Anal. Chem. 2002, 74, 5290-5296
Determination of 4-Hydroxy-3-methoxyphenylethylene Glycol 4-Sulfate in Human Urine Using Liquid Chromatography-Tandem Mass Spectrometry Peyton Jacob, III,* Margaret Wilson, Lisa Yu, John Mendelson, and Reese T. Jones
Division of Clinical Pharmacology, Building 100, Room 235, San Francisco General Hospital Medical Center, Department of Medicine, and Drug Dependence Research Center, Department of Psychiatry, University of California, San Francisco, California 94110
A major metabolite of norepinephrine (NE) in brain is 4-hydroxy-3-methoxyphenylethylene glycol (MHPG). In many species, a large fraction of MHPG formed in brain is converted to the sulfate conjugate. Consequently, MHPG sulfate has been proposed as a biomarker for NE metabolism in the central nervous system. As part of the clinical trials of the monoamine oxidase inhibitor selegiline for treating cocaine addiction, we required a method for measuring urine concentrations of MHPG sulfate. Using a deuterium-labeled analogue as an internal standard, we developed a liquid chromatography-electrospray ionization tandem mass spectrometry (LC-MS/ MS) method for determination of MHPG sulfate in human urine. Sample preparation involves simply diluting 50 µL of urine with 1 mL of ammonium formate buffer and adding the internal standard. The sample is centrifuged, the supernate is transferred to an autosampler vial, and 10 µL is injected into the LC-MS/MS system. Standard curves from 50 to 10 000 ng/mL are generated. Only one sample of 277 clinical samples analyzed had a concentration outside of this range. Precision (coefficient of variation) ranged from 1.9 to 9.7%, and accuracy ranged from 97 to 103% of expected values for controls prepared by spiking sulfatase-treated urine with MHPG sulfate. A major metabolite of norepinephrine (NE) in brain is 4-hydroxy-3-methoxyphenylethylene glycol (MHPG; Figure 1). In many species, a large fraction of MHPG formed in the brain is converted to the sulfate conjugate. Consequently, determination of MHPG sulfate in urine has been proposed as a measure of norepinephrine metabolism in the cental nervous system.1,2 Since one of the steps in the biotransformation of NE to MHPG is catalyzed by monoamine oxidases (MAO), inhibitors of MAO might decrease the excretion of MHPG and its sulfate conjugate by shunting metabolism to other pathways.3 In contrast, drugs that release norepinephrine from neuronal stores or block its * To whom correspondence should be addressed. E-mail:
[email protected], Phone: (415) 282-9495. Fax: (415) 206-5080. (1) Peyrin, L. J. Neural Transm. Gen. Sect. 1990, 80, 51-65. (2) Filser, J. G.; Spira, J.; Fischer, M.; Gattaz, W. F.; Muller, W. E. J. Psychiatr. Res. 1988, 22, 171-81. (3) Nagatsu, T. Biochemistry of Catecholamines; University Park Press: Baltimore, MD, 1973.
5290 Analytical Chemistry, Vol. 74, No. 20, October 15, 2002
reuptake into nerve terminals might be expected to increase norepinephrine turnover and increase production of MHPG. As part of the clinical trials of the MAO inhibitor selegiline (levodeprenyl) for treating cocaine addiction, we required a method for measuring concentrations of MHPG sulfate in human urine. Various methods for determination of MHPG sulfate have been reported, including HPLC,4,5 GC,6 and GC/MS7 methods. All of the methods that we are aware of are indirect in that they involve determination of MHPG released by cleaving the sulfate conjugate, either chemically or enzymatically. This is timeconsuming, because free MHPG and the glucuronide conjugate of MHPG are also present in urine. Therefore, in these methods, the sulfate must either be separated from free MHPG and its glucuronide or two determinations must be carried out on each sample: one for free MHPG and one following enzymatic cleavage of the sulfate to MHPG using an arylsulfatase. In the latter case, concentrations of MHPG sulfate are calculated as the difference of the two determinations,6 and in the former case, separation using a selective solid-phase extraction procedure followed by cleavage of the sulfate, either chemically or enzymatically, is generally used.5,7 The advent of electrospray ionization (ESI) mass spectrometry has made possible determination of low concentrations of a wide range of readily ionizable, thermally labile substances in complex matrixes. We found that MHPG sulfate can be detected with high sensitivity using ESI-MS in the negative ion mode. Using a deuterium-labeled analogue of MHPG sulfate as the internal standard, we developed a liquid chromatography-tandem mass spectrometry (LC-MS/MS) method for determination of MHPG sulfate in urine. The method is extremely simple to carry out and appears to be the first reported truly direct method for quantitation of MHPG sulfate in a biologic fluid. EXPERIMENTAL SECTION Apparatus. LC-MS and LC-MS/MS analyses were carried out with a Hewlett-Packard (Palo Alto, CA) 1090 HPLC interfaced with a Finnigan TSQ 7000 triple-stage quadrupole mass spectrom(4) Muller, H. U.; Riemann, D.; Berger, M.; Muller, W. E. Acta Psychiatr. Scand. 1993, 88, 16-20. (5) Inagaki, H.; Hirata, K.; Minami, M. J. Chromatogr., B: Biomed. Sci. Appl. 1998, 707, 9-15. (6) Piletz, J. E.; Halaris, A. Clin. Chim. Acta 1989, 185, 165-76. (7) Murray, S.; Baillie, T. A.; Davies, D. S. J. Chromatogr. 1977, 143, 541-51. 10.1021/ac020101a CCC: $22.00
© 2002 American Chemical Society Published on Web 09/17/2002
Figure 1. Norepinephrine metabolic pathways. Abbreviations: MAO, monoamine oxidase; COMT, catecholamine-O-methyltransferase.
Figure 2. Synthesis of MHPG-d3 sulfate.
eter with an API2 ion source (Thermo-Finnigan, San Jose, CA). A Phenomenex (Torrence, CA) Synergi Polar column, 250 mm × 2 mm, with an Upchurch (Oak Harbor, WA) 2-mm C-18 guard column, was used for the chromatography. Reagents. MHPG sulfate potassium salt and arylsulfatase type VI from Aerobacter aerogenes were purchased from Sigma Chemical Co. (St Louis, MO). The HPLC mobile phase was prepared from HPLC grade water, HPLC grade methanol, and formic acid (ACS reagent grade) obtained from Fisher Chemical Co. (Pittsburgh, PA). Anhydrous tetrahydrofuran (THF) was obtained from Aldrich Chemical Co. (Milwaukee, WI). Sulfur trioxide pyridine complex was obtained from Acros Organics/Fisher Scientific (Pittsburgh, PA). The internal standard, MHPG-d3 sulfate, was synthesized as described below.
Synthesis of 4-Hydroxy-3-trideuteriomethoxyphenylethylene Glycol 4-Sulfate Potassium Salt (MHPG-d3 Sulfate). MHPG-d3 was synthesized by the method of Markey et al.8 (Figure 2). The crude product was used without the further purification (silica gel chromatography) described by the authors. A magnetically stirred solution of 90 mg (0.5 mmoL) of MHPG-d3 in 4 mL of dry THF, under an atmosphere of argon, was cooled in a dry ice-acetone bath. A 1 M solution of potassium tert-butoxide in THF (0.5 mL, 0.5 mmol) was added, which caused the mixture to turn cloudy. Sulfur trioxide pyridine complex (80 mg, 0.5 mmol) was added in one portion, and the mixture was stirred (8) Markey, S. P.; Powers, K.; Dubinsky, D.; Kopin, I. J. J. Labelled Compd. Radiopharm. 1980, XVII, 103-14.
Analytical Chemistry, Vol. 74, No. 20, October 15, 2002
5291
for a few minutes with cooling in the dry ice-acetone bath. The cooling bath was removed, and the mixture was stirred while warming to room temperature, during which time a granular white precipitate formed. After stirring 30 min at room temperature, 5 mL of anyhdrous ether was added, and the mixture was centrifuged. The supernate was removed, and the precipitate was washed twice with 2-mL portions of ether. The product was then dried with a stream of nitrogen at ∼50 °C to give 98 mg of white powder. LC-MS analysis (ESI, negative ion mode) confirmed the identity of the product: m/z 266, M - K. LC-MS analysis in the APCI mode indicated that the product was ∼50% pure, the major impurity being vanillomandelic-d3 acid, a synthetic precursor of MHPG-d3. From extracted ion chromatograms, it was estimated that the product contained less than 0.2% MHPG-d0 sulfate. The product was stored in a desiccator in a freezer, since a sample turned brown after a few months at room temperature. Cleavage of MHPG Sulfate in Urine with Arylsulfatase. The pH of a pooled urine specimen (36 mL) from four people was adjusted to 7 with 4 mL of 1 M sodium acetate followed by concentrated sodium hydroxide added dropwise. To this was added 1.05 mL (18 units) of arylsulfatase type VI from A. aerogenes, and the mixture was incubated at 37 °C for 20 h in a stoppered flask. The flask was then heated in a boiling water bath for 30 min to inactivate the enzyme. Absence of MHPG sulfate was verified by LC-MS/MS analysis (see below). Preparation of Standards and Controls. Standards were prepared by diluting a stock aqueous solution of MHPG sulfate with HPLC grade water to concentrations of 50, 100, 200, 500, 1000, 2000, 5000, and 10 000 ng/mL. Controls were prepared from (1) the pooled urine specimen that was treated with arylsulfatase (see above), (2) the pooled urine specimen spiked with MHPG sulfate, or (3) the pooled urine specimen diluted with HPLC grade water. Sample Preparation. Fifty microliters of urine sample, standard, or control was added to 1 mL of 0.1% formic acid in 10 mM ammonium formate. The internal standard, MHPG-d3 sulfate, ∼100 ng (assuming the preparation was ∼50% pure) in 100 µL of water, was added. The solution was briefly mixed, and centrifuged, and the supernate was transferred to an autosampler vial for LCMS/MS analysis. Liquid Chromatography. The instrument was operated in the isocratic mode, with a flow rate of 0.2 mL/min at ambient temperature. The mobile phase consisted of 95% 0.1% aqueous formic acid and 5% methanol, degassed by sparging with helium. The injection volume was 10 µL. A divert valve was programmed to shunt the eluate to waste prior to and following elution of the analyte. Mass Spectrometry. The electrospray voltage was 4.5 kV, and the heated capillary was set at 250 °C. The sheath gas and auxiliary gas (both nitrogen) were set at 80 psi and 40 arbitrary units, respectively. For collision-induced dissociation (CID), the collision gas (argon) pressure in the second (rf-only) quadrupole was set at ∼2.5 mTorr. The resolution was set at 0.6 amu. Quantitative analyses were carried out using selected reaction monitoring (SRM). For quantitation, the SRM transitions monitored were m/z 263 (M - H) to 150 for MHPG sulfate and m/z 266 (M - H) to 150 for the internal standard, using a collision energy of 30 eV. The transitions m/z 263 (M - H) to 165 for 5292
Analytical Chemistry, Vol. 74, No. 20, October 15, 2002
MHPG sulfate and m/z 266 (M - H) to 168 for the internal standard, using a collision energy of 25 eV, were monitored for confirmation. Data Analysis. Finnigan XCalibur/LC Quan software was used to generate calibration curves (linear regression, equal weighting) and calculate concentrations using peak area ratios of analyte/ internal standard. Standard curves were generated from 0 to 2000 ng/mL and from 0 to 10 000 ng/mL. The latter range was used for concentrations above 2000 ng/mL. Clinical Study. The protocol involved 12 human volunteers, all of whom had previously used cocaine but were not cocaine dependent, studied for 14 days on a research ward at the University of California, San Francisco. Urine was collected in 24-h blocks. On the first day, the subjects were given an intravenous infusion of cocaine, 2.5 mg/kg over 4 h. On the fourth day through the thirteenth day, the subjects were administered transdermal selegiline, one STS patch (20 mg) per day. On the tenth day, subjects were again administered cocaine by intravenous infusion, 2.5 mg/kg over 4 h. Urine collection continued for 3 days following this second infusion. RESULTS AND DISCUSSION API Mass Spectrometry of MHPG Sulfate. Initially, analysis using atmospheric pressure chemical ionization (APCI) was attempted. This was done because APCI is relatively insensitive to suppression of ionization by substances derived from the sample matrix.9 The APCI spectrum (negative ion mode) of MHPG sulfate is shown in Figure 3. The relative abundance of the pseudomolecular ion (M - H)-, m/z 263, was only 3% and the base peak in the spectrum was m/z 245, resulting from loss of water. Collisioninduced dissociation of m/z 245 resulted in the formation of two major product ions, m/z 165 and 150 (Figure 3). Using the transitions of m/z 245 to 165 and m/z 245 to 150, an attempt was made to develop a quantitative method for determination of MHPG sulfate, but sensitivity was inadequate to measure the concentrations found in human urine, which have been reported to average ∼500 ng/mL.1,2 This lack of sensitivity was presumably due to extensive decomposition of MHPG sulfate in the heated vaporizer of the APCI source. Although more prone to suppression of ionization by substances derived from the sample matrix than is APCI, ESI is often applicable to thermally labile, ionic substances such as MHPG sulfate. Indeed, ESI has been used for mass spectral determination of low concentrations of steroidal sulfates.10 In the ESI mode, the negative ion spectrum of MHPG sulfate displayed one major ion, the pseudomolecular ion m/z 263. CID of this ion produced two major ions, m/z 165 and 150, the same two product ions observed from CID of the APCI-generated precursor ion m/z 245 (Figure 4). The m/z 165 ion presumably results from loss of SO3 and H2O and m/z 150 from additional loss of the methyl group (Figure 5). Using ESI and operating the mass spectrometer in the SRM mode, experiments were conducted to determine whether adequate sensitivity could be obtained for the determination of MHPG sulfate in human urine. Injection of low-picogram amounts produced a peak significantly above noise level. Consequently, (9) Matuszewski, B. K.; Constanzer, M. L.; Chavez-Eng, C. M. Anal. Chem. 1998, 70, 882-9. (10) Murphy, R. C.; Fiedler, J.; Hevko, J. Chem. Rev. 2001, 101, 479-526.
Figure 3. APCI mass spectra of MHPG sulfate. Figure 4. ESI mass spectra of MHPG sulfate.
efforts were directed toward developing a quantitative method using ESI. Sample Preparation. Initially, solid-phase anion exchange extraction procedures were explored, since the high polarity of MHPG sulfate suggested that liquid-liquid extraction would result in poor recoveries and that sample cleanup might be necessary to remove matrix impurities that could interfere directly or cause suppression of ionization. Although these experiments were encouraging, elution of MHPG sulfate from anion exchange columns required solvents with a high ionic strength containing nonvolatile salts, and such eluates would not be desirable for injection into an LC-MS system. It occurred to us that if there was sufficient retention of MHPG sulfate on the HPLC column, it might be possible to dilute urine with HPLC mobile phase, inject without extraction, and achieve sufficient separation from matrix impurities for quantitation. Using a Phenomenex Synergi Polar column, which contains a phenoxypropylsilane group to facilitate retention of polar compounds, MHPG sulfate was well retained and eluted with good peak symmetry. A mobile phase of 10 mM ammonium formate in 95:5 water/methanol was used, and the mass spectrometer was operated in the SRM mode, monitoring the transitions m/z 263 to 165 and m/z 263 to 150. Injecting a pooled (four persons) urine specimen that had been diluted with mobile phase gave a peak with the same retention time as an MHPG sulfate standard. The ratio of intensities of the two transitions was ∼0.5 for both the urine sample and the standard. Furthermore, injection of a sample of the pooled urine that had been treated with arylsulfatase in order to hydrolyze MHPG sulfate resulted in a chromatogram that
had only baseline noise at the retention time of MHPG sulfate. These results indicated that MHPG sulfate was being detected without significant interference. To investigate the possibility of matrix suppression of ionization, the arylsulfatase-treated urine was heated to ∼100 °C to inactivate the enzyme and then spiked with 1000 ng/mL MHPG sulfate. The sample was analyzed, and the peak area was compared to the peak area of an aqueous standard containing the same concentration of MHPG sulfate. This experiment indicated that the matrix suppressed ionization by ∼50%. Synthesis of MHPG-d3 Sulfate. The use of a stable isotopelabeled internal standard is generally advantageous in any quantitative mass spectrometric method, but in the present case, it was considered to be essential in order to correct for matrix effects. Coeluting with the analyte and having essentially the same ionization characteristics as the analyte, the internal standard should undergo the same degree of supression of ionization. Therefore, the ratio of signal intensities for analyte/internal standard should be independent of the degree of ionization suppression (which might vary among different samples and standards) and provide a reliable basis for quantitation. Consequently, we undertook the synthesis of a deuterium-labeled analogue of MHPG sulfate. MHPG sulfate, deuterium-labeled on the side chain, has been synthesized previously, in eight steps, and used in a GC/MS method for determination of MHPG sulfate in urine.7 A considerably shorter synthesis of a deuterium-labeled analogue appeared to be possible based on the synthesis of methoxy-labeled MHPG reported by Markey et al.8 We found that, indeed, methoxy-labeled Analytical Chemistry, Vol. 74, No. 20, October 15, 2002
5293
Figure 5. Proposed fragmentation pathways for MHPG sulfate.
Figure 6. ESI mass spectra of MHPG-d3 sulfate.
MHPG prepared by this method could be readily converted to the sulfate ester by treatment with potassium tert-butoxide followed by sulfur trioxide pyridine complex (Figure 2). The labeled analogue was characterized by its ESI-MS and MS/MS spectra (Figure 6). The MS/MS spectrum confirms that the m/z 150 ion results from loss of the methyl group, since m/z 150 is formed from both MHPG-d0 sulfate and MHPG-d3 sulfate (Figure 4). Although the product was obtained in only ∼50% purity (a major impurity being vanillomandelic-d3 acid, one of the synthetic intermediates), it was suitable for use as an internal standard, since there was no significant formation of m/z 263, the pseudomolecular ion of the analyte, formed under ESI conditions. 5294 Analytical Chemistry, Vol. 74, No. 20, October 15, 2002
Matrix Suppression of Ionization: Effect of HPLC Mobile Phase. Using a 2 mm × 250 mm Phenomenex Synergi column with a mobile phase consisting of 10 mM ammonium formate in 95:5 water/methanol, suppression of ionization was ∼50%, based on the peak area for the internal standard in urine specimens compared to that in aqueous standards. A proposed mechanism for matrix effects is competition between the analyte and other ionic species during ion evaporation.11 It was reasoned that, if this were the case, it might be possible to diminish the matrix effect by lowering the pH of the HPLC mobile phase. At the pH of ammonium formate (∼6), a wide range of weak acids, likely to be present in urine, would be expected to be ionized. Lowering the pH should convert many weak acids to neutral species. But, since MHPG sulfate is a fairly strong acid, it should remain ionized at pH values of ∼2-3, a pH that could be tolerated by reversedphase HPLC columns and would protonate the anions derived from many weak acids. To test this hypothesis, a mobile phase consisting of 0.1% formic acid in 95:5 water/methanol was employed, and the internal standard peak area for aqueous standards was compared to the internal standard peak area for urine samples. Under these conditions, the matrix effect was virtually eliminated! Comparing peak areas for the internal standard in chromatograms from clinical study samples with those in aqueous standards, there was only ∼7% suppression of ionization. However, from this result we cannot say with certainty that lowering the pH per se was the mechanism for elimination of the matrix effect: the retention time for the analyte was lengthened compared to that using the ammonium formate mobile phase, and more effective separation from matrix components is another possible mechanism. Regardless of the mechanism, this result suggested that use of the formic acid-containing mobile phase would be advantageous. Method Validation. SRM chromatograms from a human urine specimen and an arylsulfatase-treated pooled (four persons) urine are shown in Figure 7. For the arylsulfatase-treated urine, the SRM chromatogram of the transitions m/z 263 to 165 and m/z 263 to 150 (MHPG-d0 sulfate) revealed only background noise at the retention time of MHPG sulfate, as did the m/z 266 to 150 (11) Niessen, W. M. J. Chromatogr., A 1999, 856, 179-97.
Table 1. Intraday Precision and Accuracy actual concn (ng/mL)
measd mean (ng/mL)
0b 50c 100c d 1000c e 10000c
8.5 56.9 107 245 1005 1722 9953
accuracya (%)
precision (% CV)
replicate analyses
97 99
4.7 6.4 2.4 2.9 4.4 3.2
2 6 6 6 6 6 6
100 99
a Determined from blank-corrected mean. b Arylsulfatase-treated pooled (n ) 4) urine. c Arylsulfatase-treated pooled urine spiked with MHPG sulfate. d Pooled (n ) 4) urine diluted 1/5 with water. e Pooled urine spiked with 500 ng/mL MPHG sulfate.
Table 2. Interday Precision and Accuracy actual concn (ng/mL)
measd mean (ng/mL)
0b 100c d 10002 c 5000c
6.4 109 257 991 1776 4953
accuracya (%)
precision (% CV)
replicate analyses
103
9.7 3.4 1.9 3.6 5.6
6 6 6 6 6 4
99 99
a Determined from blank-corrected mean. b Arylsulfatase-treated pooled (n ) 4) urine. c Arylsulfatase-treated pooled urine spiked with MHPG sulfate. d Pooled (n ) 4) urine diluted 1/5 with water. e Pooled urine spiked with 500 ng/mL MPHG sulfate.
Figure 7. SRM chromatograms of untreated urine (upper panel) and arylsulfatase-treated urine (lower panel). No internal standard added.
transition at the retention time of the internal standard, MHPGd3 sulfate. However, there was a significant peak at the retention time of MHPG sulfate in the SRM chromatogram for m/z 266 to 168 transition of the internal standard in both the sulfatase-treated and untreated urine specimens (Figure 7). Therefore, the transitions m/z 263 to 150 for MHPG sulfate and m/z 266 to 150 for the internal standard were used for quantitation. Standard curves were generated from 0 to 2000 ng/mL and from 0 to 10 000 ng/mL, using aqueous standards. The latter range was used for concentrations above 2000 ng/mL. Equations for typical standard curves are Y ) -0.0115 + 0.00107X, r2 ) 0.9972 (0-2000 ng/mL); Y ) 0.0255 + 0.000981X, r2 ) 0.9992 (0-10 000 ng/mL). The concentrations of the standards and the back-calculated values (in parentheses) for this run were as follows (in ng/mL): 50 (52.2, 64.5); 100 (108, 111); 200 (212, 195); 500 (518, 452); 1000 (943, 971); 2000 (2090, 1960); 5000 (4950, 4800); 10 000 (10 100, 9950). To determine precision and accuracy, sulfatase-treated urine was spiked with concentrations of MHPG sulfate (following heat inactivation of the enzyme) that spanned the expected range, and replicate samples were analyzed. In addition, a urine pool obtained from four individuals was analyzed as such, diluted 1:2 with water, 1:5 with water, and spiked with 500 ng/mL. Six replicate analyses were carried out on all urine specimens. Intraday precision
(percent coefficient of variation) ranged from 2.4 to 4.7%, and intraday accuracy (percent of expected values) ranged from 97 to 100% (Table 1). Between-run (4 or 6 days) precision and accuracy was determined from QC samples analyzed along with clinical study samples (Table 2). Precision ranged from 1.9 to 9.7%, and accuracy ranged from 99 to 103%. The concentrations used for the QC samples were chosen to span the expected range based on published studies,1,2 and the results of our clinical study, discussed below, confirmed that this was the case. Clinical Study. The method was used to determine MHPG sulfate concentrations in 277 urine specimens from a clinical study of the MAO inhibitor selegiline (levodeprenyl), which is being investigated as a pharmacotherapy for treating cocaine dependence. Concentrations ranged from
