Application of polymer based stationary phases in high performance liquid chromatography and capillary high performance liquid chromatography hyphenated to microcoil 1H nuclear magnetic resonance spectroscopy

Journal of Chromatography A, 1156 (2007) 80–86

Application of polymer based stationary phases in high performance liquid chromatography and capillary high performance liquid chromatography hyphenated to microcoil 1H nuclear magnetic resonance spectroscopy Marc David Grynbaum, Christoph Meyer, Karsten Putzbach, Jens Rehbein, Klaus Albert ∗ Institute of Organic Chemistry, University of Tuebingen, Auf der Morgenstelle 18, 72076 Tuebingen, Germany Available online 28 November 2006

Abstract The increased demand for chromatographic materials that are able to achieve a fast separation of large quantities of structure analogues is a great challenge. It is known that polymer based chromatographic materials have a higher loadability, compared to silica based sorbents. Unfortunately these polymer materials cannot be used under high pressure which is necessary in order to obtain high flow rates, and hence long times are needed to perform a separation. However, by immobilizing a polymer on a mechanically stable porous silica core, this problem can be circumvented and higher flows become feasible on these materials. Especially for capillary liquid chromatography hyphenated with nuclear magnetic resonance a high loadability is of great importance in order to obtain sharp, resolved, and concentrated peaks thus resulting in a good signal to noise ratio in the NMR experiment. Therefore, a highly shape selective chromatographic sorbent was developed by covalently immobilizing a poly(ethylene-co-acrylic) acid copolymer (–CH2 CH2 –)x [CH2 CH(CO2 H)–]y (x = 119, y = 2.4) with a mass fraction of acrylic acid of 5% as stationary phase on silica via a spacer molecule (3-glycidoxypropyltrimethoxysilane). First, the loadability of this sorbent compared to C30 is demonstrated by the HPLC separation of two xanthophyll isomers. Subsequently, it has been successfully employed in the hyphenation of capillary HPLC with microcoil 1 H NMR spectroscopy by separating and identifying a highly concentrated solution of the tocopherol homologues. © 2006 Elsevier B.V. All rights reserved. Keywords: Poly(ethylene-co-acrylic) acid stationary phase; Capillary HPLC-NMR; Hyphenated techniques; Vitamin E

1. Introduction The increased demand for chromatographic materials that are able to achieve fast and well-resolved separations of large quantities of structure analogues is a challenge. The chromatographic triangle illustrates the most important criteria in chromatography, namely resolution, speed, and loadability. Optimizing any of these parameters comes at the expense of the others. A fast separation for example may result in bad resolution and loadability and vice versa. Variable parameters are column dimensions

∗

Corresponding author. Tel.: +49 7071 2978727; fax: +49 7071 295875. E-mail address: [email protected] (K. Albert).

0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2006.11.032

(length, inner diameter), mobile phase (type, flow rate), column temperature (starting temperature, temperature program) and stationary phase (type, particle and pore size, phase volume ratio). Especially for capillary HPLC-NMR a high loadability is of great importance in order to obtain sharp, well-resolved, and highly concentrated peaks. Since NMR sensitivity is still an issue [1,2] this will lead to a good signal to noise ratio in the subsequent NMR experiment. Approaches to improve the signal-to-noise ratio in NMR experiments and therefore the sensitivity have included on one hand the development of cryogenic flow-probes [3]. Another approach is the employment of transversally aligned solenoidal radio frequency coils. In contrast to conventional double-saddle-Helmholtz NMR probes, a

M.D. Grynbaum et al. / J. Chromatogr. A 1156 (2007) 80–86

perpendicular arrangement to the outside magnetic field B0 is realized. The sending and receiving coil is wrapped directly around the capillary [4]. To prevent susceptibility broadening by the copper coil, the whole arrangement is placed in a fluorocarbon (FC-43; perfluorotributylamine). This set-up generates a three-fold increase in 1 H sensitivity compared to conventional NMR probes [5–7]. The residence time of the analyte in the detection cell is limited when working in the continuous flow mode. The active detection volume of this cell is extremely small, only 1.5 ␮L compared to 500 ␮L in conventional 5 mm NMR tubes. The microcoils are well size-matched to the volume of the eluting chromatographic peaks from capillary HPLC [8]. Therefore, measurements of mass-limited samples in the low nanogram range are possible. However, in order to obtain a good signal-to-noise ratio in the NMR experiments, large concentrations are needed. For the separation and analysis of mass-limited nanolitervolume samples, miniaturized systems like capillary HPLC have the advantage of higher efficiencies with sharper peaks, resulting in a higher concentration at the eluting peak maxima. The low solvent consumption (3–5 ␮L min−1 ) makes the use of fully deuterated solvents economically feasible and thus no or only weak suppression of the residual solvent signals is needed in the NMR experiments. The capillary HPLC-NMR system was applied, for example, for the determination of regulatory phosphorylation sites in nanogram amounts of a synthetic fragment of ZAP-70 [9] or the determination of various carotenoids in a small-sized spinach sample [10]. The shape selectivity of C30 sorbents is accounted for the successful separation of the tocopherol isomers, which is ␤and ␥-tocopherol. The isomers cannot be separated on C18 sorbents [11–13]. Recently, the C30 phase which is well adapted for the separation of isomers (shape selectivity) [14,15] was also employed by Krucker et al. in capillary HPLC-NMR hyphenation experiments [1]. The separation and on-line 1 H NMR spectroscopic identification of tocopherol homologues was achieved. Thereby, the C30 sorbents prove to have the high loading capacity described before. Polymer based chromatographic materials are known to have a higher loadability, compared to silica based sorbent materials [16]. Therefore, a highly shape selective stationary phase based on a poly(ethylene-co-acrylic) acid copolymer (–CH2 CH2 –)x [CH2 CH(CO2 H)–]y (x = 119, y = 2.4) with a mass fraction of acrylic acid of 5% was developed. This polymer was immobilized on a porous silica core resulting in a highly loadable and pressure stable chromatographic sorbent [4]. Here, we show the suitability of such a polymer based sorbent for application in LC-NMR experiments. This is exemplified by the separation and on-line NMR-detection of the tocopherol homologues. Tocopherol homologues are known to have antioxidative effects [17]. ␣-Tocopherol has been shown to have anti-inflammatory effects both in vitro and in vivo [18]. However, it can be sensitive to air in the presence of light [19]. The usage of a closed system set-up such as capillary HPLC-NMR hyphenation prevents the degradation of tocopherol homologues during their analysis.

81

2. Experimental 2.1. Chemicals 3-Glycidoxypropyltrimethoxysilane (98%) and poly(ethylene-co-acrylic) acid copolymers with mass fraction of acrylic acid of 5% were obtained from Aldrich (Milwaukee, WI, USA). ProntoSIL-300-3-SI silica (3 ␮m particle size, 30 nm pore size) with a surface area of 150 m2 g−1 was obtained from Bischoff Chromatography (Leonberg, Germany). 1-Hydroxybenzotriazole (99%) was obtained from Aldrich. N,N -Diisopropylcarbodiimide (99%) was obtained from Sigma–Aldrich (Steinheim, Germany). Acetonitrile and methanol (LiChrosolv, gradient grade), [2 H4 ] methanol (methanol-d4 (Uvasol, 99.8%)), and D2 O (Uvasol, 99.8%) were purchased from Merck (Darmstadt, Germany). Water was purified using a Milli-Q water purification system (Millipore, Billerica, MA, USA). Lutein and zeaxanthin were a kind gift from BASF Aktiengesellschaft (Ludwigshafen, Germany). The xanthophyll isomers lutein and zeaxanthin were dissolved in acetonitrile to yield a concentration of 100 ␮g mL−1 of each analyte. The tocopherol homologues were obtained from Calbiochem (San Diego, CA, USA), ␣-tocopherol acetate from Merck. The tocopherol standards were prepared by dissolving the tocopherols in methanol-d4 and D2 O until the desired concentrations were reached, 1.67 and 5.66 mg mL−1 (each tocopherol), respectively. 2.2. Synthesis of the poly(ethylene-co-acrylic) acid stationary phase The synthesis was performed according to Meyer et al. [14]. In brief, the silica was dried under vacuum at 453 K for 4 h and stored under a nitrogen atmosphere. The silica was suspended in dry toluene. A three-fold excess of 3-glycidoxypropyltrimethoxysilane was then added and the mixture was heated under reflux under a nitrogen atmosphere for 12 h. The hot slurry was then filtered and washed with aliquots of toluene, acetone, and hexane. The 3-glycidoxypropylsilica was dried at room temperature for 24 h. Poly(ethylene-co-acrylic) acid with a 5% mass fraction of acrylic acid was dissolved in dry xylene. After heating under reflux for 30 min under a nitrogen atmosphere the 3-glycidoxypropylsilica was added and heating under reflux was continued for 24 h. The slurry was then filtered hot and washed with aliquots of hot xylene, acetone, methanol, methanol/water (50/50), methanol, acetone, and pentane. The resulting white colored chromatographic sorbent (Fig. 1) was dried for 24 h. 2.3. Chromatography The poly(ethylene-co-acrylic) acid sorbent was slurrypacked into a 125 mm × 4.6 mm stainless steel column (Bischoff Chromatography) following standard procedures, the C30 ˚ was column (ProntoSIL C30 , 125 mm × 4.6 mm, 3 ␮m, 200 A obtained from Bischoff Chromatography. The capillaries were packed according to following procedure. Twenty milligrams of the stationary phase (Bischoff

82

M.D. Grynbaum et al. / J. Chromatogr. A 1156 (2007) 80–86

Fig. 1. Polymer based chromatographic sorbent (poly(ethylene-co-acrylic) acid immobilized on silica).

˚ 3 ␮m, C30 or poly(ethylene-co-acrylic) acid ProntoSIL 200 A, stationary phase) were suspended in 300 ␮L of carbon tetrachloride and put in an ultrasonic bath for 10 min. Afterward, the slurry was transferred into a slurry chamber (empty 125 mm × 1 mm cm HPLC column) and forced downward into a 15 cm × 250 ␮m fused-silica capillary with a pneumatic HPLC pump (Knauer GmbH, Berlin-Zehlendorf, Germany). Initially a pressure of 400 bar was used and increased to 650 bar within 5 min. This final pressure was maintained for 30 min. The capillary end fittings consisted of zero dead volume unions ZU1C, steel screens 2SR1, and ferrules FS1.4-5 (Vici AG Valco Int., Schenkon, Switzerland). The carotenoid separations were carried out on an HP1100 system (Agilent Technologies, Waldbronn, Germany) using a UV diode array detection (DAD) system monitoring at 455 nm. Acetonitrile was used as mobile phase, at a flow-rate of 1 mL min−1 . The tocopherol capillary separations were performed on a ternary modular capillary HPLC pump (Waters, Milford, MA, USA). UV detection was performed on-column (150 ␮m i.d.) at 285 nm on a Bischoff Lambda 1010 UV detector (Bischoff Chromatography). Samples were introduced via a microinjection valve kit (Upchurch Scientific, Oak Harbor, WA, USA) with a 500 nL fused-silica injection loop for the capillary HPLC-NMR measurements.

transformation, a squared sine bell function was applied to the FID. 3. Results and discussion First, the suitability of this polymer based sorbent in regard to better loadability was demonstrated by comparing the separation of xanthophyll isomers to those obtained with a C30 sorbent. The xanthophyll isomers lutein and zeaxanthin were dissolved in acetonitrile to yield a concentration of 100 ␮g mL−1 of each analyte. Acetonitrile was employed as mobile phase to elute these analyte molecules from the polymer based chromatographic sorbent and the C30 sorbent. Depicted in Fig. 2 are the chromatograms for each sorbent material.

2.4. Capillary HPLC-NMR For coupling the capillary HPLC system to the NMR instrument (AMX 600, Bruker BioSpin GmbH, Rheinstetten, Germany), the outlet of the UV detector was connected with the 1.5 ␮L active volume 1 H selective capillary NMR probe (Protasis Corp., Marlboro, MA, USA) inlet using a 3 m fused-silica transfer capillary (360 ␮m o.d., 50 ␮m i.d.) to prevent possible interference of the NMR magnetic stray field with the capillary HPLC system. As mobile phase for the continuous-flow experiment, fully deuterated solvents and an isocratic solvent mixture of methanol-d4 :deuterium oxide 90:10 (v/v) was applied at a flow rate of 5 ␮L. The continuous-flow experiment was recorded with the pulse program lc2pnps. In this way, 24 transients with 4 K complex data points and a spectral width of 9090.91 Hz were accumulated with a relaxation delay of 2 s. The pulse angle was set to 30◦ . During the separation, 39 rows with an acquisition time of 58 s per row were recorded. Prior to Fourier

Fig. 2. Chromatograms of a highly concentrated solution of all-trans lutein (Lall-trans ) and all-trans zeaxanthin (Zall-trans ) eluting from a C30 sorbent (i) and the polymer based chromatographic sorbent (ii) (both columns: 4.6 mm × 125 mm), UV detection at 455 nm, mobile phase: acetonitrile, [cL ] = [cZ ] = 100 ␮g mL−1 , 100 ␮L injected.

M.D. Grynbaum et al. / J. Chromatogr. A 1156 (2007) 80–86

Acetonitrile is an easy to evaporate mobile phase. This is important in regard to possible applications of this polymer based chromatographic sorbent in preparative chromatography, which requires highly loadable sorbent materials to achieve fast separations of large sample amounts. Acetonitrile can be removed rapidly from the isolated target molecules. Thus, we employed this organic modifier without the addition of water and therefore the retention times are not matched. On the C30 sorbent no baseline separation could be achieved using these conditions; moreover the column lost resolution due to column overloading. In contrast, the polymer based sorbent resulted in baseline separated, well-shaped peaks, with three times better resolution, see also Table 1. Baseline separation of carotenoids (including lutein and zeaxanthin) on a C30 sorbent is generally achieved by employing a gradient of methanol, tert-butylmethyl ether and water [20,21]; however, the loadability is still inferior to the poly(ethylene-coacrylic) acid stationary phase. In addition these solvent mixtures

83

Table 1 Resolutions and plate heights of the HPLC separation of xanthophyll isomers eluting from the polymer based chromatographic sorbent (P) and a C30 sorbent; [c] = 100 mg mL−1 , injected 100 ␮L; (*) not baseline separated H (×10−5 m)

R P Lutein Zeaxanthin

2.95

C30 1.03

P

C30

5.80 3.19

* *

and gradients are not favorable for HPLC-NMR experiments (many NMR signals, altering chemical shifts of the analytes with the different solvent compositions). With respect to these findings and owing to the system requirements of the hyphenation of capillary HPLC to NMR, described above, the application of the polymer based sorbent in capillary HPLC-NMR becomes apparent. Fig. 3 shows the experimental set-up.

Fig. 3. Capillary HPLC-NMR system.

Fig. 4. Chemical structure of the tocopherol homologues and ␣-tocopherol acetate; (*) denotes chiral C-atoms.

84

M.D. Grynbaum et al. / J. Chromatogr. A 1156 (2007) 80–86 Table 2 Resolutions and plate heights of the capillary HPLC separation of the tocopherol homologues eluting from polymer based chromatographic sorbent (P): (i) [c1 ] = 1.6 mg mL−1 200 nL injected, (ii) [c2 ] = 5.66 mg mL−1 500 nL injected

␦-Tocopherol ␥-Tocopherol ␤-Tocopherol ␣-Tocopherol ␣-Tocopherol acetate

R (i)

R (ii)

1.55 – 1.43 2.30

1.46 – 1.42 2.26

H (i) (×10−5 m)

H (ii) (×10−5 m)

14.99

23.79

* *

10.48 8.88

* *

11.33 10.15

Table 2 shows the plate heights, proving the loadability of this sorbent. The sorbent was not able to separate ␤-tocopherol and ␥tocopherol, therefore the UV-detection could not distinguish them. Mass spectrometry can distinguish ␣-tocopherol, ␦tocopherol, and ␣-tocopherol acetate from each other and also from ␤-tocopherol and ␥-tocopherol; however the isomers ␤and ␥- have the same fragmentation pattern. The only method to differentiate between ␥-tocopherol and ␤-tocopherol is NMR spectroscopy. Despite the fact that these analytes co-eluted, the front of the peak contained more of ␥-tocopherol, whereas the end of the peak contained more of ␤-tocopherol (the elution order of these tocopherol homologues on RP stationary phases has been determined previously [1]). Thus, by creating a pseudo 2D NMR contour plot of the continuous-flow NMR spectra of the tocopherol homologues, it is possible to distinguish between ␥-tocopherol and ␤-tocopherol by their different chemical shifts. The substitution of the aromatic ring of ␥-tocopherol and ␤tocopherol differs only in the position of the methyl group either in position 7 (␥-tocopherol; R2 = CH3 ) or position 5 (␤tocopherol; R1 = CH3 ). The contour plot depicted in Fig. 6 shows

Fig. 5. Chromatograms of the capillary HPLC separation of the tocopherol homologues eluted from the polymer based chromatographic sorbent (15 cm × 250 ␮m), UV detection at 285 nm, mobile phase: methanol/water 90/10 (v/v), flow rate: 5 ␮L min−1 , (i) [c1 ] = 1.6 mg mL−1 200 nL injected, (ii) [c2 ] = 5.66 mg mL−1 500 nL injected. ␦: ␦-Tocopherol, ␥: ␥-tocopherol, ␤: ␤tocopherol, ␣: ␣-tocopherol, ␣-ac: ␣-tocopherol acetate.

Here we applied it for the capillary HPLC separation and microcoil 1 H NMR structure elucidation of the tocopherol homologues. Fig. 4 shows the structures of the tocopherol homologues. It must be noted that each homologue contains three chiral centers resulting in 8 stereoisomers which cannot be separated when eluting them from the polymer based sorbent. The separations of tocopherol homologues, depicted in the chromatograms in Fig. 5, presents the loadability of the polymer based sorbent material by comparing the elution of a low concentrated with a highly concentrated solution of these analytes. The injection volume was increased, too, when the highly concentrated solution was injected. Then the peaks widths only increased minimally, which means that the desired high concentration in the NMR probe can indeed be achieved.

Fig. 6. Pseudo 2D plot of the continuous-flow capillary HPLC-NMR measurement of tocopherol homologues.

M.D. Grynbaum et al. / J. Chromatogr. A 1156 (2007) 80–86

85

Fig. 7. Extracted 1 H NMR spectra of the tocopherol isomers at the corresponding signal maxima.

the 1 H chemical shift axis (F2 dimension) and the retention time (F1 dimension). In the aromatic chemical shift region (light gray box) the signal for the aromatic proton of ␥-tocopherol (R1 ; δ = 6.4 ppm) overlaps with the one of ␤-tocopherol (R2 ; δ = 6.5 ppm), displaying the co-elution of these analytes. However, the aromatic signal corresponding to ␥-tocopherol (R1 ) appears earlier than the one of ␤-tocopherol (R2 ), corresponding to the principle elution order. ␦-tocopherol shows two aromatic proton signals corresponding to protons 5 and 7. Characteristic differences of the methyl group signals bound to the aromatic ring between 2.0 and 2.2 ppm (gray box) allow additional discrimination between the tocopherol homologues. ␣-Tocopherol acetate shows methylene protons of the acetate group at 2.3 ppm (dark gray box). In order to obtain a more detailed interpretation, conventional 1 H NMR spectra of the two isomers were extracted at the NMR signal maxima of the capillary HPLC-NMR separation, see Fig. 7. Even though the signal-to-noise ratio is moderate (due to the fact that only 16 scans account for each spectrum), all resonance signals can be identified. The signal of proton 7 of ␤-tocopherol is superimposed with the shift at 6.4 ppm rising from the previously eluted ␥-tocopherol. The same applies for ␥-tocopherol where the signal of proton 5 is superimposed from the signal arising from the later eluting ␤-tocopherol. Furthermore, the signal of the proton at position 4 at 2.7 ppm experiences an upfield shift in ␤-tocopherol due to the neighboring methyl group attached to the aromatic ring in position 5. The other tocopherols can be unambiguously identified, too (data not shown). Krucker et al. [1] separated the tocopherol homologues on a C30 HPLC column within 25 and 12 min, respectively, by capil-

lary HPLC. However, in the capillary HPLC-NMR experiment they had to lengthen the run to 90 min by using more water in the mobile phase to uncompress the ␤- and ␥-tocopherol peaks under the overloading conditions, which are necessary to obtain good NMR spectra. Employing a highly loadable polymer based chromatographic sorbent we were able to accomplish a separation within 20 min. Though a chromatographic separation of ␤- and ␥-tocopherol could not be achieved, we were able to obtain the same NMR spectra for an unequivocal assignment of each tocopherol within less than a quarter of the time. 4. Conclusion The obtained results demonstrate once more the capability of the hyphenation of capillary HPLC to solenoidal microprobe NMR detection, allowing a full and unambiguous structure elucidation of natural compounds. The employment of highly loadable polymer based sorbents is a powerful tool to overcome limitations regarding NMR sensitivity, allowing the immediate identification even in the continuous-flow mode, as shown in this paper with tocopherol homologues including the structural isomers of ␤- and ␥-tocopherol. Especially the shape selectivity of the employed poly(ethylene-co-acrylic) acid sorbent is very promising towards future HPLC-NMR experiments for the identification of isomers of other small molecules (e.g. flavonoids/isoflavonoids). Acknowledgements The authors would like to thank Bischoff Chromatography (Leonberg, Germany) for donation of the silica and

86

M.D. Grynbaum et al. / J. Chromatogr. A 1156 (2007) 80–86

Dr. Norbert Welsch from Welsch & Partner (T¨ubingen, Germany) for assistance with the preparation of the figures. This work was supported by the Deutsche Forschungsgemeinschaft (Graduiertenkolleg “Chemie in Interphasen” Grant 441), BonnBad-Godesberg, Germany. References [1] M. Krucker, A. Lienau, K. Putzbach, M.D. Grynbaum, P. Schuler, K. Albert, Anal. Chem. 76 (2004) 2623. [2] K. Albert (Ed.), On-line LC NMR and Related Techniques, Wiley, Chichester UK, 2002. [3] M. Spraul, A.S. Freund, R.E. Nast, R.S. Withers, W.E. Mass, O. Corcoran, Anal. Chem. 75 (2003) 1536. [4] N. Wu, T.L. Peck, A.G. Webb, R.L. Magin, J. Sweedler, Anal. Chem. 66 (1994) 3849. [5] D.I. Hoult, R.E. Richards, J. Magn. Res. 24 (1976) 71. [6] B. Behnke, G. Schlotterbeck, U. Tallarek, S. Strohschein, L.H. Tseng, T. Keller, K. Albert, E. Bayer, Anal. Chem. 68 (1996) 1110. [7] D.L. Olson, T.L. Peck, A.G. Webb, R.L. Magin, J. Sweedler, Science 270 (1995) 1967. [8] D.L. Olson, M.E. Lacey, J. Sweedler, Anal. Chem. 70 (1998) 257. [9] P. Hentschel, M. Krucker, M.D. Grynbaum, K. Putzbach, K. Albert, R. Bischoff, Magn. Reson. Chem. 43 (9) (2005) 747.

[10] K. Putzbach, M. Krucker, M.D. Grynbaum, P. Hentschel, A.G. Webb, K. Albert, J. Pharm. Biomed. Anal. 38 (5) (2005) 910. [11] S.L. Abidi, J. Chromatogr. A 881 (2000) 197. [12] S. Strohschein, M. Pursch, D. Lubda, K. Albert, Anal. Chem. 70 (1998) 13. [13] M. Dachtler, T. Glaser, H. H¨andel, T. Lacker, L.H. Tseng, K. Albert, in: I.D. Wilson, E.R. Adlard, M. Cooke, C.F. Poole (Eds.), Encyclopdia of Separation Science, Level II, Academic Press, London, UK, 2000, p. 747. [14] C. Meyer, U. Skogsberg, N. Welsch, K. Albert, Anal. Bioanal. Chem. 382 (2005) 679. [15] C. Meyer, O. Pascui, L. Sander, S. Wise, D. Reichert, K. Albert, J. Sep. Sci. 29 (2005) 820. [16] K. Unger, Packings and Stationary Phases in Chromatographic Techniques, Chromatographic Science Series, vol. 47, Marcel Dekker Inc., New York, 1990. [17] W.A. Pryor, B.A. Bowman, R.M. Russel (Eds.), Present Knowledge in Nutrition, International Life Sciences Institute Press, Washington, DC, USA, 2001. [18] U. Singh, I. Jialal, Ann. N. Y. Acad. Sci. 1031 (2004) 195. [19] D. Kristensen, E. Hansen, A. Arndal, R.A. Trinderup, L.H. Skibsted, Int. Dairy J. 11 (2001) 837. [20] M.D. Grynbaum, P. Hentschel, K. Putzbach, J. Rehbein, M. Krucker, G. Nicholson, K. Albert, J. Sep. Sci. 28 (2005) 1685. [21] K.-J. Yeum, S.L. Booth, J.A. Sadowski, C. Liu, G. Tang, N.I. Krinskv, R.M. Russell, Am. J. Clin. Nutr. 64 (1996) 594.