 ppm) between 1/2 and 3/4, the shifting downfield singlet, and the pair of polarized AB doublets mentioned above do indicate the presence of an iridoid moiety. This moiety lacks the C-11 carbonyl group at C-4 (unsubstituted double bond C-3/

C-4) and bears two substituents at C-8, since the C-10 methyl group proton signal does not show a line-splitting due to a coupling partner. A structure-based literature search for the gathered partial structures (p-coumaric acid, glycoside, and iridoid backbone) within the Chemical Abstracts (CA)27 yielded a set of 26 hits of different acylated iridoid glycosides from different plant genera. Restricting this set with the molecular mass of the analytes gave a subset of seven entries including four epoxide derivatives with the molecular mass of M ) 508, obviously undistinguishable by the software used. The remaining three analytes were harpagide derivatives bearing p-coumaryl moieties at either the C-6′ (glucose) or C-8 (iridoid) hydroxy function. None of the derivatives obtained showed an acylation at the C-6 hydroxy group. To exclude the possibility of having overlooked these derivatives, a comprehensive structure-based literature search for iridoid glycosides with a harpagide oxidation pattern and acyl residues exceeding acetate at position 6 was performed. It resulted in only a single hit of biological origin, isoharpagoside, which was reported only once.28 The eight CA citations obtained for the isobaric hits covered three (27) SciFinder Scholar, Chemical Abstract Service (CAS), American Chemical Society, 2002. (28) Lichti, H.; Von Wartburg, A. Helv. Chim. Acta 1966, 49, 1552-1580.

Analytical Chemistry, Vol. 77, No. 3, February 1, 2005

883

Table 2. 1H NMR Shift Values, Multiplicities (mult), and Coupling Constants (J) of Compounds 1-4 Measured in CD3CN at 600 MHz and 300 K δH (ppm) mult (J (Hz)) H-

1

2

3

4

1 3 4 6 7a 7b 9 10 1′ 2′ 3′ 4′ 5′ 6′a 6′b 2′′/6′′ 3′′/5′′ 7′′ 8′′

5.61 s 6.28 d, (6.2) 4.90 dd, (6.4, 1.5) 3.66 s 1.76 dd, (4.1, 13.6) 1.68 d, (13.6) 2.44 s 1.13 s 4.55 d, (8.1) 3.17 m 3.37 m 3.35 m 3.51 m 4.42 dd, (12.1, 2.2) 4.35 dd, (12.1, 5.5) 7.50 m 6.83 m 7.64 d, (16.2) 6.38 d, (16.2)

5.60 s 6.29 d, (6.2) 4.91 dd, (6.2, 1.5) 3.68 br s 1.78 dd, (14.3, 3.3) 1.70 d, (14.3) 2.48 s 1.15 s 4.54 d, (8.2) 3.14 m 3.32 m 3.28 m 3.50 m 4.38 dd, (12.1, 2.2) 4.30 dd, (12.1, 5.2) 7.60 m 6.80 m 6.88 d, (12.8) 5.82 d, (12.8)

6.08 s 6.35 d, (6.3) 4.91 dd, (1.1, 6.2) 3.67 br s n.d. 1.84 dd, (3.3,15.8) 2.76 s 1.46 s 4.55 d, (8.1) 3.13 ddd, (3.7, 8.1, 11.7) 3.30 m 3.24 ddd, (3.7, 9.2, 9.2) 3.30 m 3.75 ddd, (2.5, 7.3, 11.2) 3.58 dd, (11.2, 5.5) 7.51 m 6.79 m 6.82 d, (12.8) 5.73 d, (12.8)

6.18 s 6.37 d, (6.8) 4.91 dd, (6.8, 1.4) 3.70 d, (3.3) 2.20 d, (15.4) 1.88 dd, (3.6, 15.4) 2.81 s 1.45 s 4.57 d, (8.1) 3.14 m 3.31 m 3.20 m 3.33 m 3.78 (2.5, 8.4, 11.8) 3.59 (4.8, 6.6, 11.8) 7.47 m 6.82 m 7.57 d, (15.8) 6.26 d, (15.8)

Figure 5. Basic structures and substituents of the elucidated isobaric iridoid regioisomers 1-4.

of four derivatives found. Comparison of 1H NMR spectra from the relevant publications29,30 with the experimental data (Table 2) allowed us to assign 1/2 as (E)- and (Z)-6′-O-(p-coumaroyl)harpagide and 4/3 as (E)- and (Z)-8-O-(p-coumaroyl)harpagide (Figure 5). The substitution at 6′ in the glycoside moiety is responsible for the discussed downfield shift of the CH2OR signal, whereas the substitution at position 8 leads to comparable shifts of the neighbored H-1, H-9, and CH3-10 signals. Compounds 1, 2, and 4 can be considered rare metabolites. Until now, the (E/Z) pair 1/2 has been isolated only once from Rogeria adenophylla,30 thus, constituting new iridoid derivatives for H. procumbens. However, closely related corresponding procumbide derivatives with an additional 6,7-epoxy moiety have been previously described from this source.29,31 Analyte 4 has been isolated repeatedly from both plant sources29,30,32,33 and has been (29) Kikuchi, T.; Matsuda, S.; Kubo, Y.; Namba, T. Chem. Pharm. Bull. 1983, 31, 2296-2301. (30) Potterat, O.; Saadou, M.; Hostettmann, K. Phytochemistry 1991, 30, 889892. (31) Burger, J. F. W.; Brandt, E. V.; Ferreira, D. Phytochemistry 1987, 26, 14531457. (32) Guillerault, L.; Ollivier, E.; Elias, R.; Balansard, G. J. Liq. Chromatogr. RT 1994, 17, 2951-2960. (33) Boje, K.; Lechtenberg, M.; Nahrstedt, A. Planta Med. 2003, 69, 820-825.

884

Analytical Chemistry, Vol. 77, No. 3, February 1, 2005

reported in Harpagophytum zeyheri.34 In this context, the ratio harpagoside (HS)/8-O-(p-coumaroyl)harpagide (8-PCHG) was described as a chemotaxonomical marker allowing the differentiation between H. procumbens and H. zeyheri. Compound 3, the (Z) isomer of 8-PCHG, represents a new natural product with structurally closely related derivatives already described from different sources.30,35,36 The co-occurrence of (Z) and (E) cinnamoyl derivatives of iridoids and phenylethanoids has been frequently reported and has been unequivocally associated with light-induced isomerization processes in the case of p-methoxycinnamoyl derivatives.37 The biological relevance of this isomerization process is unknown, and it is not understood whether the (Z) derivatives are indeed biogenic entities. In the presented case, the (Z) isomers 2 and 3 are not necessarily chromatographic workup artifacts since they can be traced back to the methanolic crude extract (Figure 1). However, a report on the remarkable (34) Baghdikian, B.; Lanhers, M. C.; Fleurentin, J.; Ollivier, E.; Maillard, C.; Balansard, G.; Mortier, F. Planta Med. 1997, 63, 171-176. (35) Fernandez, L.; Diaz, A. M.; Ollivier, E.; Faure, R.; Balansard, G. Phytochemistry 1995, 40, 1569-1571. (36) Kim, S. R.; Lee, K. Y.; Koo, K. A.; Sung, S. H.; Lee, N. G.; Kim, J.; Kim, Y. C. J. Nat. Prod. 2002, 65, 1696-1699. (37) Cogne, A. L.; Queiroz, E. F.; Wolfender, J. L.; Marston, A.; Mavi, S.; Hostettmann, K. Phytochem. Anal. 2003, 14, 67-73.

neuro-protective activity of (E/Z)-8- and 6′-O-p-methoxycinnamoylharpagides from Scrophularia buergeriana did not show stereochemical or regiochemical differentiations of the found effects.36 Thus, at least in this case, any biological significance of the observed isomerization can be ruled out. CONCLUSION Rapid analysis of minor secondary plant metabolites can be achieved by combining rapid and simple separation steps with powerful hyphenated LC-MS and LC NMR methodology. Strict monitoring of the fractionation process with LC-DAD-MS allows selecting subsamples of limited complexity but with similar physicochemical properties such as lipophily, pKa, or molecular weight. The structural characterization of the analytes was facilitated by combining the LC-UV and LC-MS data with LC NMR based 1H NMR spectra now obtainable within a short time due to sample enrichment accompanying the fractionation process. Within this contribution, the novel LC-DAD-MS/SPE-NMR hyphenation was used for NMR data retrieval. The 1H NMR spectra was obtained for all four analytes within one working day including

the optimization of chromatographic and SPE trapping/transferring parameters. Working with different transfer parameters did show that a trial-and-error-based search for an optimal combination of trapping phase and transfer solvent leads to superior results. For all analytes investigated, 1H NMR spectra with sufficient S/N was obtained within 10 min, even the analyte present in the lowest amount (2, 10 nmol). Consequently, data mining and structurebased literature searches led to the rapid identification of the four analytes: one new natural product (3), two metabolites previously not described from H. procumbens (1, 2), and a prominent chemotaxonomical discriminator in the genus Harpagophytum (4). ACKNOWLEDGMENT The authors thank Kathryn Jill Chavez for carefully reading and correcting the manuscript.

Received for review August 17, 2004. Accepted November 8, 2004. AC048772R

Analytical Chemistry, Vol. 77, No. 3, February 1, 2005

885