Analysis of by Micro High-Performance Liquid Chromatography-Mass Spectrometry with a Particle Beam Interface

Achille Cappiello, Giorgio Farniglini, and Filippo Mangani lstituto di 5cienze Chlmiche. Universita d i Urbino, Urbina, Italy

Bruno Tirillini Istituto di Botanica e arlo Botanico, Universita di Urbin o, Urbino, Italy

Coumarins are a large group of compounds that are naturally present in plant tissues and that exhibit a wide range of pharmacological properties. Analytical methods based on chromatographic techniques and conventional detectors are inadequate to accurately anal yze coumarins in complex matrices such as plant extracts. In this article a new method based on a modified particle beam liqu id chromatography-mass spectrometry interface is described. The method allows specific and accurate determination of several coumarins in biological ma trices. An application rega rding the anal ysis of 18 coumarins in the extract of Snrfrnium perfoliatum L. is also reported. (J Am Soc Mass Spectrom 1995, 6, 132-139)

oumarins are a large family of substances that ionization ten [4, 5], and electron attachment [6] have are usually extracted from plants, where they been employed successfull y. Cpursue a distinct biological role: production of Traldi and co-workers [7-10] have investigated the coumarins appears to increa se when the plant becomes structures of furocoumarin isomers, which cannot be the object of external aggression. Most of the interest in distingui shed with conventional mass spectrometric these compounds arises from their pharmacological techniques. The y established a new approach in the properties and some toxicological risks to humans. For investigation of these compounds based on high and example, has a stabilizing effect on biolog i­ low energy collision-activated dissociation . cal membranes with a resultant hepatoprotective activ­ Some of the analytical methods described in the ity. Scopoletin also inhibits platelet aggregation and, literature rely on gas chromatographic-mass spectro­ moreover, it blocks the neurotransmitters of the cholin­ metric analyses of derivatized coumarins [11]. Meth­ ergic and adrenergic neuroeffector junctions. Pso­ ods that employ direct gas chromatographic injections larens, another group of cournarins, are characterized of untreated material also were proposed for selected by photobiological effects and are used in the pho­ compounds [12-14], Liquid chromatographic methods tochemotherapeutic treatment (PUVA therapy) of some are extensively used for the analysis of rather complex skin diseases with hyperproliferative conditions [1]. mixtures of different coumarins. These methods in­ Unfortunately, the pharmacological effectiveness of clude thin layer chromatography (TLC) [15] and high­ these drugs is impaired by side effects such as severe performance liquid chromatography (HPLC) methods dermatitis, erythema, phototoxicity, and possible in­ with reversed-phase and direct-phase separations [16, creased risk of sk in cancer, which is mainly related to 17]. TLC also is widely used for sample preparation the capacity of psolarens to form light-induced mono and clean-up procedures for plant extracts. Alcock et or bis adducts with DNA pyrimidine bases [2]. aJ. [18] described an interesting and pioneering appli­ Mass spectrometry has shown a great potential for cation of liquid chromatography-mass spectrometry structure elucidation and isomer characterization of analysis (LC/MS) of timbo powder extracts. They used several coumarins. Different ionization techniques like a glass-lined stainless steel microbore column coupled electron impact (En [3], pos itive and negative chemical to a mass spectrometer via a moving belt LC/MS interface. Current LC/MS interfaces allow realization of the inherent potential of both techniques, which thus ex­ Address reprint request s to Dr. Achi lle Cap piell o, Ist ituto di Scie nze Chimiche, Un iversita d i Urb ina, Piazza Rina scimento 6, 61029 Urbina, pands the application capabilities and greatly simpli­ Italy . fies the overall analytical procedures. In particular, the

© 1995 American Society for Mass Spectrometry Received March 14, 1994 1044-0305/95/$9.50 Revised July 1, 1994 55011044-0305(94)00096-1 Accepted July 20, 1994 J Am Soc Mass Spectrom 1995,6,132-139 MICRO-HPLC/MS WITH A PARTICLE BEAM INTERFACE 133 particle beam (PB) interface [19-28] has proved valu­ ture was 40°C for all the experiments. The pressure able for LC/MS analysis of small molecules. It is fully was reduced to about 0.5 attn in the desolvation cham­ compatible with electron impact ionization (ED or ber, 0.3 torr in the second stage of the momentum chemical ionization (CD, with no significant constraints separator, and 5-8 X 10- 5 torr in the manifold of the regarding the liquid chromatography requirements or ion source. The ion source temperature was set at 250 analyte polarity, which makes it an invaluable tech­ °C and the analyzer was at 120 "C. The mass spectrom­ nique for analyses of complex mixtures. The analysis eter tuning and calibration were performed automati­ of coumarins by LC-PB/MS may offer a valid method cally by using perfluorotributylamine as the reference for evaluation of the presence of certain coumarins in compound. The repeller potential was adjusted manu­ specific plants or for monitoring variable concentra­ ally to monitor fragment ions with mass-to-charge tions when coumarins are used as pharmaceutical ratios close to sample values. The mobile phase was preparations. allowed into the ion source during calibration. The Our research group recently has modified a particle mass spectrometer was scanned from m /': 50 to 300 beam interface to make it compatible with mobile with a threshold of 50 counts. The scan speed was 1.2 phase flow rates as low as 1 ,u.L/min [29-31]. This scans per second, which gave a mean of about 10 new coupling device is fully compatible with preexist­ acquisition samples for each HPLC peak. The complete ing instrumentation and offers several advantages over mass selection report of the selected ion monitoring a conventional interface: drastic reduction in solvent (SIM) program used in this work is listed in Table 1. consumption with negligible contamination of the mass With the exception of a few compounds, to maximize spectrometer by column effluent; wider choice of sol­ the dwell time, the identification criteria were based vents and buffers that are potentially harmful to the on one characteristic ion for each substance. Peak area instrument; better signal response for high water con­ values were calculated with automatic integration. The tent mobile phases with improved sensitivity and electron energy was set at 70 eV in the positive ion chromatographic performance during gradient analy­ mode, ses; easier and more effective tunability of the interface Signal optimization with the new interface was less for chemically different mobile phases and analytes. critical and consistently easier than with the conven­ In this work an LC-PB/MS method for the analysis tional interface. A specific combination of the position of 18 different coumarins, based on a micro particle of the fused silica capillary inside the coaxial gas beam interface, is reported. A sample of Smurnium perfoliaium L. (Umbelliferae) was extracted and ana­ lyzed with this method and several coumarins were Table 1. SIM acquisition program for the analysis identified and quantified. This species can be consid­ of coumarins ered a typical example of -containing plants. Dwell time Group Channels m/z (rnsl 13 73.00 150 Experimental 208.00 150 Particle Beam Interface and Mass Spectrometer 178.00 150 202.00 150 A modified Hewlett-Packard (Palo Alto, CA) 59980B 146.00 150 particle beam interface, coupled with a Hewlett­ 148.00 150 Packard 5989A quadrupole mass spectrometer, was 162.00 150 employed. The original nebulizer was replaced by a 222.00 150 laboratory-made micronebulizer, which has been de­ 192.00 150 scribed in a previous paper [29]. The new device does 206.00 150 not require modifications to the desolvation chamber, 150.00 150 momentum separator, or their original assembly. The 120.00 150 final transfer tube prior to the ion source was in the fully retracted position. A 50-,u.m-i.d., 180-,u.m-o.d. 118.00 150 fused silica capillary tubing (Polymicro Technologies, 2 8 176.00 250 Phoenix, AZ) was used as the nebulizer tip and to 173.00 250 connect the chromatographic column. The nebulizing 246.00 250 gas was helium 5.6 purity grade (> 99.9996%) and was 186.00 250 obtained from SOL (Milano, Italy). The helium flow 160.00 250 rate was about 0.2 Lyrrun when maximum signal re­ 216.00 250 sponse was monitored by the ion source with a mobile 206.00 250 phase flow rate of 2 ,u.L/min. This value corresponds 231.00 250 to a gas pressure of 30 Ib/in.2 and to a linear velocity at the nebulizer tip of 200 my's. The gas temperature 3 2 174.00 900 was ambient while the desolvation chamber tempera- 202.00 900 134 CAPPIELLO ET AL. J Am Soc Mass Spectrom 1995,6.132-139

COMPOUND STRUCTURE COMPOUND NAME MIZ VALUES REL ABUNDANCE scoparone 206. /9/. /63 /00.50,30 0O© bydrocoumarin /48, /20, 9/ /00,80,50 OH OHQOH esculin /78, 73, /50 70, 70,50 /76, /33, /48 /00,60,50

CH: 0 OH/ oro0 -c san/omn /73, /35,246 /00,50,40 OH 0 0

CH~0!¥O fraxeun 208,59, /93 /00,50,30 OH 0 0 0 psolarene /86, 158, /02 100, 70,30 OH \o~oAo~ daphne/in /78, /50,59 /00,40,30 CH~OyO 0/900 r-methyicoumann 160, /32, 13/ /00, 70,50 OH 'GvCH, CH"Oro scopoleun /92, /77, /49 /00.60,40 I OH 0 0 0 o 00Cr0 xamhotoxin 2/6, /73,20/ 100,40,30 OH@00 umbe:lrferon /62, /34. 78 100,80,20

CH-<°ro fraxidin 222,207, /23 /00,40,20 5,7·dimethoxlcoum. 206, /78, /63 /00, 70, 40

...... 0 I 0 a CH) OH

4.merhylumbelltJeron 176,148,147 100, 70,50 ofuOH 231,246, /75 /00, 100,30 :@:yo xanthotoxol 202, /74,89 100, 40,20 ~ 0 A

~ 202, /74,89 /00,20, /0 "o~oAo o Ct:CHI 11 /C, CH) CH,

Figure 1. Coumarin structures with characteristic EI mass-to-charge ratio values and their relative intensities.

tubing and the helium pressure offered the highest sharp restriction of the gas tubing or slightly with­ signal intensity no matter what compound or mobile drawn. phase composition was used. This allowed reproduc­ tion of the same operating conditions throughout the Liquid Chromatography chromatographic run. The optimization procedure was carried out with a flow injection analysis technique for The packed capillary column used in this work was preliminary adjustment, followed by HPLC mass spec­ made in our laboratory from IjI6-in. o.d., 250-p,m i.d. trometry under the required chromatographic condi­ PEEK tubing (Alltech Associates Inc., Deerfield, IL) tions. An important property of the new interface is and was packed with 5-p,m particle size reversed-phase the good performance even with high concentrations C18 purchased from Phase Sep (Queensferry, UK) [32]. of water in the mobile phase. The signal response The 25-cm-long column had an efficiency of about remained nearly constant over a wide range of mobile 15,000 theoretical plates measured at 1 p,Ljmin flow phase concentrations. Tuning, for example, was not rate. Liquid chromatography was carried out with a affected by the water content in the mobile phase, Kontron 420 dual-pump binary-gradient conventional which thus simplified the overall procedure. All the HPLC system (Kontron Instrument, Milano, Italy). Mi­ tuning tests were conducted with a 50% water­ croliter flow rates were obtained with a laboratory­ acetonitrile solution. A Vernier dial allowed IjlO-mm made splitter that was placed between the pumping shifts of the nebulizer capillary tip; the most appropri­ system and the injector [33, 34]. The splitter allowed ate position was with the tip end aligned with the accurate and stable micro flow rates and rapid delivery J Am Soc Mass Speclrom 1995,6, 132-139 MICRO-HPLC jMS WITH A PARTICLE BEAM INTERFACE 135 of solvent concentration changes for reliable and repro­ overall performance with methanol, which gave a ducible gradients. A motor-assisted solvent mixer was higher signal response with the particle beam inter­ placed after the pumps and before the splitter device. face, it was rejected because of its higher viscosity. The For sample injection, a zero-volume Va1co injector micro-HPLC assembly is particularly vulnerable to equipped with a 60-nL internal loop was employed very high operating pressure especially at the connec­ (Valco, Houston, TX). Larger loops are not advisable tion points, where fused silica capillary tubing is held for flow rates lower than 5 p.Ljmin because of loss of in place by weak plastic ferrules. Moreover a mixture chromatographic efficiency. of water and acetonitrile usually gave a more stable Stock solutions were prepared by dissolving 10 mg and constant response during a common gradient run. of each compound in 1 mL of methanol (l0,000 ppm), To confirm these results a specific test was performed. The separation of the coumarin mixture was optimized Figure 2 shows the system signal response versus with a UV detector equipped with a micro flow cell. variations in the mobile phase composition: plot a is For the chromatographic separation of the coumarins, obtained with a combination of methanol and water, a mixture of water and acetonitrile was used as the whereas plot b is obtained with acetonitrile and water. mobile phase. The mobile phase flow rate was set at 2 Sixty nanograms of imperatorin was injected five times p.Ljmin. The acetonitrile relative concentration was for each mobile phase composition. The two curves kept 5 min at 0%, sharply increased to 40%, and then show similar effects of solvent on the ion signal. Char­ increased from 40 to 80% in 25 min for the gradient acteristic and critical relationships between substances run. The initial isocratic step achieved solute band and mobile phases are part of the particle beam perfor­ focusing and improved separation efficiency. mance. The performance of the interface is strongly All solvents were HPLC grade from Farmitalia Carlo dependent on the choice of the mobile phase and on Erba (Milano, Italy) and were filtered and degassed the structure of the analyte . Imperatorin was chosen before use. Reagent water was obtained from a Milli-Q arbitrarily and the results safely may be extended for water purification system (Millipore Corp., Bedford, similar compounds. Chromatographic separation of MA). coumarins was performed with a C18 reversed-phase packed capillary column. The 250-p.m i.d. columns Extraction Procedure Fresh plant (0.150 g) was extracted by manual shaking 250 a with 10 mL of methanol. The crude extract was con­ centrated under vacuum (40 °C ) to 0.5 mL. The concen­ 200 trated extract was separated by TLC (silica gel 60; eluent: toluene-diethyl ether-acetic acid 50:45:5). The 150 zone from start to Rf = 0.6 was scraped and the gel was extracted with methanol. This first cleanup elimi­ 100 nated chlorophyll pigments from the extract. After vacuum concentration the extract was further sepa­ rated by TLC (silica gel 60; eluent: chloroform­ 50 methanol 85:15). The zone from Rf = 0.2 to front was 0 scraped and the gel was extracted with methanol. 0 10 20 30 40 50 60 70 80 90 100 Finally the methanolic extract was filtered (0.2 p.m) organic solvent X and concentrated to 100 p.L. Snujrnium perfoliatum L. was collected from cultivated plants in the garden at 250 the Institute of Botany, University of Urbino, Italy . b Voucher specimens have been deposited in the herbar­ 200 ium of the Institute of Botany, Urbino University, Italy (S.P. 1-93). 150

Results and Discussion 100

A number of different coumarins were analyzed dur­ 50 ing this study. All of them were suitable for reversed­ phase liquid chromatography and gave reproducible o+--I---+---+----+--+--l----+---+--t-~ and interpretable EI mass spectra. Figure 1 shows the o 10 20 30 40 50 60 70 80 90 100 structural formulas of the 18 components, three charac­ organic solvent X teristic EI mass spectral fragments, and their relative Figure 2. Instrument signal response (1) versus variation in intensities. A solution of water and acetonitrile was mobile phase composition: (a) methanol and water; (b) acetoni­ used as the mobile phase. Despite a slightly better trile and water. 136 CAPPIELLO ET AL. J Am Soc Mass Speclrom 1995, 6, 132-139

Abundance

6 20000- 7

18000-

15

16000 16 6 9

5 16 14000

12

12000 J 'I

10 II 10000- 13

2 11 17 8000- ~ W\.J AJ '-' \ I I 8 12 16 20 24 28 TIme (min.) Figure 3, Reconstructed ion chromatogram of coumarin standard mixture: (1) hyd rocoumarin; (2) escul in; (3 ) ; (4) daphnetin; (5) scopo letin; (6) umbelliferon; (7) fraxidin; (S) -l-methyl­ umbelliferon : (9) xanthotoxol; (10) scoparone; OIJ herniarin; (12) santon in; (l3l psolarene ; ( 4) 7-methy icoumarin; (15) xanthotoxin; (16) 5,7-d imethoxicoumarin; ( 7) isopimpinellin; OS) impera­ torin. The area counts intensity is reported on the !I axis.

performed best with a mobile phase flow rate of about a linear and reproducible gradient. For a correct identi­ 2 p.L/min. The lowest height of the reduced theoreti­ fication, precise and reliable retention times are essen­ cal plate h is obtained with a reduced linear velocity v tial. The micro-HPLC system assured retention time 1.87, which corresponds to a mobile phase flow rate of variations lower than ±1% for a retained compound 1 p.L/min. Reduced chromatographic parameters, such during a gradient anal ysis [34]. Compound transfer as h and u, were introduced several years ago by delay between column output and the appearance of a Bristow and Knox [35] to allow comparison among mass spectrometric signal was less than 1 min . This columns packed with different materials and packing travel time does not influence the chromatographic methods, and when eluents of different viscosities or band shape and does not compromise separation effi­ solutes of different diffusion coefficients were used. In ciency even for critical separations. fact, for example, the plate height (H) is expressed in Figure 3 shows a reconstructed ion chromatogram terms of particle diameter units (d p ), h = H/dp ; the obtained from a SIM analysis of the coumarin standard eluent velocity (u) is expressed relati ve to the rate of mixture with amounts that ranged between 30 and 100 diffusion across a particle (Om)' V = udp/Om. These ng. Figure 4 shows the SIM traces that correspond to parameters are dimensionless and allow read y com­ some partially overlapped peaks eluted between 13:00 parison of results obtained in different experimental and 16:00 min. All signal interferences were virtually conditions. Slightly higher flow rates can speed up the eliminated. Variable signal response was observed with analysis with a negligible loss of chromatographic ef­ different coumarins. Table 2 shows detection limits for ficiency. This situation is due to a flatter van Deemter the 18 coumarins analyzed with the program described curve around the lowest flow rates . The splitter device previously. The limits were determined by analysis of adopted for the micro flow rate generation guarantees the specified quantity and for a signal-to-noise ratio of J Am Soc Mass Spectrom 1995, 6,132-139 MICRO-HPLC/MS WITH A PARTICLE BEAM INTERFACE 137

Table 2. Instrument detection limits for the 18 coumarins 5:1. Calibration and injection reproducibility were considered; signal-to-noise ratio 5:1 checked with the method proposed by using impera­ Quantity torin. Each substance was injected five times in differ­ Compound (ng) ent amounts that ranged between 1 and 60 ng to be 1. Hydrocoumarin 2.0 consistent with the actual sample concentration. Each 2. Esculin 50.0 sample was introduced without the column, directly 3. Fraxetin 70.0 into the mass spectrometer via the particle beam inter­ 4. Daphnetin 40.0 face and with a mobile phase composed of an equal 5. Scopoletin 7.0 concentration of water and acetonitrile. The equation 6. Umbelliferon 1.5 relative to the linear regression plot for the concentra­ 7. Fraxidin 1.5 tion calibration is y = 35.2x - 34.1 with a correlation 8.4-Methylumbelliferon 1.0 coefficient of 0.998. The mean standard deviation, cal­ culated by using the average of the peak area values 9. Xanthotoxol 4.0 for each concentration experiment, was ±10.5%. The 10. Scoparone 1.0 mass spectrometer was operated in SIM mode at m/z 11. Herniarin 3.0 202. The mobile phase was kept at 2 JLL/min con­ 12. Santonin 1.5 stantly for all experiments. Excellent linearity can be 13. Psoralene 0.7 observed in the calibration plot and for the concentra­ 14. 7 -Methylcoumarin 5.0 tion range considered. The addition of buffers, usually 15. Xanthotoxin 0.7 employed to improve the particle beam carrier mecha­ 16.5,7-Dimethoxicoumarin 1.0 nism and consequently the response linearity, was not 17. Isopimpinellin 3.0 considered in this case. Different neutral pH buffers 18. Imperatorin 0.6 were tested (i.e., ammonium acetate and potassium

Ion 192.00 amu Abundance Ion 162.00 amu Ion 222.00 amu Ion 17 00 amu 10n20 00 amu 9000 222

8000 176 192 7000

6000 162 202

5000

3000

2000

1000

12.0 13.0 14.0 15.0 16.0 17.0 TIme (min.)

Figure 4. Ion chromatograms relative to scopoletin (III/Z 192), umbelliferon (III/Z 162), fraxidin (III/Z 222), 4-methylumbelliferon (III/Z 176), and xanthotoxol (III/Z 202). 138 CAPPIELLO ET AL. J Am Soc Mass Spectrorn 1'1'15, 6, 132-139

Ion 216.00 amu Abundance

372

368

364

360

356

352

Figure 5. SIM trace at 111/: 216 (xanthotoxin) obtained from the analysis of plant extract.

dihydrogen orthophosphate}, but they all reduced (±1O.5%). Figure 5 shows the SIM trace relative to an chromatographic resolution. Furthermore, linearity for identified coumarin (xanthotoxin). the instrument response was excellent in that concen­ Some of the amounts calculated were equal to or tration range and any buffer addition was unneces­ lower than the corresponding detection limits. In par­ sary. The addition of nonvolatile buffers is tolerated by ticular, scoparone and xanthotoxol gave signal re­ the instrument because of the very low liquid intake sponses close to the signal-to-noise ratio considered for [31]. the evaluation of the detection limits (5:1) and conse­ Methanol plant extracts were concentrated to 100 quently their detection limits correspond to the calcu­ JLL and injected into the system. Identification criteria lated amounts in the sample. Although all other com­ were based on retention times and specific responses pounds fall slightly below these limits, their amounts to characteristic mass-to-charge ratio values. Because have been estimated by a careful comparison of the of the extremely low concentration of the compounds peak areas with those obtained from the diluted stan­ detected in the extract, the instrument sensitivity was dard solution. enhanced by selecting the lowest number of character­ This method can be considered as a valid aid in the istic ions and the highest dwell time for a given pro­ investigation of the composition of complex matrices gram. Most of the compounds were thus identified by such as plant extracts. Coumarins are a very important using only the highest characteristic ion. The conse­ group of active substances, and better knowledge of quent loss of detection specificity was balanced by an their presence in biological samples is always desir­ additional cleanup step as described in the Experimen­ able. tal section. Quantitation was obtained by direct com­ parison between analyte peak areas and a conveniently diluted standard solution of each coumarin. We as­ References sumed, for each coumarin, a linear response slightly above the detection limit as shown by the results 1. Ceska, 0.; Chaudhary, S. K.; Warrington, P. J.; Ashwood­ obtained from the imperatorin calibration. The results Smith, M. J. Plrl/loclremisln/ 1987, 26, 165-169. are summarized in Table 3. The confidence level of the 2. Rodighiero, G.; Dall'Acqua, F.; Pathak, M. A. In Topics in presented results was extrapolated from the mean Photonicdicinc; Smith, K. C, Ed.; Plenum: New York, 1984; p standard deviation calculated for the calibration curve 319. 3. Murray, R. D. H.; Mendez, J.; Brown, S. A. In Tire Natural COl/1I1arins; Wiley: New York, 1982; pp 45-53. Table 3. Amounts and sample concentrations relative 4. Harrison, A. G. Chcniical lonisalion Mass SpcClnJlllcfrl/; CRC to detected coumarins in Smyrnilllll pcrfoliatum L. Press, Boca Raton, FL, 1983. plant extract 5. Plattner, R. D.; Spencer, G. F. Org. Mass Spcclrom. 1988, 23, Quantity Sample 624. Compound (ng) cone. (J.Lg!g) 6. Voight, D. '. Praki, Cirelli. 1977, 319,767. 4-Methylumbelliferon 0.7 7.3 7. Evans, C; Traldi, P.; Chi lin, A.; Pastorini, G.; Rodighiero, P. Org. Mass Speclrol1l. 1991, 26,688-694. Xanthotoxol 3.8 40.3 8. Bravo, P.; Ticozzi, C; Daolio, S.; Traldi, P. Org. Mass Spec­ Scoparone 1.0 10.6 troni. 1985, 20,740-747. Xanthotoxin 0.6 6.3 9. Pelli, B.; Traldi, P.; Rodighiero, P.; Guiotto, A. Biomcd. Entii­ Fraxidin 1.0 10.6 nm. Mass Spectrum. 1986,13,417-422. J Am Soc Mass Spectrom 1995,6, 132-139 MICRO-HPLC/MS WITH A PARTICLE BEAM INTERFACE 139

10, Horning, S. R.; Bier, M. E.; Cooks, R. G.; Brusini, G.; Traldi, 22. Jones, T. L.; Betowski, L. D.; Lesnik, B.; Chiang, T. C; Teberg, P.; Guiotto, A.; Rodighiero, P. Biomed. Ellviroll. Mass Spec­ J. E. Ellviroll. Sci. Tec/mo/. 1991, 25, 1880-1884. trent. 1989, 18, 927-934. 23. Jones, G. G.; Pauls, R. E.; Willoughby, R. C Anal. Gem. 1991, 11. De Vries, J. X.; Taucher, B.; Wurzel, G. Biomed, Ellviroll. Mass 63, 460-463. Spectrom.1988, 15,413-417. 24. Julien-Larose, C; Voirin, P.; Mas-Chamberlin, C; Dufour, A. 12. Yamamoto, K.; Kato, S.; Shimornura, H. I. Chromatogr. 1986, f. Gromatogr. 1991, 562, 39-45. 362, 274-277. 25. Hsu, J. Allal. CllCm. 1992, 64, 434-443. 13. Marles, R. J.; Compadre, C M.; Farnsworth, N. R. Ecoll. Bot. 1987, 41, 41-47. 26. Bagheri, H.; Slobodnik, J.; Marce Recasens, R. M.; Ghijsen, R. T.; Brinkman, U. A. Th. Chromatographia 1993, 37, 159-167. 14. Weinberg, D. S.; Manier, M. L.; Richardson, M. D.; Haibach, F. G. f. Agric. Food Choll. 1993, 41, 48-51. 27. Voyksner, R. D.; Straub, R.; Keever, J. Ellviroll. Sci. Technol. 15. Glowniak, K.; Matysik, G.; Bieganowska, M.; Soczewinski, E. 1993, 27, 1665-1672. Chromntographia 1986, 22,307-310. 28. Behymer, T. D.; Bellar, T. A.; Budde, W. L. Anal. Chem. 1990, 16. Vande Casteele, K.; Geiger, H.; Van Surnere, C F. I. Chro­ 62, 1686-1690. matogr. 1983,258,111-124. 29. Cappiello, A.; Bruner, F. Anal. Chem. 1993, 65, 1281-1287. 17. Zogg, G. C; Nyiredy, Sz.; Sticher, O. Cltromaiographia 1989, 30. Cappiello, A.; Famiglini, G.; Bruner, F. Alia/. Chem., 1994, 66, 27, 591-595. 1416-1423. 18. Alcock, N. J.; Corbelli, L.; Games, D. E.; Lant, M. S.; West­ 31. Cappiello, A.; Famiglini, G. Allal. Chem., 1994,66,3970-3976. wood, S. A. Biomed. Mass Spectrom. 1982, 9,499-504. 32. Cappiello, A.; Palma, P.; Mangani, F. Chromatographia 1991, 19. Yinon, J.; Jones, T. L.; Betowski, L. D. Chromatogr. 1989, I. 32, 389-391. 482,75-85. 33. Palma, P.; Cappiello, A. AIIII. Chimica 1992, 82,371-377. 20. Girault, J.; Istin, B.; Malgouyat, J. M.; Brisson, A. M.; Four­ tillan, J. B. I. Chromaiogr. 1991, 564, 43-53. 34. Berloni, A.; Cappiello, A.; Famiglini, G.; Palma, P. Chro­ 21. In Suk Kim; Sasinos, F. I.; Stephens, R. D.; Brown, M. A. I. maiographia, submitted. Agric. Food Chem. 1990, 38, 1223-1226. 35. Bristow, P. A.; Knox, J. H. Chromaiographia 1977, 10,279-289.