Anal. Chem. 2003, 75, 1546-1551
Advancing NMR Sensitivity for LC-NMR-MS Using a Cryoflow Probe: Application to the Analysis of Acetaminophen Metabolites in Urine M. Spraul,† A. S. Freund,‡ R. E. Nast,‡ R. S. Withers,‡ W. E. Maas,§ and O. Corcoran*,‡
Bruker BioSpin GmbH, Silberstreifen, 76287 Rheinstetten, Germany, Bruker BioSpin Corporation, 47697 Westinghouse Drive, Fremont, California 94539, and Bruker BioSpin Corporation, 15 Fortune Drive, Manning Park, Billerica, Massachusetts 01821-3991
Cryogenic cooling of the NMR radio frequency coils and electronics to give greatly enhanced sensitivity is arguably the most significant recent advance in NMR spectroscopy. Here we report the first cryogenic probe built in flow configuration and demonstrate the application to LCNMR-MS studies. This probe provides superior sensitivity over conventional noncryogenic flow NMR probes, allowing the use of 100 µL of untreated urine (40% less material than previous studies that required preconcentration) and yet revealing drug metabolites hitherto undetected by LC-NMR-MS at 500 MHz. Besides the known sulfate and glucuronide metabolites, previously undetected metabolites of acetaminophen were directly observable in a 15-min on-flow experiment. Simultaneous MS data also provided knowledge on the NMR-silent functional moieties. Further, stop-flow LC-NMR-MS experiments were conducted for greater signal-to-noise ratios on minor metabolites. The cryoflow probe enables the NMR analysis of lower concentrations of metabolites than was previously possible for untreated biofluids. This strategy is generally applicable for samples containing mass-limited analytes, such as those from drug metabolism studies, biomarker and toxicity profiling, impurity analysis, and natural product analysis. NMR spectroscopy often provides the ultimate spectroscopic tool for structure elucidation, although it is still widely perceived as being a low-sensitivity technique. Fortunately, in recent years, NMR detection limits have advanced significantly with the introduction of so-called CryoProbes. In these NMR probes, the electronic components are cryogenically cooled to ∼20 K. By operating the electronic components at these temperatures, while the sample remains at ambient temperature, the electronic noise is greatly reduced.1,2 As a result, the signal-to-noise ratio for CryoProbes is increased on average 4-fold over that of conven* Corresponding author. Present address: Department of Pharmacy, King’s College London, Franklin-Wilkins Building, 150 Stamford St., London SE1 9NN, U.K. Phone: + 44 (0) 207 848 4049. Fax: +44 (0) 207 848 4800. E-mail:
[email protected]. † Bruker BioSpin GmbH. ‡ Bruker BioSpin Corp., Fremont, CA. § Bruker BioSpin Corp., Billerica, MA. (1) Styles, P.; Soffe, N. F.; Scott, C. A.; Cragg, D. A.; White, D. J.; White, P. C. J. J. Magn. Reson. 1984, 60, 397-404.
1546 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003
tional probes. This increase in signal-to-noise ratio has significant implications for the NMR measurements; a 4-fold increase in sensitivity enables a 4-fold lower detection limit (for a given experiment time) and for a given amount of sample, the experiment time is reduced by a factor of 16 over that of a conventional probe. Both advantages have great importance for the detection of metabolites in biofluids where (1) the analyte in a sample is of limited mass, (2) analytes are chemically unstable, (3) NMR experiments of low intrinsic sensitivity are required, or (4) NMR experimental time is necessarily short such as in a highthroughput analytical regime. Pharmaceutical laboratories worldwide use directly coupled LCNMR-MS to avoid traditional isolation of analytes in order to concentrate them for off-line NMR analysis.3-6 By means of a 95/ 5% postcolumn split to the NMR/MS spectrometers, NMR and MS data are readily obtained on the same analyte. MS data are sometimes critical for full structural elucidation since the common metabolic products of small drug molecules typically contain phenol, amide, nitro, N-oxide, sulfate, and glucuronide groups and “NMR-silent” heteroatoms such as N, Cl, and O. By combining LC, MS, and NMR instrumentation under full automation minimal sample is required, the risk of analyte degradation is minimized, and the fraction that is analyzed by NMR can be recovered for further spectroscopic analysis or bioactivity testing. LC-NMR-MS can be used in on-flow mode (often called continuous-flow mode) to acquire 1H NMR spectra as the solvent elutes from the chromatographic column through the NMR detection cell via a microbore capillary. However, the intrinsic sensitivity of the on-flow LC-NMR experiment at routinely used magnetic field strengths such as 500 MHz is somewhat limited by the residence time of the analyte in the magnet flowing at 1 mL/min through a 120-µL cell (4-mm o.d.). This gives a typical working detection limit of 10 µg per (2) Serber, Z.; Richter, C.; Moskau, D.; Bohlen, J. M.; Gerfin, F.; Marek, D.; Haberli, M.; Baselgia, L.; Laukien, F.; Stern, A. S.; Hoch, J. C.; Dotsch, V. J. Am. Chem. Soc. 2000, 122, 3554-3555. (3) Albert, K.; Kunst, M.; Bayer, E.; Spraul, M.; Bermel, W. J. Chromatogr. 1989, 463, 355-363. (4) Spraul, M.; Hofmann, M.; Lindon, J. C.; Farrant, R. D.; Seddon, M. J.; Nicholson, J. K.; Wilson, I. D. NMR Biomed. 1994, 7, 295-303. (5) Shockcor, J. P.; Unger, S. E.; Wilson, I. D.; Foxall, P. J. D.; Nicholson, J. K.; Lindon, J. C. Anal. Chem. 1996, 68, 4431-4435. (6) Godejohann, M.; Preiss, A.; Mugge, C.; Wunsch, G. Anal. Chem. 1997, 69, 3832-3837. 10.1021/ac026203i CCC: $25.00
© 2003 American Chemical Society Published on Web 02/14/2003
analyte on-column.6 A common strategy to increase the analyte residence time is to reduce the flow rate to 0.1-0.4 mL/min and accumulate more free induction decays (FIDs) per increment. However, the total experimental acquisition times can increase to several hours and on-column diffusion degrades the chromatographic resolution over time. To combat this, typical sample preparation for on-flow experiments includes lyophilization or solid-phase extraction (SPE).4,5,7 This step is generally followed by reconstitution in an NMR-friendly solvent and thereby concentration of the biofluid. However, certain reactive metabolites such as glucuronic acid esters or glutathione conjugates can be degraded and lost in this lengthy process. Furthermore, gross overload of the chromatographic column precludes better NMR sensitivity, again due to band broadening on the chromatographic column. New methods for increasing the NMR sensitivity of the onflow LC-NMR-MS experiment are of utmost importance in pharmaceutical analysis of biofluids.Thus, as the logical progression of both cryogenic and LC-NMR-MS technologies, we present the first CryoProbe built in flow configuration. Acetaminophen (APAP) was chosen here to test the novel cryoflow probe because the urinary metabolites are well documented by conventional LCNMR and LC-NMR-MS.4,5 EXPERIMENTAL SECTION Human urine was obtained from a healthy female volunteer (31 years old) 4 h after a single oral dose of 500 mg of APAP. After centrifugation to remove particulates, a 500-µL aliquot was reserved for immediate LC-NMR-MS analysis. Aliquots (100 µL) were directly injected onto the HPLC column. The HPLC instrument consisted of an Agilent 1100 series quaternary HPLC pump using an Agilent VWD UV detector set to measure 210 nm. The chromatographic column used was a YMC-Pack FL-ODS (4.6 mm × 50 mm) with 5-µm particle size. Separation was effected at 25 °C using D2O and acetonitrile-d3 (Cambridge Isotope Laboratories, Inc., Andover, MA) both containing 0.05% trifluoroacetic acid. Gradient elution was employed, starting at 2% acetonitrile-d3 and increasing to 60% over 15 min, at a flow rate of 0.4 mL/min. The total chromatographic runtime of APAP metabolites from the column under these conditions was 15 min. The chromatography was controlled by Hystar software (Bruker Daltonik, Bremen, Germany). Chromatographic detection was by UV at 210 nm, positive ion electrospray mass spectrometry, and 1H NMR spectroscopy at 500.17 MHz. Following UV detection of the analytes, a 95/5% NMR/MS split of the chromatographic flow was achieved using a commercially available mass spectrometry/NMR interface consisting of a delay loop and double dilutor syringe pump (BNMI interface, Bruker BioSpin GmBH). A total of 95% of the flow was directed to the cryoflow NMR probe, and the remainder was split to the Bruker Esquire 3000 ion trap mass spectrometer (Bruker Daltonics, Billerica, MA). The NMR spectra were measured at 500.17-MHz 1H observation frequency using a Bruker Avance 500 spectrometer. The cryoflow probe has an active cell volume of ∼40 µL and was built in a dual inverse 1H/13C configuration, with z-gradients and a deuterium lock. The glass flow cell geometry is based on (7) Wilson, I. D.; Nicholson, J. K. Anal. Chem. 1987, 59, 2830-2832.
Figure 1. Representative 500-MHz 1H NMR spectrum of the whole untreated urine obtained 4 h after a 500-mg APAP dose. This spectrum was acquired over 128 summed FIDs, in a conventional 5-mm triple inverse resonance probe (500 µL) and 100-µL aliquots were used for LC-NMR-MS analysis.
conventional flow probe designs used for LC-NMR-MS at the analytical scale, with a vertical cell attached to a PEEK capillary inlet at the bottom and outlet capillary at the top. This cell is positioned inside the ambient temperature bore of the probe and isolated from the radio freqency coils, which are cooled to 20 K, situated only a few millimeters away from the cell. For on-flow NMR detection, successive NMR spectra were acquired automatically with each spectrum comprising the summation of 16 FIDs collected into 4K data points, with an acquisition time of 0.41 s/scan. For stop-flow LC-NMR-MS experiments, typically 16-256 FIDs were collected using an acquisition time of 1.63 s/scan (32K data points). A pulse width of 10.25 µs at 18 dB (90°) was used. The residual HDO and acetonitrile resonances were suppressed using preirradiation based on the 1D version of the NOESY sequence. Sample temperature in the probe was maintained at 300 K throughout the NMR experiments. The FIDs were zerofilled to 64K data points and, prior to Fourier transformation, were multiplied by a line-broadening function corresponding to 1 Hz. Chemical shifts were internally referenced to the acetonitrile signal at δ ) 2.0. The minor proportion (5%) of the chromatographic eluant was mixed on-line with 150 µL/min LC-grade water containing 0.1% formic acid and simultaneously directed to the inlet of the ion trap MS. Separate injections were used to acquire both positive and negative ion electrospray mass spectra up to mass m/z ) 600. RESULTS AND DISCUSSION The signal-to-noise ratio advantage of the novel cryoflow probe over a typical conventional LC flow probe used for LC-NMR-MS studies is demonstrated by manual injection of 2 mM sucrose (90% D2O solution). 1H NMR spectra obtained in one scan at 500 MHz with a line-broadening factor of 1 Hz give a signal-to-noise ratio of 215:1 for the anomeric sugar proton using the novel cryoflow probe. This compares to 80:1 for a typical 4-mm LC selective inverse 1H-13C noncryoflow probe of 120-µL volume, and 20:1 for a typical noncryo 30-µL LC probe. Analytical Chemistry, Vol. 75, No. 6, March 15, 2003
1547
Figure 2. UV-detected chromatogram for untreated 4-h urine after a 500-mg APAP dose, obtained at 210 nm. Vertical lines indicate peaks for which stopped-flow spectra were obtained.
Figure 3. 500-MHz onflow-1H NMR-detected LC-NMR-MS chromatogram of whole human urine following administration of 500 mg of APAP. The horizontal axis represents chemical shifts, and the vertical axis denotes retention time in minutes. The eluting peaks are observed as contours. 1548 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003
Table 1. NMR Assignments and Chemical Shifts of Endogenous and APAP-Related Analytes Found in 4-h Urine after a 500-mg APAP Dosea Tr (min)
hydrogen
(δ 1H)
multiplicity
creatinine
2.5
glycine citrate
2.5 3.5
APAP sulfate
6.5
APAP glucuronide
7.5
APAP
8.5
methoxyAPAPd glucuronide (previously undetected)
8.5
APAP-related (previously undetected)
9.4
3-methoxy-APAP glucuronide (previously undetected)
9.4
indoxyl sulfate
10.5
hippurate
13.5
phenylacetylglutamine
13.5
CH2 CH3 CH2 CH2 CH2 H2, H6 H3, H5 N-acetyl H2, H6 H3, H5 H1′ H5′ H2′-H4′ N-acetyl H2, H6 H3, H5 N-acetyl H5 H-aromatic H1′ H5′ H2′-H4′ OMe N-acetyl H2, H6 H3, H5 N-acetyl H2 H6 H5 OMe H1′ H5′ H2′-H4′ N-acetyl H7 H4 H2 H5/H6 H6/H5 H2, H6 H4 H3, H5 methylene H3, H5 H2, H4, H6 R-CH methylene R-CH2β-CH2-
4.27 3.11 3.53 2.52 2.68 7.4 7.25 2.11 7.31 7.08 5.12 4.13 3.55-3.65 2.10 7.32 7.08 2.09 6.93 6.84-7.2 5.11 4.13 3.55-3.65 3.81 2.11 7.20 6.83 2.07 7.17 7.10 6.89 3.80 5.09 4.09 3.54-3.64 2.10 7.63 7.43 7.32 7.21 7.13 7.83 7.65 7.55 4.18 7.4 7.34 4.37 3.65 2.34 2.19
s s s d d d d s d d d d m s d d s dd d/m d d m s s d d s d d dd s d d m s d d s t t d t t s d d m s m m
component
m/z b 115 + NDND( 217+Na/+c ND233 + 352+Na/330/153+ 350-355-
175+Na/153+ ND(
ND( 360-
ND(
203+Na/181/105+ ND289+Na/267+ ND-
aGradient chromatography was used as described in the Experimental Section. Shifts were internally referenced to the residual HDO signal at δ 4.8. b m/z is reported for the observed [M + Dn]+ or [M + Dn]-. ND( denotes not detectable by either positive or negative electrospray ionization under the experimental conditions used. c X + Na denotes the sodium adduct of M. d Coelution of APAP and a methoxyAPAP glucuronide leads to overlap of aromatic resonance.
APAP was chosen here as a well-documented test compound whose urinary metabolic profile has been studied exhaustively using LC-NMR spectroscopy.4,5 The metabolites in human urine comprise principally the phenolic glucuronide and phenolic sulfate, and urinary recovery is the major route of elimination for APAPrelated material. In humans, ∼40% of the dose is recovered via the urine in a 4-h period postadministration. A typical 500-MHz 1H NMR spectrum of human urine (untreated except for the addition of ∼5% deuterated water for a field lock) sampled 4 h after a 500-mg dose of APAP is shown in Figure 1. On horizontal and vertical expansion, some thousand resonances arise from hundreds of endogenous metabolites. This spectrum was obtained
in 128 scans from 500 µL of the untreated human urine in a conventional 5-mm triple inverse resonance NMR probe, and the sample was further used for the LC-NMR-MS analysis reported here. Severe spectral overlap is observed in the conventional NMR spectrum of untreated urine, and further assignment of metabolite resonances is very difficult without available synthetic standards of putative metabolites. Clearly some form of sample purification is required in order to analyze the minor APAP metabolites in untreated urine. Traditional LC-NMR-MS at 500 MHz still presents signal-tonoise limitations that require lyophilization and reconstitution of sample to concentrate 5-6.7-fold prior to NMR analysis.4,5 The Analytical Chemistry, Vol. 75, No. 6, March 15, 2003
1549
direct injection of untreated urine maintains the integrity of the sample and allows a more accurate profile of APAP metabolites. Using a cryoflow probe as shown here results in rapid characterization as well as enhanced signal-to-noise ratios, enabling the identifcation of metabolites previously undetected at this routine magnetic field strength. Aliquots (100 µL) of untreated human urine were injected onto the HPLC column for on-flow LC-NMR-MS experiments. Figure 2 illustrates the resulting UV-detected chromatogram monitored at 210 nm. Major UV-absorbing components are observed, although the LC-UV trace provides no useful structural data apart from an indication of the elution profile. A corresponding control urine was not measured as the main metabolites of APAP have been characterized using LC-NMR-MS previously.4,5 Within a total on-flow NMR experiment time of 15 min, individual 1H NMR spectra were obtained sequentially as the HPLC eluant flowed through the cryogenic NMR probe, as shown in the on-flow LCNMR-MS chromatogram Figure 3. In this plot, the Fourier transformed FIDs are depicted as rows of a two-dimensional plot with 1H NMR chemical shift on the horizontal axis and HPLC retention time in minutes as the vertical axis. The entire acquisition interval is shown. The resonances from acetonitrile (δ ) 2.0) and residual HDO (δ ) 4.8) are due to the residual solvents in the mobile phase, after selective presaturation at these two frequencies. Usually, in LC-NMR, protonated acetonitrile would have been sufficient for the mobile phase, but here, acetonitriled3 was employed to allow detection of the N-acetyl groups of acetaminophen metabolites. The NMR assignments and chemical shift data for the endogenous and APAP-related material are given in Table 1. Figure 4 illustrates typical on-flow and stopped-flow spectra. The first analytes to elute from the chromatographic column are highly polar and mostly endogenous analytes: sugar-type molecules, creatinine, glycine, citrate, and other such small amino acid components. Panels c and d of Figure 4 show the excellent quality spectra for APAP phenolic glucuronide and phenolic sulfate obtained with only 16 FIDs during the onflow experiment. For better signal-to-noise ratios, further stopped-flow analysis was carried out. The stop-flow spectra in Figure 4e-g indicate up to six coeluting components in these three HPLC peaks. It is important to note that in the assignment of many urinary components neither NMR nor MS data alone could provide definitive assignment. In certain cases, components gave poor MS ionization by either positive or negative mode, or both. Only a combination of the data obtained by LC-NMR-MS could allow assignment. Three APAP metabolites, including two glucuronides, are identified here for the first time at 500 MHz, due to the superior NMR sensitivity achieved using the cryoflow probe. CONCLUSIONS We have demonstrated the characterization of APAP metabolites seen in vivo using 40% less material than previously reported. Particularly for drugs that are biotransformed to unstable or reactive metabolites, the direct injection of whole untreated urine provides the true profile of urinary metabolites without losses otherwise incurred by solvent extraction, lyophilization, or SPE purification. For chemically and biologically reactive drug metabolites such as acyl glucuronides and glutathione conjugates, which have been associated with idiosyncratic toxicity of certain 1550 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003
Figure 4. (a, b, h) Individual 1H NMR spectra extracted from the on-flow LC-NMR-MS experiment in Figure 3. Each extracted experiment corresponds to one NMR increment of 16 accumulated FIDs. (c-g) Stopped-flow spectra of a repeated injection of the same sample with the flow stopped at the indicated retention times, with 16 (c, d), 128 (e), and 256 (f, g) accumulated FIDs for improved signal/ noise ratio. The residual HDO signal was attenuated by filtering the δ 4.5-5.2 region postacquisition.
clinical drugs, the reduction in NMR analysis time due to cryoflow probes will allow for dynamic studies to be carried out on the most reactive metabolites.8 Clearly, APAP dose in humans at 500 mg provides a good test compound for this study as the urinary metabolism is extensive and 80% consists of two major metabolites. Of course, for most potent modern drugs the clinical therapeutic dose may be significantly lower (1-5 mg) and there may be several urinary metabolites present in clinical samples. As yet there are few literature publications on metabolites in plasma samples by LCNMR,9 presumably limited by the NMR sensitivity of the LC-NMR experiments at routinely used field strengths of 500 and 600 MHz. It is for these real-world scenarios that cryoflow probes and LCNMR-MS technology will be most beneficial, providing for NMR characterization of drug metabolites in a faster time frame and at lower detection quantities then previously possible. For in vitro drug metabolism studies at the stage of candidate selection in (8) Corcoran, O.; Mortensen, R. W.; Hansen, S. H.; Troke, J.; Nicholson, J. K. Chem. Res. Toxicol. 2001, 14, 1363-1370. (9) Daykin, C. A..; Corcoran, O.; Hansen, S. H.; Bjornsdottir, I.; Cornett, C.; Connor, S. C.; Lindon, J. C.; Nicholson, J. K. Anal. Chem. 2001, 73, 10841090.
rational drug design strategies, this instrumentation should also see increasing use for identifying novel metabolites in microsomal, hepatocyte, and liver slice screening.10,11 The cryoflow probe will be applicable not only to the hyphenated techniques described herein, but pioneering advances have already incorporated an on-line postcolumn solid-phase extraction system for on-line preconcentration of analytes prior to transfer to the NMR probe (LC SPE-NMR).12 The eluting volumes from (10) Corcoran, O.; Spraul, M.; Hofmann, M.; Ismail, I. M.; Lindon, J. C.; Nicholson, J. K. J. Pharm. Biomed. Anal. 1997, 16, 481-489. (11) Shockcor, J. P.; Silver, I. S.; Wurm, R. M.; Sanderson, P. N.; Farrant, R. D.; Sweatman, B. C.; Lindon, J. C. Xenobiotica 1996, 26, 41-48. (12) Corcoran, O.; Wilkinson, P. S.; Godejohann, M.; Braumann, U.; Hofmann, M.; Spraul, M. Am. Lab. 2002, 34, 18-21. (13) Spraul. M.; Hofmann, M.; Ackermann, M.; Nicholls, A. W.; Damment, S. J. P.; Hasleden, J. N.; Shockcor, J. P.; Nicholson, J. K.; Lindon, J. C. Anal. Commun. 1997, 34, 339-341.
the in-line SPE traps are closely matched to the 40-µL NMR probe to provide an additional 4-fold NMR sensitivity enhancement over that of the 4-fold cryoprobe advantage. Furthermore, the probe can be used for flow injection applications including but not limited to combinatorial library profiling and toxicity profiling.13 This cryoflow probe provides an important addition to the NMR toolbox used in rational drug design strategies. ACKNOWLEDGMENT The authors are grateful to Stephen Grimaldi, Floy Smith, and Peidong Zhao for technical assistance. Received for review October 6, 2002. Accepted January 13, 2003. AC026203I
Analytical Chemistry, Vol. 75, No. 6, March 15, 2003
1551
