Anal. Chem. 1998, 70, 3280-3285
On-Line Coupling of Capillary Electrochromatography, Capillary Electrophoresis, and Capillary HPLC with Nuclear Magnetic Resonance Spectroscopy Klaus Pusecker, Jens Schewitz, Petra Gfro 1 rer, Li-Hong Tseng, Klaus Albert, and Ernst Bayer*
Institut fu¨ r Organische Chemie, Universita¨ t Tu¨ bingen, Auf der Morgenstelle 18, D 72076 Tu¨ bingen, Germany
A novel capillary NMR coupling configuration, which offers the possibility of combining capillary zone electrophoresis (CZE), capillary HPLC (CHPLC), and for the first time capillary electrochromatography (CEC) with nuclear magnetic resonance (NMR), has been developed. The hyphenated technique has a great potential for the analysis of chemical, pharmaceutical, biological, and environmental samples. The versatile system allows facile changes between these three different separation methods. A special NMR capillary containing an enlarged detection cell suitable for on-line NMR detection and measurements under high voltage has been designed. The acquisition of 1D and 2D NMR spectra in stopped-flow experiments is also possible. CHPLC NMR has been performed with samples of hop bitter acids. The identification and structure elucidation of humulones and isohumulones by on-line and stopped-flow spectra has been demonstrated. The suitability of the configuration for electrophoretic methods has been investigated by the application of CZE and CEC NMR to model systems. Nuclear magnetic resonance spectroscopy (NMR) is one of the most powerful analytical methods for identification and structure elucidation of organic compounds. However, overlapping of the NMR signals makes identification of compounds in complex mixtures very difficult. This problem can be overcome by combining NMR detection with a separation technique. The first experiments for an on-line coupling of HPLC with NMR were carried out in the late 1970s,1,2 but wide application of the method was hindered by the low sensitivity of NMR and technical problems arising from the use of protonated solvents. A decade later, sensitivity of NMR had enhanced dramatically and NMR spectrometers with an increased dynamic receiver range became available. Thus, registration of the signals of the samples diluted in protonated solvents became possible, and HPLC NMR coupling can today be used routinely for a wide range of applications.3-9 (1) Watanabe, N.; Niki, E. Proc. Jpn. Acad., Ser. B 1978, 54, 194-199. (2) Bayer, E.; Albert, K.; Nieder, M.; Grom, E.; Keller, T. J. Chromatogr. 1979, 186, 497-507. (3) Albert, K. J. Chromatogr., A 1995, 703, 123-147. (4) Albert, K.; Bayer, E. Anal. Methods Instrum. 1995, 2, 302-314. (5) Ho ¨ltzel, A.; Schlotterbeck, G.; Albert, K.; Bayer, E. Chromatographia 1996, 42, 499-505.
3280 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
The major drawback of this method remains the necessity for elaborate solvent signal suppression leading to distortion of parts of the spectra. Problems associated with protonated solvents are prevented by the use of capillary separation techniques. The extremely small volume of eluent consumed makes the use of relatively costly deuterated solvents economically feasible. However, this means a reduction of the detection volume by a factor of ∼1000 from 60 to 200 µL in conventional LC NMR to 5-400 nL in capillary NMR. Despite this problem, efforts have been made in the last years to increase sensitivity of NMR detection in order to enable capillary coupling techniques.10 Capillary zone electrophoresis (CZE) is a well established analytical technique that can achieve rapid and efficient separation of charged analytes and biomolecules even in small volumes.11 The combination of CZE and NMR has been achieved using a solenoidal radio frequency (rf) coil attached to the capillary.12,13 In these publications, the microcoil was directly mounted on the separation capillary since this configuration yields to a favorable filling factor. The sample was externally loaded after which the probe head was inserted and shimmed. This necessitates the preparation of a new microcoil for each column used and time consuming shimming for each new analysis. With this configuration, changing of either the capillary or the mode of separation is cumbersome and limits sample throughput. Capillary electrochromatography (CEC) is a new and emerging technique with great potential, combining the selectivity of HPLC with the high speed and high efficiency of capillary electrophoresis.14-23 CEC is carried out in packed capillaries with an (6) Strohschein, S; Pursch, M., Handel, H.; Albert, K. Fresenius’ J. Anal. Chem. 1997, 357, 498-502. (7) Siedelmann, U. G.; Braumann, U.; Hofinann, M.;Spraul, M.; Lindon, J. C.; Nicholoson, J. K.; Hansen, S. H. Anal. Chem. 1997, 69, 607-612. (8) Lindon, J. C.; Nicholson, J. K.; Sidelmann, U. G.; Wilson, I. D. Drug Metab. Rev. 1997, 29, 705-746. (9) Lindon, J. C.; Nicholson, J. K., Wilson, I. D. Adv. Chromatogr. 1996, 36, 315-382. (10) Olson, D. L.; Peck, T. L.; Webb, A. G.; Magin, R. L.; Sweedler, J. V. Science 1995, 270, 1967-1970. (11) Jorgensen, J.W.; Lukacs, K.D. Science 1983, 222, 266-272. (12) Wu, N.; Peck, T. L.; Webb, A. G.; Magin, R. L.; Sweedler, J. V. Anal. Chem. 1994, 66, 3849-3857. (13) Wu, N.; Peck, T. L.; Webb, A. G.; Magin, R. L.; Sweedler, J. V. J. Am. Chem. Soc. 1994, 116, 7929-7930. (14) Pretorious, V.; Hopkins, B. J.; Schieke, J. D. J. Chromatogr. 1974, 99, 2330. S0003-2700(98)00063-8 CCC: $15.00
© 1998 American Chemical Society Published on Web 07/03/1998
electric field applied across the length of the column. If reversedphase materials are used, the retention of uncharged analytes is governed solely by their hydrophobic interaction with the stationary phase whereas the retention of charged analytes is additionally influenced by their electrophoretic mobility in the mobile phase. The mobile phase in CEC is driven by the electroosmotic flow (EOF) or by a combination of EOF and pressure (pressurized electrochromatography). The EOF provides a 2-fold advantage over conventional pump systems: first, the EOF is independent of the particle diameter allowing the use of long capillaries packed with stationary phases of relatively small particle size (e.g., 1.5 µm). Second, CEC provides the high separation efficiency typical of capillary electrophoresis since the electroosmotic flow has a plug flow profile. As a result, improvement of the separation efficiency, expressed as number of theoretical plates, by a factor of 2-3 over pressure-driven CHPLC is attainable.16 An advantage of CEC over CZE is the higher sample capacity which makes CEC especially attractive for hyphenation with NMR spectroscopy. In this work, we report on the development of a new configuration for the coupling of miniaturized separation methods with NMR. The method provides facile handling and is suitable for CEC, CZE, and CHPLC. For this purpose we designed a special NMR detection capillary and investigated the effect of the strong electric fields in the NMR cavity upon statically measured NMR spectra. Additionally we present the results of experiments with continuous- and stopped-flow NMR coupled to CEC, CZE, and CHPLC. The system was tested with model systems for its suitability for the electrophoretic techniques and applied to the analysis of hop bitter acids by CHPLC. EXPERIMENTAL SECTION Materials. A Bruker AMX 600 NMR spectrometer (Bruker, Rheinstetten, FRG) was connected with a modular capillary electrophoresis system (Grom, Herrenberg, FRG) and a HPLC system (Sykam, Gilching, FRG) for CZE and CHPLC, respectively. For CEC as a hybrid of CZE and CHPLC, both systems were combined. Fused-silica capillaries of 50- and 250-µm i.d. and 360µm o.d. were obtained from Polymicro Technology (Phoenix, AZ). The packing materials, 5 µm Grom-sil ODS-4 HE, 3-µm Grom-sil Si, and 5-µm Grom-sil 100 ODS-0 AB, were donated by Grom (Herrenberg, FRG). All buffers were made with fully deuterated solvents purchased from Deutero GmbH (Herresbach, FRG); NaH2PO4 and H3PO4 were obtained from Merck (Darmstadt, FRG). Detection Capillary. A 50-µm-i.d. fused-silica capillary was treated for 15 min with 2 M HCl, after which the capillary was flushed with 2 M NaOH for 15 min to clean the internal surface. The capillary was then rinsed for 3-4 h with 10% HF solution while the area of the detection window was heated in a water bath to 90 °C over a length of 1.5 cm. Finally, the capillary was flushed (15) Jorgenson, J. W.; Lukacs, K. D. J. Chromatogr. 1981, 218, 209-216. (16) Knox, J. H.; Grant, I. H. Chromatographia 1991, 32, 317-328. (17) Tsuda, T. LC-GC Int. 1992, 5, 26-36. (18) Yamamoto, H.; Baumann, J.; Erni, F. J. Chromatogr. 1992, 593, 313-319. (19) Yan, C.; Schautelberg, D.; Erni, F. J. Chromatogr., A 1994, 670, 15-23. (20) Smith, N. W.; Evans, M. B. Chromatographia 1994, 38, 649-675. (21) Boughtflower, R.J.; Underwood, T.; Paterson, C. J. Chromatographia 1995, 40, 329-335. (22) Rebscher, H.; Pyell, U. Chromatographia 1996, 42, 171-176. (23) Behnke, B.; Bayer, E. J. Chromatogr., A 1994, 680, 93-98.
Figure 1. (A) I. Capillary with enlarged detection cell for NMR detection. The capillary is suitable for CZE separations. II. By connecting a packed capillary in front of the NMR detection capillary, the system is adapted to use with CEC and CHPLC. (B) Schematic diagram of the apparatus for the construction of the detection capillary.
with distilled water. In the detection area, the polyimide coating was burned and was removed by cleaning with methanol. Column Packing. The column packing method was based on that of Boughtflower et al.21 Fused-silica capillaries with 250µm i.d. and 350-µm o.d were first tapped into a slurry of silica gel and water. After drying, a temporary end frit was sintered using a homemade heater. Then the capillary was packed at 400 bar (liquid chromatographic pump from Sykam, Gilching, FRG) with a slurry of Grom-sil ODS-2 dp ) 5 µm and 10% Grom-sil Si dp ) 3 µm (column 1) in acetone for use in CEC measurements or with Grom-sil 100 ODS-0 AB dp ) 5 µm (column 2) in acetone for CHPLC analysis of hop bitter acids. Finally, the outlet- and inlet frits were sintered 20 cm apart under a flow of distilled water and the surplus parts of the capillary cut off directly after the frits. CZE NMR. For CZE measurements, both ends of the detection capillary were immersed in the buffer reservoirs constructed of poly(ether ether ketone) (PEEK) which were positioned 15 cm below the NMR magnet. The power supply was ∼3 m from the magnet. High voltage was applied to the anode and the cathode was grounded. The detection capillary (60-µm i.d.) itself was used as column for the separation. The overall length was 200 cm with an effective length to the detection window of 90 cm. Before each run the capillary was flushed with 0.2 M NaOH and filled with eluent (20 mM Na2HPO4 D2O/10% CD3CN). Samples were electrokinetical loaded and separated with an applied voltage of 20 kV. CHPLC NMR. When operating in the CHPLC mode, the detection capillary (Figure 1 AI) is connected outside the NMR probe to a packed capillary by means of a Teflon tube (Figure 1AII). The other end of this capillary was immersed into the injection device and the end of the detection capillary to the outlet vial (cf. CZE-NMR). A HPLC pump was connected via a 250µm-i.d. capillary to the injection device. The sample was injected Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3281
by filling the 2-µL reservoir of the injection device with sample and connecting it to the pump. By means of a 100-µm-i.d. resistance capillary, the stream to the column was split, yielding an effective injection volume of 1 µL which contained ∼20 µg of the analytes. A 20% solution of hop bitter acids was separated on column 2 with a mixture of CD3CN (60%) and 0.05% phosphoric acid in D2O (40%) as eluent. CEC NMR. In this case, the CHPLC system was combined with the CE power supply, as described previously.23 The alkyl benzoates were separated on column 1 with 2 mM borate D2O/ CD3CN (20/80) at a pressure of 8 bar and an applied voltage of 20 kV. NMR Spectroscopy. NMR spectra were recorded on a Bruker AMX 600 spectrometer equipped with a 2.00-mm capillary microprobe matched with an rf coil selective for protons and a detection capillary with a 240-nL detection cell. Prior to measurement, the NMR spectrometer was shimmed to the chloroform signal of a hump test (3% chloroform in 97% acetone-d6) resulting in a half-height width of 2.5 Hz. Static NMR Experiments. Static NMR spectra were performed by injecting the respective solution directly into the detection capillary. A total of 16 000 data points with a spectral width of 3521 Hz, resulting from an acquisition time of 2.32 s, were recorded. The relaxation delay was set to 1.2 s, and 64 transients were coadded, with a total acquisition time of 3 min 51 s. On-Line NMR Experiments. For continuous-flow experiments, 64 transients with 8000 complex data points and a spectral width of 6024 Hz were recorded. A relaxation delay of 0.8 s and a acquisition time of 0.679 s/transient were used. Successive 1H NMR spectra were recorded using accumulations of 64-128 FIDs for each spectrum. The total acquisition time per row was 1 min 34 s (for 64 FIDs) and 3 min 8 s (for 128 FIDs). Data were treated as a 2D matrix. Fourier transformation was performed in the t2 direction only. Stopped-Flow 2D Experiments. A total of 256 t1 increments with 88 transients and 2000 K complex data points were acquired with a spectral width of 6024 Hz in both dimensions. The total acquisition time amounted to 13 h 5 min. RESULTS AND DISCUSSION The present work utilizes a configuration recently developed in our group for capillary HPLC NMR coupling24,25 and modified here for electrophoretic techniques. When one is measuring under electrical current, safety considerations necessitate the avoidance of leaks within the NMR probe, and in this case, a one-piece capillary devoid of Teflon tubing inside the NMR probe was developed (Figure 1A). Additionally, an enlarged detection cell (V ) 240 nL) was found to be advantageous in order to increase the volume of detection and thus the residence times. Longer residence times enable more transients to be coadded for the NMR spectrum and reduce the effects of the flow upon the spin-spin relaxation times.3,9 The increased detection volume made shimming easier, because a (24) Behnke, B.; Schlotterbeck, G.; Tallarek, U.; Strohschein, S.; Tseng, L.-H.; Keller, T.; Albert, K.; Bayer, E. Anal. Chem. 1996, 68, 1110-1115. (25) Schlotterbeck, G.; Tseng, L.-H.; Handel, H.; Braumann, U.; Albert, K. Anal. Chem. 1997, 69, 1421-1425.
3282 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
Figure 2. Scheme of the instrumentation for the coupling of capillary separation techniques with NMR.
more stable lock and a better signal to noise ratio for the hump test is obtained. Enlargement of the internal diameter of a standard fused-silica capillary was achieved by etching with an HF solution (Figure 1B). By varying duration of etching, the diameter of the capillary in the detection area could be increased up to a maximum of 240 µm, while the rest of the capillary still had a internal diameter of ∼60 µm. The resulting detection cells were highly symmetrical. The used capillary had an internal diameter of ∼190 µm. Regarding the length of the rf coil of 9 mm a detection volume of 240 nL was calculated. The capillary probe was constructed for a Bruker AMX 600 NMR spectrometer. A saddle-type rf coil with a diameter of 2 mm was used. The capillary is thus positioned parallel to the B0 field. Since the direction of the electric current in the capillary (µA range) resulting from the applied voltage is parallel to the main magnetic field, the induced magnetic field has no component in the B0 direction and effects upon the NMR spectra due to perturbation of the B0 field and the current-induced magnetic field should be negligible. The comparatively large diameter of the coil results on the one hand in a less favorable filling factor, but on the other hand, problems due to the effects of magnetic susceptibility caused by proximity of the rf coil to the sample are not to be expected. The moderate filling factor with concomitant reduction in sensitivity is partly compensated by the increased detection volume. The rf coil is attached to a glass tube (2-mm o.d. and 1.7-mm i.d.), allowing simple exchange of the capillary without disturbing the coil. After its installation in the probe, the detection capillary was connected to the chromatography system (Figure 2). Static Measurements. Experience with strong electric fields (e.g., 10 kV/m) in a NMR cavity is limited. For that reason, the assumption that the effect on the NMR spectra of our arrangement would be negligible was checked by measuring static spectra of 2.5 and 5% lysine phosphate buffer over a range of applied voltages (0 V - 20 kV) (Figure 3A). The voltage was increased in steps of 2.5 kV. The lysine spectra thus obtained contained all
Figure 3. (A) Static 600-MHz 1H NMR spectra of lysine under CZE conditions: (I) without voltage, (II) with 10 kV and 29 µA, and (III) with 20 kV and 36 µA. The 240-nL cell contained 41 nM amino acid. (B) CZE NMR chromatogram of the separation of a lysine (1) and histidine (2) mixture.
information of a conventional NMR spectrum. Chemical shift values, coupling constants, and intensities are accessible for structure elucidation. No peak broadening was observed with increasing voltage, indicating that perturbation of the B0 field is negligible. The effect of the Joule heating was however detected as a slight chemical shift of the whole spectrum, the extent of which was dependent upon the electrical current. The 240-nL cell contained 41 nM of the amino acid. A signal to noise ratio (S/N) of 54 was calculated for the -CH2 group. This yields a limit of detection (S/N ) 3) of 2.3 nM (336 ng). This sensitivity in the lower nanomolar range is only 1 order of magnitude inferior to the range given by Sweedler (limit of detection ∼100 pM),4 this being due to the more favorable filling factor of the solenoidal rf coil configuration. CZE NMR. The suitability of the configuration for on-flow detection was investigated by CZE NMR coupling. In this case, the separation can be performed within the detection capillary. However, the combination of CZE with NMR requires careful
selection of the buffer system which should provide a high EOF under proton-free conditions. The NMR contour plot in Figure 3B illustrates the separation of lysine (1) and histidine (2). Identification of the amino acids is possible from the aromatic and aliphatic signals. A background subtraction was applied during the processing of the spectra. In this way, the residual water signal was eliminated in front of the EOF. The higher water concentration with the EOF was used as a flow marker. CEC NMR. The advantages of CEC in comparison to CHPLC are apparent from the separation of an alkyl benzoate mixture (Figure 4A). With CHPLC, no baseline separation was achieved due to the high loading necessary for the NMR detection. Nevertheless, the additional information of the NMR, the chemical shift values, enables identification of all five alkylbenzoates. For example, no distinction of the homologues is possible from the three continuous signals in the aromatic region, but differences in the aliphatic region reflect the different side chains. In addition to the signals of the analytes, the signal from the nondeuterated acetonitrile in which the analytes were dissolved is observed. This also caused the shift of the water signal. CEC was carried out under the same conditions as above, but with a voltage of 20 kV applied across the length of the capillary. As illustrated in the contour plot (Figure 4B), the elution time could be reduced by a factor of 2. This significant shortening of the analysis time is a result of the additive effects of pressure and electroosmotic force upon the flow rate. Furthermore, dramatic band sharpening with improved resolution resulting from the plug flow is observed.8 Despite the overloading needed for NMR detection, baseline separation could be achieved, noticeable in the aromatic signals region. To obtain comprehensive structural information, the acquisition of stopped-flow 2D experiments is often advantageous. The configuration described allows measurement under stopped-flow conditions simply by switching off pressure and voltage. In Figure 4C, a CEC stopped-flow H,H COSY spectrum of the last peak (pentyl benzoate) is shown. The aliphatic region contains the cross-peaks of the pentyl side chain. The coupling pattern of the aromatic part is typical for a monosubstituted aromatic hydrocarbon. In the case of a disubstituted compound, NMR would enable a decision about the position of the second substitutent. Capillary HPLC. The usefulness of the method for the analysis of natural products is demonstrated by the on-line CHPLC NMR measurements of hop bitter acids. The n-, co-, and adhumulones are the most important constituents of hops. During the brewing process, they are converted into the corresponding cis- and trans-isohumulones, which are responsible for the typical bitter taste of beer. As reported previously, the conventional LC NMR is well suited for the investigation of these unstable stereoisomeres and allows unequivocal identification of the compounds.5 The H,H COSY spectrum of the n-humulone demonstrates the quality for spectra in capillary NMR (Figure 5A). The spectrum was recorded under stopped-flow conditions. A humulone standard consisting of 89% n-humulone, 5% cohumulone, and 3% adhumulone was injected. By means of CHPLC the n-humulone was separated from the other isomers and impurities. The total acquisition time amounted to 13 h. The spectrum contained all expected 2D connectivities for the molecule, i.e., the structure Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3283
Figure 4. (A) CHPLC NMR chromatogram of the separation of five alkyl benzoates. The signals 1-5 correspond to methyl, ethyl, propyl, butyl, and pentyl benzoate, respectively. (B) CEC NMR chromatogram of the separation of the five alkyl benzoates. Same conditions as in Figure 4A, additionally an applied voltage of 20 kV. (C) Stopped-flow H,H COSY of the pentyl benzoate (5) from the CEC separation.
elucidation and assignment of the 1H NMR signals of such a natural product is possible. The isohumulones are an example of a complex mixture in natural product chemistry. It consists of six isohumulones including stereoisomers, homologues, and isomers. Furthermore, the mixture contains impurities caused by oxidation and photolysis. With the CHPLC NMR, identification of the different compounds in the on-flow experiments was possible. The NMR contour plot of the aliphatic region of the separation in Figure 5B enables distinction between the four main constituents trans/cisisohumulone (2, 4) and trans-/cis-isocohumulone (1, 3). By the additional NMR chemical shift dimension, an identification of the compounds even under difficult separation conditions caused by high loading of the column was possible. A stopped-flow H,H COSY NMR spectrum was used for assignment of the signals (Figure 5D). In contrast to conventional LC NMR, there is no 3284 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
need for solvent suppression, since fully deuterated solvents can be used due to the miniaturization. Thus, the signal in the 3'position (Figure 5C) of the n-isohumulone, normally masked by the solvent signals, can also be detected in a 2D NMR experiment. This demonstrates one of the important advantages of CHPLC NMR over the conventional HPLC NMR. CONCLUSIONS The novel configuration provides the possibility of coupling the most important miniaturized capillary separation techniques with NMR. The ease of change between the various separation methods renders the system suitable for a large variety of analyses. The static measurements proved the advantage of the used arrangement. The high voltage required for CZE and CEC did not create problems with the NMR spectroscopy. The online experiments demonstrated that the LOD of the system is
Figure 5. (A) Stopped-flow H,H COSY of n-humulone from a CHPLC separation of a humulone standard. (B) CHPLC NMR chromatogram of the separation of a isohumulone mixture; (1) trans-isocohumulone, (2) cis-isocohumulone, (3) trans-isohumulone, and (4) cis-isohumulone. (C) Structures of hop bitter acids. (D) Stopped-flow H,H COSY of n-isohumulone from a CHPLC separation of a isohumulone mixture.
sufficient for the detection of analytes in all miniaturized separation techniques. Also, the measurements confirm the advantages of coupling CEC to NMR: increased sample capacity in comparison to CZE and increased separation efficiency and reduced elution time compared to CHPLC NMR coupling. The investigation of hop bitter acids illustrated the suitability of the system for the analysis of real samples. Further developments will be predominantly focused on improvement of the NMR sensitivity by reducing the diameter of the rf coil. This in turn will improve the chromatographic efficiency by eliminating the necessity for column overloading.
ACKNOWLEDGMENT The authors gratefully acknowledge the valuable technical assistance of Bruker (Germany), W. Schaal, and E. Grom, the helpful advice of G. J. Nicholson, Go¨tz Schlotterbeck, and M. Maier, and financial support by the DFG Graduiertenkolleg “Analytische Chemie”.
Received for review January 23, 1998. Accepted May 13, 1998. AC980063O
Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3285
