Determination of Trace Isoflavone Phytoestrogens in Biological

four adult subjects (three females and one male) were recruited into the study following the Informed Consent Procedures approved by the Instituti...
0 downloads 0 Views 126KB Size
Anal. Chem. 2002, 74, 5998-6005

Determination of Trace Isoflavone Phytoestrogens in Biological Materials by Capillary Electrochromatography Jason A. Starkey,† Yehia Mechref,† Chang Kyu Byun,† Rosemary Steinmetz,‡ John S. Fuqua,‡ Ora H. Pescovitz,‡ and Milos V. Novotny*,†

Department of Chemistry, Indiana University, Bloomington, Indiana 47405, and Department of Pediatrics, Indiana University School of Medicine, Indianapolis, Indiana 46202

Capillary electrochromatography using a specialty monolithic matrix was utilized in developing a rapid and highly efficient separation of isoflavones in biological materials. Without a preconcentration technique, it is relatively easy to reach ppm-ppb concentrations of these compounds in soy-based foods and verify them structurally using a photodiode array detector. With on-column preconcentration, we were able to measure low-ppb levels in human serum. Using blood samples from human volunteers, whose diet was supplemented by a soy-based product, the method has been validated for high-throughput screening of isoflavones in clinical studies. Various natural products with biological activity (bioactive nonnutrients) are consumed by humans as a part of their regular diet. Some of these may play a beneficial role in disease prevention,1 while others could have adverse effects in the recipients. Isoflavones are among the widespread bioactive nonnutrients that (depending on age and endocrine status) could have either beneficial effects in hormone-dependent afflictions such as breast and colon cancer, osteoporosis, menopausal symptoms, and coronary heart disease2-7 in adults or potentially far-reaching adverse effects in infants due to their phytoestrogenic activities.8 Among the rich sources of isoflavones are soybean products that are widely consumed by adults and also form the basis of numerous infant formulas.9,10 Since the bioactive isoflavone derivatives present in soy-based formulas feature a wide range of * To whom correspondence should be addressed. Tel: (812) 855-4532. Fax: (812) 855-8300. E-mail: [email protected]. † Department of Chemistry. ‡ Department of Pediatrics. (1) Adlercreutz, H.; Mazur, W. Ann. Med. 1997, 29 (2), 95-120. (2) Adlercreutz, H.; Honjo, H.; Higashi, A.; Fotsis, T.; Hamalainen, E.; Hasegawa, T.; Okada H. Am. J. Clin. Nutr 1991, 54 (6), 1093-1100. (3) Lee, H. P.; Gourley, L.; Duffy, S. W.; Esteve, J.; Lee, J.; Day, N. E. Lancet 1991, 337 (8751), 1197-1200. (4) Watanabe, S.; Koessel, S. J. Epidemiol. 1993, 3, 47-61. (5) Knight, D. C.; Eden, J. A. Obstet. Gynecol. 1996, 87, 897-904. (6) Adlercreutz, H.; Hamalainen, E.; Gorbach, S.; Goldin, B. Lancet 1992, 339, 1233. (7) Clarkson, T. B.; Anthony, M. S.; Hughes, C. L. Trends Endocrinol. Metab. 1995, 6, 11-16. (8) Setchell, K. D. R.; Zimmer-Nechemias, L.; Cai, J.; Heubi, J. E. Lancet 1997, 350, 23-27. (9) Coward, L.; Barnes, N. C.; Setchell, K. D.; Barnes, S. J. Agric. Food Chem. 1993, 41, 1961-1967.

5998 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

hormonal and nonhormonal activities,11 concerns arise about the differences between developing infants and adults in terms of tissue differentiation and gene expression. These concerns are further amplified by the known susceptibility of infants to the effects of hormonally active agents.8 During the past decade, correlations have been sought between the growth and development of infants fed soy protein-based formulas and the levels of phytoestrogens.12,13 Future and more definitive studies invariably necessitate development of sensitive analytical methods for measuring the levels of various isoflavone derivatives in human blood. The sensitivity requirements are given by the fact that the ingested soy products contain ppm levels of isoflavones, resulting in serum levels at only ppb concentrations. The measurement task is further complicated by the structural variation in the dietary isoflavones and metabolic processes that could result in a number of structural relatives and different biological conjugates. Genistein, daidzein, and glycitein (Figure 1) are the principal isoflavones found in soybean products together with various conjugates. Consequently, the most desirable analytical methods will typically combine an efficient separation technique with a sensitive solute measurement principle. Moreover, the speed of analysis becomes essential to large-scale clinical analyses and for meaningful epidemiological studies. We report here a new analytical methodology for isoflavone determinations in biological samples. It features highly efficient solute separation, sensitivity down to low-ppb levels, and rapid analysis (10-15-min runs), providing a good alternative to the previously used methods. The methodology features capillary electrochromatography (with sample preconcentration on a hydrophobic monolithic column) and photodiode array optical detection. This methodology has been validated here for determinations at ppm levels (infant formulas) and ppb levels (serum samples from human volunteers consuming soybean products). The previously reported analytical approaches to isoflavone analysis involve either chromatographic separations or capillary zone electrophoresis (CZE). Gas chromatography/mass spectrometry in the single-ion monitoring mode has been reported,14,15 (10) Franke, A. A.; Custer, L. J.; Cerna, C. M.; Narala, K., Soc. Exp. Biol. Med. 1995, 208, 18-26. (11) North, K.; Golding, J. BJU Int. 2000, 85, 107-113. (12) Irvine, C.; Fitzpatrick, M.; Robertson, I.; Woodhams, D. N. Z. Med. J. 1995, 108, 208-209. (13) Robertson, I. Proc. Nutr. Soc. N. Z. 1995, 20, 35-42. 10.1021/ac025929b CCC: $22.00

© 2002 American Chemical Society Published on Web 11/02/2002

Figure 1. Structures of the core isoflavone skeleton, main naturally occurring isoflavones, and the flavone apigenin, which was used as an internal standard.

where the isoflavones of interest had to be derivatized with a tertbutyldimethylsilyl donor reagent for an enhanced volatility. This derivatization can be tedious and relatively nonreproducible.15 High-performance liquid chromatography (HPLC) using UV absorbance,16-18 electrochemical detection,19 or mass spectrometric measurements20-23 has been extensively explored. However, the reported procedures feature long retention times and, with (14) Heinonen, S.; Wa¨ha¨la¨, K.; Adlercreutz, H. Anal. Biochem. 1999, 274, 211219. (15) Tekel, J.; Daeseleire, E.; Heeremans, A.; van Peteghem, C. J. Agric. Food Chem. 1999, 47, 3489-3494. (16) Casteele, K. V.; Geiger, H.; van Sumere, C. F. J. Chromatogr., A 1982, 240, 81-94. (17) Hutabarat, L. S.; Mulholland, M.; Greenfield, H. J. Chromatogr., A 1998, 795, 377-382. (18) Hutabarat, L. S.; Greenfield, H.; Mullholland, M. J. Chromatogr., A 2000, 886, 55-63. (19) Gamache, P. H.; Acworth, I. N. Proc. Soc. Exp. Biol. Med. 1998, 217, 274280. (20) Griffith, A. P.; Collison, M. W. J. Chromatogr., A 2001, 913, 397-413. (21) Barnes, S.; Kirk, M.; Coward, L. J. Agric. Food Chem. 1994, 42, 24662474.

the exception of mass spectrometry, a marginal sensitivity for dealing with small samples. While the CZE-based procedures provided short migration times (and, consequently, times of analysis), they do not permit adequate resolution of certain sample components, in addition to the well-known problems with sample preconcentration. 24-28 Capillary electrochromatography (CEC) is a relatively uncommon, highly versatile separation technique of high efficiency that has been derived from the flat flow profile of the electroosmotic (22) Andlauer, W.; Martena, M. J.; Furst, P. J. Chromatogr., A 1999, 849, 341348. (23) Satterfield, M.; Black, D. M.; Brodbelt, J. S. J. Chromatogr., B 2001, 759, 33-41. (24) Adlercreutz, H. Proc. Soc. Exp. Biol. Med. 1998, 217, 386-392. (25) Shihabi, Z. K.; Kute, T.; Garcia, L. L.; Hinsdale, M. J. Chromatogr., A 1994, 680, 181-185. (26) Aramendia, M. A.; Garcia, I.; Lafont, F.; Marinas, J. M. J. Chromatogr., A 1995, 707, 327-333. (27) Aussenac, T.; Lacombe, S.; Dayde´, J. Am. J. Clin. Nutr 1998, 68, 1480S1485S. (28) Mellenthin, O.; Galensa, R. J. Agric. Food Chem. 1999, 47, 594-602.

Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

5999

flow (EOF), with its selectivity being dependent on the chemical nature of the packing materials. To maximize the separation potential of CEC, we have recently developed29-33 monolithic macroporous columns (featuring up to 500 000 plates/m efficiencies) and applied them to a wide range of relatively small biological molecules, including steroids, bile acids,34 and saccharides.29-33 Without any unusual procedural modifications, this type of CEC is easily applicable to reach sample concentrations at low-ppm levels with conventional concentration-sensitive detectors and subppm levels with highly sensitive detectors such as laser-induced fluorescence or mass spectrometry. Unlike with CZE, the CECbased procedures permit greater sensitivity enhancement through on-line preconcentration methods, as was recently demonstrated with environmental applications at low-ppb levels.35 Such a preconcentration step permits implementation of the relatively simple and inexpensive means of detection, such as UV absorbance and diode array detectors (which would otherwise be restricted by small injection volumes and short optical path lengths). In this article, the merits of CEC (with and without an on-line sample preconcentration) were assessed using various concentrations of standard isoflavones, a reproducible quantification of genistein, daidzein, glycitein, and their conjugates in soy-based infant formulas at low-ppm levels and, finally, at low-ppb levels in the peripheral circulation system. The overall procedure has been developed for use with the low sample volumes (250 µL of human serum), as would be needed in large-scale studies of isoflavones in human infants. Solute identification at all measured levels was provided through a coincidence of retention times with those of standards and their UV (diode array) spectra. MATERIALS AND METHODS Materials. Capillary columns were purchased from Polymicro Technologies (Phoenix, AZ). Acrylamide and N,N′-methylenebisacrylamide were purchased from Bio-Rad Laboratories (Hercules, CA). HPLC grade acetonitrile, apigenin, human male serum, type H2 glucuronidase, formic acid (96%), ammonium persulfate, N,N,N′,N′-tetramethylenediamine (TEMED), (3-methacryloxypropyl)trimethoxysilane (Bind-Silane), poly(ethylene glycol) (PEG, MW 10 000), vinylsulfonic acid (sodium salt, 25% (v/v)), lauryl acrylate, N-methylformamide (99%), and methyl tert-butyl ether (MTBE) were received from Sigma-Aldrich Co. (Milwaukee, WI). Genistin, genistein, glycitin, glycitein, daidzin, and daidzein were purchased from LC Labs (Woburn, MA). Column Preparation. The columns were prepared according to our previously published procedure.29 Briefly, fused-silica tubing with 100-µm inner diameter and 360-µm outer diameter were used as the separation capillaries. Removing a 1.0-cm section of the polyimide coating using a razor blade formed the basis of an optical detection window. The inner wall of the tubing was then treated with 1.0 M sodium hydroxide for 40 min, flushed with 0.1 (29) Palm, A.; Novotny, M. V. Anal. Chem. 1997, 69, 4499-4507. (30) Que, A. H.; Palm, A.; Baker, A. G.; Novotny, M. V. J. Chromatogr., A 2000, 887, 379-391. (31) Starkey, J. A.; Mechref, Y.; Novotny, M. V., to be submitted. (32) Que, A. H.; Novotny, M. V. Anal. Chem. 2002, 74, 5184-5191. (33) Que, A. H.; Novotny, M. V., Anal. Bioanal. Chem., submitted. (34) Que, A. H.; Konse, T.; Baker, A. G.; Novotny, M. V. Anal. Chem. 2000, 72, 2703-2710. (35) Yang, C.; El Rassi, Z. Electrophoresis 1999, 20, 2337-2342.

6000

Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

M hydrochloric acid for 30 min, and, finally, rinsed with deionized water for 30 min. Thereafter, a 50% (v/v) Bind-Silane solution in acetone was introduced and left inside the capillary for 20 min, followed by another treatment with the Bind-Silane for an additional 20-min period. Finally, the capillary was briefly rinsed with acetone and water, separately. A lauryl acrylate stationary phase of (according to Hjerte´n’s definition36) 5% T, 60% C, 14.5% lauryl acrylate (40.5 mM), 10% vinylsulfonic acid (28 mM), and 3% PEG was prepared by mixing 30.2 mg of acrylamide, 60 mg of bisacrylamide, 24.8 µL of vinylsulfonic acid, 60 mg of poly(ethylene glycol), and 24.5 µL of lauryl acrylate in a solution made of 1.85 mL of N-methylformamide, 50 µL of 0.4 M Tris buffer (pH 8.7), and 50 µL of 0.6 M boric acid. A 0.5-mL aliquot of this solution was degassed for 40 min prior to the addition of 4 µL of 100% TEMED. Next, the solution was heated at 50 °C for 5 min prior to the addition of 10 µL of 40% ammonium persulfate. This solution was then immediately purged through the freshly activated capillary tubing up to the detection window and allowed to polymerize for at least 5 h. Column Conditioning. Packed columns were washed with several void volumes using a mobile phase consisting of 30% acetonitrile prepared with 2.5 mM of sodium phosphate buffer (pH 3.0). This mobile phase was purged through the columns using compressed nitrogen gas at 100-120 psi to remove the excess reagents and small bubbles formed during polymerization. Prior to use, 3-cm sections at the capillary ends were removed, and the capillary was subsequently conditioned for 3-5 h (depending on the capillary length) at increasing field strengths (50-800 V/cm), with currents not exceeding 10 µA using the same mobile phase. Preparation of Isoflavone Standards. Stock solutions of isoflavone standards were prepared by dissolving 1.0 mg of each of the six commercially available standards in 1.0 mL of 50% aqueous acetonitrile prepared in 0.04 M sodium hydroxide. Further dilutions of these stock solutions were made with 30% aqueous acetonitrile prepared with 2.4 mM formate buffer (pH 2.9). The calibration curves generated with the standard isoflavones yielded a linear relationship with correlation coefficient values of 0.98 or better for all five standards. The limit of detection under standard injection conditions was 500 ppb for each isoflavone, with a linear dynamic range extending up to 500 ppm without any significant loss of column efficiency. Instrumentation. An Agilent G1600A capillary electrophoresis system (Waldbronn, Germany) equipped with a photodiode array detector and ChemStation data-handling software was utilized in this study. Electrochromatograms were monitored at a single wavelength or as spectra with a response time of 0.2 s for the UV-visible range of 190-370 nm with a 30-nm bandwidth. A reference was set at 450 nm with an 80-nm bandwidth. All separations were performed at 30 °C with a capillary that was 32.5 cm as its total length and 24.5 cm as its effective length. A pressure of 4.0 mbar was applied to the inlet and outlet vials to prevent air bubble formation. Extraction of Isoflavones from Soy-Based Infant Formulas and Human Breast Milk. The isoflavone composition of a selection of several commercial brands of soy-based infant formulas and a sample of human breast milk (matrix control) were (36) Hjerte´n, S. Arch. Biochem. Biophys. 1962, (Suppl. 1), 147-151.

extracted according to the following procedure. Briefly, 1.0 mL of the liquid formula or breast milk was mixed with 4.0 mL of ice-cold 80% methanol for 2 h. Next, lipids were removed twice by liquid extraction with 4.0 mL of hexane. The aqueous layer was evaporated to dryness in vacuo. The pellet was resuspended in 1.0 mL of 90% cold ethanol and placed in an ethanol/dry ice bath for 5 min, then centrifuged for 15 min at 10000g. The supernatant was retained, again evaporated to dryness in vacuo, and finally resuspended in 100 µL of 2.4 mM ammonium formate buffer (pH 2.7) prior to CEC analysis. Human Isoflavone Dosing Study. To validate phytoestrogen determinations in human serum, four adult subjects (three females and one male) were recruited into the study following the Informed Consent Procedures approved by the Institutional Review Board at Indiana University/Purdue University at Indianapolis. Subjects were asked to continue their regular dietary habits for the study duration. Over the course of 5 days, the subjects consumed increasing amounts of soymilk according to the following schedule: Day 1, no soy-milk was taken; day 2, 3 × 8-oz glasses; day 3, 4 × 8-oz glasses; day 4, 5 × 8-oz glasses; day 5, 6 × 8-oz glasses. Soymilk was consumed between 0800 and 2400 h on all days. Ten-milliliter blood samples were obtained from the subjects at 0800 h on days 1-5. The blood samples were collected in plain vacuum container tubes and allowed to clot. Samples were centrifuged, and the serum aliquots were collected and frozen at -20 °C prior to analysis. Extraction of Isoflavones from Human Blood Serum. For quantitative analyses of the total isoflavones in human serum, removal of sulfate and glucuronide conjugation appeared desirable. As the internal standard, apigenin was added to the samples to be analyzed prior to the sulfatase/glucuronidase addition. A 250µL volume of serum was mixed with 2.0 mL of methyl tert-butyl ether in a conical tube and continuously agitated for 30 min. Next, the sample was centrifuged for 10 min at ∼2000g, while the organic layer containing isoflavones was collected and dried in a vacuum concentrator. The pellet was resuspended in 400 µL of water and applied to a Waters OASIS-AEX solid-phase extraction (SPE) cartridge (Millford, MA) that had been preconditioned with 2 mL of acetonitrile and water. The cartridge was washed with 1.0 mL of 10% methanol, 1.0 mL of 10% methanol in 2.0 M ammonium hydroxide, and again with 1.0 mL of methanol. The sample was eluted with 1.5 mL of 20% methanol, 78% acetonitrile, and 2% trifluoroacetic acid. The eluent was dried in a vacuum concentrator and then resuspended in 50 µL of 2.4 mM ammonium formate buffer (pH 2.7). The calibration curves used for quantitative analyses were constructed using commercial serum spiked with different concentrations of standard isoflavones. RESULTS AND DISCUSSION Initial Evaluation of CEC for Isoflavone Separations. The aglycon and glycoside isoflavones of potential interest in biomedical and clinical studies (Figure 1) have a range of physical and chemical properties that required attention before a fairly comprehensive extraction and CEC methodology was developed. The range of polarities for a hydrophobic-phase CEC requires solubility of these compounds in aqueous/organic media. The glycosidic conjugates are sparingly soluble in 50% aqueous acetonitrile, and aglycons are even less so. The addition of dimethyl sulfoxide

Figure 2. Electrochromatogram of isoflavone standards on a C12 CEC column using acetonitrile (A) or methanol (B). Conditions: voltage, 800 V/cm; column, 32-cm column with a 24.5-cm effective length; mobile phase, 30% acetonitrile (A) and 50% methanol (B) prepared in 2.4 mM ammonium formate buffer (pH 2.7); injection, 5 s at 5 kV; sample, mixture of six isoflavones at 33.3 ppm. Peaks: (1) daidzin, (2) genistin, (3) daidzein, (4) genistein, and (5) apigenin.

(DMSO) enhanced the solubility of aglycons but interfered with a good initial separation and complicated UV spectroscopic detection and spectral identification of isoflavones. An alternative was to prepare isoflavone solutions using an alkaline buffer. Isoflavones are highly soluble in alkaline buffers due to the partial ionization of their hydroxy groups. Moreover, this pH solubilityinducing adjustment appears reversible, with no adverse effect on the spectroscopic properties, as no visible changes were observed in the UV spectrum of daidzin (a potentially labile glucuronide conjugate) when the solution pH was increased from 4.0 to 12.0 and brought back to 4.0 after 4 h (data not shown). There was no apparent loss in spectral information or sign of degradation (absorbance decrease) with all studied isoflavones as a result of this pH manipulation. The CEC separation of four standard isoflavones using hydrophobic poly(acrylamide)/poly(ethylene glycol) columns33 was optimized as a prelude to the analysis of biological samples. The mobile phases containing either the aqueous acetonitrile or methanol were tested for component resolution and overall efficiency (number of theoretical plates). As shown in Figure 2, superior resolution and shorter analysis time were attained using acetonitrile (Figure 2 A vs B). On average, the efficiency values with acetonitrile were ∼3-fold higher than for methanol, which may be partly attributable to the higher viscosity of methanol (less solute diffusivity) or some matrix-swelling phenomenon. Figure 3 represents a plot of the logarithm of capacity factor (k) against percentage of acetonitrile in the mobile phase for all standards separated on a lauryl acrylate column. While the linear relationship is indicative of a reversed-phase behavior, as is the case with the two nonpolar aglycons, a nonlinear relationship for the more polar glucuronide conjugates at a higher organic percentage appears due to a mixed interaction process. When two distinct retention processes, i.e., hydrophilic and hydrophobic interaction, occur, the capacity factor is given as

k ) χiki + χjkj where χi and χj are the hydrophilic and hydrophobic phase ratios, respectively, and ki and kj are the correlated solute distribution values.35 According to the above equation, at a low content of organic mobile phase, the reversed-phase retention mechanism Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

6001

Figure 3. Dependence of capacity factor (k) on the acetonitrile concentration (v/v) in the mobile phase for a macroporous polyacrylamide/poly(ethylene glycol) matrix. Conditions as in Figure 2. Labels: (1) daidzin, (2) genistin, (3) daidzein, and (4) genistein. Table 1. Column Reproducibility Studiesa run-to-run (n ) 6)

day-to-day (n ) 6)

column-to-column (n ) 6)

analyte

tR (min)

RSD (%)

tR (min)

RSD (%)

tR (min)

RSD (%)

daidzin genistin daidzein genistein apigenin

4.5 5.2 7.0 9.7 12.4

0.3 0.2 0.3 0.3 0.3

4.5 5.2 7.0 9.8 12.4

0.3 0.4 0.1 0.3 0.2

4.5 5.2 6.7 9.7 12.4

0.3 0.2 0.3 0.3 0.3

a

tr, retention time; RSD, relative standard deviation

predominates when log k decreases linearly with an increasing percentage of acetonitrile. Conversely, a hydrophilic interactionbased mechanism at the higher organic content becomes more significant for the more polar solutes, originating presumably from the interactions with the amide backbone of the polymer matrix. This feature does not pose any problem for useful applications of the lauryl acrylate CEC columns, as the optimized separations occur typically below 40% organic mobile-phase content. The columns utilized in our investigations were highly efficient and reproducible between runs, days, and the individually prepared columns. On average, separation efficiencies of ∼250 000 plates/m were achieved with an average RSD of 2.4%. Table 1 summarizes the reproducibility data obtained with the standard compounds, detailing the error values: RSD for the retention times of five isoflavones, compiled from six consecutive runs (