Anal. Chem. 2000, 72, 2066-2074
Electrochemically Modulated Preconcentration and Matrix Elimination for Organic Analytes Coupled On-Line with Electrospray Mass Spectrometry Jack R. Pretty,† Haiteng Deng,‡ Douglas E. Goeringer, and Gary J. Van Berkel*
Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6365
Demonstrated for the first time is the use of electrochemically modulated preconcentration and sample matrix elimination combined on-line with electrospray mass spectrometry (EMPM/ES-MS) for the enhanced analysis of organics by ES-MS. EMPM is similar to adsorptive stripping analysis. Accumulation of the targeted analytes at the working electrode of an on-line electrochemical flow cell is accomplished via a nonelectrolytic adsorption process that is controlled through the proper combination of the solvent system, the working electrode material, and applied potential. Once on the electrode, the analyte may be washed free of sample matrix components detrimental to mass spectrometric detection. The potential applied to the electrode during the detection step is chosen to release or strip the analytes unaltered back into the solvent stream for mass spectrometric detection rather than to oxidize or reduce them as would be the case for electrochemical detection. Thus, retention and elution of a target analyte with EMPM are controlled by switching the working electrode potential, rather than via a switch in mobile-phase composition, as is done in more traditional preconcentration and cleanup schemes used online with ES-MS. The proof-of-principle studies described here use the breast cancer drug tamoxifen and a metabolite, 4-hydroxytamoxifen, as the target analytes. A thinlayer, flow-by electrode cell with a glassy carbon working electrode is used as the preconcentration device. The nature of the working electrode, the solvent systems, and the electrode potentials necessary to accumulate and strip tamoxifen and 4-hydroxytamoxifen are discussed. Calibration curves were fitted using the Langmuir isotherm. Detection limits (DLs) using a 5.0 min preconcentration period with selected reaction monitoring for tamoxifen (m/z 372 f 72) were bracketed as 0.010 nM < DL< 0.025 nM. The ability to simultaneously detect low nanomolar levels of both tamoxifen and 4-hydroxytamoxifen in pristine solution and 1/10 diluted urine is also demonstrated. The utility of electrospray mass spectrometry (ES-MS)1 in the analysis of “real world” samples, like many analytical techniques, can be compromised by the nature of the sample matrix. On-line 2066 Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
schemes for separation of targeted analytes from sample matrixes detrimental to ES-MS operation, often with simultaneous preconcentration to maintain or improve concentration detection limits, are used routinely in analytical ES-MS.2 These schemes also minimize manual sample handling, which usually increases sample throughput, decreases cost, and limits exposure of the analyst to potentially hazardous samples. Popular routines make use of a preconcentration device, such as a short chromatographic column packed with a suitable stationary phase, to capture or retain the analytes. Following capture, the analytes are washed free of matrix and subsequently flushed from the column with a different “clean” solvent for downstream mass spectrometric detection. Valving and plumbing are configured so that the interfering matrix components (and nontargeted analytes) are diverted away from the mass spectrometer to “waste” during the retention and washing steps. An analytical separation column is also often included in this setup to separate the various components that may be present in the preconcentrated sample prior to detection. With the approach just outlined, the retentive interaction between the analyte and the stationary phase is altered by a change in mobile phase composition. Our group at Oak Ridge,3-8 among other groups,9-17 has been exploring a promising alterna* Corresponding author. Phone: 423-574-1922. Fax: 423-576-8559. E-mail:
[email protected]. † Current address: Mail Stop C-26, Applied Biology Branch, Division of Biomedical and Behavioral Sciences, National Institute of Occupational Safety and Health, 4676 Columbia Parkway, Cincinnati, Ohio 45226. ‡ Current address: Laboratory for Macromolecular Analysis and Proteomics, Albert Einstein College of Medicine, Bronx, NY 10461. (1) Electrospray Ionization Mass Spectrometry; Cole, R. B., Ed.; Wiley: New York, 1997. (2) Weston, A.; Brown, P. R. HPLC and CE; Academic Press: San Diego, CA, 1997. (3) Zhou, F.; Van Berkel, G. J.; Morton, S. J.; Duckworth, D. C.; Adeniyi, W. K.; Keller, J. M. In Applications of Inductively Coupled Plasma-Mass Spectrometry to Radionuclide Determinations; Report ASTM STP 1291; Morrow, R. W., Crain, J. S., Eds.; American Society for Testing and Materials: West Conshohocken, PA, 1995; pp 82-98. (4) Zhou, F.; Van Berkel, G. J. Anal. Chem. 1995, 67, 3643-3649. (5) Pretty, J. R.; Duckworth, D. C.; Van Berkel, G. J. Anal. Chem. 1997, 69, 3544-3551. (6) Pretty, J. R.; Duckworth, D. C.; Van Berkel, G. J. Anal. Chem. 1998, 70, 1141-1148. (7) Pretty, J. R.; Van Berkel, G. J.; Duckworth, D. C. Int. J. Mass Spectrom. Ion Processes 1998, 178, 51-63. (8) Pretty, J. R.; Van Berkel, G. J. Rapid Commun. Mass Spectrom. 1998, 12, 1644-1652. (9) Long, S. E.; Snook, R. D. Analyst 1983, 108, 1331-1338. (10) Ogaram, D. A.; Snook, R. D. Analyst 1984, 109, 1597-1601. 10.1021/ac990813+ CCC: $19.00
© 2000 American Chemical Society Published on Web 03/17/2000
tive that makes use of an on-line electrochemical flow cell, in place of the chromatographic column in this generic scheme, to effect analyte retention. In the ideal case, the affinity of the analyte for the stationary phase (i.e., an electrically conducting material acting as the working electrode in the cell) is turned on, turned off, or otherwise adjusted (so as to preconcentrate, wash, or separate the targeted species) through application of the appropriate potentials or currents to that phase. Therefore, retention and elution of a target analyte are controlled through the choice of working electrode material and potential, rather than via mobilephase composition. This provides the possibility of liberating the analytes into a solvent matrix ideal for mass spectrometric detection. In the case of ES-MS, this would be a predominantly organic solvent matrix containing a sufficient quantity of an electrolyte that affords control of the electrochemical cell but that does not inhibit gas-phase ion generation.18 This also provides for simpler plumbing and solvent control than does a system requiring a change in mobile phase. Moreover, reequilibration of the stationary phase following elution is not needed, which increases sample throughput. Our approach also maintains or, in some cases, improves concentration detection limits through analyte preconcentration simultaneous with the elimination of problematic sample matrixes.3,8 Electrochemically modulated preconcentration and matrix elimination (EMPM) has previously been coupled on-line with mass spectrometry,3-8,11,12,14-16 including ES-MS,4,8 but the application of those combinations was limited to the analysis of elemental species. The elemental species were deposited in metallic form onto or into (mercury film electrode) the working electrode of the cell by electrolytic reduction. Following deposition and washing, the deposited metals were oxidized, through control of electrode potential, releasing them in ionic form back into the solution for subsequent detection downstream by the mass spectrometer. Such an analytical scheme is akin to anodic stripping analysis with matrix exchange.19 However, with the EMPM/mass spectrometry combination, the analyte is detected as a gas-phase ion with mass-to-charge (m/z) specificity rather than via a measured potential or current at the working electrode owing to oxidation/reduction of the accumulated analyte. As a result, the EMPM/mass spectrometry combination improves mass spectrometric performance, provides highly specific detection (e.g., isotopic analysis is possible), and may also provide lower limits of detection than could be obtained with the traditional electrochemical approach alone. In this paper, we present the proof-of-principle experiments demonstrating the application of the EMPM/ES-MS combination for the enhanced analysis of organic analytes by ES-MS. The (11) Pretty, J. R.; Evans, E. H.; Blubaugh, E. A.; Shen, W. L.; Caruso, J. A.; Davidson, T. M. J. Anal. At. Spectrom. 1990, 5, 437-443. (12) Pretty, J. R.; Evans, E. H.; Blubaugh, E. A.; Caruso, J. A.; Davidson, T. M. J. Anal. At. Spectrom. 1992, 7, 1131-1137. (13) Pretty, J. R.; Caruso, J. A. J. Anal. At. Spectrom. 1993, 8, 545-550. (14) Pretty, J. R.; Blubaugh E. A.; Caruso, J. A. Anal. Chem. 1993, 65, 33963404. (15) Pretty, J. R.; Blubaugh, E. A.; Caruso, J. A.; Davidson, T. M. Anal. Chem. 1994, 66, 1540-1547. (16) Hwang, T. J.; Jiang, S. J. J. Anal. At. Spectrom. 1996, 11, 353-357. (17) Deinhammer, R. S.; Porter, M. D.; Shimazu, K. J. Electroanal. Chem. Interfacial Electrochem. 1995, 387, 35-46. (18) Kebarle, P.; Ho, Y. In Electrospray Ionization Mass Spectrometry; Cole, R. B., Ed.; Wiley: New York, 1997; Chapter 1, pp 3-63. (19) Wang, J. Stripping Analysis; VCH Publishers: Deerfield Beach, FL, 1985.
experiments described here are similar to adsorptive stripping analysis20 in that retention (deposition or accumulation) of the targeted analytes at the working electrode involves a nonelectrolytic adsorption process that is controlled through the proper combination of the solvent system, the working electrode material, and applied potential. Unlike that of traditional adsorptive stripping, however, the potential applied to the electrode during the stripping step is chosen to release the analytes unaltered back into the solvent stream for mass spectrometric detection rather than to oxidize or reduce them as is necessary for electrochemical detection. Electrolytic alterations of organic analytes during deposition or stripping that result in a change in mass are, in most cases, to be avoided. Specific mass spectrometric detection of an analyte is dependent upon measuring the abundance of analyte ions of a characteristic m/z ratio. The studies described here use as model analytes the breast cancer drug tamoxifen (1) and a major metabolite, 4-hydroxytamoxifen (2). Adsorptive stripping
protocols already established for tamoxifen were used as a starting point to optimize the EMPM portion of our experiments.21 A thinlayer, flow-by electrode cell, with a glassy carbon working electrode, was used as the preconcentration device. The nature of the working electrode, the solvent systems, and the applied potentials necessary to accumulate and strip these analytes are discussed. Detection limits with single-ion-monitoring and selectedreaction-monitoring tandem mass spectrometry experiments are determined. The ability to simultaneously detect low levels of 1 and 2 in pristine solution and in diluted urine at low nanomolar levels is also demonstrated. EXPERIMENTAL SECTION Reagents. All solutions were prepared in HPLC grade methanol (J. T. Baker, Phillipsburg, NJ) and deionized water (Milli-RO 12 Plus, Millipore, Bedford, MA). The medium used as carrier solutions in the electrochemical flow system and for most samples was 1/1 (v/v) H2O/CH3OH which contained 5.0 mM ammonium acetate (NH4OAc, 99.999%, Aldrich, Milwaukee, WI) and 0.75% (v/v) double-distilled PPB/Teflon grade acetic acid (HOAc, Aldrich). Stock solutions of 1.0 mM tamoxifen (96%, Aldrich), 4-hydroxytamoxifen (98% minimum, 70% Z isomer, Sigma), toremifene citrate (Z isomer, Orion Pharma International, Helsinki, Finland), clomiphene citrate (Z,E isomer mixture, Sigma), promethazine hydrochloride (Sigma), and chlorpromazine hydrochloride (Sigma) were prepared in CH3OH. Analytical standards were prepared by dilution of stock solution aliquots with the electrolyte solution. Fresh urine from laboratory personnel was (20) Wang, J. In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 1989; Vol. 16, pp 1-88. (21) Wang, J.; Cai, X.; Fernandes, J. R.; Ozsoz, M.; Grant, D. H. Talanta 1997, 45, 273-278.
Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
2067
Figure 1. Schematic of the EMPM/ES-MS setup showing the pump and valve system, thin-layer electrochemical flow cell, and TurboIonSpray ES ion source as well as an expanded view of the cell. Double-arrow lines indicate communication links. Valve 1 (V1) and valve 2 (V2) are both shown in the inject mode (solid-line flow paths within valve), which delivers the analyte sample from the loop through the cell and directly into the ES ion source. Switching V2 to the load mode (dotted-line flow path within valve) is used to divert the solvent effluent from the cell to waste while maintaining a constant flow of carrier solvent/electrolyte to the ES ion source. The solvent flow rate was 30 µL/min. Drawing is not to scale.
spiked as necessary with aliquots of the analyte stock solutions, and the resulting solutions were diluted 10-fold with the electrolyte solution. Instrumentation. A schematic of the instrumental setup is shown in Figure 1. All experiments were performed on either an API165 single-quadrupole or an API365 triple-quadrupole mass spectrometer (PE SCIEX Concord, Ontario, Canada) using a TurboIonSpray source. A 30-cm-long, Teflon-encapsulated fusedsilica (TEFS) transfer tube (75-µm-i.d. fused silica encapsulated in 1/16-in.-o.d. Teflon, CETAC Technologies, Inc., Omaha, NE) connected a 3.5-cm-long stainless steel ES emitter (400-µm o.d., 100-µm i.d.) to a stainless steel 254-µm-i.d. bore-through bulkhead grounding port built into the source assembly. The orifice (60 V) and ring electrode (300-325 V) in the atmospheric sampling interface were set to maximize signal and minimize fragmentation by collision-induced dissociation in this high pressure (∼1 Torr) region. The emitter was held at 4.5 kV, placed 1.5-2.5 cm from the N2 curtain gas plate aperture (1.0 kV), and angled to spray across the aperture. Nitrogen was used for sample nebulization and for the heated cross-flow Turbo gas (typically 8.0 L/min, 200250 °C). Single-ion monitoring was carried out using a 100- or 350-ms dwell time. Full-scan spectra were obtained using a 0.1 m/z step size (1.0-ms dwell time). Single- and multiple-reactionmonitoring tandem mass spectrometry experiments were carried out by monitoring the m/z 372 f 72 and/or m/z 388 f 72 2068
Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
transitions (350 ms dwell time each) for the protonated molecules of 1 and 2, respectively. The laboratory frame-of-reference collision energy was 30 eV with an N2 collision gas thickness of 1.8 × 1015 cm-2. A gas displacement pump (GDP, pressurized with N2) delivered solvent and sample through two microprocessor-actuated sixport PEEK valves (V1 and V2, Microneb 2000, CETAC) and the cell to the ES ion source at a flow rate of 30 µL/min. The cell was connected between the valving using 2-10-cm-long, 75-µm-i.d. TEFS tubes. The outlet of V2 to the ES-MS was an 18-cm-long, 75-µm-i.d. TEFS tube. The sample loop of valve V1 (1.0 mL) was loaded by pulling sample through the loop with a syringe. Valve V2 was used as a switching valve to divert detrimental matrixes flowing through the cell to waste during analyte accumulation and washing. During the stripping portion of the experiment, V2 was switched to allow the cell effluent to enter the ES ion source. With pristine solutions, valve V2 was set to deliver effluent from the cell to the ES source throughout the experiment. A schematic of the thin-layer electrochemical cell used in this work is also shown in Figure 1.8 The working electrode was an offset 6.0-mm-diameter glassy carbon (GC) disk (Bioanalytical Systems, Inc. (BAS), West Lafayette, IN) positioned midway between the reference electrode port (model RE-4 reference electrode, BAS) and outlet port in the assembled cell. The working electrode and counter electrode (7.0-mm-wide Pt foil) were
separated from one another using a 16 µM Teflon spacing gasket (BAS, total cell volume ∼1.1 µL). A PAR model 173 potentiostat and model 175 universal programmer (Princeton Applied Research Corp., Princeton, NJ) were used to step (10 V/s) or scan (5 mV/s) among the various potentials applied to the working electrode during the course of an experiment. During some experiments, potential and current readouts from the potentiostat were directed to the mass spectrometer computer system, and stored there, by means of a dual-channel PE Nelson model 970A A/D interface (Perkin Elmer, San Jose, CA) and PE SCIEX sample control software (Version 1.3). RESULTS AND DISCUSSION Mechanism of Accumulation and Stripping. Typically, adsorptive stripping methodologies using carbon electrodes rely on the hydrophobic nature of analytes to enhance their adsorption onto the electrode.20 Analytes with an extended π-electron structure have a particular affinity for a carbon electrode surface because of the similar structures. For analytes that carry a positive charge in solution, such as those under study here (protonated in acidic solution), accumulation may also be facilitated through electrostatics by application of a potential to the electrode that is negative of the potential of zero charge (PZC).20 A potentiometric adsorptive stripping protocol for tamoxifen reported by Wang and co-workers21 utilized a GC working electrode with 80/20 (v/v) H2O/CH3OH as the solvent system, the small fraction of CH3OH being necessary to dissolve the tamoxifen. This solvent mixture, with the appropriate electrolyte, allowed us to accumulate and retain tamoxifen very efficiently at the GC working electrode when fairly negative potentials (ca. -1.0 V) were applied. However, because of the low solubility of tamoxifen in H2O, cell washout times following accumulation were 5 min or longer and stripping peaks tailed severely. Moreover, the large fraction of H2O suppressed ES generation of tamoxifen gas-phase ions even in continuous-infusion experiments.18 As higher proportions of CH3OH were used, the washout of unretained tamoxifen became more rapid, stripping peaks were sharper, and the gas-phase ion signals improved. However, the overall accumulation was adversely affected because of the increased solubility of tamoxifen in CH3OH. We explored a variety of CH3OH/H2O combinations, with NH4OAc and/or HOAc added as electrolyte. A compromise solvent of CH3OH/H2O (1/1 v/v) containing 5.0 mM NH4OAc and 0.75% HOAc by volume as the solvent for the washing and stripping portions of all experiments was eventually chosen. For comparison, continuous-infusion ES-MS signals for tamoxifen in this compromise solvent were about 20% of those registered when the same concentration was sprayed out of 5/95 (v/v) H2O/CH3OH. The latter is an excellent ES-MS solvent system for tamoxifen but provides no retention of this analyte at the working electrode at any applied potential. Along with the choice of solvent composition, selection of both the accumulation potential and stripping potential was critical to making the EMPM/ES-MS experiment work. This is demonstrated by the experimental data in Figure 2, which explores the fate of the tamoxifen in the cell during the injection, deposition, washing, and stripping stages of the EMPM/ES-MS experiment. Time-synchronized plots of the working electrode potential and corresponding working electrode current (Figure 2a), the mass spectrometric ion current for tamoxifen ((1 + H)+, m/z 372,
Figure 2. Time-synchronized plots of (a) the working electrode potential and corresponding current and (b) the mass spectrometric ion current recorded in the SIM mode (100-ms dwell time for each ion) for the protonated molecule of tamoxifen ((1 + H)+, m/z 372) and (c) the summed ion current for the four major oxidation products of tamoxifen (m/z 299 + 358 + 388 + 402), each recorded over the course of tamoxifen injection, deposition, wash out, and stripping. The potential scan rate was 5.0 mV/s. Note that valve V2 was eliminated from the setup shown in Figure 1 for this experiment and the cell exit was connected directly to the ES source grounding port with a 5-cmlong, 75-µm-i.d. TEFS tube. The cell response time at a solvent flow rate of 30 µL/min was 5.1 s, resulting in about a 25 mV offset between the time registered for a given applied potential (and measured current) and the ion current measured by the mass spectrometer.
Figure 2b) and the summed mass spectrometric ion current for protonated molecules of four of the major oxidation products22 of 1 (Figure 2c) were recorded over the course of the experiment. To better illustrate the phenomena, the potential was altered during the experiment by slow scanning (5.0 mV/s) rather than by potential stepping (10 V/s), which was used in the optimized analytical experiments (see below). The data in Figure 2 were obtained by first adjusting the working electrode potential to 0 V (Figure 2a). At this potential, the tamoxifen was not retained at either the working or the counter electrode. The tamoxifen sample was injected into the cell (30 s), and in less than 1 min, a steady-state mass spectrometric ion signal for (1 + H)+ at m/z 372 was reached (Figure 2b). If the potential of the working electrode was maintained at 0 V, this (1 + H)+ signal remained constant until the sample in the injection loop was exhausted. In the present example, the potential was scanned from 0 to -1.0 V, beginning at 2 min. One notes that, almost immediately following the start of the scan to negative potential, the (1 + H)+ signal began to decrease slightly. The (22) Deng, H.; Van Berkel, G. J. In Proceedings of the 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, June 13-17, 1999; ASMS: Santa Fe, NM, 1999; pp 1781-1782.
Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
2069
tamoxifen signal reached an initial minimum at 4 min (-0.6 V). The signal then began to slowly increase until about 5 min (-0.9 V). At this point, a more significant and sustained drop in the tamoxifen signal began, reaching a constant minimum value at -1.0 V (ca. 5.5 min), the most negative working electrode potential applied. The significant decrease in the tamoxifen signal at potentials more negative than -0.9 V was accompanied by a dramatic increase in the current at the working electrode that also maximized at -1.0 V (Figure 2a). It is most probable that reduction of the solvents or electrolytes is the main source of the increase in current. Similar reductive currents were measured in blank experiments (no tamoxifen injected), supporting this hypothesis. Moreover, no new signals were observed in the ES mass spectra obtained that might be assigned as tamoxifen reduction products. At 6.5 min, the injection of sample was stopped and the unretained tamoxifen was allowed to wash from the cell. This resulted in the continued drop in (1 + H)+ signal to nearly baseline levels from 6.5 to 8.5 min. Much of the drop in (1 + H)+ signal that is observed as the working electrode potential is scanned to negative potentials is attributed to accumulation of the tamoxifen at the working electrode (Figure 2b). This accumulation forms the basis for the preconcentration and matrix elimination experiments. However, the mass spectrometric ion current in Figure 2c illustrates that a portion of the drop in (1 + H)+ signal is due to the oxidation of the tamoxifen. The summed ion signal shown in Figure 2c is that of the four most prominent products of the electrochemical oxidation of tamoxifen21,22 in aqueous CH3OH. The oxidation of tamoxifen that produces these products must occur at the counter electrode because it would be infeasible to oxidize the tamoxifen at the negative potentials applied to the working electrode during this portion of the experiment. The data in Figure 2c also show that the abundances of the oxidation products rapidly rise (5 min) and then fall and plateau (6 min) at potentials not far beyond those that lead to their initial generation. This behavior can be explained if portions of the tamoxifen oxidation products generated at the counter electrode subsequently deposit on the working electrode. In our present cell design, when tamoxifen is introduced to the cell, it has access to the counter electrode prior to the working electrode (Figure 1). Because the counter electrode will be at potential positive of that of the working electrode during accumulation, it appears that it is possible for some fraction of the tamoxifen to be oxidized at the counter electrode before it reaches the working electrode. The tamoxifen oxidation products created at the counter electrode and the remaining tamoxifen then flow over the working electrode, where some fraction of each deposits. The detection or stripping period of this experiment involved a potential scan from -1.0 to 1.2 V that began at 8.5 min (Figure 2a). Release of the tamoxifen accumulated at the working electrode was detected by the mass spectrometer as a broad, lowlevel stripping peak that began at -0.85 V (9 min) and was essentially complete by -0.25 V (ca. 11 min) (Figure 2b). The mass spectrometer also detected ions in the stripping peak corresponding in m/z to oxidized tamoxifen (Figure 2c). These ions are believed to be those tamoxifen oxidation products formed at the counter electrode, during the accumulation step, that deposited at the working electrode. The potential of the working electrode during this part of the stripping period is insufficiently 2070 Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
Figure 3. Plots of mass spectrometer responses for the protonated molecule of tamoxifen ((1 + H)+, m/z 372) recorded in the SIM mode (dwell time 100 ms) during replicate EMPM/ES-MS experiments using 25 nM tamoxifen deposited from and stripped into a solution composed of H2O/CH3OH (1/1 v/v) containing 5.0 mM NH4OAc and 0.75% (v/v) HOAc; 100 ) 4.13 × 105 Hz: (a) expansion of the first replicate in (b) illustrating the timing sequence of the experiment and the various figures of merit used to define the performance of the system (hatch-marked areas were used to calculate percent recovery); (b) (1 + H)+ signals recorded for three replicate experiments demonstrating the rate of sample throughput (∼5 samples/h) and the reproducibility (2.5% RSD) of the stripping peak heights for a 2.5min deposition and a 2.5-min washout. Estimated percent recovery ≈ 38 ( 3%. See text for a detailed description of the experimental figures of merit.
positive to oxidize the deposited tamoxifen and thus form these species. A small peak at m/z 358 (demethylated tamoxifen)22 was observed beyond the stripping peak as the potential was scanned more positive (ca. 15 min) (Figure 2c). Most probably this species was formed by oxidation at the working electrode (ca 1.0 V) of trace amounts of tamoxifen that remained in solution or on the working electrode. Optimization of Accumulation and Stripping. The slow potential scans in the experiment shown in Figure 2 resulted in relatively inefficient tamoxifen accumulation for most of the deposition period and in a broad tamoxifen stripping peak. Stepping from the initial working electrode potential (0 V) to the deposition potential (-1.0 V) increases deposition efficiency. Stepping from the deposition potential to just beyond the potential at which tamoxifen stripping is essentially complete (i.e., 0 V) results in a much sharper stripping peak (Figure 3 below) with no modification of the analyte during stripping. Electrostatics may play a role in sharpening the stripping peak, effectively pushing the positively charged analyte from the electrode at potentials positive of the PZC. However, beyond ca. 0.5 V, the rate of oxidation of tamoxifen on the electrode competes effectively with the rate of release of tamoxifen from the electrode. As a result, a
substantial fraction of the analyte is oxidized. This is the mechanism of detection in electrochemical stripping techniques but is to be avoided with the EMPM/ES-MS coupling. Oxidation in this case alters the mass of the targeted analyte (see above). The results for a representative optimized EMPM/ES-MS experiment are shown in Figure 3. The plot of the (1 + H)+ signal intensity at m/z 372 in Figure 3a is an expansion of the first of the three replicate experiments shown in Figure 3b. A 25 nM tamoxifen standard passed continuously through the cell and on to the mass spectrometer so that the (1 + H)+ signal level could be monitored throughout the experiment. Notice the rapid and efficient deposition of tamoxifen and the sharp transient stripping peak obtained by stepping rather than slowly scanning the working electrode potential appropriately. Using these data, one may define several of the metrics used to evaluate the performance of our method. The maximum continuous-infusion steady-state signal recorded for analyte during sample injection, without a deposition potential applied, is termed CS. The minimum continuous-infusion steady-state signal recorded during sample injection, with a deposition/accumulation potential applied, is termed DS, and the analyte stripping peak height is termed PH. The gain in signal owing to the preconcentration effect is defined as the signal enhancement factor, EF ) PH/CS. In our other work with inorganics,5-8 we judged the efficiency of analyte deposition or accumulation on the basis of percent deposition efficiency defined as DE (%) ) [(CS - DS)/CS] × 100. More meaningful in the present work is the metric percent recovery defined as [(area of the analyte stripping peak/area of analyte signal reduction during accumulation) × 100]. The relevant signal areas used in this calculation are hatch-marked in Figure 3a. As shown above, some of the signal loss during deposition is due to modification of the analyte at the counter electrode. This distorts the value of DS and, therefore, the value of DE (%). Also, analyte initially deposited onto the electrode may be washed off during the wash portion of the experiment. In fact, for a 3.0 min deposition of 50 nM tamoxifen, the stripping peak heights decreased by a total of 45% in an approximately linear fashion for washout times ranging from 5.0 s to 6.0 min. Detrimental effects owing to this phenomenon in analytical determinations are minimized by using short and consistent washout times. Both analyte modification and analyte “washoff” are reflected in the percent recovery. In the present example, the reproducibility of the stripping peak heights was 2.5% RSD, the average EF for the 2.5 min deposition was 4.1 (i.e., ca. 1.6 min-1), and the recovery was 38 ( 3%. Stripping Peak Signals and Signal Enhancement. Plots of mass spectral stripping peak heights versus tamoxifen solution concentration, for deposition times of 2.5, 5.0, 7.5, and 10 min, are shown in Figure 4a. The mass spectrometric stripping peak height (ion current) plotted, iMS, is related to the quantity of tamoxifen on and liberated from the electrode surface during the stripping process, N, by eq 1, where the proportionality constant
iMS ) KMSN
(1)
KMS incorporates mass spectral ionization efficiency, ion-sampling effects, etc. Moreover, N is equal to the product of the effective electrode area A and the electrode surface density of adsorbed
Figure 4. (a) Relative stripping peak heights for protonated tamoxifen ((1 + H)+, m/z 372) plotted versus solution concentration of tamoxifen (1) at deposition times of 2.5, 5.0, 7.5, and 10 min; 100 ) 4.4 × 105 counts/s. (b) Calculated enhancement factors, for the same experimental data, plotted versus deposition time for the various solution concentrations of 1. Solid-line curves in (a) were obtained using least-squares fits of the data to eq 4. Data were recorded in the SIM mode (dwell time 100 ms). Individual data points represent a single analysis. Deposit ) -0.95 V; strip ) 0 V; washout ) 2.5 min.
analyte Γ (eq 2). Because analyte accumulation is via adsorption,
iMS ) KMSAΓ
(2)
Γ, in turn, can be related to analyte solution concentration CA by an adsorption isotherm, the most often used being the Langmuir isotherm given by eq 3, which assumes monolayer coverage of
(
Γ ) ΓM
BACA 1 + BACA
)
(3)
the electrode surface by the analyte.20,23 Here BA is the surface adsorption coefficient of the analyte and ΓM the maximum surface density for the adsorbing analyte species. Substitution for Γ in eq 2 then yields eq 4, which, fitted to the peak height versus
(
iMS ) KMSAΓM
BACA 1 + BACA
)
(4)
concentration data in Figure 4a, produces the solid line curves (23) Trasatti, S. J. Electroanal. Chem. Interfacial Electrochem. 1974, 53, 335363.
Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
2071
Table 1. Experimental Detection Levels (DLs) for Tamoxifena
SIM (m/z 372) SRM (m/z 372 f 72)
ES-MSb
EC/ES-MSc
0.3 nM 0.2 nM
0.025 nM < DL < 0.050 nM 0.010 nM < DL < 0.025 nM
Table 2. Simultaneous EMPM/ES-MS Determinations of Tamoxifen (1) and 4-Hydroxytamoxifen (2) in Pristine Solutions sample 2, 10 nM
Dwell time ) 350 ms for both SIM and SRM. b Detection limit calculated as the mean (3sy/x/slope) from several replicate calibration curves spanning 0-15 nM tamoxifen. c Detection limit range determined using the t test (95%) to determine the sample concentration for which the stripping peak height was greater than the height of the stripping peak for the blank (three replicates for each concentration and for the blank). a
2, 25 nM 1,10 nM 1, 25 nM 1 and 2, 10 nM 1 and 2, 25 nM
shown in that plot. Thus, the nonlinearity exhibited by the EMPM/ES-MS stripping peak analytical curves in Figure 4a is precisely that expected. Furthermore, as seen in the inset plot, the curves are linear at low solution concentrations of the analyte, as predicted from eq 4 (i.e., BACA , 1). The calculated EFs for this same data set, plotted versus tamoxifen solution concentration, are shown in Figure 4b. These data demonstrate that the longer the deposition, the greater the EF. However, except for very lowest tamoxifen concentrations in solution, prolonging preconcentration time beyond 5.0 min, at any one concentration, does not produce a linear improvement in the EF. Also, the EFs are substantially greater at low solution concentrations of the analyte. This observed behavior results again from the fact that the analyte accumulation is an adsorption process. At low analyte solution concentration, the initial slopes of the analytical curves for peak height (PH) versus concentration (Figure 4a) increase in order of increasing deposition time. As solution concentration increases, the slopes of all the PH curves decrease to varying degrees. Therefore, the EFs either remain constant (PH curve essentially linear, 2.5 min deposition) or diminish as solution concentration increases. Detection Levels. The continuous-infusion ES-MS (single-ionmonitoring, SIM) and ES-MS/MS (selected-reaction-monitoring, SRM) detection limits (DLs) for tamoxifen in our electrolyte solvent system were estimated using the mean 3sy/x/slope24 value determined from several replicate calibration curves generated from blanks and tamoxifen standards ranging up to 15 nM. The m/z 372 f 72 transition for (1 + H)+ was monitored in the SRM mode.25 The detection levels obtained in both types of detection modes are listed in Table 1 and show that the SRM level (0.2 nM) is slightly better than the level obtained in SIM mode (0.3 nM). For the EMPM/ES-MS and EMPM/ES MS/MS experiments, the DLs were bracketed by statistical evaluation of the stripping peak heights of replicate blank and three low-level tamoxifen standards using a 5.0 min deposition period (Table 1). The limited number of standards used in these experiments made this method of estimating the DLs more reliable than the regression analysis used above.24 Again, DLs in the SRM mode (0.010 nM < DL < 0.025 nM) were better than those in the SIM mode (0.0250 nM < DL < 0.050 nM) and were about 20-8 and 12-6 times better, (24) Miller, J. C.; Miller, J. N. Statistics for Analytical Chemistry, 2nd ed.; Ellis Horwood Ltd: Chichester, U.K., 1988. (25) Lim, C. K.; Yuan, Z.-X.; Jones, R. M.; White, I. N. H.; Smith, L. L. J. Pharm. Biomed. Anal. 1997, 15, 1335-1342.
2072 Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
1, 10 nM 2, 25 nM 1, 25 nM 2, 10 nM
(1 + H)+, m/z 372 PHa-c
(2 + H)+, m/z 388 PHa,b,d
19 625 (19.4%) 19 267 (2.4%) 488 167 (3.4%) 1 187 667 (2.5%) 408 833 (1.9%) 1 182 333 (5.8%) 473 000e
251 125 (13.8%) 661 833 (8.9%) 25 500 (36.4%) 20 833 (4.5%) 257 500 (5.9%) 505 000 (9.6%) 493 000e
1 111 667 (3.2%)
23 1833 (8.3%)
a PH ) stripping peak height (counts/s). 3.0 min deposition at -1.05 V; 3.0 min washout; strip at 0 V. Data were acquired in the scanning mode; m/z 300-425; dwell time ) 1.0 ms every 0.1 m/z unit. b Numbers in parentheses are RSDs for three replicate analyses. c EFs ranged from 5.1 to 6.7. d EFs ranged from 4.6 to 6.4. e Average value for two replicates.
respectively, than the DLs obtained without the preconcentration. Note also that the EMPM/ES-MS levels of detection for tamoxifen with SIM and SRM are a minimum of 8 and 16 times better, respectively, than those DLs reported for pristine solutions using potentiometric adsorptive stripping analysis (i.e., 0.4 nM).21 Simultaneous Determination of Tamoxifen and 4-Hydroxytamoxifen. Pristine Solution. The potentiometric adsorptive stripping protocol reported by Wang and co-workers21 mentioned, but did not confirm, that simultaneous preconcentration and detection of tamoxifen and its metabolites might be feasible. In clinical analyses, for example, quantification of both parent and metabolites is often desired. The m/z-specific detection afforded by the mass spectrometer facilitates detection of multiple target analytes. We chose to attempt the simultaneous analysis of tamoxifen (1) and 4-hydroxytamoxifen (2), a commercially available metabolite, to test this possibility. The molecular mass of this metabolite (387 Da) raised the issue of distinguishing the protonated molecule from the oxidation product of tamoxifen at m/z 388. An ion at m/z 388 was always observed at low levels in tamoxifen stripping peaks ( 150 µA). Essentially, we operated the deposition part of the experiment with the urine samples in a controlled- or constant-current mode rather than a controlled-potential mode. Switching valve V2 (Figure 1) diverted the urine matrix from the ES-MS detector during deposition and washing. Flow from the cell was diverted back to the mass spectrometer 45 s before application of the stripping potential. Urine samples, diluted 10-fold with our electrolyte solution, but with no other off-line pretreatment, were thus analyzed. The data in Figure 5 were acquired in the analysis of the diluted urine samples containing various but equal concentrations of both tamoxifen and 4-hydroxytamoxifen. Multiple-reaction-monitoring tandem mass spectrometry was used as the detection mode (MRM, m/z 372 f 72 and m/z 388 f 72, respectively, for tamoxifen and 4-hydroxytamoxifen). Figure 5a shows the stripping peak response for (1 + H)+ and Figure 5b the response for (2 + H)+ determined in four replicate runs with a sample containing 5.0 nM of both analytes. Note that little signal was observed between stripping peaks because the sample exiting the cell was diverted to waste during these periods while clean carrier solution was supplied to the ES source. A slight decline in the peak heights for both analytes was noticed over the course of several replicate runs (Figure 5a,b). This suggests a moderate degree of electrode fouling. Nonetheless, plots of blank-corrected peak height versus analyte concentration for this data set, shown in Figure 5c, were fitted well by the isotherm in eq 4. The signal levels in the case of tamoxifen indicate that it certainly could be detected at levels lower than 5.0 nM in the diluted urine. It is interesting that the response of 4-hydroxytamoxifen is much less than expected on the basis of the work with the pristine standards (Table 2). Apparently, some component of the urine matrix is affecting accumulation or the use of the more negative deposition potentials resulted in modification (oxidation) of this species at the counter electrode. 4-Hydroxytamoxifen is, in fact, significantly easier to oxidize than tamoxifen (0.75 V vs 1.1 V peak potential).22
Figure 5. Relative stripping peak heights for (a) tamoxifen (m/z 372 f 72) and (b) 4-hydroxytamoxifen (m/z 388 f 72) recorded for four replicate analyses involving the simultaneous determination of 5.0 nM of each analyte in 1/10 diluted urine. Both plots were normalized to the maximum stripping peak height in (a); 100 ) 2348 counts/s. Part c shows plots of blank-subtracted relative stripping peak heights for tamoxifen (b) and 4-hydroxytamoxifen (1) in 1/10 diluted urine at concentrations of 5, 10, and 20 nM; 100 ) 5894 counts/s. All data were recorded in the MRM mode (dwell time 350 ms per transition). Deposit ) 3.0 min at -1.4 V; washout ) 3.25 min at -1.25 V; strip ) 0 V. Solid-line curves were obtained using least-squares fits of the data to eq 4. Error bars represent (1 standard deviation for four replicate analyses.
CONCLUSIONS The proof-of-principle results presented in this paper demonstrate the successful on-line coupling of electrochemically modulated preconcentration and matrix elimination for organic analytes with ES-MS and ES-MS/MS detection. This hybrid methodology is potentially an alternative approach to traditional on-line sample cleanup coupled with mass spectrometric detection. Retention and elution of a target analyte are controlled through the choice of the potential (or current) at the stationary phase in the preconcentration device rather than via a change in mobile-phase composition. In the ideal case, which has not yet been fully Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
2073
realized, this provides the possibility of liberating the analytes for detection into a solvent matrix ideal for mass spectrometric detection, with simpler plumbing and solvent control, and eliminates the time required for reequilibration of the stationary phase following elution. In the present example, the EMPM/ES-MS/MS concentration detection limits for tamoxifen using selected reaction monitoring ranged from 10 to 25 pM in the pristine solution. Detection levels in diluted urine were indicated to be in the low-nanomolar range. Our results suggest that it might be possible to use the EMPM/ ES-MS/MS combination to determine both tamoxifen and 4-hydroxytamoxifen in diluted urine at low-nanomolar levels. It should be pointed out that the 5-7 min analyte solution that actually flowed through the EMPM system (injection time plus deposition time of 3 or 5 min) consumed only 150-210 µL of total sample volume. In the case of the original urine, which was diluted 1/10 before analysis, this translates to just 15-21 µL of urine consumed per analysis. Additionally, cursory studies found that the same protocols applied to the analysis of tamoxifen and the metabolite are successfully applied to the analysis of the other triphenylethylene-based drugs clomifen and toremifen as well as the phenothiazene-based drugs promethazine and chlorpromazine. Thus, the possibility exists for selective analysis of other analytes that have the appropriate solubility and structural characteristics without much modification of the current approach. In fact, starting-point protocols for the application of EMPM/ES-MS to the analysis of a multitude of different organic and biological analytes might be obtained from the extensive adsorptive stripping analysis literature.20 Improvements in the current EMPM setup and further experimental optimization will be needed to make this samplehandling approach competitive with the more traditional preconcentration and cleanup methods used on-line with ES-MS. A major modification needed is the removal of the counter electrode from direct contact with the analyte solution. This will eliminate the problem of modifying the analyte during deposition (improving percent recovery) and the deposition of these modified analytes (eliminating false positive responses). Until surfaces can be tailored with greater specificity so that the solvent system has (26) Mohr, P.; Pommerening, K. Affinity Chromatography; Marcel Dekker: New York, 1985. (27) Kriz, D.; Ramstro¨m, O.; Mosbach, K. Anal. Chem. 1997, 69, 345A-349A. (28) Henry, C. Anal. Chem. 1997, 69, 359A-361A. (29) Freemantle, M. Chem. Eng. News 1999, 77 (Feb 22), 27-36.
2074
Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
less effect on analyte accumulation (see below), switching media during the strip, although more complicated, might be used to obtain superior peak heights, enhancement factors, and faster washout times than illustrated under the conditions used in the work presented here. For example, several trials injecting 95/5 (v/v) CH3OH/H2O with added electrolyte in synchronization with application of the stripping potential resulted in sharper, more intense tamoxifen stripping peaks than those generated without the solvent change. We anticipate that our EMPM/ES-MS approach, using electrodes appropriately modified for enhanced analyte selectivity, might be viewed and used in the same way as state-of-the-art bioaffinity-based separations coupled on-line with mass spectrometry for the selective isolation, separation, and highly specific detection of targeted analytes.26 A hoped for advantage over traditional bioaffinity-based separations would be increased robustness and infinite regeneration of the binding surface, i.e., the working electrode. This might be achieved by molecular imprinting of polymeric electrode surfaces.27 Another advantage might be the ease of miniaturization of this electrode-based system, as witnessed by the proliferation of miniature electrochemical sensor technology,27 making this particular approach readily transferable to lab-on-a-chip or micro-total analysis systems (µ-TAS)28,29 that might be coupled on-line with mass spectrometers. ACKNOWLEDGMENT J.R.P. and H.D. acknowledge support through an appointment to the Oak Ridge National Laboratory (ORNL) Postdoctoral Research Associates Program administered jointly by the Oak Ridge Institute for Science and Education and ORNL. Prof. J. Martin E. Quirke (Florida International University) is thanked for helpful discussions and for critical review of the original manuscript. The ES-MS instrumentation used in the work was provided through a Cooperative Research and Development Agreement with Perkin-Elmer SCIEX Instruments (CRADA No. ORNL960458). This research was sponsored by the National Cancer Institute under Interagency Agreements DOE 0485-F053-A1 and NCI Y1-CB-0016-01. ORNL is managed by the Lockheed Martin Energy Research Corp. Received for review July 22, 1999. Accepted January 29, 2000. AC990813+
