Microchip Capillary Electrophoresis with Electrochemical Detection of

Department of Chemistry and Biochemistry, New Mexico State University, Las Cruces, New Mexico 88003. Ashok Mulchandani. Department of Chemical and ...
2 downloads 0 Views 108KB Size
Anal. Chem. 2004, 76, 4721-4726

Microchip Capillary Electrophoresis with Electrochemical Detection of Thiol-Containing Degradation Products of V-Type Nerve Agents Joseph Wang,* Jiri Zima,† Nathan S. Lawrence, and Madhu Prakash Chatrathi

Department of Chemistry and Biochemistry, New Mexico State University, Las Cruces, New Mexico 88003 Ashok Mulchandani

Department of Chemical and Environmental Engineering, University of California, Riverside, California 92521 Greg E. Collins

Chemistry Division, Naval Research Laboratory, Washington D.C. 20375

A microchip protocol for the capillary electrophoresis separation and electrochemical detection of thiol-containing degradation products of V-type nerve agents is described. The microchip assay relies on the derivatization reaction of 2-(dimethylamino)ethanethiol (DMAET), 2-(diethylamino)ethanethiol (DEAET), and 2-mercaptoethanol (ME) with o-phthaldialdehyde in the presence of the amino acid valine along with amperometric monitoring of the isoindole derivatives. Both off-chip and on-chip derivatization reactions have led to highly sensitive and rapid detection of the thiol degradation products. Various parameters influencing the derivatization, separation, and detection processes were examined and optimized. These include the amino acid co-reagent, reagent-mixing ratio, reaction time, injection time, separation voltage, and detection potential. The chip microsystem offers a rapid (+1.0 V), and fouling problems associated with the detection of thiols.27 In the following sections we will demonstrate the optimization and attractive analytical performance of the CE microchip for monitoring thiol-containing hydrolysis products of V-type nerve agents.

EXPERIMENTAL SECTION Apparatus. The glass microchips used in this study were fabricated by Micralyne (Edmonton, Canada) and are depicted in Figure 1. The simple cross (Figure 1A) chip,28 used in off-chip derivatization experiments, had a 74-mm-long separation channel and a 5-mm-long injection channel. The microchip involving the on-chip reaction was based on our custom design,20 the layout of which is depicted in Figure 1B. Briefly, this chip consisted of the reagent (R), sample (S), running-buffer (RB), and unused waste (W) reservoirs. A reaction chamber (200 µm wide and 3.6 mm long) was connected through 50-µm-wide, 20-µm-deep channels to the reagent and sample reservoirs at one side and to a fourway injection cross at the other side. The injection cross was followed by a 74-mm-long separation channel. Pipet tips were cut and placed in each of the fluidic ports of the reservoirs. (19) Jacobson, S. C.; Koutny, L. B.; Hergenroder, R.; Moore, A. W., Jr.; Ramsey, J. M. Anal. Chem. 1994, 66, 3472. (20) Wang, J.; Chatrathi, M. P.; Tian, B. Anal. Chem. 2000, 72, 5774. (21) Ro, K. W.; Lim, K.; Kim, H.; Hahn, J. H. Electrophoresis 2002, 23, 1129. (22) Colyer, C. L.; Mangru, S. D.; Harrison, D. J. J. Chromatogr., A 1997, 781, 271. (23) Liu, Y.; Foote, R. S.; Jacobson, S. C.; Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 2000, 72, 4608. (24) Inoue, T.; Kirchhoff, J. R. Anal. Chem. 2002, 74, 1349. (25) Vandeberg, P. J.; Johnson, D. C. Anal. Chem. 1993, 65, 2713. (26) Carvalho, F. D.; Remiao, F.; Vale, P.; Timbrell, J. A.; Bastos, M. L.; Ferreira, M. A. Biomed. Chromatogr. 1994, 8, 134. (Rev. III-13) (27) White, P. C.; Lawrence, N. S.; Davis, J.; Compton, R. G. Electroanalysis 2002, 14, 89. (28) Wang, J.; Tian, B.; Sahlin, E. Anal. Chem. 1999, 71, 5436.

Figure 1. Schematic of (A) simple cross and (B) precolumn microchips used in the present study for carrying out off-chip and on-chip derivatization reactions, respectively. Abbreviations: (R) reagent reservoir, (S) sample, (RB) running buffer, (W) unused waste reservoirs, and (D) end-column amperometric detector.

The layout of the CE/amperometric microsystem used in this study has been described previously.28 Briefly, the CE microchip was placed in a laboratory-built Plexiglas holder fabricated for housing the separation chip and the detector, thereby allowing their convenient replacement. Platinum wires, inserted into each reservoir, served as contacts for the high-voltage power supply. A platinum wire and an Ag/AgCl wire (prepared by electrolytic oxidation of silver wire in 0.1 M HCl) were inserted into the detection reservoir to serve as counter and reference electrodes, respectively, for the amperometric detection. The screen-printed working electrodes were fabricated with a semiautomatic printer (model TF 100, MPM, Franklin, MA), using an Acheson carbon ink (Electrodag 440B/49AB90, Acheson Colloids, Ontario, CA). Details of the printing processes were described previously.28 Briefly, the printing was performed through a patterned stencil (100 µm thick; Global Stencil, Grand Praire, TX) onto 100 × 100 × 0.64 mm alumina ceramic plates. Each plate consisted of 30 strips (33.3 × 10.0 × 0.64 mm) in which each strip were defined by a laser precut. The printing procedure consisted of the following steps: A carbon ink working electrode layer (0.3 × 8 mm) was printed on each of the strips. A silver ink (Ercon R-421; DRE-68) layer (1.5 × 21 mm) was subsequently printed for contacting the carbon layer. An insulating ink (Ercon R488CI-GI Insulator Green) was then printed to cover the carbon-silver junction and define the working electrode area on one side while exposing the contact area (0.3 × 2.5 mm) on the other side. The carbon and silver layers were cured for 60 min and the insulator layer for 120 min at 100 °C. The screen-printed carbon electrode was further held in place by a plastic screw pressing the strip against the channel outlet. The distance of the working electrode from the chip outlet (50 µm) was controlled by a thick tape spacer. The homemade high-voltage power supply had an adjustable voltage range between 0 and +4000 V. Amperometric detection was performed with an electrochemical analyzer 621A (CH Instruments, Austin, TX) connected to a Pentium 166 MHz computer with 32 MB RAM. The electropherograms were recorded with a time resolution of 0.1 s while applying a detection potential of +0.8 V (vs Ag/AgCl pseudo reference wire electrode). Sample injections were performed after stabilization of the baseline. The raw data of electropherograms were digitally filtered using built-in 15-point least-squares smoothing using CHI Version 3.27 software (CH Instruments).

Reagents. DMAET, DEAET, sodium tetraborate decahydrate, and sodium hydroxide were purchased from Aldrich. Methanol (HPLC grade), 2-mercaptoethanol, OPA, L-valine, L-histidine, L-isoleucine, L-leucine, and sodium dodecyl sulfate were obtained from Sigma. All chemicals were used without further purification. The running buffer, a mixture of 10 mM sodium borate and 20 mM sodium dodecyl sulfate (pH 9.45), was prepared in deionized water and filtered through an RC 0.20-µm membrane filter (ColeParmer) using a 20-mL syringe and sonicated for 20 min. Stock solutions were prepared daily in deionized water (for thiols), in methanol (30 mM OPA), or in the running buffer (50 mM L-valine). The sample solutions were prepared by diluting the corresponding stock solutions in the electrophoresis buffer. The river water sample, collected from the Rio Grande River in Las Cruces, NM, was filtered through a membrane filter prior to use. Tap water (collected on the NMSU campus) was used directly without any processing. All experiments were carried out at room temperature. Derivatization Procedure. Prior to the analysis, sample solutions were prepared in accordance to Alvarez-Coque et al.29 by first mixing 5-150 µL of the thiol solution with 100 µL of OPA solution (30 mM). A 100-µL aliquot of valine (50 mM) was then added to this mixture, followed by dilution to 1 mL by the addition of running buffer. Unless otherwise mentioned, off-chip reaction was carried out for 60 s prior to injection. Using on-chip experiments, the sample solution consisted of the thiol mixture, while the reagent solution contained a mixture of OPA and valine. The on-chip assay was performed by injecting the sample (mixture of thiols) and reagent (OPA and valine) into the precolumn reaction chamber for derivatization.20 Off-chip assays involving tap and river water samples were prepared by mixing 500 µL of the appropriate sample with a 100µL aliquot of OPA reagent, followed by addition of 100 µL of the valine reagent solution; the resulting solution was diluted to 1 mL with running buffer. Electrophoresis Procedure. Prior to use, the channels were flushed with 0.1 M sodium hydroxide and deionized water for 20 and 5 min, respectively. Each of the reservoirs in the chip holder and the corresponding pipet tips on the microchannel chip were filled with their respective solutions. The sample reservoir was filled with the off-chip reaction mixture. For on-chip derivatization assays, the sample reservoir was filled with the mixture of thiols and the reagent reservoir was filled with a mixture of OPA (6 mM) and valine (30 mM). The injection was performed by applying +2000 V for 2 s to the sample reservoirs, with the detection reservoir grounded and other reservoirs floating. The separation was usually performed by applying +2000 V to the running buffer reservoir with the detection reservoir grounded and other reservoirs floating. Safety Considerations: The high-voltage power supply and associated open electrical connections must be handled with extreme care. Stock solutions of the thiol compounds must be handled in the fumehood. RESULTS AND DISCUSSION The present study targeted the chip-based measurements of degradation products of priority chemical weapons, including (29) Alvarez-Coque, M. C. G.; Hernandez, M. J. M.; Camanas, R. M. V.; Fernandez, C. M. Anal. Biochem. 1989, 178, 1.

Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

4723

Figure 3. Influence of the OPA/DMAET concentration ratio on the current response of the valine-DMAET isoindole derivative. Other conditions, as in Figure 2B. Inset: OPA-thiol derivatization reaction. Figure 2. Influence of histidine (A), valine (B), isoleucine (C), and leucine (D) on the CE separation and detection of the isoindole derivatives of 250 µM ME (a), DMAET (b), and DEAET (c). Running buffer, 10 mM sodium tetraborate and 20 mM sodium dodecyl sulfate (pH 9.45); off-chip derivatization using 3.0 mM OPA and 5.0 mM amino acid; injection and separation voltages, +2000 V; injection time, 2 s; detection potential, +0.8 V (vs Ag/AgCl wire reference electrode).

DMAET and DEAET hydrolysis products, that correspond to VX and R-VX nerve agents following the reaction pathway shown in Scheme 1. The new microchip protocol relies on the derivatization reaction of DMAET, DEAET, and related thiol ME, with OPA (in the presence of the amino acid valine), along with amperometric monitoring of the isoindole derivatives. The OPA-based route offered improved stability and sensitivity over direct anodic measurements of these thiol compounds or over those based on derivatization with 1,4-benzoquinone (not shown). The sensitive determination of the resulting isoindole derivatives of the OPA reaction depends on the specific amino acid coreagent. Therefore, we compared the influence of different amino acids upon the response for a mixture of different thiol-containing compounds. Figure 2 displays electropherograms for a mixture of ME (a), DMAET (b), and DEAET (c) recorded under identical conditions, but using different amino acid co-reagents. The use of valine (B), isoleucine (C), and leucine (D) resulted in three well-separated peaks, while histidine (A) led to two major peaks and two smaller unresolved ones. The latter indicates additional reaction products, as was supported by an immediate color change (colorless to brown) of the resulting mixture in the presence of histidine. Among the other three amino acids, valine (B) yielded the most favorable signal-to-noise characteristics, along with shorter migration times (B vs C, D), and hence was used for all subsequent work. The reaction was performed in a 10 mM sodium borate buffer solution (pH 9.45) containing 20 mM sodium dodecyl sulfate. These conditions were reported as the optimum medium for formation of both the highly fluorescent and oxidizable isoindole derivatives,29 which are stable enough to enable quantitative analytical measurements with good reproducibility. As the derivatization reaction (Figure 3, inset) may proceed with different efficiencies when various analyte-to-reagent ratios are used,18 the influence of the OPA/DMAET concentration ratio on the current response of the isoindole reaction product was examined. Figure 3 displays the effect of the OPA-to-DMAET concentration ratio on the DMAET current response. For this purpose, the concentration of OPA was varied while maintaining a constant (0.5 mM) DMAET level. The concentration of valine, 4724 Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

Figure 4. Hydrodynamic voltammograms for 250 µM DMAET (A) and DEAET (B). Other conditions, as in Figure 2B.

which must always be in excess to the OPA level29 (to ensure a stable isoindole product), was kept constant at 5 mM. The peak height increases rapidly with the [OPA]/[DMAET] ratio up to 2.0 and levels off thereafter. Higher ratios resulted in a slight increase in the baseline current and the corresponding noise. Accordingly, an OPA/DMAET concentration ratio of at least 2.0 was maintained for all subsequent work. We also examined different off-chip reaction times (20-120 s) and found 60 s to be optimal. The sensitive monitoring of V-type nerve agent degradation products requires optimization of the detection potential. Hydrodynamic voltammograms were constructed for this purpose and are depicted in Figure 4. The current signals obtained for DMAET (A) and DEAET (B) were monitored with a stepwise increase in the detection potential over the +0.2 to +1.3 V range. For both compounds, the oxidation starts above +0.4 V, with the current rising sharply to +0.8 V and then more slowly. A marked increase in the baseline current and the associated noise was observed at detection potentials higher than +0.9 V, indicating incomplete isolation of detection potential from high separation voltages. For example, the baseline current increases rapidly from ∼80 nA at +1.0 V to almost 500 nA at +1.3 V. A detection potential of +0.8 V was therefore used for all further measurements, as it produced the optimal signal-to-noise characteristics. Higher operating potentials would be required while working with higher separation voltages that may shift the voltammetric profile to the anodic direction.30 (30) Wallenborg, S. R.; Nyholm, L.; Lunte, C. E. Anal. Chem. 1999, 71, 544.

Figure 5. Influence of the separation voltage on the current response for a mixture containing 250 µM ME (a), DMAET (b), and DEAET (c). Also shown (in the inset) is the dependence of the plate number (N) of DMAET and the resolution (R) between DMAET and DEAET. Separation voltage: (A) 3000, (B) 2500, (C) 2000, (D) 1500, and (E) 1000 V. Other conditions, as in Figure 2B.

As expected, the separation voltage has a profound effect on both the separation and detection efficiencies of the V-type nerve agent degradation products and on the overall assay time. This effect is examined in Figure 5 for a mixture containing ME (a), DMAET (b), and DEAET (c), using separation voltages ranging from +1000 to +3000 V. Such an increase of the electrical field decreases the migration time for the DEAET isoindole derivative from 430 s at +1000 V to 123 s at +3000 V. Similarly, the migration time of DMAET and ME decreases from 330 to 98 s and from 212 to 64 s, respectively. The flat baseline and the low noise level observed over the entire separation voltage range indicate an effective isolation of the detector. Also shown (in the inset) is the influence of the high voltage upon both the separation efficiency, i.e., on the plate number (N) calculated for the DMAET signal, and on the resolution (R) between DMAET (b) and DEAET (c). Both the plate number and resolution decrease from 3150 to 1100 and from 3.1 to 1.4, respectively, upon raising the separation voltage (between +1000 and 3000 V). The peak widths at halfmaximum also decrease upon increasing the separation voltage as indicated from the DMAET plate number versus separation voltage profile (inset). A separation voltage of +2000 V, which offered a good tradeoff between sensitivity and speed, was used for all subsequent work. Using this electrical field, a complete assay requires ∼3 min, which is shorter compared to conventional CE protocols.9-11 The ratio of the migration times is nearly independent of the separation voltage. For example, tb/ta and tc/ ta are 1.57 and 1.98, respectively (with RSD of ∼3%; n ) 7). Therefore, ME can be used as an internal standard for confirming the peak identity and improving the reproducibility. The amperometric detection of the isoindole derivatives of the above thiols results in well-defined concentration dependences. Figure 6 displays electropherograms for mixtures containing increasing quantities of ME (a), DMAET (b), and DEAET (c), in 50 µM steps (1-5). Defined peaks, with heights proportional to the concentration of the analytes, are observed. The resulting calibration plots are linear with sensitivities of 0.076, 0.062, and 0.036 µA/mM for ME, DMAET, and DEAET, respectively (correlation coefficients, 0.997, 0.993, and 0.991). The detection limits, based on signal-to-noise ratio (S/N) of 3, were found to be 5 and 8 µM for DMAET and DEAET, respectively. Practical security/

Figure 6. (A) Electropherograms detailing the response to increasing concentrations of ME (a), DMAET (b), and DEAET (c) in 50 µM steps (1-5). (B) The resulting calibration plots. Other conditions, as in Figure 2B.

environmental applications may require lower detection limits (down to the nanomolar range), that could be achieved by incorporating on-chip preconcentration schemes. The high sensitivity and speed of the CE/thick-film detector microsystem are coupled to good reproducibility. A series of eight repetitive injections of a mixture containing 100 µM DMAET and DEAET (using the same detector strip) resulted in relative standard deviations of 4.0 and 8.0%, respectively, for the peak heights and of ∼1.0% for the migration times. Good precision was observed for different detector strips, reflecting the reproducibility of the thick-film fabrication and detector positioning. Whenever needed, the design of the microsystem permits rapid (5-10 s) and facile replacement of the detector strip. The derivatization of thiols by OPA/valine leads to excellent selectivity. The suitability of the microchip CE/amperometric detection for measuring low levels of V-type nerve agent breakdown products in real water sample matrixes is demonstrated in Figure 7. The electrophoregrams for tap (A) and river (B) water samples, spiked with 100-800 µM (2-5) DMAET (a) and DEAET (b), are characterized with two well-defined and baseline-resolved peaks. The total assay time is ∼3 min. Note the flat baseline and the absence of any background interferences (1) [A(1) and B(1)] indicates low likelihood of “false positives”. Apparently, no naturally occurring species (in these matrixes) are oxidizable at the modest potential of +0.8 V. The minimal sample preparation makes the method suitable for routine field screening work. We also examined the ability to conduct the entire assay of V-type nerve agent degradation products, including the derivatization reaction, directly on the microchip (in a precolumn mixing chamber). Figure 8A displays electropherograms for sample mixtures containing 50 (1), 100 (2), and 150 (3) µM DMAET (a) and DEAET (b). Well-defined peaks, proportional to the analyte concentration, are observed. The resulting calibration plots for DMAET and DEAET are linear with sensitivities of 0.031 and 0.018 Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

4725

Figure 7. Electropherogram for tap (A) and river (B) water samples before (1) and after additions of 100 (2), 200 (3), 400 (4), and 800 (5) µM DMAET (a) and DEAET (b). Injection time, 3 s; other conditions, as in Figure 2B. The river water sample was filtered while the tap water was analyzed directly.

warning against possible chemical attack and prioritizing samples for more rigorous analysis.

Figure 8. On-chip derivatization: (A) electropherograms detailing the response to 50 µM increments (1-3) in the level of DMAET (a) and DEAET (b). (B) Electropherogram for the river water sample before (1) and after (2) spiking with 250 µM DMAET (a) and DEAET (b). Reagent solution, 6.0 mM OPA and 30 mM valine (in 40% methanol and 60% running buffer). Other conditions, as in Figure 2B.

µA/mM (correlation coefficients, 0.993 and 0.991). The corresponding detection limits (based on S/N ) 3) were found to be 6 and 11 µM for DMAET and DEAET, respectively. The suitability of on-chip derivatization protocol for measurements of thiolcontaining nerve agent degradation products in a river water matrix is demonstrated in Figure 8B. The on-chip procedure results in well-defined and separated peaks, with no apparent interference from the sample matrix. The response, however, is slightly lower compared to the off-chip derivatization, possibly due to an insufficient reaction time. For example, the signal obtained for DMAET is ∼40% lower in the case of on-chip derivatization [compare Figure 6A (1) vs Figure 8A (1)]. The slightly lower signal obtained using on-chip reaction can be offset considering the advantages in carrying out all the sample manipulations (derivatization in this case) on the microchip. Furthermore, it should be pointed out that while offering favorable performance characteristics, such signals are not stable (as indicated from their diminution in subsequent runs, e.g., 30 and 60% in the second and third injections, respectively). The exact reason for the decreased signals is not clear at this stage, considering the high stability observed using the off-chip reaction. Nevertheless, this is not a major concern for rapid screening applications aimed at

4726

Analytical Chemistry, Vol. 76, No. 16, August 15, 2004

CONCLUSIONS We have described a CE microchip method for monitoring degradation products of V-type nerve agents based on the derivatization with o-phthaldialdehyde and valine. The analytical utility of such a protocol was demonstrated with tap and river water samples. The favorable analytical performance makes the new protocol attractive for addressing the needs of various security scenarios. In particular, the new microchip is advantageous in terms of speed, efficiency, portability, sensitivity, cost, or sample size compared to conventional capillary electrophoresis or liquid chromatography systems. Such development of field-deployable inexpensive “counterterrorism” microanalyzers will enable transporting the forensic laboratory to the sample source. This could have a major impact upon the protection of first responders and military personnel and on assessing the nature of a chemical attack. The power and utility of such microfluidic assays will be greatly enhanced by designing self-contained microsystems integrating additional sample processing functions such as cleanup or preconcentration (necessary for meeting the real-life demands) onto the chip layout. ACKNOWLEDGMENT Financial support from the U.S. Department of Homeland Security (MIPT Award 2002-J-A-139), EPA (Award RD 830900), and NIH (Award R01A 1056047-01) is gratefully acknowledged. Points of view in this document are those of the authors and do not necessarily represent the official position of the U.S. Department of Homeland Security/MIPT or EPA. J.Z. was partly supported by The Grant Agency of the Czech Republic (Project 203/04/0136). Received for review March 3, 2004. Accepted May 27, 2004. AC049658B