Anal. Chem. 1998, 70, 13-18
Shape Selectivity of C30 Phases for RP-HPLC Separation of Tocopherol Isomers and Correlation with MAS NMR Data from Suspended Stationary Phases Sabine Strohschein,† Matthias Pursch,† Dieter Lubda,‡ and Klaus Albert*,†
Institut fu¨ r Organische Chemie, Universita¨ t Tu¨ bingen, Auf der Morgenstelle 18, D-72076 Tu¨ bingen, Germany, and Merck KGaA, LPRO Chrom1/Synthese, D-64271 Darmstadt, Germany
Vitamin E (tocopherol) acts in various organisms as the main free radical scavenger. This capacity, which is enhanced by the synergetic effect of vitamin C and carotenes, points to a possible application as anti-tumor agent in chemotherapy. There are several isomeric forms, namely r-, β-, γ-, and δ-tocopherol, having different antioxidative abilities, with r-tocopherol being the most biologically active. Using methanol as eluent and a C30 stationary phase, we achieved complete separation of r-, β-, γ-, and δ-tocopherol and r-tocopherol acetate by RPHPLC within 14 min. Detection was performed by UV and 1H NMR spectroscopy. The advantage of NMR is the possibility of structural identification of chromatographic peaks. Also, coeluting peaks are easily recognized. The enhanced shape recognition of the C30 phase has been attributed to the high order of the alkyl chains of the stationary phase. This has, so far, been proven by solidstate NMR spectroscopy. We now introduce the technique of 13C MAS NMR spectroscopy of suspended stationary phases. The resulting NMR spectra reveal that, in the presence of weak eluents, like methanol, the overall high order of the C30 chains is slightly altered, whereas in stronger eluents, like MTBE, the alkyl chains possess a higher mobility. Vitamin E is an important factor in human health due to its antioxidative capacity and ability to act as free radical scavenger. Especially in combination with vitamin A, carotenoides, and vitamin C, its application as an antitumor agent is attracting increased interest. It has been shown that the administration of these vitamins can significantly reduce the cancer risk and delay the progression of precancer lesions to cancer cells.1,2 All naturally occurring forms of the above-mentioned vitamins exist in different isomers. Of all these structures, one of each isomeric form has the highest biological activity, while the others are less active. Separation of vitamin A, carotenoid, and vitamin †
Universita¨t Tu ¨ bingen. Merck KGaA. (1) Lupulescu, A. Int. J. Vit. Nutr. Res. 1993, 63, 3-14. (2) Gerster, H. Int. J. Vit. Nutr. Res. 1993, 63, 93-121. ‡
S0003-2700(97)00414-9 CCC: $14.00 Published on Web 01/01/1998
© 1997 American Chemical Society
E isomers is, therefore, of great interest.3 The separation of R-, β-, γ-, and δ-tocopherol has been especially challenging so far. With reversed-phase (RP) materials, only the determination of R-, (β- and/or γ-), and δ-tocopherol was possible,4-6 whereas the separation of all four isomers could be achieved only using normalphase (NP) chromatography.7 Here we report the first separation of R-, β-, γ-, and δ-tocopherol together with R-tocopherol acetate using a C30 stationary reversed phase. This phase was specially developed for the separation of carotenoid isomers,8-11 but proves also to be extremely suitable for other separation problems. Compared to conventional UV detection, 1H NMR spectroscopy is accompanied by lower sensitivity but often enables unequivocal structural identification. The assignment of a chromatographic peak via only retention time may be ambiguous. The peak of interest could also contain coeluting substances, making an exact quantification impossible. With 1H NMR spectroscopy, the existence of coeluting substances can often be detected.12 Thus, LCNMR proves to be a valuable tool for the qualitative and quantitative analysis of sample mixtures.13,14 Solid-state NMR spectroscopy has shown to be a very effective technique for structural characterization of bonded phases. 13C CP/MAS NMR spectroscopy15,16 provides detailed information (3) Lee, B. L.; Chua, S. C.; Ong, H. Y.; Ong, C. N. J. Chromatogr. B 1992, 581, 41-47. (4) Sarzanini, C.; Mentasti, E.; Vincenti, M. J. Chromatogr. B 1993, 620, 268272. (5) Koprivinjak, J.-F.; Lum, K. R.; Sisak, M. M.; Saborowski, R. Comp. Biochem. Physiol. 1996, 113B, 143-148. (6) Go¨bel, Y.; Schaffer, C.; Koletzko, B. J. Chromatogr. B 1997, 688, 57-62. (7) Manzi, P.; Panfili, G.; Pizzoferrato, L. Chromatographia 1996, 43, 89-93. (8) Sander, L. C.; Sharpless, K. E.; Craft, N. E.; Wise, S. A. Anal. Chem. 1994, 66, 1667-1674. (9) Emenhiser, C.; Sander, L. C.; Schwartz, S. J. J. Chromatogr. A 1995, 707, 205-216. (10) Emenhiser, C.; Simunovic, N.; Sander, L. C.; Schwartz, S. J. J. Agric. Food Chem. 1996, 44, 3887-3893. (11) Strohschein, S.; Pursch, M.; Ha¨ndel, H.; Albert, K. Fresenius J. Anal. Chem. 1997, 357, 498-502. (12) Strohschein, S.; Schlotterbeck, G.; Richter, J.; Pursch, M.; Tseng, L.-H.; Ha¨ndel, H.; Albert, K. J. Chromatogr. A 1997, 765, 207-214. (13) Albert, K. J. Chromatogr. A 1995, 703, 123-147. (14) Lindon, J. C.; Nicholson, J. K.; Wilson, I. D. Prog. Nucl. Magn. Reson. Spectrosc. 1996, 29, 1-49. (15) Schaefer, J.; Steiskal, E. O. J. Am. Chem. Soc. 1976, 98, 1031-1032. (16) Sindorf, D. W.; Maciel, G. E. J. Am. Chem. Soc. 1983, 105, 1848-1851.
Analytical Chemistry, Vol. 70, No. 1, January 1, 1998 13
regarding the ligand structure and shows the presence of different alkyl chain conformations.17 A correlation of solid-state NMR data with LC results has led to an improved understanding of the retention mechanism.17,18 However, as implied by the name, solidstate NMR measurements are carried out using dry bonded phases, whereas HPLC separations are performed in contact with a solvent. NMR investigations of bonded phases in suspension have been the subject of many papers.19-24 But the resulting 13C NMR spectra acquired with conventional NMR equipment exhibited quite broad lines, so most of the research has dealt with the evaluation of T1 data. The broadened lines were due to susceptibility inhomogeneities and dipolar couplings present within the sample. Since it was found that magic angle spinning (MAS)25 can reduce 1H and 13C NMR line widths of solvent-swollen polymers,26-28 we have used this approach for acquisition of highly resolved 13C MAS NMR spectra of a C30 phase in suspension. EXPERIMENTAL SECTION Materials. R-, β-, γ-, and δ-tocopherol as well as methanol, chloroform, methyl tert-butyl ether (MTBE) (all LiChrosolv gradient grade), and methanol-d4 (99.8%) were purchased from Merck KGaA (Darmstadt, Germany). R-Tocopherol acetate was a gift from the Pharmaceutical Institute (University of Tu¨bingen). Column. A polymeric C30 bonded phase was prepared by chemical modification of LiChrospher WP 200 (3 µm particle size and 200 Å average pore diameter, research sample, Merck KGaA) with triacontyltrichlorosilane (ABCR, Karlsruhe, Germany) in a fashion similar to that described previously.8,17 The bonded phase was slurry packed with methanol into a 250 mm × 4.6 mm HPLC column (Bischoff, Leonberg, Germany). HPLC. All chromatographic separations were carried out under ambient conditions using a Merck LiChroGraph L-6200A gradient pump. Detection was carried out with a Merck LiChroGraph L-4200A UV-visible detector at 295 nm. The mobile phase was 100% methanol at a flow rate of 1 mL/min. For the LC-NMR experiments, 2% methanol-d4 was added, and the flow rate was reduced to 0.3 mL/min. Ten milligrams each of R-, β-, γ-, δ-tocopherol, and R-tocopherol acetate were dissolved in 500 µL of chloroform for the LC-NMR experiments. For the analytical separation, this solution was diluted by a factor of 200. Solid-State NMR Spectroscopy. 13C MAS NMR measurements were carried out by use of a Bruker ASX 300 spectrometer (Bruker, Rheinstetten, Germany). 13C pulses amounted to 4.5 µs, (17) Pursch, M.; Strohschein, S.; Ha¨ndel, H.; Albert, K. Anal. Chem. 1996, 68, 386-393. (18) Pursch, M.; Sander, L. C.; Albert, K. Anal. Chem. 1996, 68, 4107-4113. (19) Gilpin, R. K.; Gangoda, M. E. Anal. Chem. 1984, 56, 1470-1473. (20) Gilpin, R. K.; Gangoda, M. E. J. Magn. Reson. 1985, 64, 408-413. (21) Albert, K.; Evers, B.; Bayer, E. J. Magn. Reson. 1985, 62, 428-436. (22) Bayer, E.; Paulus, A.; Peters, B.; Laupp, G.; Reiners, J.; Albert, K. J. Chromatogr. 1986, 364, 25-37. (23) Shah, P.; Rogers, L. B.; Fetzer, J. C. J. Chromatogr. 1987, 388, 411-419. (24) Zeigler, R. C.; Maciel, G. E. In Chemically Modified Surfaces; Leyden, D. E., Collins, W. T., Eds.; Gordon and Breach Science Publishers: New York, 1988; Vol. 2, pp 319-336. (25) Andrew, E. R.; Bradbury, A.; Eades, R. G. Nature 1959, 183, 1802. (26) Sto ¨ver, H. D. H.; Frechet, J. M. J. Macromolecules 1991, 24, 883-891. (27) Fitch, W. L.; Detre, G.; Holmes, G. P.; Shoolery, J. N.; Keifer, P. A. J. Org. Chem. 1994, 59, 7955-7956. (28) Pursch, M.; Schlotterbeck, G.; Tseng, L.-H.; Albert, K.; Rapp, W. Angew. Chem., Int. Ed. Engl. 1996, 35, 2867-2869.
14 Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
Figure 1. Separation of tocopherol isomers on a C30 column, eluent 100% methanol, detection at 295 nm. Peak identification is as follows: 1, δ-tocopherol; 2, γ-tocopherol; 3, β-tocopherol; 4, R-tocopherol; 5, R-tocopherol acetate.
and delay times were 3 s. The samples were packed into 7 mm ZrO2 rotors. Magic angle spinning was performed at 3000 Hz, and the spectra were acquired under high-power decoupling of the proton resonances. The suspensions of the C30 phase in various eluents were prepared by stepwise filling the rotor with solvent and bonded phase, generating a slurry. The suspension was found to be still homogeneous after application of the MAS technique. High-Resolution NMR Spectroscopy. Experiments were recorded with a Bruker AMX 600 spectrometer (Bruker, Rheinstetten, Germany). The chromatographic equipment was controlled by Chromstar software (Bruker-Franzen, Bremen, Germany). For the data acquisition, a selective LC-NMR probe with a 120 µL detection cell was used. The temperature within the detection cell was set to 300 K. Solvent signal suppression was achieved using a NOESY-type presaturation technique. Sixteen transients with a time domain of 8K per FID were coadded using a pulse length of 8.1 µs and a sweep width of 7246.4 Hz. In all, 128 experiments were recorded in the F1 domain, with a presaturation time of 1 s and an acquisition time of 0.565 s, resulting in a time resolution of 27.3 s/single spectrum. Data processing was performed with XWINNMR and 1D WINNMR software (Bruker). Zero-filling to 32K data points and exponential multiplication with a line broadening of 0.3 Hz was applied to the FID prior to Fourier transformation in the F2 dimension. RESULTS AND DISCUSSION The separation of all four tocopherol isomers (R, β, γ, and δ) and R-tocopherol acetate is shown in Figure 1. The upper chromatogram shows the chromatographic separation with analytical concentrations, i.e., 2.5 µg of sample injected onto the column, which equals about 1.2 nmol of each isomer. With a flow rate of 1 mL/min, all substances elute within 14 min on a C30
Figure 2. Contour plot of the separation of tocopherol isomers, corresponding to the chromatogram shown in Figure 1 (bottom).
column (3 µm, 200 Å, 250 mm × 4.6 mm) as stationary phase. Peaks 2 and 3 (i.e., β- and γ-tocopherol) are well separated, with a resolution R of 1.28, which is ideal for analytical separations.29 With an isocratic eluent and a flow rate of 1 mL/min, this is a fast, efficient, and highly reproducible separation. The number of theoretical plates calculated for the γ-tocopherol peak is 60 000/ m, whereas for the peak of β-tocopherol, 30 800/m was calculated. These values show an inherent feature of C30 bonded phases. Compared to C18 columns, the theoretical plate number is lower, leading to larger peak half-widths. Nevertheless, this is more than outweighted by the increased shape selectivity of these stationary phases. Further, compared to NP separations, a minor water content in the mobile phase has no effect on the retention times, and with 100% methanol as eluent, even the solvent composition is not critical for retention time reproducibility. Therefore, the separation in Figure 1 should have a high potential for routine analysis of tocopherol levels in pharmaceutical and biological samples. The lower chromatogram of Figure 1 shows the separation of a 200-fold amount of sample compared to the analytical separation, i.e., 0.5 mg injected onto the column. Except for reduction of the flow rate to 0.3 mL/min, the chromatographic conditions were identical. It is remarkable that the separation efficiency of the column is only slightly decreased with that much sample loaded onto the column. This also renders the C30 phase ideal for preparative scale separations. An on-line HPLC-1H NMR experiment was recorded for the separation shown in Figure 1 (bottom), the contour plot of which is illustrated in Figure 2. In this representation, the 1H NMR chemical shift axis is displayed in the horizontal dimension (F2 axis), whereas the retention time is shown in the vertical axis (F1 axis), resulting in a pseudo-two-dimensional NMR contour (29) Determination of resolution R {R ) 1.18(tRb - tRg)/(w(1/2)β + w(1/2)γ)} and theoretical plate number N {N ) 5.54(tR/w1/2)2} from: Meyer, V. R. Praxis der Hochleistungs-Flu ¨ ssigchromatographie, 7. Aufl.; Salle and Sauerla¨nder, Frankfurt, FRG, 1992; pp 22-24.
map. The chromatogram along the F1 axis was reconstructed by summation of all aliphatic 1H NMR signals between 0.8 and 1.9 ppm. Although the resolution of such a reconstruction suffers from the small number of F1 data points, the separation of all five compounds is still clearly visible, even though the separation of β- and γ-tocopherol seems to be inferior to the UV-detected separation. The 1H NMR spectra extracted from single rows of the contour plot at the corresponding peak maxima of each compound are displayed in Figure 3. The spectra from bottom to top reflect the elution order of the isomers in the contour plot. For all 1H NMR spectra, an unequivocal structural assignment of the corresponding tocopherol isomer can be performed. All spectra give sufficient signal-to-noise ratio (S/N) for determination of integration ratios and coupling constants. The spectra are all free of superimposed signals from coeluting compounds, except for the spectrum of β-tocopherol, where small signals of the previously eluting γ-tocopherol are visible. The structural differences among these isomers can easily be monitored by the 1H NMR signals between 2.0 and 2.2 ppm of the methyl groups attached to the aromatic ring, or equivalently by examining the aromatic 1H NMR signals between 6.4 and 6.5 ppm. For δ-tocopherol, one 1H NMR signal of the methyl group at C-8 is visible, while two aromatic 1H NMR signals with a small splitting due to the meta coupling between H-5 and H-7 are detected. For γ- and β-tocopherol, aromatic 1H NMR signals at 6.35 and 6.45 ppm, respectively, are visible. At 2.1 ppm, 1H NMR signals of two methyl groups appear, which, in the case of β-tocopherol, have approximately the same chemical shift value. A comparison between the 1H NMR spectra of R-tocopherol and R-tocopherol acetate shows that the additional acetate group (2.30 ppm) causes an upfield shift of the 1H NMR signals of the three methyl groups attached to the aromatic system. Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
15
Figure 3. Stacked plot of the 1H NMR spectra of the separated tocopherol isomers. The aromatic region has been enlarged by a factor of 2.
The idea of shape selectivity of reversed phases has been the subject of many papers and reviews.30-33 Compared to C18 phases, C30 phases exhibit a strongly improved shape recognition of rigid, extended solutes, making them ideal stationary phases for the separation of isomers, e.g., carotenoid cis/trans isomers.8-11 This ability correlates with a high order of the stationary phase, which is mainly due to the alkyl chain length34 and the dense surface coverage of C30 chains resulting from the polymeric synthesis.18 The order of the stationary phase can be determined by solidstate 13C NMR spectroscopy. Compared to C18 phases, the C30 phases possess a much higher alkyl chain rigidity.34 The lower (30) Dorsey, J. G.; Dill, K. A. Chem. Rev. 1989, 89, 331-346. (31) Cole, S. R.; Dorsey, J. G. J. Chromatogr. 1993, 635, 177-186. (32) Tchapla, A.; He´ron, S; Lesellier, E. J. Chromatogr. A 1993, 656, 81-112. (33) Sander, L. C.; Wise, S. A. Anal. Chem. 1995, 67, 3284-3292. (34) Pursch, M.; Brindle, R.; Ellwanger, A.; Sander, L. C.; Bell, C. M.; Ha¨ndel, H.; Albert, K. Solid State NMR, in press.
16 Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
spectrum of Figure 4 shows the solid-state 13C MAS NMR spectrum of a C30-modified silica gel. The most interesting feature of this spectrum is the splitting of the signal of the main chain (C-3 to C-28), which has already been assigned to chain segments with different mobility.17,35,36 The signal at 32.8 ppm corresponds to a rigid trans conformation, whereas the signal at 30.0 ppm belongs to more mobile gauche conformations. Further, the signals of these long alkyl chains act as a sensor for applied temperature and bonding density of the sorbent17 and should also be sensitive to the nature of various solvents. The correlation between HPLC results and solid-state NMR data of the stationary phase proved to be in good agreement for the separation of vitamin A acetate isomers36 and PAHs18 on C30 (35) Ohta, H.; Jinno, K.; Saito, Y.; Fetzer, J. C.; Biggs, W. R.; Pesek, J. J.; Matyska, M. T.; Chen, Y.-L. Chromatographia 1996, 42, 56-64. (36) Albert, K.; Ha¨ndel, H.; Pursch, M.; Strohschein, S. Chemically Modified Surfaces; The Royal Chemical Society: Cambridge, 1996; Vol. 7, pp 30-45.
Figure 4. ether (top).
13C
MAS NMR spectra of a C30 interphase in the solid state (bottom) and suspended in methanol (middle) and methyl tert-butyl
and C18 phases, respectively. But the question is still whether results obtained from solid-state NMR investigations of bonded phases can actually be compared with data from HPLC separations, where the stationary phase is suspended in the mobile phase. Therefore, we employed the technique of 13C MAS NMR spectroscopy of bonded phases in the suspended state. This method has been used by Zeigler and Maciel for the study of C18 phases; however, the addition of solvents did not show any significant effect on the resulting NMR spectra.24 In our approach, a C30 phase was suspended in methanol, placed in a 7 mm rotor, and the NMR spectrum was measured in a conventional solidstate NMR probe using sample spinning rates of 3000 Hz. The resulting 13C NMR spectrum is displayed in Figure 4 (middle). In the spectra of the suspension, an additional sharp 13C NMR signal of the solvent (methanol) appears, but otherwise the lower (solid-state) and middle spectra are very similar. The resonances at 32.8 and 30.0 ppm reveal that the intensity of the gauche signals is somewhat increased, probably because of a general higher mobility of the alkyl chains due to the presence of methanol. A similar NMR spectrum was obtained with acetonitrile as solvent. Thus, the addition of such solvents leads to a small modulation of the structure of the stationary phase. But with methanol or
acetonitrile, the overall rigidity of the C30 chains is not altered. This is the main reason for the improved shape selectivity of the C30 phase for the separation of carotenoids and vitamin isomers when the chromatography is performed with methanol or acetonitrile as mobile phase.8,17 Additionally, in Figure 4 (top spectrum), the 13C MAS NMR spectrum of a C30 phase suspended in MTBE is displayed. The differences between the two suspended-state NMR spectra are quite obvious. Contrary to the suspension in methanol, in the presence of a nonpolar solvent, like MTBE, the alignment of the alkyl chains is dramatically altered by the contact between stationary phase and solvent. On the one hand, this is proven by the fact that the gauche signal at 30.0 ppm dominates the spectrum. This indicates that a high fraction of mobile units is present within the alkyl chains. On the other hand, the line widths of the terminal methyl groups also reveal differences in the mobility of the chain ends compared to the solid-state NMR spectrum. In the latter, two resonances for the terminal methyl groups appear, the one at 14.1 ppm corresponding to more rigid structures and the other one, at 12.5 ppm, indicating higher chain end mobility. In the suspended state, only the signal at 14.1 ppm is visible. It seems that the γ-gauche effect is not visible for Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
17
terminal methyl groups in suspension. This has also been observed by Pfleiderer from NMR investigations of C18 phases in suspension.37 But the line widths of these signals differ dramatically with solvent. In methanol, a 13C MAS NMR signal half-width of 68 Hz is measured, compared to 41 Hz in MTBE. In NMR spectroscopy, the line width of a given signal is inversely proportional to the mobility of the observed system. So, also the signals of the terminal methyl groups point to an overall higher mobility of the stationary phase in MTBE compared to that in methanol. Any solvent with a high solvent strength has two effects on the separation system. First, the pharmaceutical is predominantly desolved in the mobile phase, resulting in negligible interaction between solute and stationary phase. Second, solvents like MTBE cause a strong disordering of the bonded alkyl chains, as can be seen from the suspended-state MAS NMR spectra. A separation of tocopherol isomers with pure MTBE was, therefore, not possible. The 13C MAS NMR spectra in suspension have shown that weak solvents, like methanol and acetonitrile, have only a minor influence on the order of alkyl chains of the stationary phase. These solvents are mostly used as the main components of the mobile phase in RP chromatography. Strong eluents, i.e., MTBE or chloroform, often added in minor amounts as modifiers for RP separations, strongly increase the mobility of the stationary phase.
The shape selectivity of C30 phases benefits from the high order of its alkyl chains. The more ordered their structure, the better the resolution of isomeric compounds in LC separations. Therefore, by the combination of 13C MAS NMR spectroscopy in the suspended state and HPLC-NMR experiments, the synergetic effect of the per se highly ordered C30 stationary phase and methanol as mobile phase demonstrates the suitability of this system for the separation of isomeric compounds. CONCLUSION We have shown that the shape selectivity of the C30 phase allows a complete separation of tocopherol isomers. By recording of on-line 1H NMR spectra during the separation, a thorough structural assignment is possible. The introduction of 13C MAS NMR spectroscopy of C30 bonded phases in the suspended state shows promise for the study of solvent influences on alkyl chain structure and mobility. ACKNOWLEDGMENT We are grateful to Martin Raitza for recording the solid-state 13C MAS NMR spectrum and to James Sudmeier (Tufts University, Boston, MA) for helpful discussions. The support of Bischoff Chromatography (Leonberg) is gratefully acknowledged. Received for review April 21, 1997. Accepted October 7, 1997.X AC970414J
(37) Pfleiderer, B. Ph.D. Dissertation, Universita¨t Tu ¨ bingen, 1989.
18
Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
X
Abstract published in Advance ACS Abstracts, November 15, 1997.
