Chip-Based Solid-Phase Extraction Pretreatment for Direct

Frank A. Kero , Dana B. Barr , Benjamin C. Blount , Douglas B. Mawhinney .... Dieudonne A. Mair , Emil Geiger , Albert P. Pisano , Jean M. J. Fr?chet ...
0 downloads 0 Views 169KB Size
Download PDF
Anal. Chem. 2003, 75, 5504-5511

Chip-Based Solid-Phase Extraction Pretreatment for Direct Electrospray Mass Spectrometry Analysis Using an Array of Monolithic Columns in a Polymeric Substrate Aimin Tan, Salete Benetton, and Jack D. Henion*

Analytical Toxicology, New York State College of Veterinary Medicine, Cornell University, 927 Warren Drive, Ithaca, New York 14850

An array of eight porous monolithic columns, prepared in a Zeonor polymeric chip by UV-initiated polymerization of butyl methacrylate and ethylene dimethacrylate, was tested for solid-phase extraction (SPE) cleanup of biological samples prior to directly coupled electrospray mass spectrometry (ESI-MS). The chip, fabricated by hot embossing and thermal bonding, consists of eight parallel channels (10 mm long, 360 µm i.d.) connected via external fused-silica capillaries. The monomer mixture was aspirated simultaneously into the eight channels using a homemade vacuum manifold device and polymerized in parallel for 20 min under UV irradiation. The porous monolithic columns were then characterized by scanning electron microscopy and evaluated by ESI-MS applications with respect to sample capacity, recovery, reproducibility of peak area or peak height ratios, and linearity between peak height ratio and concentration using imipramine as a pharmaceutical test compound. The average sample capacity was estimated to be 0.30 µg with a relative standard deviation (RSD) of 26.5% for the eight monolithic columns on the same polymeric chip. For two chips prepared using the same monomer mixture, the difference in average sample capacity was 7.0%. The average recovery for the eight monolithic SPE columns on the same chip was 79.1% with an RSD of 7.9%. Using imipramine-d3 as an internal standard, the RSD of peak height ratios for the eight different columns was 2.0% for a standard solution containing 1 µg/mL imipramine. A linear calibration curve (R2 ) 0.9995) was obtained for standard aqueous solutions of imipramine in the range from 0.025 to 10 µg/mL. To demonstrate the analytical potential of the chip-based SPE system, two different types of real-world samples including human urine sample and P450 drug metabolism incubation mixture were tested. Similar to standard aqueous solution, a linear correlation (R2 ) 0.9995) was also found for human urine sample spiked with imipramine in the range of 0.025-10 µg/ mL. When aliquots of a human urine sample spiked with 1 µg/mL imipramine were loaded onto eight different monolithic columns, the RSD of peak height ratios was * Corresponding author: (e-mail) [email protected]; (phone) 607-253-3971; (fax) 607-253-3973.

5504 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

3.8%. For a P450-imipramine incubation mixture, the formation of the N-demethylated metabolite (m/z 267.2) and the monohydroxylated metabolite (m/z 297.2) of imipramine was observed following chip-based monolithic SPE sample cleanup and preconcentration. Solid-phase extraction (SPE) is an important and widely used sample pretreatment technique,1 which allows both the preconcentration of analytes and the reduction of interfering components. SPE is often used for desalting and preconcentration of samples prepared for GC/MS or LC/MS analysis. There is a growing interest in miniaturized sample preparation that may be directly coupled with electrospray mass spectrometry (ESI-MS). Girault et al.2 cut a poly(vinylidene difluoride) membrane into small pieces and integrated them in a microfabricated disposable polymer injector for desalting proteomic samples prior to mass spectrometric analysis. Thibault et al.3 have described a dual sprayer system for high-throughput proteome analysis that incorporated a styrene-divinylbenzene membrane for preconcentration and desalting of proteolytic digests prior to nanoelectrospray mass spectrometry analysis. Wachs and Henion4 also described a method and associated device for automated elution and spraying from a new 96-well format SPE device. In microfluidic systems, SPE is particularly attractive given the small amount of sample that can be handled within a chip without sample-transfer steps between devices. Hence, there is considerable interest in coupling SPE and other sampling pretreatment methods to chip-based systems.5 To accomplish SPE within a chip, different methods have been reported to introduce stationary phases, such as coating,6 packing,7-11 or in situ polymerization of (1) Majors, R. E. LCGC North Am. 2002, 20, 1098-1113. (2) Lion, N.; Gobry, V.; Jensen, H.; Rossier, J. S.; Girault, H. Electrophoresis 2002, 23, 3583-3588. (3) Bonneil, E.; Li, J.; Tremblay, T. L.; Bergeron, J. J.; Thibault, P. Electrophoresis 2002, 23, 3589-3598. (4) Wachs, T.; Henion, J. D. Anal. Chem. 2003, 75, 1769-1775. (5) Lichtenberg, J.; de Rooij, N. F.; Verpoorte, E. Talanta 2002, 56, 233-266. (6) Kutter, J. P.; Jacobson, S. C.; Ramsey, J. M. J. Microcolumn Sep. 2000, 12(2), 93-97. (7) Oleschuk, R. D.; Shultz-Lockyear, L. L.; Ning, Y.; Harrison, D. J. Anal. Chem. 2000, 72, 585-590. (8) Jemere, A. B.; Oleschuk, R. D.; Ouchen, F.; Fajuyigbe, F.; Harrison, D. J. Electrophoresis 2002, 23, 3537-3544. (9) Ekstrom, S.; Malmstrom, J.; Wallman, L.; Lofgren, M.; Nilsson, J.; Laurell, T.; Marko-Varga, G. Proteomics 2002, 2, 413-421. 10.1021/ac030196w CCC: $25.00

© 2003 American Chemical Society Published on Web 09/12/2003

a monolithic column.12-15 Ramsey et al.6 demonstrated solid-phase extraction with microfluidic devices by coating the channels with C-18 albeit with somewhat limited capacity. Several other groups have reported the fabrication of different microstructures on a chip, such as a weir or grids to hold stationary phases in place, and different methods to pack a channel with different stationary phases.7-11 The advantages of these methods include higher sample capacities and a wide choice of stationary phases. As an alternative to packing, in situ polymerization to form a monolithic packing in a chip offers several attractive features. These include frit-free construction, easy preparation with good control of porosity, and varied surface chemistry. More importantly, recently introduced UV-initiated polymerization enables the formation of monoliths within specific regions of a microfluidic device.12-15 Frechet et al. have shown the potential for photoinitiated monolithic porous polymers for on-chip SPE preconcentration.12 Despite this progress, more research is necessary especially to facilitate coupling multiple miniaturized SPE directly with MS. Although these monolith columns may be reused, it is preferable to employ a new SPE column for each real-world sample to avoid cross-contamination between samples and clogging of the channel. In addition, high levels of abundant endogenous compounds may limit the capacity of the SPE material when the same small SPE channel is reused. Moreover, one of the goals of microfluidic devices is to have inexpensive, disposable devices and to fabricate identical multicomponents on a chip for high-throughput parallel processing. Therefore, it is desirable to prepare an array of monolithic columns on a chip and to investigate the variation in their performance. The demonstration of on-chip SPE treatment of real-world samples is clearly important. In addition, most monolithic columns to date have been prepared in fused-silica capillaries or in channels on a glass/silica substrate. The use of polymer chips has the potential benefits of low cost, easy fabrication, and disposability. The Zeonor polymer material employed in this work has been widely used as a substrate for compact disk and digital video disk.16 It also has some attractive features, including high purity, absence of fluorescence or UV absorption, good chemical resistance, low water absorption, and high transparency.16,17 As previously reported,18 Zeonor polymer material has been successfully employed to construct microfluidic analytical devices through hot embossing and thermal bonding. In this paper, we report the fabrication of an array of eight channels on a Zeonor chip. The miniature porous monolithic SPE (10) Bergkvist, J.; Ekstrom, S.; Wallman, L.; Lofgren, M.; Marko-Varga, G.; Nilsson, J.; Laurell, T. Proteomics 2002, 2, 422-429. (11) Wolfe, K. A.; Breadmore, M. C.; Ferrance, J. P.; Power, M. E.; Conroy, J. F.; Norris, P. M.; Landers, J. P. Electrophoresis 2002, 23, 727-733. (12) Yu, C.; Davey, M. H.; Svec, F.; Frechet, J. M. J. Anal. Chem. 2001, 73, 5088-5096. (13) Rohr, T.; Yu, C.; Davey, M. H.; Svec, F.; Frechet, J. M. J. Electrophoresis 2001, 22, 3959-3967. (14) Throckmorton, D. J.; Shepodd, T. J.; Singh, A. K. Anal. Chem. 2002, 74, 784-789. (15) Ngola, S. M.; Fintschenko, Y.; Choi, W. Y.; Shepodd, T. J. Anal. Chem. 2001, 73, 849-856. (16) Lamonte, R. R.; McNally, D. Plast. Eng. 2000, 56, 51-55. (17) Kameoka, J.; Zhang, H.; Henion, J. D.; Craighead, H. G. Proceedings of SPIE on Microfluidics and BioMEMS; Society of Photo-Optical Instrumentation Engineers: San Francisco, CA, 2001; 4560, pp 227-235. (18) Kameoka, J.; Craighead, H. G.; Zhang, H.; Henion, J. D. Anal. Chem. 2001, 73, 1935-1941.

columns are prepared in situ inside each channel via UV-initiated polymerization. Using imipramine (a common antidepressant drug) and its d3 analogue as standard test compounds, the SPE chip was coupled to ESI-MS through a micro ion sprayer19 developed in this laboratory. The prepared monolithic columns were characterized with respect to sample capacity, recovery, reproducibility, and linearity. Moreover, the monolithic columns were applied to the cleanup of two different types of real-world biological samples, human urine samples and P450 drug metabolism incubation mixtures, followed by ESI-MS detection. EXPERIMENTAL SECTION Chemicals and Materials. Ethylene dimethacrylate (EDMA), butyl methacrylate (BMA), 2,2-dimethoxy-2-phenylacetophenone (DAP), 2-acrylamido-2-methyl-1-propanesulfonic acid (AMPS), hexane, formic acid, and KHCO3 were all obtained from Aldrich Chemical Co. (Milwaukee, WI). Imipramine hydrochloride and imipramine-d3 hydrochloride, β-nicotinamide adenine dinucleotide phosphate (NADPH), magnesium chloride, and Tris base were purchased from Sigma (St. Louis, MO). Pooled human liver microsomes were purchased from Gentest Corp. (Woburn, MA). Acetonitrile, methanol, and ammonium hydroxide were from VWR Scientific (West Chester, PA). Acetic acid was obtained from EM Science (Gibbstown, NJ). Prior to polymerization, EDMA and BMA were purified by passing them through a 0.45-µm PTFE syringe filter (13-mm diameter, Nalge Nunc International, Rochester, NY) filled with fresh basic alumina powder (mesh 60-325, Fisher, Pittsburgh, PA). Stock solutions of imipramine-d0 (10 mg/mL) and imipramined3 (1.1 mg/mL) were prepared by dissolving 45 mg of imipramined0 hydrochloride and 5 mg of imipramine-d3 hydrochloride in 4 mL of methanol, respectively. Chip Fabrication. The Zeonor polymeric chip was fabricated by hot embossing and thermal bonding as described in our previous publication with some modifications.18 For rapid prototyping, instead of etching a silicon wafer to make a master, eight conventional fused-silica capillaries (360-µm o.d., 50-µm i.d., 85mm length) were placed in parallel 4.5 mm apart and epoxy glued on a glass microscope slide as a “master” (Figure 1A). A Zeonor polymer plate (2 mm thick, Zeon Chemicals, Louisville, KY) was cut into a 33 mm by 50 mm substrate, which was then embossed against this capillary array “master” (Figure 1B) for 10 min with heat (132.2 °C) and pressure (129 psi) using an MTP-8 Press (Tetrahedron, San Diego, CA). After completing this embossing process, the eight capillaries were easily removed from the polymer chip, leaving eight embossed channels (360 µm) on the substrate chip. For each of the eight embossed channels, two connecting capillaries (360-µm o.d., 50-µm i.d., 110- and 180-mm length, respectively) were inserted into each channel from both ends, leaving a 10-mm gap in the middle (Figure 1C). Then, the embossed chip was covered by another plain Zeonor polymer plate (33 mm by 50 mm, 2 mm thick) to form a seal by the application of heat (105.6 °C) and pressure (86 psi) for 10 min with the press to form a cover for the completed device. After this thermal bonding process, a leakage-free system for each of the connecting capillaries was formed to the Zeonor plates. A fabricated chip with eight parallel channels and corresponding inlet and outlet connecting capillaries is shown in Figure 1C. (19) Wachs, T.; Henion, J. D. Anal. Chem. 2001, 73, 632-638.

Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

5505

Figure 1. Chip fabrication. (A) The “master” for hot embossing: CA, an array of eight fused-silica capillaries (360 µm o.d., 85 mm long, 4.5 mm apart); E, epoxy glue; G, glass microscope slide. (B) Hotembossing procedure: CA, the eight fused-silica capillaries on the “master”; Z, Zeonor polymer substrate plate (33 mm by 50 mm, 2 mm thick); Si, blank silicon wafer; GP, glass plate; P, pressure. (C) The chip (33 mm × 50 mm): M-eight channels for monolithic SPE columns, 10 mm long, 360 µm i.d., 4.5 mm apart from each other; Ci, inlet connecting capillaries, Co, outlet connecting capillaries; Z, Zeonor polymer substrate plate.

Preparation of the Monolithic SPE Columns. The monomer and porogen mixture was similar to those used by others.12,13 The volume ratio of monomer to porogen was 25 to 75, which leads to high degree of porosity. The monomer was composed of 60% BMA and 40% EDMA (v/v) as the cross-linker. AMPS and DAP represented 1 and 2% relative to monomer by weight, respectively. A typical formulation of monomer and porogen mixture was as follows. To a vial containing 5 mg of DAP and 3 mg of AMPS, 500 µL of methanol, 250 µL of hexane, 150 µL of BMA, and 100 µL of EDMA were added sequentially. After purging with nitrogen for 5 min, the mixture was ready for UV irradiation and the subsequent polymerization. To expedite filling of the eight channels with the monomer mixture and to reduce possible variations in polymerization conditions, the eight channels were filled simultaneously and polymerized in parallel (Figure 2A). A homemade device was constructed for this purpose that consisted of a small vacuum manifold (40 mm × 22 mm × 5 mm, Figure 2B) equipped with a 10-mL plastic syringe and nine nanoports (N-124S, Upchurch Scientific, Oak Harbor, WA) for connecting the eight capillaries plus a vent. After connecting the chip to this device (Figure 2A) and closing the vent, the eight channels may be filled simultaneously with the monomer mixture by aspiration via the syringe. The polymerization may be initiated by UV irradiation for 20 min under UV light (UVL-28, 365 nm, VWR). Initially, an aluminum 5506 Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

Figure 2. (A) Experimental setup for polymerization preparation of monolith: S, 10-mL plastic syringe; N, nanoports; VB, vacuum distribution box; H, vent hole; C, polymer substrate chip; Ci, Co, inlet/ outlet connecting capillaries; UV, UV light; V, vial containing monomer mixture. (B) Exploded view of the vacuum manifold showing in (A). Three Perspex plates (P1, P2, P3) were glued to form a box using chloroform solvent. Eight nanoports for connecting capillaries were epoxy glued to the eight holes (H1) on plate P1. The nanoport for the vent and the syringe were epoxy glued to the two holes on plate P2, H2 and H3, respectively.

Figure 3. Experimental setup for coupling to ESI-MS: P, syringe pump; I, manual injection valve; U, union; C, the chip; ESI, micro ion sprayer; xyz, x, y, z stage; MS, API 3000 LC/MS/MS system.

foil mask was used to cover the region where the monolith was not required. However, it was later determined that the use of a mask was not necessary since the polyimide coating on the connecting capillaries is not transparent to the UV irradiation. When the polymerization process was complete, the remaining monomer mixture inside the connecting capillaries and unpolymerized monomer in the chip was removed with methanol. Scanning electron microscopy (SEM) images of the porous monolithic column were obtained on a Leo 982 SEM (Leo, Thornwood, NY). Coupling to Mass Spectrometry. As shown in Figure 3, the chip was coupled to an API 3000 LC/MS/MS system (PE Sciex, Concord, ON, Canada) through a micro ion sprayer developed in this laboratory.19 The mass spectrometer was operated in the positive mode with unit mass resolution. For both the selected ion monitoring (SIM) and the selected reaction monitoring (SRM), dwell time was 200 ms. For the full-scan mode (m/z 250-350), the dwell time was 2 ms. The sprayer voltage was maintained at

5 kV for all experiments reported here. Syringe infusion pumps (model 22, Harvard Apparatus, Holliston, MA) were employed for the delivery of makeup liquid, sample, buffer,, and acetonitrile. A makeup liquid containing 0.5% formic acid in 1:1 methanol/water was infused into the makeup tube19 at a rate of 4-8 µL/min depending upon the experiment. In some cases, a manual injector (model 7125, Rheodyne, Rohnert Park, CA) was used for sample loop injection (Figure 3, I). The eight SPE columns on the chip were connected in sequence to the syringe pump and aligned to the sprayer with an x, y, z microposition stage (Newport, Irvine, CA). Sample Capacity Measurement for the Monolithic SPE Columns. After the polymer chip-based monolithic column was preconditioned with acetonitrile and 10 mM NH4AC (pH 9.3), a solution containing 5 µg/mL imipramine and 0.5 mM KHCO3 (10 mM NH4AC, pH 9.3) was continuously pumped through the column at 2 µL/min. Concurrently, the effluent was monitored by the SIM at m/z 281 for imipramine and m/z 39 for potassium. The potassium ion was added as an unretained marker to determine the time required for the sample solution to flow through connecting capillaries. The sample capacity was estimated based on the retention volume and the concentration of imipramine in the influent.20,21 Recovery Studies. The SPE columns were first preconditioned with ACN and then with 10 mM NH4AC (pH 9.3). After this preconditioning process, 10 µL of 2 µg/mL imipramine-d0 in 10 mM NH4AC (pH 9.3) was flow-injected into a buffer carrier stream (10 mM NH4AC, pH 9.3, 20 µL/min) that loaded the imipramine-d0 into the monolithic SPE column while purging unretained or loosely retained components to waste. Then, ACN containing 5 µg/mL imipramine-d3 and 0.1% formic acid was used to elute the retained imipramine-d0 from the monolithic SPE column. A peak was recorded for the eluted imipramine-d0 by the SRM transition (m/z 281.3 f 86.3). The area of this ion current signal is a measure of the amount of imipramine-d0 recovered. After the retained imipramine-d0 had been removed, 10 µL of 2 µg/mL imipramine-d0 in ACN (the same quantity of imipramined0 as above but not in aqueous solution) was injected into the ACN stream, producing an FIA peak for imipramine-d0 via the SRM transition (m/z 281.3 f 86.3). Since both the sample matrix and the carrier stream were the same, i.e., 100% acetonitrile, the imipramine-d0 would not be retained. Therefore, the area of this FIA peak is a measure of the total amount of imipramine-d0 injected. The recovery was calculated based on the area ratio of the first peak (retained and eluted) to the second peak (unretained FIA peak). To monitor any variations in the sprayer, 5 µg/mL imipramine-d3 was added to the ACN eluent and was concurrently monitored by the SRM transition (m/z 284.3 f 89.3). Linearity Studies. Seven standard solutions of imipramined0 were prepared at the following concentrations: 0.025, 0.05, 0.1, 0.5, 2.5, 5, and 10 µg/mL in 10 mM NH4AC (pH 9.3). Imipramined3 was added as an internal standard to all standard solutions at a concentration of 1.3 µg/mL. Each of the seven standard solutions was loaded (10 µL) to two different monolithic SPE columns, requiring a total of 14 different monolithic SPE columns on two Zeonor polymer chips. Each of the monolithic SPE columns had been preconditioned with acetonitrile and 10 mM NH4AC (pH 9.3) prior to sample loading. The retained imipramine-d0 and imipramine-d3 were coeluted with ACN. The elution process was monitored using SRM (m/z 281.3 f 86.3 and m/z 284.3 f 89.3

for imipramine-d0 and imipramine-d3, respectively). The peak height ratios of imipramine-d0 to imipramine-d3 were measured to provide a quantitative measure of the ion current response of the target components. A regression curve was constructed between the average peak height ratios (two injections for each standard solution) and the corresponding concentration of imipramine-d0. Cleanup of Human Urine Sample. A freshly collected human urine sample was adjusted to pH 9.3 using ammonium hydroxide and added (1:1) to seven aqueous solutions that had been spiked with imipramine-d0 and imipramine-d3 standards. The final concentrations of imipramine-d0 in the seven diluted human urine samples were as follows: 0.025, 0.05, 0.1, 0.5, 2.5, 5, and 10 µg/mL. The concentration of imipramine-d3 in all the seven diluted human urine samples was 1.3 µg/mL. A total of 10 µL of the spiked human urine sample was loaded to a preconditioned monolithic SPE column and washed with 100 µL of 10 mM NH4AC (pH 9.3) followed by elution with ACN. The elution process was monitored by the SRM transition (m/z 281.3 f 86.3 and m/z 284.3 f 89.3). SPE Cleanup of P450 Drug Metabolism Incubation Mixtures. A 30-µL standard incubation mixture contained 3 mM MgCl2, 20 µM imipramine, 1 mM NADPH, 60 µg of microsomal protein (human liver microsomes), and 2.5 mM Tris buffer (pH 7.4). The incubation was performed at 37 °C for 30 min and terminated by transferring the vial to a freezer. The blank control experiment was done the same way as described above but using inactivated microsomes for incubation (microsomes were inactivated by putting the vial in boiling water for 5 min). Five microliters was withdrawn from the incubation mixture and diluted with 20 µL of 10 mM ammonium hydroxide before being transferred to a monolithic SPE column. After purging with 100 µL of 10 mM NH4AC (pH 9.3), the retained compounds were eluted with acetonitrile (containing 0.1% formic acid) at a flow rate of 4 µL/min. The elution process was monitored by ESI-MS in the full-scan mode (m/z 250-350). RESULTS AND DISCUSSION Chip Fabrication. It is well known that it is preferable to use a fresh SPE column for each new, real-world sample. In addition, for higher sample throughput applications, an array of SPE devices can facilitate rapid sample preparation. We believe miniaturization can facilitate high-throughput sample preparation and analysis in the future. Therefore, the primary goal of this research work was to prepare an array of porous monolithic SPE columns on a single chip and use them for SPE cleanup of real-world biological samples. The pore size, sample capacity, and dimensions of the monolithic column should preclude clogging while providing adequate sample cleanup. The Zeonor polymer was chosen as the substrate because of its attractive characteristics and especially its hot-embossing and thermal-bonding characteristics as shown by previous research work from this laboratory.18 Although a silicon wafer could be etched as the master for hot embossing, for rapid prototyping purposes, we chose to use ordinary 360-µm-o.d. capillaries as the “master” to form channels in the Zeonor substrate for hot embossing. To facilitate the alignment, eight capillaries were epoxy glued on a glass microscope slide. In addition, it is very easy to insert connecting capillaries of the same dimensions into the embossed channels and to bond them between Zeonor polymer plates during thermal bonding. Though most channels in microfluidic devices are less Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

5507

Figure 4. (A) SEM image showing the cross section of monolithic column inside a Zeonor polymer chip; (B) SEM image showing the details of monolithic nodules and pores.

than 100 µm, we decided at this stage of our microtechnology developments to employ somewhat larger dimensions.7,9,22 The space between adjacent channels was selected as 4.5 mm for compatibility with the gap between wells on typical 384 well plates. It is contemplated that monolithic columns could be prepared in the standard 96 (8 × 12) or 384 (16 × 24) formats in a polymer block in the future. Polymerization of the Monolith. The polymerization conditions were similar to those of others except for a few minor modifications.12,13 For easier preparation, volume ratios of monomers, instead of weight ratios, were used. At the early stage, BMA and EDMA monomers were not purified. Though they were successfully polymerized with increased amounts of photoinitiator (2% relative to the weight of monomers), satisfactory consistency of polymerization was not obtained. Improved results were obtained by purification of the monomers by filtering them through basic alumina cartridges. This greatly improved the consistency of polymerization products. AMPS was added in order to support electroosmotic flow for pumping with high voltages. It was found that AMPS is also helpful for better retention of imipramine in a way similar to mixed-mode retention.23,24 The pore sizes of the monolith depend mainly on the amount of porogen and especially the ratio of methanol to hexane. The higher the ratio of methanol to hexane, the larger the pore size produced. When the methanol concentration was reduced to 60%, it was possible to produce nonporous and gellike polymers. To avoid excess loss of porogen during purging with nitrogen, the purging time was reduced to 5 min. With careful control of all experimental conditions and the purity of monomer solutions, good consistency of the polymerized product was obtained. As shown in Figure 4, the porous monolith fills the embossed channel on the Zeonor polymer chip. Since covalent bonding to the channel walls is not involved, the monolith was retained in its (20) Josefson, C. M.; Johnston, J. B.; Trubey, R. Anal. Chem. 1984, 56, 764-768. (21) Dumont, P. J.; Fritz, J. S. J. Chromatogr., A 1995, 691, 123-131. (22) Attiya, S.; Jemere, A. B.; Tang, T.; Fitzpatrick, G.; Seiler, K.; Chiem, N.; Harrison, D. J. Electrophoresis 2001, 22, 318-327. (23) Martinez, M. A.; Sanchez de la Torre, C.; Almarza, E. J. Anal. Toxicol. 2002, 26, 296-302. (24) Wu, R.; Zou, H.; Fu, H.; Jin, W.; Ye, M. Electrophoresis 2002, 23, 12391245.

5508

Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

place by mechanical forces. The monolithic column is composed of linked nodules in the 2-3-µm range. Since the pore size is larger (∼10 µm), the back pressure is low, which allows the use of a standard syringe infusion pump to deliver liquid at a flow rate as high as 100 µL/min. It is interesting to note that the cross section of the embossed channel on a bonded Zeonor polymer chip is nearly circular, while the cross section of most channels fabricated by other methods is usually either half circular or trapezoidal. The reason for this might be that the Zeonor plate is not very brittle, particularly when it was still warm after the embossing. During the hot-embossing procedure, the capillaries were buried in the Zeonor plate. When the capillaries were removed from the embossed Zeonor substrate, the embossed channels broke to let the capillaries out but restored to their original shape (circular) immediately after the capillaries were removed. The circular-shaped channel may be considered an additional advantage since this geometry is the same as a conventional capillary column rather than the half circular or trapezoidal channels in most microfluidic devices. Sample Capacity of Monolithic Columns. The capacity measurement is based on breakthrough curves.20,21 To avoid errors caused by the delay for the analytes to flow through connecting capillaries, potassium ion (added as KHCO3) was chosen as an unretained marker. A common antidepressant, imipramine, was chosen as the test analyte to be monitored during these studies. A solution containing 5 µg/mL imipramine and 0.5 mM KHCO3 was prepared in 10 mM NH4AC (pH 9.3)25 for retention on the reversed-phase monolithic SPE column housed in the chip. The test solution was continuously pumped to an SPE column, and the effluent was concurrently monitored for imipramine and potassium by SIM. Since potassium ion was not retained by the SPE column, it reached its ion current maximum very rapidly (Figure 5, curve B). However, imipramine was retained and then gradually saturated the trapping capacity of the SPE monolithic packing. The excess, unretained analyte then passed through the column reaching an ion current maximum (Figure 5, curve A). Based on the time delay to reach 50% of the ion current maximum, (25) Jinno, K.; Kawazoe, M.; Saito, Y.; Takeichi, T.; Hayashida, M. Electrophoresis 2001, 22, 3785-3790.

Figure 5. Breakthrough curve for sample capacity measurement: test analyte standard solution, 5 µg/mL imipramine and 0.5 mM K+ (pH 9.3); flow rate, 2 µL/min. (A) SIM at m/z 281 for protonated molecule of imipramine; (B) SIM at m/z 39 for potassium.

the flow rate, and the concentration of imipramine, the capacity was estimated to be 0.30 µg (1.1 × 10-9 mol).20,21 The relative standard deviation (RSD) of sample capacities for eight different monolithic SPE columns on the same polymeric chip was 26.3%. For two chips prepared using the same monomer mixture, the difference in average sample capacity was 7.0%. There are several factors for the relatively large variation in sample capacity between different SPE columns on a chip. These include variation in channel length (manual control) and differences in the flux of UV irradiation on each of the monolithic columns in the Zeonor polymer channels. However, considering that the actual sample load is usually much lower than the capacity, this variation may not have adverse influence in the analytical performance. Recovery and Reproducibility. The recovery and reproducibility of the prepared monolithic SPE columns were also evaluated using imipramine as a test standard. The average recovery for eight different monolithic SPE columns was determined to be 79.1%. Since imipramine is a representative basic drug that can exhibit adsorptive losses and hence reduced recovery, these results appear to be acceptable.23 As expected, the variation in recovery between different columns (7.9%) is much less than the capacity variation reported above. To measure the reproducibility of the same sample on different columns, a mixture of imipramined0 and imipramine-d3 was loaded onto the monolithic SPE material, washed, and eluted to the mass spectrometer. Both peak area ratios and peak height ratios were calculated. The RSD for peak area ratios was 4.0%, and the RSD for peak height ratios was 2.0%. The larger variation in peak area ratios appeared to be primarily due to errors in integrating tailing peaks, which result from direct elution of analyte with ACN. Linearity Studies. To determine the linear dynamic range for the selected test compound, each of the seven standard solutions was loaded to two respective SPE columns. A total of 14 SPE columns on two different chips was involved, and each monolithic column was used only once. Imipramine-d3 was added to all standard solutions, and peak height ratios were used as the indicator of signal intensity. Representative elution profiles are shown in Figure 6. A linear calibration curve (y ) 0.529x + 0.02,

R2 ) 0.9995) was obtained for imipramine-d0 in the range from 0.025 to 10 µg/mL. These results suggest that the SPE columns housed both on and between different chips may be used for quantitative analysis purposes when a suitable isotope internal standard is employed. Cleanup of Biological Samples. To demonstrate the analytical potential of the SPE columns for the cleanup of real-world samples, the chip-based columns were employed for the cleanup of human urine sample and P450 drug metabolism incubation mixture. These columns were sufficiently porous that no clogging was observed. However, once an SPE column has been used for the treatment of a biological sample, its capacity is greatly reduced even after extensive elution with 100% acetonitrile and is therefore discarded. Since one typically does not reuse SPE columns, this issue should not be a limitation. Similar to standard aqueous solutions, a linear calibration curve (y ) 1.38x + 0.01, R2 ) 0.9995) was also obtained for human urine samples spiked with imipramine in the range from 0.025 to 10 µg/mL. The ion current elution profile for a human urine sample spiked at the lower limit of quantitation at 0.025 µg/mL imipramine-d0 is shown in Figure 7. These experiments were performed using a syringe pump operating at low flow rates (2-4 µL/min) and a homemade micro ion spray interface. These data show a shoulder observed in Figure 7A. Since some partial separation or limited chromatography may be expected in SPE techniques, the shoulder observed in Figure 7A appears to be due to partial separation of a compound(s) that has/have the same transition as the targeted analyte. Since the Y-axis in Figure 7A is shown at a more sensitive setting for the analyte than the internal standard, the shoulder in Figure 7A is more apparent. For urine samples spiked with 1 µg/mL imipramine-d0 and imipramine-d3, the relative standard deviations for peak area ratios and peak height ratios over eight different columns were 5.0 and 3.8%, respectively. These variations are similar to those for imipramine prepared as analytical standard in aqueous solution, which were 4.1 and 2.0%, respectively. In P450 drug metabolism studies, the identification of metabolites is an important task. Mass spectrometry is a sensitive selective detection method that provides structural information, Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

5509

Figure 6. Representative SRM ion current elution profiles for imipramine isolated from standard aqueous solution. A total of 10 µL of solution containing 0.05 µg/mL imipramine-d0 and 1.3 µg/mL imipramine-d3 was loaded onto a preconditioned monolithic column followed by elution with ACNL (A) imipramine-d0 ion current profile (SRM, m/z 281.3 f 86.3); (B) imipramine-d3 ion current profile (SRM, m/z 284.3 f 89.3).

Figure 7. SRM ion current elution profiles for imipramine spiked in human urine sample. A total of 10 µL of human urine sample spiked with 0.025 µg/mL imipramine-d0 and 1.3 µg/mL imipramine-d3 was loaded onto a preconditioned monolithic column followed by washing with 10 mM NH4AC (pH 9.3) and elution with ACN: (A) imipramine-d0 ion current profile (SRM, m/z 281.3 f 86.3); (B) imipramine-d3 ion current profile (SRM, m/z 284.3 f 89.3).

which is essential for qualitative metabolite characterization. However, direct infusion of the incubation mixture to a mass spectrometer is usually problematic because the incubation mixture contains buffer (salts) and large protein molecules (microsomes). In addition, the concentrations of metabolites are often very low and hence require preconcentration. The chip-based 5510

Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

SPE monolithic columns described in this work can be used for the cleanup of the P450 drug metabolism incubation mixture and for the preconcentration of the metabolites as well. As shown in Figure 8A and B, similar TIC elution profiles were obtained for both the standard and blank control incubation mixtures. However, the abbreviated mass spectra (m/z 250-350) of the corre-

Figure 8. SPE cleanup of P450-imipramine incubation mixtures and ESI-MS detection. Incubation mixtures of imipramine with inactivated microsomes (blank control incubation) and active human liver microsomes (standard incubation) were loaded on two separate SPE columns on a chip for desalting and preconcentration prior to ESI-MS detection: (A) TIC elution profile for blank control incubation; (B) TIC elution profile for standard incubation; (C) mass spectrum for blank control incubation showing the peak of the parent drug, imipramine (m/z 281.2); (D) mass spectrum for standard incubation showing the peaks of both the parent drug (m/z 281.2) and the two main metabolites (m/z 267.2 and 297.2).

sponding elution peaks are different. In Figure 8C (blank control incubation), only the parent drug (imipramine, m/z 281.2) was observed in the mass spectrum. In Figure 8D (standard incubation), in addition to the main parent drug peak (m/z 281.2), the formation of the N-demethylated metabolite (m/z 267.2) and the monohydroxylated metabolite (m/z 297.2) was observed. These results demonstrate that the chip-based monolithic SPE sample cleanup described in this work can be employed for the subsequent direct infusion-electrospray analysis of the P450 incubation mixture and the detection of metabolites produced in this experiment. For a more complex metabolism mixture, gradient elution and longer SPE columns can be used to achieve some separation prior to ESI-MS detection.

separate monolithic columns on the same chip. The results described herein suggest that the chip-based monolithic columns are applicable to SPE cleanup of real-world biological samples such as fortified human urine and P450 drug metabolism incubation mixture. Although demonstrated here in an array of eight SPE columns on a polymer-based substrate, it is anticipated that an array of 96 (8 × 12) or 384 (16 × 24) monolithic columns in a polymer device could be fabricated and employed for larger numbers of samples. Furthermore, coupling of an array of monolithic columns to a corresponding array of sprayers26-28 may be possible to provide higher analytical ESI-MS throughput without adverse carryover issues.

CONCLUSIONS An array of eight porous monolithic columns was successfully prepared on a Zeonor polymer chip with UV-initiated polymerization. The performance of these columns for solid-phase extraction was systematically evaluated by coupling to ESI-MS. High recovery and good reproducibility was obtained between the eight

ACKNOWLEDGMENT We thank Amgen and Dr. Krys Miller for financial support of the main portion of this work and Merck Research Laboratories for financial support of the P450 portion of this work. We also thank Dr. Claudia Fleischer for initiating this project and Dr. Jun Kameoka and Ms. Yanou Yang for providing SEM analysis of the monolith SPE columns. In addition, we are grateful for many helpful discussions with Dr. Tim Wachs.

(26) Kameoka, J.; Orth, R.; Ilic, B.; Czaplewski, D.; Wachs, T.; Craighead, H. G. Anal. Chem. 2002, 74, 5897-5901. (27) Schultz, G. A.; Corso, T. N.; Prosser, S. J.; Zhang, S. Anal. Chem. 2000, 72, 4058-4063. (28) Gobry, V.; van Oostrum, J.; Martinelli, M.; Rohner, T. C.; Reymond, F.; Rossier, J. S.; Girault, H. H. Proteomics 2002, 2, 405-412.

Received for review May 15, 2003. Accepted July 15, 2003. AC030196W

Analytical Chemistry, Vol. 75, No. 20, October 15, 2003

5511